diff --git a/core/java/android/text/AndroidBidi.java b/core/java/android/text/AndroidBidi.java index e4f934e721732..eacd40d0db29e 100644 --- a/core/java/android/text/AndroidBidi.java +++ b/core/java/android/text/AndroidBidi.java @@ -16,6 +16,8 @@ package android.text; +import android.text.Layout.Directions; + /** * Access the ICU bidi implementation. * @hide @@ -44,5 +46,132 @@ package android.text; return result; } + /** + * Returns run direction information for a line within a paragraph. + * + * @param dir base line direction, either Layout.DIR_LEFT_TO_RIGHT or + * Layout.DIR_RIGHT_TO_LEFT + * @param levels levels as returned from {@link #bidi} + * @param lstart start of the line in the levels array + * @param chars the character array (used to determine whitespace) + * @param cstart the start of the line in the chars array + * @param len the length of the line + * @return the directions + */ + public static Directions directions(int dir, byte[] levels, int lstart, + char[] chars, int cstart, int len) { + + int baseLevel = dir == Layout.DIR_LEFT_TO_RIGHT ? 0 : 1; + int curLevel = levels[lstart]; + int minLevel = curLevel; + int runCount = 1; + for (int i = lstart + 1, e = lstart + len; i < e; ++i) { + int level = levels[i]; + if (level != curLevel) { + curLevel = level; + ++runCount; + } + } + + // add final run for trailing counter-directional whitespace + int visLen = len; + if ((curLevel & 1) != (baseLevel & 1)) { + // look for visible end + while (--visLen >= 0) { + char ch = chars[cstart + visLen]; + + if (ch == '\n') { + --visLen; + break; + } + + if (ch != ' ' && ch != '\t') { + break; + } + } + ++visLen; + if (visLen != len) { + ++runCount; + } + } + + if (runCount == 1 && minLevel == baseLevel) { + // we're done, only one run on this line + if ((minLevel & 1) != 0) { + return Layout.DIRS_ALL_RIGHT_TO_LEFT; + } + return Layout.DIRS_ALL_LEFT_TO_RIGHT; + } + + int[] ld = new int[runCount * 2]; + int maxLevel = minLevel; + int levelBits = minLevel << Layout.RUN_LEVEL_SHIFT; + { + // Start of first pair is always 0, we write + // length then start at each new run, and the + // last run length after we're done. + int n = 1; + int prev = lstart; + curLevel = minLevel; + for (int i = lstart, e = lstart + visLen; i < e; ++i) { + int level = levels[i]; + if (level != curLevel) { + curLevel = level; + if (level > maxLevel) { + maxLevel = level; + } else if (level < minLevel) { + minLevel = level; + } + // XXX ignore run length limit of 2^RUN_LEVEL_SHIFT + ld[n++] = (i - prev) | levelBits; + ld[n++] = i - lstart; + levelBits = curLevel << Layout.RUN_LEVEL_SHIFT; + prev = i; + } + } + ld[n] = (lstart + visLen - prev) | levelBits; + if (visLen < len) { + ld[++n] = visLen; + ld[++n] = (len - visLen) | (baseLevel << Layout.RUN_LEVEL_SHIFT); + } + } + + // See if we need to swap any runs. + // If the min level run direction doesn't match the base + // direction, we always need to swap (at this point + // we have more than one run). + // Otherwise, we don't need to swap the lowest level. + // Since there are no logically adjacent runs at the same + // level, if the max level is the same as the (new) min + // level, we have a series of alternating levels that + // is already in order, so there's no more to do. + // + boolean swap; + if ((minLevel & 1) == baseLevel) { + minLevel += 1; + swap = maxLevel > minLevel; + } else { + swap = runCount > 1; + } + if (swap) { + for (int level = maxLevel - 1; level >= minLevel; --level) { + for (int i = 0; i < ld.length; i += 2) { + if (levels[ld[i]] >= level) { + int e = i + 2; + while (e < ld.length && levels[ld[e]] >= level) { + e += 2; + } + for (int low = i, hi = e - 2; low < hi; low += 2, hi -= 2) { + int x = ld[low]; ld[low] = ld[hi]; ld[hi] = x; + x = ld[low+1]; ld[low+1] = ld[hi+1]; ld[hi+1] = x; + } + i = e + 2; + } + } + } + } + return new Directions(ld); + } + private native static int runBidi(int dir, char[] chs, byte[] chInfo, int n, boolean haveInfo); } \ No newline at end of file diff --git a/core/java/android/text/BoringLayout.java b/core/java/android/text/BoringLayout.java index 944f7354fdb2c..9309b05c6fdaa 100644 --- a/core/java/android/text/BoringLayout.java +++ b/core/java/android/text/BoringLayout.java @@ -208,11 +208,11 @@ public class BoringLayout extends Layout implements TextUtils.EllipsizeCallback * width because the width that was passed in was for the * full text, not the ellipsized form. */ - synchronized (sTemp) { - mMax = (int) (FloatMath.ceil(Styled.measureText(paint, sTemp, - source, 0, source.length(), - null))); - } + TextLine line = TextLine.obtain(); + line.set(paint, source, 0, source.length(), Layout.DIR_LEFT_TO_RIGHT, + Layout.DIRS_ALL_LEFT_TO_RIGHT, false, null); + mMax = (int) FloatMath.ceil(line.metrics(null)); + TextLine.recycle(line); } if (includepad) { @@ -276,14 +276,13 @@ public class BoringLayout extends Layout implements TextUtils.EllipsizeCallback if (fm == null) { fm = new Metrics(); } - - int wid; - synchronized (sTemp) { - wid = (int) (FloatMath.ceil(Styled.measureText(paint, sTemp, - text, 0, text.length(), fm))); - } - fm.width = wid; + TextLine line = TextLine.obtain(); + line.set(paint, text, 0, text.length(), Layout.DIR_LEFT_TO_RIGHT, + Layout.DIRS_ALL_LEFT_TO_RIGHT, false, null); + fm.width = (int) FloatMath.ceil(line.metrics(fm)); + TextLine.recycle(line); + return fm; } else { return null; @@ -389,7 +388,7 @@ public class BoringLayout extends Layout implements TextUtils.EllipsizeCallback public static class Metrics extends Paint.FontMetricsInt { public int width; - + @Override public String toString() { return super.toString() + " width=" + width; } diff --git a/core/java/android/text/Layout.java b/core/java/android/text/Layout.java index ff1f2a60f36ff..3b8f2950f326d 100644 --- a/core/java/android/text/Layout.java +++ b/core/java/android/text/Layout.java @@ -19,12 +19,10 @@ package android.text; import com.android.internal.util.ArrayUtils; import android.emoji.EmojiFactory; -import android.graphics.Bitmap; import android.graphics.Canvas; import android.graphics.Paint; import android.graphics.Path; import android.graphics.Rect; -import android.graphics.RectF; import android.text.method.TextKeyListener; import android.text.style.AlignmentSpan; import android.text.style.LeadingMarginSpan; @@ -60,9 +58,7 @@ public abstract class Layout { MIN_EMOJI = -1; MAX_EMOJI = -1; } - }; - - private RectF mEmojiRect; + } /** * Return how wide a layout must be in order to display the @@ -91,8 +87,8 @@ public abstract class Layout { next = end; // note, omits trailing paragraph char - float w = measureText(paint, workPaint, - source, i, next, null, true, null); + float w = measurePara(paint, workPaint, + source, i, next, true, null); if (w > need) need = w; @@ -122,6 +118,15 @@ public abstract class Layout { if (width < 0) throw new IllegalArgumentException("Layout: " + width + " < 0"); + // Ensure paint doesn't have baselineShift set. + // While normally we don't modify the paint the user passed in, + // we were already doing this in Styled.drawUniformRun with both + // baselineShift and bgColor. We probably should reevaluate bgColor. + if (paint != null) { + paint.bgColor = 0; + paint.baselineShift = 0; + } + mText = text; mPaint = paint; mWorkPaint = new TextPaint(); @@ -262,6 +267,7 @@ public abstract class Layout { Alignment align = mAlignment; + TextLine tl = TextLine.obtain(); // Next draw the lines, one at a time. // the baseline is the top of the following line minus the current // line's descent. @@ -373,11 +379,11 @@ public abstract class Layout { // XXX: assumes there's nothing additional to be done c.drawText(buf, start, end, x, lbaseline, paint); } else { - drawText(c, buf, start, end, dir, directions, - x, ltop, lbaseline, lbottom, paint, mWorkPaint, - hasTab, spans); + tl.set(paint, buf, start, end, dir, directions, hasTab, spans); + tl.draw(c, x, ltop, lbaseline, lbottom); } } + TextLine.recycle(tl); } /** @@ -639,7 +645,7 @@ public abstract class Layout { private float getHorizontal(int offset, boolean trailing, int line) { int start = getLineStart(line); - int end = getLineVisibleEnd(line); + int end = getLineEnd(line); int dir = getParagraphDirection(line); boolean tab = getLineContainsTab(line); Directions directions = getLineDirections(line); @@ -649,17 +655,10 @@ public abstract class Layout { tabs = ((Spanned) mText).getSpans(start, end, TabStopSpan.class); } - float wid = measureText(mPaint, mWorkPaint, mText, start, offset, end, - dir, directions, trailing, tab, tabs); - - if (offset > end) { - if (dir == DIR_RIGHT_TO_LEFT) - wid -= measureText(mPaint, mWorkPaint, - mText, end, offset, null, tab, tabs); - else - wid += measureText(mPaint, mWorkPaint, - mText, end, offset, null, tab, tabs); - } + TextLine tl = TextLine.obtain(); + tl.set(mPaint, mText, start, end, dir, directions, tab, tabs); + float wid = tl.measure(offset - start, trailing, null); + TextLine.recycle(tl); Alignment align = getParagraphAlignment(line); int left = getParagraphLeft(line); @@ -761,21 +760,15 @@ public abstract class Layout { private float getLineMax(int line, Object[] tabs, boolean full) { int start = getLineStart(line); - int end; + int end = full ? getLineEnd(line) : getLineVisibleEnd(line); + boolean hasTabs = getLineContainsTab(line); + Directions directions = getLineDirections(line); - if (full) { - end = getLineEnd(line); - } else { - end = getLineVisibleEnd(line); - } - boolean tab = getLineContainsTab(line); - - if (tabs == null && tab && mText instanceof Spanned) { - tabs = ((Spanned) mText).getSpans(start, end, TabStopSpan.class); - } - - return measureText(mPaint, mWorkPaint, - mText, start, end, null, tab, tabs); + TextLine tl = TextLine.obtain(); + tl.set(mPaint, mText, start, end, 1, directions, hasTabs, tabs); + float width = tl.metrics(null); + TextLine.recycle(tl); + return width; } /** @@ -975,165 +968,49 @@ public abstract class Layout { return getOffsetToLeftRightOf(offset, false); } - // 1) The caret marks the leading edge of a character. The character - // logically before it might be on a different level, and the active caret - // position is on the character at the lower level. If that character - // was the previous character, the caret is on its trailing edge. - // 2) Take this character/edge and move it in the indicated direction. - // This gives you a new character and a new edge. - // 3) This position is between two visually adjacent characters. One of - // these might be at a lower level. The active position is on the - // character at the lower level. - // 4) If the active position is on the trailing edge of the character, - // the new caret position is the following logical character, else it - // is the character. private int getOffsetToLeftRightOf(int caret, boolean toLeft) { int line = getLineForOffset(caret); int lineStart = getLineStart(line); int lineEnd = getLineEnd(line); + int lineDir = getParagraphDirection(line); - boolean paraIsRtl = getParagraphDirection(line) == -1; - int[] runs = getLineDirections(line).mDirections; - - int runIndex, runLevel = 0, runStart = lineStart, runLimit = lineEnd, newCaret = -1; - boolean trailing = false; - - if (caret == lineStart) { - runIndex = -2; - } else if (caret == lineEnd) { - runIndex = runs.length; - } else { - // First, get information about the run containing the character with - // the active caret. - for (runIndex = 0; runIndex < runs.length; runIndex += 2) { - runStart = lineStart + runs[runIndex]; - if (caret >= runStart) { - runLimit = runStart + (runs[runIndex+1] & RUN_LENGTH_MASK); - if (runLimit > lineEnd) { - runLimit = lineEnd; - } - if (caret < runLimit) { - runLevel = (runs[runIndex+1] >>> RUN_LEVEL_SHIFT) & RUN_LEVEL_MASK; - if (caret == runStart) { - // The caret is on a run boundary, see if we should - // use the position on the trailing edge of the previous - // logical character instead. - int prevRunIndex, prevRunLevel, prevRunStart, prevRunLimit; - int pos = caret - 1; - for (prevRunIndex = 0; prevRunIndex < runs.length; prevRunIndex += 2) { - prevRunStart = lineStart + runs[prevRunIndex]; - if (pos >= prevRunStart) { - prevRunLimit = prevRunStart + (runs[prevRunIndex+1] & RUN_LENGTH_MASK); - if (prevRunLimit > lineEnd) { - prevRunLimit = lineEnd; - } - if (pos < prevRunLimit) { - prevRunLevel = (runs[prevRunIndex+1] >>> RUN_LEVEL_SHIFT) & RUN_LEVEL_MASK; - if (prevRunLevel < runLevel) { - // Start from logically previous character. - runIndex = prevRunIndex; - runLevel = prevRunLevel; - runStart = prevRunStart; - runLimit = prevRunLimit; - trailing = true; - break; - } - } - } - } + boolean advance = toLeft == (lineDir == DIR_RIGHT_TO_LEFT); + if (caret == (advance ? lineEnd : lineStart)) { + // walking off line, so look at the line we're headed to + if (caret == lineStart) { + if (line > 0) { + --line; + } else { + return caret; // at very start, don't move + } + } else { + if (line < getLineCount() - 1) { + ++line; + } else { + return caret; // at very end, don't move } - break; - } } - } - // caret might be = lineEnd. This is generally a space or paragraph - // separator and has an associated run, but might be the end of - // text, in which case it doesn't. If that happens, we ran off the - // end of the run list, and runIndex == runs.length. In this case, - // we are at a run boundary so we skip the below test. - if (runIndex != runs.length) { - boolean rtlRun = (runLevel & 0x1) != 0; - boolean advance = toLeft == rtlRun; - if (caret != (advance ? runLimit : runStart) || advance != trailing) { - // Moving within or into the run, so we can move logically. - newCaret = getOffsetBeforeAfter(caret, advance); - // If the new position is internal to the run, we're at the strong - // position already so we're finished. - if (newCaret != (advance ? runLimit : runStart)) { - return newCaret; - } - } - } + lineStart = getLineStart(line); + lineEnd = getLineEnd(line); + int newDir = getParagraphDirection(line); + if (newDir != lineDir) { + // unusual case. we want to walk onto the line, but it runs + // in a different direction than this one, so we fake movement + // in the opposite direction. + toLeft = !toLeft; + lineDir = newDir; + } } - // If newCaret is -1, we're starting at a run boundary and crossing - // into another run. Otherwise we've arrived at a run boundary, and - // need to figure out which character to attach to. Note we might - // need to run this twice, if we cross a run boundary and end up at - // another run boundary. - while (true) { - boolean advance = toLeft == paraIsRtl; - int otherRunIndex = runIndex + (advance ? 2 : -2); - if (otherRunIndex >= 0 && otherRunIndex < runs.length) { - int otherRunStart = lineStart + runs[otherRunIndex]; - int otherRunLimit = otherRunStart + (runs[otherRunIndex+1] & RUN_LENGTH_MASK); - if (otherRunLimit > lineEnd) { - otherRunLimit = lineEnd; - } - int otherRunLevel = runs[otherRunIndex+1] >>> RUN_LEVEL_SHIFT & RUN_LEVEL_MASK; - boolean otherRunIsRtl = (otherRunLevel & 1) != 0; + Directions directions = getLineDirections(line); - advance = toLeft == otherRunIsRtl; - if (newCaret == -1) { - newCaret = getOffsetBeforeAfter(advance ? otherRunStart : otherRunLimit, advance); - if (newCaret == (advance ? otherRunLimit : otherRunStart)) { - // Crossed and ended up at a new boundary, repeat a second and final time. - runIndex = otherRunIndex; - runLevel = otherRunLevel; - continue; - } - break; - } - - // The new caret is at a boundary. - if (otherRunLevel < runLevel) { - // The strong character is in the other run. - newCaret = advance ? otherRunStart : otherRunLimit; - } - break; - } - - if (newCaret == -1) { - // We're walking off the end of the line. The paragraph - // level is always equal to or lower than any internal level, so - // the boundaries get the strong caret. - newCaret = getOffsetBeforeAfter(caret, advance); - break; - } - // Else we've arrived at the end of the line. That's a strong position. - // We might have arrived here by crossing over a run with no internal - // breaks and dropping out of the above loop before advancing one final - // time, so reset the caret. - // Note, we use '<=' below to handle a situation where the only run - // on the line is a counter-directional run. If we're not advancing, - // we can end up at the 'lineEnd' position but the caret we want is at - // the lineStart. - if (newCaret <= lineEnd) { - newCaret = advance ? lineEnd : lineStart; - } - break; - } - - return newCaret; - } - - // utility, maybe just roll into the above. - private int getOffsetBeforeAfter(int offset, boolean after) { - if (after) { - return TextUtils.getOffsetAfter(mText, offset); - } - return TextUtils.getOffsetBefore(mText, offset); + TextLine tl = TextLine.obtain(); + // XXX: we don't care about tabs + tl.set(mPaint, mText, lineStart, lineEnd, lineDir, directions, false, null); + caret = lineStart + tl.getOffsetToLeftRightOf(caret - lineStart, toLeft); + tl = TextLine.recycle(tl); + return caret; } private int getOffsetAtStartOf(int offset) { @@ -1427,373 +1304,28 @@ public abstract class Layout { return right; } - private void drawText(Canvas canvas, - CharSequence text, int start, int end, - int dir, Directions directions, - float x, int top, int y, int bottom, - TextPaint paint, - TextPaint workPaint, - boolean hasTabs, Object[] parspans) { - char[] buf; - if (!hasTabs) { - if (directions == DIRS_ALL_LEFT_TO_RIGHT) { - if (DEBUG) { - Assert.assertTrue(DIR_LEFT_TO_RIGHT == dir); - } - Styled.drawText(canvas, text, start, end, dir, false, x, top, y, bottom, paint, workPaint, false); - return; + /* package */ + static float measurePara(TextPaint paint, TextPaint workPaint, + CharSequence text, int start, int end, boolean hasTabs, + Object[] tabs) { + + MeasuredText mt = MeasuredText.obtain(); + TextLine tl = TextLine.obtain(); + try { + mt.setPara(text, start, end, DIR_REQUEST_LTR); + Directions directions; + if (mt.mEasy){ + directions = DIRS_ALL_LEFT_TO_RIGHT; + } else { + directions = AndroidBidi.directions(mt.mDir, mt.mLevels, + 0, mt.mChars, 0, mt.mLen); } - buf = null; - } else { - buf = TextUtils.obtain(end - start); - TextUtils.getChars(text, start, end, buf, 0); + tl.set(paint, text, start, end, 1, directions, hasTabs, tabs); + return tl.metrics(null); + } finally { + TextLine.recycle(tl); + MeasuredText.recycle(mt); } - - float h = 0; - - int lastRunIndex = directions.mDirections.length - 2; - for (int i = 0; i < directions.mDirections.length; i += 2) { - int here = start + directions.mDirections[i]; - int there = here + (directions.mDirections[i+1] & RUN_LENGTH_MASK); - boolean runIsRtl = (directions.mDirections[i+1] & RUN_RTL_FLAG) != 0; - - if (there > end) - there = end; - - int segstart = here; - for (int j = hasTabs ? here : there; j <= there; j++) { - int pj = j - start; - if (j == there || buf[pj] == '\t') { - h += Styled.drawText(canvas, text, - segstart, j, - dir, runIsRtl, x + h, - top, y, bottom, paint, workPaint, - i != lastRunIndex || j != end); - - if (j != there) - h = dir * nextTab(text, start, end, h * dir, parspans); - - segstart = j + 1; - } else if (hasTabs && buf[pj] >= 0xD800 && buf[pj] <= 0xDFFF && j + 1 < there) { - int emoji = Character.codePointAt(buf, pj); - - if (emoji >= MIN_EMOJI && emoji <= MAX_EMOJI) { - Bitmap bm = EMOJI_FACTORY. - getBitmapFromAndroidPua(emoji); - - if (bm != null) { - h += Styled.drawText(canvas, text, - segstart, j, - dir, runIsRtl, x + h, - top, y, bottom, paint, workPaint, - i != lastRunIndex || j != end); - - if (mEmojiRect == null) { - mEmojiRect = new RectF(); - } - - workPaint.set(paint); - Styled.measureText(paint, workPaint, text, - j, j + 1, - null); - - float bitmapHeight = bm.getHeight(); - float textHeight = -workPaint.ascent(); - float scale = textHeight / bitmapHeight; - float width = bm.getWidth() * scale; - - mEmojiRect.set(x + h, y - textHeight, - x + h + width, y); - - canvas.drawBitmap(bm, null, mEmojiRect, paint); - h += width; - - j++; - segstart = j + 1; - } - } - } - } - } - - if (hasTabs) - TextUtils.recycle(buf); - } - - /** - * Get the distance from the margin to the requested edge of the character - * at offset on the line from start to end. Trailing indicates the edge - * should be the trailing edge of the character at offset-1. - */ - /* package */ static float measureText(TextPaint paint, - TextPaint workPaint, - CharSequence text, - int start, int offset, int end, - int dir, Directions directions, - boolean trailing, boolean hasTabs, - Object[] tabs) { - char[] buf = null; - - if (hasTabs) { - buf = TextUtils.obtain(end - start); - TextUtils.getChars(text, start, end, buf, 0); - } - - float h = 0; - - int target = trailing ? offset - 1 : offset; - if (target < start) { - return 0; - } - - int[] runs = directions.mDirections; - for (int i = 0; i < runs.length; i += 2) { - int here = start + runs[i]; - int there = here + (runs[i+1] & RUN_LENGTH_MASK); - if (there > end) { - there = end; - } - boolean runIsRtl = (runs[i+1] & RUN_RTL_FLAG) != 0; - - int segstart = here; - for (int j = hasTabs ? here : there; j <= there; j++) { - int codept = 0; - Bitmap bm = null; - - if (hasTabs && j < there) { - codept = buf[j - start]; - } - - if (codept >= 0xD800 && codept <= 0xDFFF && j + 1 < there) { - codept = Character.codePointAt(buf, j - start); - - if (codept >= MIN_EMOJI && codept <= MAX_EMOJI) { - bm = EMOJI_FACTORY.getBitmapFromAndroidPua(codept); - } - } - - if (j == there || codept == '\t' || bm != null) { - float segw; - - boolean inSegment = target >= segstart && target < j; - - if (inSegment) { - if (dir == DIR_LEFT_TO_RIGHT && !runIsRtl) { - h += Styled.measureText(paint, workPaint, text, - segstart, offset, - null); - return h; - } - - if (dir == DIR_RIGHT_TO_LEFT && runIsRtl) { - h -= Styled.measureText(paint, workPaint, text, - segstart, offset, - null); - return h; - } - } - - // XXX Style.measureText assumes LTR? - segw = Styled.measureText(paint, workPaint, text, - segstart, j, - null); - - if (inSegment) { - if (dir == DIR_LEFT_TO_RIGHT) { - h += segw - Styled.measureText(paint, workPaint, - text, - segstart, - offset, null); - return h; - } - - if (dir == DIR_RIGHT_TO_LEFT) { - h -= segw - Styled.measureText(paint, workPaint, - text, - segstart, - offset, null); - return h; - } - } - - if (dir == DIR_RIGHT_TO_LEFT) - h -= segw; - else - h += segw; - - if (j != there && buf[j - start] == '\t') { - if (offset == j) - return h; - - h = dir * nextTab(text, start, end, h * dir, tabs); - - if (target == j) { - return h; - } - } - - if (bm != null) { - workPaint.set(paint); - Styled.measureText(paint, workPaint, text, - j, j + 2, null); - - float wid = bm.getWidth() * - -workPaint.ascent() / bm.getHeight(); - - if (dir == DIR_RIGHT_TO_LEFT) { - h -= wid; - } else { - h += wid; - } - - j++; - } - - segstart = j + 1; - } - } - } - - if (hasTabs) - TextUtils.recycle(buf); - - return h; - } - - /** - * Measure width of a run of text on a single line that is known to all be - * in the same direction as the paragraph base direction. Returns the width, - * and the line metrics in fm if fm is not null. - * - * @param paint the paint for the text; will not be modified - * @param workPaint paint available for modification - * @param text text - * @param start start of the line - * @param end limit of the line - * @param fm object to return integer metrics in, can be null - * @param hasTabs true if it is known that the line has tabs - * @param tabs tab position information - * @return the width of the text from start to end - */ - /* package */ static float measureText(TextPaint paint, - TextPaint workPaint, - CharSequence text, - int start, int end, - Paint.FontMetricsInt fm, - boolean hasTabs, Object[] tabs) { - char[] buf = null; - - if (hasTabs) { - buf = TextUtils.obtain(end - start); - TextUtils.getChars(text, start, end, buf, 0); - } - - int len = end - start; - - int lastPos = 0; - float width = 0; - int ascent = 0, descent = 0, top = 0, bottom = 0; - - if (fm != null) { - fm.ascent = 0; - fm.descent = 0; - } - - for (int pos = hasTabs ? 0 : len; pos <= len; pos++) { - int codept = 0; - Bitmap bm = null; - - if (hasTabs && pos < len) { - codept = buf[pos]; - } - - if (codept >= 0xD800 && codept <= 0xDFFF && pos < len) { - codept = Character.codePointAt(buf, pos); - - if (codept >= MIN_EMOJI && codept <= MAX_EMOJI) { - bm = EMOJI_FACTORY.getBitmapFromAndroidPua(codept); - } - } - - if (pos == len || codept == '\t' || bm != null) { - workPaint.baselineShift = 0; - - // XXX Styled.measureText assumes the run direction is LTR, - // but it might not be. Check this. - width += Styled.measureText(paint, workPaint, text, - start + lastPos, start + pos, - fm); - - if (fm != null) { - if (workPaint.baselineShift < 0) { - fm.ascent += workPaint.baselineShift; - fm.top += workPaint.baselineShift; - } else { - fm.descent += workPaint.baselineShift; - fm.bottom += workPaint.baselineShift; - } - } - - if (pos != len) { - if (bm == null) { - // no emoji, must have hit a tab - width = nextTab(text, start, end, width, tabs); - } else { - // This sets up workPaint with the font on the emoji - // text, so that we can extract the ascent and scale. - - // We can't use the result of the previous call to - // measureText because the emoji might have its own style. - // We have to initialize workPaint here because if the - // text is unstyled measureText might not use workPaint - // at all. - workPaint.set(paint); - Styled.measureText(paint, workPaint, text, - start + pos, start + pos + 1, null); - - width += bm.getWidth() * - -workPaint.ascent() / bm.getHeight(); - - // Since we had an emoji, we bump past the second half - // of the surrogate pair. - pos++; - } - } - - if (fm != null) { - if (fm.ascent < ascent) { - ascent = fm.ascent; - } - if (fm.descent > descent) { - descent = fm.descent; - } - - if (fm.top < top) { - top = fm.top; - } - if (fm.bottom > bottom) { - bottom = fm.bottom; - } - - // No need to take bitmap height into account here, - // since it is scaled to match the text height. - } - - lastPos = pos + 1; - } - } - - if (fm != null) { - fm.ascent = ascent; - fm.descent = descent; - fm.top = top; - fm.bottom = bottom; - } - - if (hasTabs) - TextUtils.recycle(buf); - - return width; } /** @@ -1898,6 +1430,7 @@ public abstract class Layout { * line is ellipsized, not getLineStart().) */ public abstract int getEllipsisStart(int line); + /** * Returns the number of characters to be ellipsized away, or 0 if * no ellipsis is to take place. diff --git a/core/java/android/text/MeasuredText.java b/core/java/android/text/MeasuredText.java new file mode 100644 index 0000000000000..e3a113da2de30 --- /dev/null +++ b/core/java/android/text/MeasuredText.java @@ -0,0 +1,250 @@ +/* + * Copyright (C) 2010 The Android Open Source Project + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +package android.text; + +import com.android.internal.util.ArrayUtils; + +import android.graphics.Paint; +import android.icu.text.ArabicShaping; +import android.text.style.MetricAffectingSpan; +import android.text.style.ReplacementSpan; +import android.util.Log; + +/** + * @hide + */ +class MeasuredText { + /* package */ CharSequence mText; + /* package */ int mTextStart; + /* package */ float[] mWidths; + /* package */ char[] mChars; + /* package */ byte[] mLevels; + /* package */ int mDir; + /* package */ boolean mEasy; + /* package */ int mLen; + private int mPos; + private float[] mWorkWidths; // temp buffer for Paint.measureText, arrgh + private TextPaint mWorkPaint; + + private MeasuredText() { + mWorkPaint = new TextPaint(); + } + + private static MeasuredText[] cached = new MeasuredText[3]; + + /* package */ + static MeasuredText obtain() { + MeasuredText mt; + synchronized (cached) { + for (int i = cached.length; --i >= 0;) { + if (cached[i] != null) { + mt = cached[i]; + cached[i] = null; + return mt; + } + } + } + mt = new MeasuredText(); + Log.e("MEAS", "new: " + mt); + return mt; + } + + /* package */ + static MeasuredText recycle(MeasuredText mt) { + mt.mText = null; + if (mt.mLen < 1000) { + synchronized(cached) { + for (int i = 0; i < cached.length; ++i) { + if (cached[i] == null) { + cached[i] = mt; + break; + } + } + } + } + return null; + } + + /** + * Analyzes text for + * bidirectional runs. Allocates working buffers. + */ + /* package */ + void setPara(CharSequence text, int start, int end, int bidiRequest) { + mText = text; + mTextStart = start; + + int len = end - start; + mLen = len; + mPos = 0; + + if (mWidths == null || mWidths.length < len) { + mWidths = new float[ArrayUtils.idealFloatArraySize(len)]; + mWorkWidths = new float[mWidths.length]; + } + if (mChars == null || mChars.length < len) { + mChars = new char[ArrayUtils.idealCharArraySize(len)]; + } + TextUtils.getChars(text, start, end, mChars, 0); + + if (text instanceof Spanned) { + Spanned spanned = (Spanned) text; + ReplacementSpan[] spans = spanned.getSpans(start, end, + ReplacementSpan.class); + + for (int i = 0; i < spans.length; i++) { + int startInPara = spanned.getSpanStart(spans[i]) - start; + int endInPara = spanned.getSpanEnd(spans[i]) - start; + for (int j = startInPara; j < endInPara; j++) { + mChars[j] = '\uFFFC'; + } + } + } + + if (TextUtils.doesNotNeedBidi(mChars, 0, len)) { + mDir = 1; + mEasy = true; + } else { + if (mLevels == null || mLevels.length < len) { + mLevels = new byte[ArrayUtils.idealByteArraySize(len)]; + } + mDir = AndroidBidi.bidi(bidiRequest, mChars, mLevels, len, false); + mEasy = false; + + // shape + if (mLen > 0) { + byte[] levels = mLevels; + char[] chars = mChars; + byte level = levels[0]; + int pi = 0; + for (int i = 1, e = mLen;; ++i) { + if (i == e || levels[i] != level) { + if ((level & 0x1) != 0) { + AndroidCharacter.mirror(chars, pi, i - pi); + ArabicShaping.SHAPER.shape(chars, pi, i - pi); + } + if (i == e) { + break; + } + pi = i; + level = levels[i]; + } + } + } + } + } + + float addStyleRun(TextPaint paint, int len, Paint.FontMetricsInt fm) { + int p = mPos; + float[] w = mWidths, ww = mWorkWidths; + int count = paint.getTextWidths(mChars, p, len, ww); + int width = 0; + if (count < len) { + // must have surrogate pairs in here, pad out the array with zero + // for the trailing surrogates + char[] chars = mChars; + for (int i = 0, e = mLen; i < count; ++i) { + width += (w[p++] = ww[i]); + if (p < e && chars[p] >= '\udc00' && chars[p] < '\ue000' && + chars[p-1] >= '\ud800' && chars[p-1] < '\udc00') { + w[p++] = 0; + } + } + } else { + for (int i = 0; i < len; ++i) { + width += (w[p++] = ww[i]); + } + } + mPos = p; + if (fm != null) { + paint.getFontMetricsInt(fm); + } + return width; + } + + float addStyleRun(TextPaint paint, MetricAffectingSpan[] spans, int len, + Paint.FontMetricsInt fm) { + + TextPaint workPaint = mWorkPaint; + workPaint.set(paint); + // XXX paint should not have a baseline shift, but... + workPaint.baselineShift = 0; + + ReplacementSpan replacement = null; + for (int i = 0; i < spans.length; i++) { + MetricAffectingSpan span = spans[i]; + if (span instanceof ReplacementSpan) { + replacement = (ReplacementSpan)span; + } else { + span.updateMeasureState(workPaint); + } + } + + float wid; + if (replacement == null) { + wid = addStyleRun(workPaint, len, fm); + } else { + // Use original text. Shouldn't matter. + wid = replacement.getSize(workPaint, mText, mTextStart + mPos, + mTextStart + mPos + len, fm); + float[] w = mWidths; + w[mPos] = wid; + for (int i = mPos + 1, e = mPos + len; i < e; i++) + w[i] = 0; + } + + if (fm != null) { + if (workPaint.baselineShift < 0) { + fm.ascent += workPaint.baselineShift; + fm.top += workPaint.baselineShift; + } else { + fm.descent += workPaint.baselineShift; + fm.bottom += workPaint.baselineShift; + } + } + + return wid; + } + + int breakText(int start, int limit, boolean forwards, float width) { + float[] w = mWidths; + if (forwards) { + for (int i = start; i < limit; ++i) { + if ((width -= w[i]) < 0) { + return i - start; + } + } + } else { + for (int i = limit; --i >= start;) { + if ((width -= w[i]) < 0) { + return limit - i -1; + } + } + } + + return limit - start; + } + + float measure(int start, int limit) { + float width = 0; + float[] w = mWidths; + for (int i = start; i < limit; ++i) { + width += w[i]; + } + return width; + } +} \ No newline at end of file diff --git a/core/java/android/text/StaticLayout.java b/core/java/android/text/StaticLayout.java index bfa0ab61f3e4d..0c6c545dfb46b 100644 --- a/core/java/android/text/StaticLayout.java +++ b/core/java/android/text/StaticLayout.java @@ -23,8 +23,6 @@ import android.graphics.Paint; import android.text.style.LeadingMarginSpan; import android.text.style.LineHeightSpan; import android.text.style.MetricAffectingSpan; -import android.text.style.ReplacementSpan; -import android.util.Log; /** * StaticLayout is a Layout for text that will not be edited after it @@ -96,13 +94,13 @@ extends Layout mLineDirections = new Directions[ ArrayUtils.idealIntArraySize(2 * mColumns)]; + mMeasured = MeasuredText.obtain(); + generate(source, bufstart, bufend, paint, outerwidth, align, spacingmult, spacingadd, includepad, includepad, ellipsize != null, ellipsizedWidth, ellipsize); - mChdirs = null; - mChs = null; - mWidths = null; + mMeasured = MeasuredText.recycle(mMeasured); mFontMetricsInt = null; } @@ -113,6 +111,7 @@ extends Layout mLines = new int[ArrayUtils.idealIntArraySize(2 * mColumns)]; mLineDirections = new Directions[ ArrayUtils.idealIntArraySize(2 * mColumns)]; + mMeasured = MeasuredText.obtain(); } /* package */ void generate(CharSequence source, int bufstart, int bufend, @@ -130,38 +129,22 @@ extends Layout Paint.FontMetricsInt fm = mFontMetricsInt; int[] choosehtv = null; - int end = TextUtils.indexOf(source, '\n', bufstart, bufend); - int bufsiz = end >= 0 ? end - bufstart : bufend - bufstart; - boolean first = true; + MeasuredText measured = mMeasured; - if (mChdirs == null) { - mChdirs = new byte[ArrayUtils.idealByteArraySize(bufsiz + 1)]; - mChs = new char[ArrayUtils.idealCharArraySize(bufsiz + 1)]; - mWidths = new float[ArrayUtils.idealIntArraySize((bufsiz + 1) * 2)]; - } - - byte[] chdirs = mChdirs; - char[] chs = mChs; - float[] widths = mWidths; - - AlteredCharSequence alter = null; Spanned spanned = null; - if (source instanceof Spanned) spanned = (Spanned) source; int DEFAULT_DIR = DIR_LEFT_TO_RIGHT; // XXX - for (int start = bufstart; start <= bufend; start = end) { - if (first) - first = false; + int paraEnd; + for (int paraStart = bufstart; paraStart <= bufend; paraStart = paraEnd) { + paraEnd = TextUtils.indexOf(source, '\n', paraStart, bufend); + if (paraEnd < 0) + paraEnd = bufend; else - end = TextUtils.indexOf(source, '\n', start, bufend); - - if (end < 0) - end = bufend; - else - end++; + paraEnd++; + int paraLen = paraEnd - paraStart; int firstWidthLineCount = 1; int firstwidth = outerwidth; @@ -170,19 +153,20 @@ extends Layout LineHeightSpan[] chooseht = null; if (spanned != null) { - LeadingMarginSpan[] sp; - - sp = spanned.getSpans(start, end, LeadingMarginSpan.class); + LeadingMarginSpan[] sp = spanned.getSpans(paraStart, paraEnd, + LeadingMarginSpan.class); for (int i = 0; i < sp.length; i++) { LeadingMarginSpan lms = sp[i]; firstwidth -= sp[i].getLeadingMargin(true); restwidth -= sp[i].getLeadingMargin(false); if (lms instanceof LeadingMarginSpan.LeadingMarginSpan2) { - firstWidthLineCount = ((LeadingMarginSpan.LeadingMarginSpan2)lms).getLeadingMarginLineCount(); + firstWidthLineCount = + ((LeadingMarginSpan.LeadingMarginSpan2)lms) + .getLeadingMarginLineCount(); } } - chooseht = spanned.getSpans(start, end, LineHeightSpan.class); + chooseht = spanned.getSpans(paraStart, paraEnd, LineHeightSpan.class); if (chooseht.length != 0) { if (choosehtv == null || @@ -194,7 +178,7 @@ extends Layout for (int i = 0; i < chooseht.length; i++) { int o = spanned.getSpanStart(chooseht[i]); - if (o < start) { + if (o < paraStart) { // starts in this layout, before the // current paragraph @@ -208,135 +192,48 @@ extends Layout } } - if (end - start > chdirs.length) { - chdirs = new byte[ArrayUtils.idealByteArraySize(end - start)]; - mChdirs = chdirs; - } - if (end - start > chs.length) { - chs = new char[ArrayUtils.idealCharArraySize(end - start)]; - mChs = chs; - } - if ((end - start) * 2 > widths.length) { - widths = new float[ArrayUtils.idealIntArraySize((end - start) * 2)]; - mWidths = widths; - } + measured.setPara(source, paraStart, paraEnd, DIR_REQUEST_DEFAULT_LTR); + char[] chs = measured.mChars; + float[] widths = measured.mWidths; + byte[] chdirs = measured.mLevels; + int dir = measured.mDir; + boolean easy = measured.mEasy; - TextUtils.getChars(source, start, end, chs, 0); - final int n = end - start; - - boolean easy = true; - boolean altered = false; - int dir = DEFAULT_DIR; // XXX pass value in - - for (int i = 0; i < n; i++) { - if (chs[i] >= FIRST_RIGHT_TO_LEFT) { - easy = false; - break; - } - } - - // Ensure that none of the underlying characters are treated - // as viable breakpoints, and that the entire run gets the - // same bidi direction. - - if (source instanceof Spanned) { - Spanned sp = (Spanned) source; - ReplacementSpan[] spans = sp.getSpans(start, end, ReplacementSpan.class); - - for (int y = 0; y < spans.length; y++) { - int a = sp.getSpanStart(spans[y]); - int b = sp.getSpanEnd(spans[y]); - - for (int x = a; x < b; x++) { - chs[x - start] = '\uFFFC'; - } - } - } - - if (!easy) { - // XXX put override flags, etc. into chdirs - // XXX supply dir rather than force - dir = AndroidBidi.bidi(DIR_REQUEST_DEFAULT_LTR, chs, chdirs, n, false); - - // Do mirroring for right-to-left segments - - for (int i = 0; i < n; i++) { - if (chdirs[i] == Character.DIRECTIONALITY_RIGHT_TO_LEFT) { - int j; - - for (j = i; j < n; j++) { - if (chdirs[j] != - Character.DIRECTIONALITY_RIGHT_TO_LEFT) - break; - } - - if (AndroidCharacter.mirror(chs, i, j - i)) - altered = true; - - i = j - 1; - } - } - } - - CharSequence sub; - - if (altered) { - if (alter == null) - alter = AlteredCharSequence.make(source, chs, start, end); - else - alter.update(chs, start, end); - - sub = alter; - } else { - sub = source; - } + CharSequence sub = source; int width = firstwidth; float w = 0; - int here = start; + int here = paraStart; - int ok = start; + int ok = paraStart; float okwidth = w; int okascent = 0, okdescent = 0, oktop = 0, okbottom = 0; - int fit = start; + int fit = paraStart; float fitwidth = w; int fitascent = 0, fitdescent = 0, fittop = 0, fitbottom = 0; boolean tab = false; - int next; - for (int i = start; i < end; i = next) { + int spanEnd; + for (int spanStart = paraStart; spanStart < paraEnd; spanStart = spanEnd) { if (spanned == null) - next = end; + spanEnd = paraEnd; else - next = spanned.nextSpanTransition(i, end, - MetricAffectingSpan. - class); + spanEnd = spanned.nextSpanTransition(spanStart, paraEnd, + MetricAffectingSpan.class); + + int spanLen = spanEnd - spanStart; + int startInPara = spanStart - paraStart; + int endInPara = spanEnd - paraStart; if (spanned == null) { - paint.getTextWidths(sub, i, next, widths); - System.arraycopy(widths, 0, widths, - end - start + (i - start), next - i); - - paint.getFontMetricsInt(fm); + measured.addStyleRun(paint, spanLen, fm); } else { - mWorkPaint.baselineShift = 0; - - Styled.getTextWidths(paint, mWorkPaint, - spanned, i, next, - widths, fm); - System.arraycopy(widths, 0, widths, - end - start + (i - start), next - i); - - if (mWorkPaint.baselineShift < 0) { - fm.ascent += mWorkPaint.baselineShift; - fm.top += mWorkPaint.baselineShift; - } else { - fm.descent += mWorkPaint.baselineShift; - fm.bottom += mWorkPaint.baselineShift; - } + MetricAffectingSpan[] spans = + spanned.getSpans(spanStart, spanEnd, MetricAffectingSpan.class); + measured.addStyleRun(paint, spans, spanLen, fm); } int fmtop = fm.top; @@ -344,27 +241,17 @@ extends Layout int fmascent = fm.ascent; int fmdescent = fm.descent; - if (false) { - StringBuilder sb = new StringBuilder(); - for (int j = i; j < next; j++) { - sb.append(widths[j - start + (end - start)]); - sb.append(' '); - } - - Log.e("text", sb.toString()); - } - - for (int j = i; j < next; j++) { - char c = chs[j - start]; + for (int j = spanStart; j < spanEnd; j++) { + char c = chs[j - paraStart]; float before = w; if (c == '\n') { ; } else if (c == '\t') { - w = Layout.nextTab(sub, start, end, w, null); + w = Layout.nextTab(sub, paraStart, paraEnd, w, null); tab = true; - } else if (c >= 0xD800 && c <= 0xDFFF && j + 1 < next) { - int emoji = Character.codePointAt(chs, j - start); + } else if (c >= 0xD800 && c <= 0xDFFF && j + 1 < spanEnd) { + int emoji = Character.codePointAt(chs, j - paraStart); if (emoji >= MIN_EMOJI && emoji <= MAX_EMOJI) { Bitmap bm = EMOJI_FACTORY. @@ -387,13 +274,13 @@ extends Layout tab = true; j++; } else { - w += widths[j - start + (end - start)]; + w += widths[j - paraStart]; } } else { - w += widths[j - start + (end - start)]; + w += widths[j - paraStart]; } } else { - w += widths[j - start + (end - start)]; + w += widths[j - paraStart]; } // Log.e("text", "was " + before + " now " + w + " after " + c + " within " + width); @@ -429,12 +316,12 @@ extends Layout if (c == ' ' || c == '\t' || ((c == '.' || c == ',' || c == ':' || c == ';') && - (j - 1 < here || !Character.isDigit(chs[j - 1 - start])) && - (j + 1 >= next || !Character.isDigit(chs[j + 1 - start]))) || + (j - 1 < here || !Character.isDigit(chs[j - 1 - paraStart])) && + (j + 1 >= spanEnd || !Character.isDigit(chs[j + 1 - paraStart]))) || ((c == '/' || c == '-') && - (j + 1 >= next || !Character.isDigit(chs[j + 1 - start]))) || + (j + 1 >= spanEnd || !Character.isDigit(chs[j + 1 - paraStart]))) || (c >= FIRST_CJK && isIdeographic(c, true) && - j + 1 < next && isIdeographic(chs[j + 1 - start], false))) { + j + 1 < spanEnd && isIdeographic(chs[j + 1 - paraStart], false))) { okwidth = w; ok = j + 1; @@ -451,7 +338,7 @@ extends Layout if (ok != here) { // Log.e("text", "output ok " + here + " to " +ok); - while (ok < next && chs[ok - start] == ' ') { + while (ok < spanEnd && chs[ok - paraStart] == ' ') { ok++; } @@ -461,9 +348,9 @@ extends Layout v, spacingmult, spacingadd, chooseht, choosehtv, fm, tab, - needMultiply, start, chdirs, dir, easy, + needMultiply, paraStart, chdirs, dir, easy, ok == bufend, includepad, trackpad, - widths, start, end - start, + chs, widths, here - paraStart, where, ellipsizedWidth, okwidth, paint); @@ -487,7 +374,7 @@ extends Layout if (ok != here) { // Log.e("text", "output ok " + here + " to " +ok); - while (ok < next && chs[ok - start] == ' ') { + while (ok < spanEnd && chs[ok - paraStart] == ' ') { ok++; } @@ -497,9 +384,9 @@ extends Layout v, spacingmult, spacingadd, chooseht, choosehtv, fm, tab, - needMultiply, start, chdirs, dir, easy, + needMultiply, paraStart, chdirs, dir, easy, ok == bufend, includepad, trackpad, - widths, start, end - start, + chs, widths, here - paraStart, where, ellipsizedWidth, okwidth, paint); @@ -513,18 +400,19 @@ extends Layout v, spacingmult, spacingadd, chooseht, choosehtv, fm, tab, - needMultiply, start, chdirs, dir, easy, + needMultiply, paraStart, chdirs, dir, easy, fit == bufend, includepad, trackpad, - widths, start, end - start, + chs, widths, here - paraStart, where, ellipsizedWidth, fitwidth, paint); here = fit; } else { // Log.e("text", "output one " + here + " to " +(here + 1)); - measureText(paint, mWorkPaint, - source, here, here + 1, fm, tab, - null); + // XXX not sure why the existing fm wasn't ok. + // measureText(paint, mWorkPaint, + // source, here, here + 1, fm, tab, + // null); v = out(source, here, here+1, @@ -533,18 +421,18 @@ extends Layout v, spacingmult, spacingadd, chooseht, choosehtv, fm, tab, - needMultiply, start, chdirs, dir, easy, + needMultiply, paraStart, chdirs, dir, easy, here + 1 == bufend, includepad, trackpad, - widths, start, end - start, + chs, widths, here - paraStart, where, ellipsizedWidth, - widths[here - start], paint); + widths[here - paraStart], paint); here = here + 1; } - if (here < i) { - j = next = here; // must remeasure + if (here < spanStart) { + j = spanEnd = here; // must remeasure } else { j = here - 1; // continue looping } @@ -561,7 +449,7 @@ extends Layout } } - if (end != here) { + if (paraEnd != here) { if ((fittop | fitbottom | fitdescent | fitascent) == 0) { paint.getFontMetricsInt(fm); @@ -574,20 +462,20 @@ extends Layout // Log.e("text", "output rest " + here + " to " + end); v = out(source, - here, end, fitascent, fitdescent, + here, paraEnd, fitascent, fitdescent, fittop, fitbottom, v, spacingmult, spacingadd, chooseht, choosehtv, fm, tab, - needMultiply, start, chdirs, dir, easy, - end == bufend, includepad, trackpad, - widths, start, end - start, + needMultiply, paraStart, chdirs, dir, easy, + paraEnd == bufend, includepad, trackpad, + chs, widths, here - paraStart, where, ellipsizedWidth, w, paint); } - start = end; + paraStart = paraEnd; - if (end == bufend) + if (paraEnd == bufend) break; } @@ -602,9 +490,9 @@ extends Layout v, spacingmult, spacingadd, null, null, fm, false, - needMultiply, bufend, chdirs, DEFAULT_DIR, true, + needMultiply, bufend, null, DEFAULT_DIR, true, true, includepad, trackpad, - widths, bufstart, 0, + null, null, bufstart, where, ellipsizedWidth, 0, paint); } } @@ -714,28 +602,6 @@ extends Layout } */ - private static int getFit(TextPaint paint, - TextPaint workPaint, - CharSequence text, int start, int end, - float wid) { - int high = end + 1, low = start - 1, guess; - - while (high - low > 1) { - guess = (high + low) / 2; - - if (measureText(paint, workPaint, - text, start, guess, null, true, null) > wid) - high = guess; - else - low = guess; - } - - if (low < start) - return start; - else - return low; - } - private int out(CharSequence text, int start, int end, int above, int below, int top, int bottom, int v, float spacingmult, float spacingadd, @@ -744,7 +610,7 @@ extends Layout boolean needMultiply, int pstart, byte[] chdirs, int dir, boolean easy, boolean last, boolean includepad, boolean trackpad, - float[] widths, int widstart, int widoff, + char[] chs, float[] widths, int widstart, TextUtils.TruncateAt ellipsize, float ellipsiswidth, float textwidth, TextPaint paint) { int j = mLineCount; @@ -752,8 +618,6 @@ extends Layout int want = off + mColumns + TOP; int[] lines = mLines; - // Log.e("text", "line " + start + " to " + end + (last ? "===" : "")); - if (want >= lines.length) { int nlen = ArrayUtils.idealIntArraySize(want + 1); int[] grow = new int[nlen]; @@ -840,122 +704,12 @@ extends Layout if (easy) { mLineDirections[j] = linedirs; } else { - int startOff = start - pstart; - int baseLevel = dir == DIR_LEFT_TO_RIGHT ? 0 : 1; - int curLevel = chdirs[startOff]; - int minLevel = curLevel; - int runCount = 1; - for (int i = start + 1; i < end; ++i) { - int level = chdirs[i - pstart]; - if (level != curLevel) { - curLevel = level; - ++runCount; - } - } - - // add final run for trailing counter-directional whitespace - int visEnd = end; - if ((curLevel & 1) != (baseLevel & 1)) { - // look for visible end - while (--visEnd >= start) { - char ch = text.charAt(visEnd); - - if (ch == '\n') { - --visEnd; - break; - } - - if (ch != ' ' && ch != '\t') { - break; - } - } - ++visEnd; - if (visEnd != end) { - ++runCount; - } - } - - if (runCount == 1 && minLevel == baseLevel) { - if ((minLevel & 1) != 0) { - linedirs = DIRS_ALL_RIGHT_TO_LEFT; - } - // we're done, only one run on this line - } else { - int[] ld = new int[runCount * 2]; - int maxLevel = minLevel; - int levelBits = minLevel << RUN_LEVEL_SHIFT; - { - // Start of first pair is always 0, we write - // length then start at each new run, and the - // last run length after we're done. - int n = 1; - int prev = start; - curLevel = minLevel; - for (int i = start; i < visEnd; ++i) { - int level = chdirs[i - pstart]; - if (level != curLevel) { - curLevel = level; - if (level > maxLevel) { - maxLevel = level; - } else if (level < minLevel) { - minLevel = level; - } - // XXX ignore run length limit of 2^RUN_LEVEL_SHIFT - ld[n++] = (i - prev) | levelBits; - ld[n++] = i - start; - levelBits = curLevel << RUN_LEVEL_SHIFT; - prev = i; - } - } - ld[n] = (visEnd - prev) | levelBits; - if (visEnd < end) { - ld[++n] = visEnd - start; - ld[++n] = (end - visEnd) | (baseLevel << RUN_LEVEL_SHIFT); - } - } - - // See if we need to swap any runs. - // If the min level run direction doesn't match the base - // direction, we always need to swap (at this point - // we have more than one run). - // Otherwise, we don't need to swap the lowest level. - // Since there are no logically adjacent runs at the same - // level, if the max level is the same as the (new) min - // level, we have a series of alternating levels that - // is already in order, so there's no more to do. - // - boolean swap; - if ((minLevel & 1) == baseLevel) { - minLevel += 1; - swap = maxLevel > minLevel; - } else { - swap = runCount > 1; - } - if (swap) { - for (int level = maxLevel - 1; level >= minLevel; --level) { - for (int i = 0; i < ld.length; i += 2) { - if (chdirs[startOff + ld[i]] >= level) { - int e = i + 2; - while (e < ld.length && chdirs[startOff + ld[e]] >= level) { - e += 2; - } - for (int low = i, hi = e - 2; low < hi; low += 2, hi -= 2) { - int x = ld[low]; ld[low] = ld[hi]; ld[hi] = x; - x = ld[low+1]; ld[low+1] = ld[hi+1]; ld[hi+1] = x; - } - i = e + 2; - } - } - } - } - linedirs = new Directions(ld); - } - - mLineDirections[j] = linedirs; + mLineDirections[j] = AndroidBidi.directions(dir, chdirs, widstart, chs, + widstart, end - start); // If ellipsize is in marquee mode, do not apply ellipsis on the first line if (ellipsize != null && (ellipsize != TextUtils.TruncateAt.MARQUEE || j != 0)) { - calculateEllipsis(start, end, widths, widstart, widoff, + calculateEllipsis(start, end, widths, widstart, ellipsiswidth, ellipsize, j, textwidth, paint); } @@ -966,7 +720,7 @@ extends Layout } private void calculateEllipsis(int linestart, int lineend, - float[] widths, int widstart, int widoff, + float[] widths, int widstart, float avail, TextUtils.TruncateAt where, int line, float textwidth, TextPaint paint) { int len = lineend - linestart; @@ -986,7 +740,7 @@ extends Layout int i; for (i = len; i >= 0; i--) { - float w = widths[i - 1 + linestart - widstart + widoff]; + float w = widths[i - 1 + linestart - widstart]; if (w + sum + ellipsiswid > avail) { break; @@ -1002,7 +756,7 @@ extends Layout int i; for (i = 0; i < len; i++) { - float w = widths[i + linestart - widstart + widoff]; + float w = widths[i + linestart - widstart]; if (w + sum + ellipsiswid > avail) { break; @@ -1019,7 +773,7 @@ extends Layout float ravail = (avail - ellipsiswid) / 2; for (right = len; right >= 0; right--) { - float w = widths[right - 1 + linestart - widstart + widoff]; + float w = widths[right - 1 + linestart - widstart]; if (w + rsum > ravail) { break; @@ -1030,7 +784,7 @@ extends Layout float lavail = avail - ellipsiswid - rsum; for (left = 0; left < right; left++) { - float w = widths[left + linestart - widstart + widoff]; + float w = widths[left + linestart - widstart]; if (w + lsum > lavail) { break; @@ -1047,7 +801,7 @@ extends Layout mLines[mColumns * line + ELLIPSIS_COUNT] = ellipsisCount; } - // Override the baseclass so we can directly access our members, + // Override the base class so we can directly access our members, // rather than relying on member functions. // The logic mirrors that of Layout.getLineForVertical // FIXME: It may be faster to do a linear search for layouts without many lines. @@ -1156,10 +910,8 @@ extends Layout private static final char FIRST_RIGHT_TO_LEFT = '\u0590'; /* - * These are reused across calls to generate() + * This is reused across calls to generate() */ - private byte[] mChdirs; - private char[] mChs; - private float[] mWidths; + private MeasuredText mMeasured; private Paint.FontMetricsInt mFontMetricsInt = new Paint.FontMetricsInt(); } diff --git a/core/java/android/text/Styled.java b/core/java/android/text/Styled.java deleted file mode 100644 index 513b2cd444373..0000000000000 --- a/core/java/android/text/Styled.java +++ /dev/null @@ -1,434 +0,0 @@ -/* - * Copyright (C) 2006 The Android Open Source Project - * - * Licensed under the Apache License, Version 2.0 (the "License"); - * you may not use this file except in compliance with the License. - * You may obtain a copy of the License at - * - * http://www.apache.org/licenses/LICENSE-2.0 - * - * Unless required by applicable law or agreed to in writing, software - * distributed under the License is distributed on an "AS IS" BASIS, - * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. - * See the License for the specific language governing permissions and - * limitations under the License. - */ -package android.text; - -import android.graphics.Canvas; -import android.graphics.Paint; -import android.text.style.CharacterStyle; -import android.text.style.MetricAffectingSpan; -import android.text.style.ReplacementSpan; - -/** - * This class provides static methods for drawing and measuring styled text, - * like {@link android.text.Spanned} object with - * {@link android.text.style.ReplacementSpan}. - * - * @hide - */ -public class Styled -{ - /** - * Draws and/or measures a uniform run of text on a single line. No span of - * interest should start or end in the middle of this run (if not - * drawing, character spans that don't affect metrics can be ignored). - * Neither should the run direction change in the middle of the run. - * - *

The x position is the leading edge of the text. In a right-to-left - * paragraph, this will be to the right of the text to be drawn. Paint - * should not have an Align value other than LEFT or positioning will get - * confused. - * - *

On return, workPaint will reflect the original paint plus any - * modifications made by character styles on the run. - * - *

The returned width is signed and will be < 0 if the paragraph - * direction is right-to-left. - */ - private static float drawUniformRun(Canvas canvas, - Spanned text, int start, int end, - int dir, boolean runIsRtl, - float x, int top, int y, int bottom, - Paint.FontMetricsInt fmi, - TextPaint paint, - TextPaint workPaint, - boolean needWidth) { - - boolean haveWidth = false; - float ret = 0; - CharacterStyle[] spans = text.getSpans(start, end, CharacterStyle.class); - - ReplacementSpan replacement = null; - - // XXX: This shouldn't be modifying paint, only workPaint. - // However, the members belonging to TextPaint should have default - // values anyway. Better to ensure this in the Layout constructor. - paint.bgColor = 0; - paint.baselineShift = 0; - workPaint.set(paint); - - if (spans.length > 0) { - for (int i = 0; i < spans.length; i++) { - CharacterStyle span = spans[i]; - - if (span instanceof ReplacementSpan) { - replacement = (ReplacementSpan)span; - } - else { - span.updateDrawState(workPaint); - } - } - } - - if (replacement == null) { - CharSequence tmp; - int tmpstart, tmpend; - - if (runIsRtl) { - tmp = TextUtils.getReverse(text, start, end); - tmpstart = 0; - // XXX: assumes getReverse doesn't change the length of the text - tmpend = end - start; - } else { - tmp = text; - tmpstart = start; - tmpend = end; - } - - if (fmi != null) { - workPaint.getFontMetricsInt(fmi); - } - - if (canvas != null) { - if (workPaint.bgColor != 0) { - int c = workPaint.getColor(); - Paint.Style s = workPaint.getStyle(); - workPaint.setColor(workPaint.bgColor); - workPaint.setStyle(Paint.Style.FILL); - - if (!haveWidth) { - ret = workPaint.measureText(tmp, tmpstart, tmpend); - haveWidth = true; - } - - if (dir == Layout.DIR_RIGHT_TO_LEFT) - canvas.drawRect(x - ret, top, x, bottom, workPaint); - else - canvas.drawRect(x, top, x + ret, bottom, workPaint); - - workPaint.setStyle(s); - workPaint.setColor(c); - } - - if (dir == Layout.DIR_RIGHT_TO_LEFT) { - if (!haveWidth) { - ret = workPaint.measureText(tmp, tmpstart, tmpend); - haveWidth = true; - } - - canvas.drawText(tmp, tmpstart, tmpend, - x - ret, y + workPaint.baselineShift, workPaint); - } else { - if (needWidth) { - if (!haveWidth) { - ret = workPaint.measureText(tmp, tmpstart, tmpend); - haveWidth = true; - } - } - - canvas.drawText(tmp, tmpstart, tmpend, - x, y + workPaint.baselineShift, workPaint); - } - } else { - if (needWidth && !haveWidth) { - ret = workPaint.measureText(tmp, tmpstart, tmpend); - haveWidth = true; - } - } - } else { - ret = replacement.getSize(workPaint, text, start, end, fmi); - - if (canvas != null) { - if (dir == Layout.DIR_RIGHT_TO_LEFT) - replacement.draw(canvas, text, start, end, - x - ret, top, y, bottom, workPaint); - else - replacement.draw(canvas, text, start, end, - x, top, y, bottom, workPaint); - } - } - - if (dir == Layout.DIR_RIGHT_TO_LEFT) - return -ret; - else - return ret; - } - - /** - * Returns the advance widths for a uniform left-to-right run of text with - * no style changes in the middle of the run. If any style is replacement - * text, the first character will get the width of the replacement and the - * remaining characters will get a width of 0. - * - * @param paint the paint, will not be modified - * @param workPaint a paint to modify; on return will reflect the original - * paint plus the effect of all spans on the run - * @param text the text - * @param start the start of the run - * @param end the limit of the run - * @param widths array to receive the advance widths of the characters. Must - * be at least a large as (end - start). - * @param fmi FontMetrics information; can be null - * @return the actual number of widths returned - */ - public static int getTextWidths(TextPaint paint, - TextPaint workPaint, - Spanned text, int start, int end, - float[] widths, Paint.FontMetricsInt fmi) { - MetricAffectingSpan[] spans = - text.getSpans(start, end, MetricAffectingSpan.class); - - ReplacementSpan replacement = null; - workPaint.set(paint); - - for (int i = 0; i < spans.length; i++) { - MetricAffectingSpan span = spans[i]; - if (span instanceof ReplacementSpan) { - replacement = (ReplacementSpan)span; - } - else { - span.updateMeasureState(workPaint); - } - } - - if (replacement == null) { - workPaint.getFontMetricsInt(fmi); - workPaint.getTextWidths(text, start, end, widths); - } else { - int wid = replacement.getSize(workPaint, text, start, end, fmi); - - if (end > start) { - widths[0] = wid; - for (int i = start + 1; i < end; i++) - widths[i - start] = 0; - } - } - return end - start; - } - - /** - * Renders and/or measures a directional run of text on a single line. - * Unlike {@link #drawUniformRun}, this can render runs that cross style - * boundaries. Returns the signed advance width, if requested. - * - *

The x position is the leading edge of the text. In a right-to-left - * paragraph, this will be to the right of the text to be drawn. Paint - * should not have an Align value other than LEFT or positioning will get - * confused. - * - *

This optimizes for unstyled text and so workPaint might not be - * modified by this call. - * - *

The returned advance width will be < 0 if the paragraph - * direction is right-to-left. - */ - private static float drawDirectionalRun(Canvas canvas, - CharSequence text, int start, int end, - int dir, boolean runIsRtl, - float x, int top, int y, int bottom, - Paint.FontMetricsInt fmi, - TextPaint paint, - TextPaint workPaint, - boolean needWidth) { - - // XXX: It looks like all calls to this API match dir and runIsRtl, so - // having both parameters is redundant and confusing. - - // fast path for unstyled text - if (!(text instanceof Spanned)) { - float ret = 0; - - if (runIsRtl) { - CharSequence tmp = TextUtils.getReverse(text, start, end); - // XXX: this assumes getReverse doesn't tweak the length of - // the text - int tmpend = end - start; - - if (canvas != null || needWidth) - ret = paint.measureText(tmp, 0, tmpend); - - if (canvas != null) - canvas.drawText(tmp, 0, tmpend, - x - ret, y, paint); - } else { - if (needWidth) - ret = paint.measureText(text, start, end); - - if (canvas != null) - canvas.drawText(text, start, end, x, y, paint); - } - - if (fmi != null) { - paint.getFontMetricsInt(fmi); - } - - return ret * dir; // Layout.DIR_RIGHT_TO_LEFT == -1 - } - - float ox = x; - int minAscent = 0, maxDescent = 0, minTop = 0, maxBottom = 0; - - Spanned sp = (Spanned) text; - Class division; - - if (canvas == null) - division = MetricAffectingSpan.class; - else - division = CharacterStyle.class; - - int next; - for (int i = start; i < end; i = next) { - next = sp.nextSpanTransition(i, end, division); - - // XXX: if dir and runIsRtl were not the same, this would draw - // spans in the wrong order, but no one appears to call it this - // way. - x += drawUniformRun(canvas, sp, i, next, dir, runIsRtl, - x, top, y, bottom, fmi, paint, workPaint, - needWidth || next != end); - - if (fmi != null) { - if (fmi.ascent < minAscent) - minAscent = fmi.ascent; - if (fmi.descent > maxDescent) - maxDescent = fmi.descent; - - if (fmi.top < minTop) - minTop = fmi.top; - if (fmi.bottom > maxBottom) - maxBottom = fmi.bottom; - } - } - - if (fmi != null) { - if (start == end) { - paint.getFontMetricsInt(fmi); - } else { - fmi.ascent = minAscent; - fmi.descent = maxDescent; - fmi.top = minTop; - fmi.bottom = maxBottom; - } - } - - return x - ox; - } - - /** - * Draws a unidirectional run of text on a single line, and optionally - * returns the signed advance. Unlike drawDirectionalRun, the paragraph - * direction and run direction can be different. - */ - /* package */ static float drawText(Canvas canvas, - CharSequence text, int start, int end, - int dir, boolean runIsRtl, - float x, int top, int y, int bottom, - TextPaint paint, - TextPaint workPaint, - boolean needWidth) { - // XXX this logic is (dir == DIR_LEFT_TO_RIGHT) == runIsRtl - if ((dir == Layout.DIR_RIGHT_TO_LEFT && !runIsRtl) || - (runIsRtl && dir == Layout.DIR_LEFT_TO_RIGHT)) { - // TODO: this needs the real direction - float ch = drawDirectionalRun(null, text, start, end, - Layout.DIR_LEFT_TO_RIGHT, false, 0, 0, 0, 0, null, paint, - workPaint, true); - - ch *= dir; // DIR_RIGHT_TO_LEFT == -1 - drawDirectionalRun(canvas, text, start, end, -dir, - runIsRtl, x + ch, top, y, bottom, null, paint, - workPaint, true); - - return ch; - } - - return drawDirectionalRun(canvas, text, start, end, dir, runIsRtl, - x, top, y, bottom, null, paint, workPaint, - needWidth); - } - - /** - * Draws a run of text on a single line, with its - * origin at (x,y), in the specified Paint. The origin is interpreted based - * on the Align setting in the Paint. - * - * This method considers style information in the text (e.g. even when text - * is an instance of {@link android.text.Spanned}, this method correctly - * draws the text). See also - * {@link android.graphics.Canvas#drawText(CharSequence, int, int, float, - * float, Paint)} and - * {@link android.graphics.Canvas#drawRect(float, float, float, float, - * Paint)}. - * - * @param canvas The target canvas - * @param text The text to be drawn - * @param start The index of the first character in text to draw - * @param end (end - 1) is the index of the last character in text to draw - * @param direction The direction of the text. This must be - * {@link android.text.Layout#DIR_LEFT_TO_RIGHT} or - * {@link android.text.Layout#DIR_RIGHT_TO_LEFT}. - * @param x The x-coordinate of origin for where to draw the text - * @param top The top side of the rectangle to be drawn - * @param y The y-coordinate of origin for where to draw the text - * @param bottom The bottom side of the rectangle to be drawn - * @param paint The main {@link TextPaint} object. - * @param workPaint The {@link TextPaint} object used for temporal - * workspace. - * @param needWidth If true, this method returns the width of drawn text - * @return Width of the drawn text if needWidth is true - */ - public static float drawText(Canvas canvas, - CharSequence text, int start, int end, - int direction, - float x, int top, int y, int bottom, - TextPaint paint, - TextPaint workPaint, - boolean needWidth) { - // For safety. - direction = direction >= 0 ? Layout.DIR_LEFT_TO_RIGHT - : Layout.DIR_RIGHT_TO_LEFT; - - // Hide runIsRtl parameter since it is meaningless for external - // developers. - // XXX: the runIsRtl probably ought to be the same as direction, then - // this could draw rtl text. - return drawText(canvas, text, start, end, direction, false, - x, top, y, bottom, paint, workPaint, needWidth); - } - - /** - * Returns the width of a run of left-to-right text on a single line, - * considering style information in the text (e.g. even when text is an - * instance of {@link android.text.Spanned}, this method correctly measures - * the width of the text). - * - * @param paint the main {@link TextPaint} object; will not be modified - * @param workPaint the {@link TextPaint} object available for modification; - * will not necessarily be used - * @param text the text to measure - * @param start the index of the first character to start measuring - * @param end 1 beyond the index of the last character to measure - * @param fmi FontMetrics information; can be null - * @return The width of the text - */ - public static float measureText(TextPaint paint, - TextPaint workPaint, - CharSequence text, int start, int end, - Paint.FontMetricsInt fmi) { - return drawDirectionalRun(null, text, start, end, - Layout.DIR_LEFT_TO_RIGHT, false, - 0, 0, 0, 0, fmi, paint, workPaint, true); - } -} diff --git a/core/java/android/text/TextLine.java b/core/java/android/text/TextLine.java new file mode 100644 index 0000000000000..8ab481be7199f --- /dev/null +++ b/core/java/android/text/TextLine.java @@ -0,0 +1,1053 @@ +/* + * Copyright (C) 2010 The Android Open Source Project + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +package android.text; + +import com.android.internal.util.ArrayUtils; + +import android.graphics.Bitmap; +import android.graphics.Canvas; +import android.graphics.Paint; +import android.graphics.RectF; +import android.graphics.Paint.FontMetricsInt; +import android.icu.text.ArabicShaping; +import android.text.Layout.Directions; +import android.text.style.CharacterStyle; +import android.text.style.MetricAffectingSpan; +import android.text.style.ReplacementSpan; +import android.text.style.TabStopSpan; +import android.util.Log; + +/** + * Represents a line of styled text, for measuring in visual order and + * for rendering. + * + *

Get a new instance using obtain(), and when finished with it, return it + * to the pool using recycle(). + * + *

Call set to prepare the instance for use, then either draw, measure, + * metrics, or caretToLeftRightOf. + * + * @hide + */ +class TextLine { + private TextPaint mPaint; + private CharSequence mText; + private int mStart; + private int mLen; + private int mDir; + private Directions mDirections; + private boolean mHasTabs; + private TabStopSpan[] mTabs; + + private char[] mChars; + private boolean mCharsValid; + private Spanned mSpanned; + private TextPaint mWorkPaint = new TextPaint(); + private int mPreppedIndex; + private int mPreppedLimit; + + private static TextLine[] cached = new TextLine[3]; + + /** + * Returns a new TextLine from the shared pool. + * + * @return an uninitialized TextLine + */ + static TextLine obtain() { + TextLine tl; + synchronized (cached) { + for (int i = cached.length; --i >= 0;) { + if (cached[i] != null) { + tl = cached[i]; + cached[i] = null; + return tl; + } + } + } + tl = new TextLine(); + Log.e("TLINE", "new: " + tl); + return tl; + } + + /** + * Puts a TextLine back into the shared pool. Do not use this TextLine once + * it has been returned. + * @param tl the textLine + * @return null, as a convenience from clearing references to the provided + * TextLine + */ + static TextLine recycle(TextLine tl) { + tl.mText = null; + tl.mPaint = null; + tl.mDirections = null; + if (tl.mLen < 250) { + synchronized(cached) { + for (int i = 0; i < cached.length; ++i) { + if (cached[i] == null) { + cached[i] = tl; + break; + } + } + } + } + return null; + } + + /** + * Initializes a TextLine and prepares it for use. + * + * @param paint the base paint for the line + * @param text the text, can be Styled + * @param start the start of the line relative to the text + * @param limit the limit of the line relative to the text + * @param dir the paragraph direction of this line + * @param directions the directions information of this line + * @param hasTabs true if the line might contain tabs or emoji + * @param spans array of paragraph-level spans, of which only TabStopSpans + * are used. Can be null. + */ + void set(TextPaint paint, CharSequence text, int start, int limit, int dir, + Directions directions, boolean hasTabs, Object[] spans) { + mPaint = paint; + mText = text; + mStart = start; + mLen = limit - start; + mDir = dir; + mDirections = directions; + mHasTabs = hasTabs; + mSpanned = null; + mPreppedIndex = 0; + mPreppedLimit = 0; + + boolean hasReplacement = false; + if (text instanceof Spanned) { + mSpanned = (Spanned) text; + hasReplacement = mSpanned.getSpans(start, limit, + ReplacementSpan.class).length > 0; + } + + mCharsValid = hasReplacement || hasTabs || + directions != Layout.DIRS_ALL_LEFT_TO_RIGHT; + + if (mCharsValid) { + if (mChars == null || mChars.length < mLen) { + mChars = new char[ArrayUtils.idealCharArraySize(mLen)]; + } + TextUtils.getChars(text, start, limit, mChars, 0); + + if (hasTabs) { + TabStopSpan[] tabs = mTabs; + int tabLen = 0; + if (mSpanned != null && spans == null) { + TabStopSpan[] newTabs = mSpanned.getSpans(start, limit, + TabStopSpan.class); + if (tabs == null || tabs.length < newTabs.length) { + tabs = newTabs; + } else { + for (int i = 0; i < newTabs.length; ++i) { + tabs[i] = newTabs[i]; + } + } + tabLen = newTabs.length; + } else if (spans != null) { + if (tabs == null || tabs.length < spans.length) { + tabs = new TabStopSpan[spans.length]; + } + for (int i = 0; i < spans.length; ++i) { + if (spans[i] instanceof TabStopSpan) { + tabs[tabLen++] = (TabStopSpan) spans[i]; + } + } + } + + if (tabs != null && tabLen < tabs.length){ + tabs[tabLen] = null; + } + mTabs = tabs; + } + } + } + + /** + * Renders the TextLine. + * + * @param c the canvas to render on + * @param x the leading margin position + * @param top the top of the line + * @param y the baseline + * @param bottom the bottom of the line + */ + void draw(Canvas c, float x, int top, int y, int bottom) { + if (!mHasTabs) { + if (mDirections == Layout.DIRS_ALL_LEFT_TO_RIGHT) { + drawRun(c, 0, 0, mLen, false, x, top, y, bottom, false); + return; + } + if (mDirections == Layout.DIRS_ALL_RIGHT_TO_LEFT) { + drawRun(c, 0, 0, mLen, true, x, top, y, bottom, false); + return; + } + } + + float h = 0; + int[] runs = mDirections.mDirections; + RectF emojiRect = null; + + int lastRunIndex = runs.length - 2; + for (int i = 0; i < runs.length; i += 2) { + int runStart = runs[i]; + int runLimit = runStart + (runs[i+1] & Layout.RUN_LENGTH_MASK); + if (runLimit > mLen) { + runLimit = mLen; + } + boolean runIsRtl = (runs[i+1] & Layout.RUN_RTL_FLAG) != 0; + + int segstart = runStart; + char[] chars = mChars; + for (int j = mHasTabs ? runStart : runLimit; j <= runLimit; j++) { + int codept = 0; + Bitmap bm = null; + + if (mHasTabs && j < runLimit) { + codept = mChars[j]; + if (codept >= 0xd800 && codept < 0xdc00 && j + 1 < runLimit) { + codept = Character.codePointAt(mChars, j); + if (codept >= Layout.MIN_EMOJI && codept <= Layout.MAX_EMOJI) { + bm = Layout.EMOJI_FACTORY.getBitmapFromAndroidPua(codept); + } else if (codept > 0xffff) { + ++j; + continue; + } + } + } + + if (j == runLimit || codept == '\t' || bm != null) { + h += drawRun(c, i, segstart, j, runIsRtl, x+h, top, y, bottom, + i != lastRunIndex || j != mLen); + + if (codept == '\t') { + h = mDir * nextTab(h * mDir); + } else if (bm != null) { + float bmAscent = ascent(j); + float bitmapHeight = bm.getHeight(); + float scale = -bmAscent / bitmapHeight; + float width = bm.getWidth() * scale; + + if (emojiRect == null) { + emojiRect = new RectF(); + } + emojiRect.set(x + h, y + bmAscent, + x + h + width, y); + c.drawBitmap(bm, null, emojiRect, mPaint); + h += width; + j++; + } + segstart = j + 1; + } + } + } + } + + /** + * Returns metrics information for the entire line. + * + * @param fmi receives font metrics information, can be null + * @return the signed width of the line + */ + float metrics(FontMetricsInt fmi) { + return measure(mLen, false, fmi); + } + + /** + * Returns information about a position on the line. + * + * @param offset the line-relative character offset, between 0 and the + * line length, inclusive + * @param trailing true to measure the trailing edge of the character + * before offset, false to measure the leading edge of the character + * at offset. + * @param fmi receives metrics information about the requested + * character, can be null. + * @return the signed offset from the leading margin to the requested + * character edge. + */ + float measure(int offset, boolean trailing, FontMetricsInt fmi) { + int target = trailing ? offset - 1 : offset; + if (target < 0) { + return 0; + } + + float h = 0; + + if (!mHasTabs) { + if (mDirections == Layout.DIRS_ALL_LEFT_TO_RIGHT) { + return measureRun( 0, 0, target, mLen, false, fmi); + } + if (mDirections == Layout.DIRS_ALL_RIGHT_TO_LEFT) { + return measureRun(0, 0, target, mLen, true, fmi); + } + } + + char[] chars = mChars; + int[] runs = mDirections.mDirections; + for (int i = 0; i < runs.length; i += 2) { + int runStart = runs[i]; + int runLimit = runStart + (runs[i+1] & Layout.RUN_LENGTH_MASK); + if (runLimit > mLen) { + runLimit = mLen; + } + boolean runIsRtl = (runs[i+1] & Layout.RUN_RTL_FLAG) != 0; + + int segstart = runStart; + for (int j = mHasTabs ? runStart : runLimit; j <= runLimit; j++) { + int codept = 0; + Bitmap bm = null; + + if (mHasTabs && j < runLimit) { + codept = chars[j]; + if (codept >= 0xd800 && codept < 0xdc00 && j + 1 < runLimit) { + codept = Character.codePointAt(chars, j); + if (codept >= Layout.MIN_EMOJI && codept <= Layout.MAX_EMOJI) { + bm = Layout.EMOJI_FACTORY.getBitmapFromAndroidPua(codept); + } else if (codept > 0xffff) { + ++j; + continue; + } + } + } + + if (j == runLimit || codept == '\t' || bm != null) { + boolean inSegment = target >= segstart && target < j; + + boolean advance = (mDir == Layout.DIR_RIGHT_TO_LEFT) == runIsRtl; + if (inSegment && advance) { + return h += measureRun(i, segstart, offset, j, runIsRtl, fmi); + } + + float w = measureRun(i, segstart, j, j, runIsRtl, fmi); + h += advance ? w : -w; + + if (inSegment) { + return h += measureRun(i, segstart, offset, j, runIsRtl, null); + } + + if (codept == '\t') { + if (offset == j) { + return h; + } + h = mDir * nextTab(h * mDir); + if (target == j) { + return h; + } + } + + if (bm != null) { + float bmAscent = ascent(j); + float wid = bm.getWidth() * -bmAscent / bm.getHeight(); + h += mDir * wid; + j++; + } + + segstart = j + 1; + } + } + } + + return h; + } + + /** + * Draws a unidirectional (but possibly multi-styled) run of text. + * + * @param c the canvas to draw on + * @param runIndex the index of this directional run + * @param start the line-relative start + * @param limit the line-relative limit + * @param runIsRtl true if the run is right-to-left + * @param x the position of the run that is closest to the leading margin + * @param top the top of the line + * @param y the baseline + * @param bottom the bottom of the line + * @param needWidth true if the width value is required. + * @return the signed width of the run, based on the paragraph direction. + * Only valid if needWidth is true. + */ + private float drawRun(Canvas c, int runIndex, int start, + int limit, boolean runIsRtl, float x, int top, int y, int bottom, + boolean needWidth) { + + if ((mDir == Layout.DIR_LEFT_TO_RIGHT) == runIsRtl) { + float w = -measureRun(runIndex, start, limit, limit, runIsRtl, null); + handleRun(runIndex, start, limit, limit, runIsRtl, c, x + w, top, + y, bottom, null, false, PREP_NONE); + return w; + } + + return handleRun(runIndex, start, limit, limit, runIsRtl, c, x, top, + y, bottom, null, needWidth, PREP_NEEDED); + } + + /** + * Measures a unidirectional (but possibly multi-styled) run of text. + * + * @param runIndex the run index + * @param start the line-relative start of the run + * @param offset the offset to measure to, between start and limit inclusive + * @param limit the line-relative limit of the run + * @param runIsRtl true if the run is right-to-left + * @param fmi receives metrics information about the requested + * run, can be null. + * @return the signed width from the start of the run to the leading edge + * of the character at offset, based on the run (not paragraph) direction + */ + private float measureRun(int runIndex, int start, + int offset, int limit, boolean runIsRtl, FontMetricsInt fmi) { + return handleRun(runIndex, start, offset, limit, runIsRtl, null, + 0, 0, 0, 0, fmi, true, PREP_NEEDED); + } + + /** + * Prepares a run for measurement or rendering. This ensures that any + * required shaping of the text in the run has been performed so that + * measurements reflect the shaped text. + * + * @param runIndex the run index + * @param start the line-relative start of the run + * @param limit the line-relative limit of the run + * @param runIsRtl true if the run is right-to-left + */ + private void prepRun(int runIndex, int start, int limit, + boolean runIsRtl) { + handleRun(runIndex, start, limit, limit, runIsRtl, null, 0, 0, 0, + 0, null, false, PREP_ONLY); + } + + /** + * Walk the cursor through this line, skipping conjuncts and + * zero-width characters. + * + *

This function cannot properly walk the cursor off the ends of the line + * since it does not know about any shaping on the previous/following line + * that might affect the cursor position. Callers must either avoid these + * situations or handle the result specially. + * + *

The paint is required because the region around the cursor might not + * have been formatted yet, and the valid positions can depend on the glyphs + * used to render the text, which in turn depends on the paint. + * + * @param paint the base paint of the line + * @param cursor the starting position of the cursor, between 0 and the + * length of the line, inclusive + * @param toLeft true if the caret is moving to the left. + * @return the new offset. If it is less than 0 or greater than the length + * of the line, the previous/following line should be examined to get the + * actual offset. + */ + int getOffsetToLeftRightOf(int cursor, boolean toLeft) { + // 1) The caret marks the leading edge of a character. The character + // logically before it might be on a different level, and the active caret + // position is on the character at the lower level. If that character + // was the previous character, the caret is on its trailing edge. + // 2) Take this character/edge and move it in the indicated direction. + // This gives you a new character and a new edge. + // 3) This position is between two visually adjacent characters. One of + // these might be at a lower level. The active position is on the + // character at the lower level. + // 4) If the active position is on the trailing edge of the character, + // the new caret position is the following logical character, else it + // is the character. + + int lineStart = 0; + int lineEnd = mLen; + boolean paraIsRtl = mDir == -1; + int[] runs = mDirections.mDirections; + + int runIndex, runLevel = 0, runStart = lineStart, runLimit = lineEnd, newCaret = -1; + boolean trailing = false; + + if (cursor == lineStart) { + runIndex = -2; + } else if (cursor == lineEnd) { + runIndex = runs.length; + } else { + // First, get information about the run containing the character with + // the active caret. + for (runIndex = 0; runIndex < runs.length; runIndex += 2) { + runStart = lineStart + runs[runIndex]; + if (cursor >= runStart) { + runLimit = runStart + (runs[runIndex+1] & Layout.RUN_LENGTH_MASK); + if (runLimit > lineEnd) { + runLimit = lineEnd; + } + if (cursor < runLimit) { + runLevel = (runs[runIndex+1] >>> Layout.RUN_LEVEL_SHIFT) & + Layout.RUN_LEVEL_MASK; + if (cursor == runStart) { + // The caret is on a run boundary, see if we should + // use the position on the trailing edge of the previous + // logical character instead. + int prevRunIndex, prevRunLevel, prevRunStart, prevRunLimit; + int pos = cursor - 1; + for (prevRunIndex = 0; prevRunIndex < runs.length; prevRunIndex += 2) { + prevRunStart = lineStart + runs[prevRunIndex]; + if (pos >= prevRunStart) { + prevRunLimit = prevRunStart + + (runs[prevRunIndex+1] & Layout.RUN_LENGTH_MASK); + if (prevRunLimit > lineEnd) { + prevRunLimit = lineEnd; + } + if (pos < prevRunLimit) { + prevRunLevel = (runs[prevRunIndex+1] >>> Layout.RUN_LEVEL_SHIFT) + & Layout.RUN_LEVEL_MASK; + if (prevRunLevel < runLevel) { + // Start from logically previous character. + runIndex = prevRunIndex; + runLevel = prevRunLevel; + runStart = prevRunStart; + runLimit = prevRunLimit; + trailing = true; + break; + } + } + } + } + } + break; + } + } + } + + // caret might be == lineEnd. This is generally a space or paragraph + // separator and has an associated run, but might be the end of + // text, in which case it doesn't. If that happens, we ran off the + // end of the run list, and runIndex == runs.length. In this case, + // we are at a run boundary so we skip the below test. + if (runIndex != runs.length) { + boolean runIsRtl = (runLevel & 0x1) != 0; + boolean advance = toLeft == runIsRtl; + if (cursor != (advance ? runLimit : runStart) || advance != trailing) { + // Moving within or into the run, so we can move logically. + prepRun(runIndex, runStart, runLimit, runIsRtl); + newCaret = getOffsetBeforeAfter(runIndex, cursor, advance); + // If the new position is internal to the run, we're at the strong + // position already so we're finished. + if (newCaret != (advance ? runLimit : runStart)) { + return newCaret; + } + } + } + } + + // If newCaret is -1, we're starting at a run boundary and crossing + // into another run. Otherwise we've arrived at a run boundary, and + // need to figure out which character to attach to. Note we might + // need to run this twice, if we cross a run boundary and end up at + // another run boundary. + while (true) { + boolean advance = toLeft == paraIsRtl; + int otherRunIndex = runIndex + (advance ? 2 : -2); + if (otherRunIndex >= 0 && otherRunIndex < runs.length) { + int otherRunStart = lineStart + runs[otherRunIndex]; + int otherRunLimit = otherRunStart + + (runs[otherRunIndex+1] & Layout.RUN_LENGTH_MASK); + if (otherRunLimit > lineEnd) { + otherRunLimit = lineEnd; + } + int otherRunLevel = (runs[otherRunIndex+1] >>> Layout.RUN_LEVEL_SHIFT) & + Layout.RUN_LEVEL_MASK; + boolean otherRunIsRtl = (otherRunLevel & 1) != 0; + + advance = toLeft == otherRunIsRtl; + if (newCaret == -1) { + prepRun(otherRunIndex, otherRunStart, otherRunLimit, + otherRunIsRtl); + newCaret = getOffsetBeforeAfter(otherRunIndex, + advance ? otherRunStart : otherRunLimit, advance); + if (newCaret == (advance ? otherRunLimit : otherRunStart)) { + // Crossed and ended up at a new boundary, + // repeat a second and final time. + runIndex = otherRunIndex; + runLevel = otherRunLevel; + continue; + } + break; + } + + // The new caret is at a boundary. + if (otherRunLevel < runLevel) { + // The strong character is in the other run. + newCaret = advance ? otherRunStart : otherRunLimit; + } + break; + } + + if (newCaret == -1) { + // We're walking off the end of the line. The paragraph + // level is always equal to or lower than any internal level, so + // the boundaries get the strong caret. + newCaret = getOffsetBeforeAfter(-1, cursor, advance); + break; + } + + // Else we've arrived at the end of the line. That's a strong position. + // We might have arrived here by crossing over a run with no internal + // breaks and dropping out of the above loop before advancing one final + // time, so reset the caret. + // Note, we use '<=' below to handle a situation where the only run + // on the line is a counter-directional run. If we're not advancing, + // we can end up at the 'lineEnd' position but the caret we want is at + // the lineStart. + if (newCaret <= lineEnd) { + newCaret = advance ? lineEnd : lineStart; + } + break; + } + + return newCaret; + } + + /** + * Returns the next valid offset within this directional run, skipping + * conjuncts and zero-width characters. This should not be called to walk + * off the end of the run. + * + * @param runIndex the run index + * @param offset the offset + * @param after true if the new offset should logically follow the provided + * offset + * @return the new offset + */ + private int getOffsetBeforeAfter(int runIndex, int offset, boolean after) { + // XXX note currently there is no special handling of zero-width + // combining marks, since the only analysis involves mock shaping. + + boolean offEnd = offset == (after ? mLen : 0); + if (runIndex >= 0 && !offEnd && mCharsValid) { + char[] chars = mChars; + if (after) { + int cp = Character.codePointAt(chars, offset, mLen); + if (cp >= 0x10000) { + ++offset; + } + while (++offset < mLen && chars[offset] == '\ufeff'){} + } else { + while (--offset >= 0 && chars[offset] == '\ufeff'){} + int cp = Character.codePointBefore(chars, offset + 1); + if (cp >= 0x10000) { + --offset; + } + } + return offset; + } + + if (after) { + return TextUtils.getOffsetAfter(mText, offset + mStart) - mStart; + } + return TextUtils.getOffsetBefore(mText, offset + mStart) - mStart; + } + + /** + * Utility function for measuring and rendering text. The text must + * not include a tab or emoji. + * + * @param wp the working paint + * @param start the start of the text + * @param limit the limit of the text + * @param runIsRtl true if the run is right-to-left + * @param c the canvas, can be null if rendering is not needed + * @param x the edge of the run closest to the leading margin + * @param top the top of the line + * @param y the baseline + * @param bottom the bottom of the line + * @param fmi receives metrics information, can be null + * @param needWidth true if the width of the run is needed + * @return the signed width of the run based on the run direction; only + * valid if needWidth is true + */ + private float handleText(TextPaint wp, int start, int limit, + boolean runIsRtl, Canvas c, float x, int top, int y, int bottom, + FontMetricsInt fmi, boolean needWidth) { + + float ret = 0; + + int runLen = limit - start; + if (needWidth || (c != null && (wp.bgColor != 0 || runIsRtl))) { + if (mCharsValid) { + ret = wp.measureText(mChars, start, runLen); + } else { + ret = wp.measureText(mText, mStart + start, + mStart + start + runLen); + } + } + + if (fmi != null) { + wp.getFontMetricsInt(fmi); + } + + if (c != null) { + if (runIsRtl) { + x -= ret; + } + + if (wp.bgColor != 0) { + int color = wp.getColor(); + Paint.Style s = wp.getStyle(); + wp.setColor(wp.bgColor); + wp.setStyle(Paint.Style.FILL); + + c.drawRect(x, top, x + ret, bottom, wp); + + wp.setStyle(s); + wp.setColor(color); + } + + drawTextRun(c, wp, start, limit, runIsRtl, x, y + wp.baselineShift); + } + + return runIsRtl ? -ret : ret; + } + + /** + * Utility function for measuring and rendering a replacement. + * + * @param replacement the replacement + * @param wp the work paint + * @param runIndex the run index + * @param start the start of the run + * @param limit the limit of the run + * @param runIsRtl true if the run is right-to-left + * @param c the canvas, can be null if not rendering + * @param x the edge of the replacement closest to the leading margin + * @param top the top of the line + * @param y the baseline + * @param bottom the bottom of the line + * @param fmi receives metrics information, can be null + * @param needWidth true if the width of the replacement is needed + * @param prepFlags one of PREP_NONE, PREP_REQUIRED, or PREP_ONLY + * @return the signed width of the run based on the run direction; only + * valid if needWidth is true + */ + private float handleReplacement(ReplacementSpan replacement, TextPaint wp, + int runIndex, int start, int limit, boolean runIsRtl, Canvas c, + float x, int top, int y, int bottom, FontMetricsInt fmi, + boolean needWidth, int prepFlags) { + + float ret = 0; + + // Preparation replaces the first character of the series with the + // object-replacement character and the remainder with zero width + // non-break space aka BOM. Cursor movement code skips over the BOMs + // so that the replacement character is the only character 'seen'. + if (prepFlags != PREP_NONE && limit > start && + (runIndex > mPreppedIndex || + (runIndex == mPreppedIndex && start >= mPreppedLimit))) { + char[] chars = mChars; + chars[start] = '\ufffc'; + for (int i = start + 1; i < limit; ++i) { + chars[i] = '\ufeff'; // used as ZWNBS, marks positions to skip + } + mPreppedIndex = runIndex; + mPreppedLimit = limit; + } + + if (prepFlags != PREP_ONLY) { + int textStart = mStart + start; + int textLimit = mStart + limit; + + if (needWidth || (c != null && runIsRtl)) { + ret = replacement.getSize(wp, mText, textStart, textLimit, fmi); + } + + if (c != null) { + if (runIsRtl) { + x -= ret; + } + replacement.draw(c, mText, textStart, textLimit, + x, top, y, bottom, wp); + } + } + + return runIsRtl ? -ret : ret; + } + + /** + * Utility function for handling a unidirectional run. The run must not + * contain tabs or emoji but can contain styles. + * + * @param p the base paint + * @param runIndex the run index + * @param start the line-relative start of the run + * @param offset the offset to measure to, between start and limit inclusive + * @param limit the limit of the run + * @param runIsRtl true if the run is right-to-left + * @param c the canvas, can be null + * @param x the end of the run closest to the leading margin + * @param top the top of the line + * @param y the baseline + * @param bottom the bottom of the line + * @param fmi receives metrics information, can be null + * @param needWidth true if the width is required + * @param prepFlags one of PREP_NONE, PREP_REQUIRED, or PREP_ONLY + * @return the signed width of the run based on the run direction; only + * valid if needWidth is true + */ + private float handleRun(int runIndex, int start, int offset, + int limit, boolean runIsRtl, Canvas c, float x, int top, int y, + int bottom, FontMetricsInt fmi, boolean needWidth, int prepFlags) { + + // Shaping needs to take into account context up to metric boundaries, + // but rendering needs to take into account character style boundaries. + // So we iterate through metric runs, shape using the initial + // paint (the same typeface is used up to the next metric boundary), + // then within each metric run iterate through character style runs. + float ox = x; + for (int i = start, inext; i < offset; i = inext) { + TextPaint wp = mWorkPaint; + wp.set(mPaint); + + int mnext; + if (mSpanned == null) { + inext = limit; + mnext = offset; + } else { + inext = mSpanned.nextSpanTransition(mStart + i, mStart + limit, + MetricAffectingSpan.class) - mStart; + + mnext = inext < offset ? inext : offset; + MetricAffectingSpan[] spans = mSpanned.getSpans(mStart + i, + mStart + mnext, MetricAffectingSpan.class); + + if (spans.length > 0) { + ReplacementSpan replacement = null; + for (int j = 0; j < spans.length; j++) { + MetricAffectingSpan span = spans[j]; + if (span instanceof ReplacementSpan) { + replacement = (ReplacementSpan)span; + } else { + span.updateDrawState(wp); // XXX or measureState? + } + } + + if (replacement != null) { + x += handleReplacement(replacement, wp, runIndex, i, + mnext, runIsRtl, c, x, top, y, bottom, fmi, + needWidth || mnext < offset, prepFlags); + continue; + } + } + } + + if (prepFlags != PREP_NONE) { + handlePrep(wp, runIndex, i, inext, runIsRtl); + } + + if (prepFlags != PREP_ONLY) { + if (mSpanned == null || c == null) { + x += handleText(wp, i, mnext, runIsRtl, c, x, top, + y, bottom, fmi, needWidth || mnext < offset); + } else { + for (int j = i, jnext; j < mnext; j = jnext) { + jnext = mSpanned.nextSpanTransition(mStart + j, + mStart + mnext, CharacterStyle.class) - mStart; + + CharacterStyle[] spans = mSpanned.getSpans(mStart + j, + mStart + jnext, CharacterStyle.class); + + wp.set(mPaint); + for (int k = 0; k < spans.length; k++) { + CharacterStyle span = spans[k]; + span.updateDrawState(wp); + } + + x += handleText(wp, j, jnext, runIsRtl, c, x, + top, y, bottom, fmi, needWidth || jnext < offset); + } + } + } + } + + return x - ox; + } + + private static final int PREP_NONE = 0; + private static final int PREP_NEEDED = 1; + private static final int PREP_ONLY = 2; + + /** + * Prepares text for measuring or rendering. + * + * @param paint the paint used to shape the text + * @param runIndex the run index + * @param start the start of the text to prepare + * @param limit the limit of the text to prepare + * @param runIsRtl true if the run is right-to-left + */ + private void handlePrep(TextPaint paint, int runIndex, int start, int limit, + boolean runIsRtl) { + + // The current implementation 'prepares' text by manipulating the + // character array. In order to keep track of what ranges have + // already been prepared, it uses the runIndex and the limit of + // the prepared text within that run. This index is required + // since operations that prepare the text always proceed in visual + // order and the limit itself does not let us know which runs have + // been processed and which have not. + // + // This bookkeeping is an attempt to let us process a line partially, + // for example, by only shaping up to the cursor position. This may + // not make sense if we can reuse the line, say by caching repeated + // accesses to the same line for both measuring and drawing, since in + // those cases we'd always prepare the entire line. At the + // opposite extreme, we might shape and then immediately discard only + // the run of text we're working with at the moment, instead of retaining + // the results of shaping (as the chars array is). In this case as well + // we would not need to do the index/limit bookkeeping. + // + // Technically, the only reason for bookkeeping is so that we don't + // re-mirror already-mirrored glyphs, since the shaping and object + // replacement operations will not change already-processed text. + + if (runIndex > mPreppedIndex || + (runIndex == mPreppedIndex && start >= mPreppedLimit)) { + if (runIsRtl) { + int runLen = limit - start; + AndroidCharacter.mirror(mChars, start, runLen); + ArabicShaping.SHAPER.shape(mChars, start, runLen); + + // Note: tweaked MockShaper to put '\ufeff' in place of + // alef when it forms lam-alef ligatures, so no extra + // processing is necessary here. + } + mPreppedIndex = runIndex; + mPreppedLimit = limit; + } + } + + /** + * Render a text run with the set-up paint. + * + * @param c the canvas + * @param wp the paint used to render the text + * @param start the run start + * @param limit the run limit + * @param runIsRtl true if the run is right-to-left + * @param x the x position of the left edge of the run + * @param y the baseline of the run + */ + private void drawTextRun(Canvas c, TextPaint wp, int start, int limit, + boolean runIsRtl, float x, int y) { + + // Since currently skia only renders text left-to-right, we need to + // put the shaped characters into visual order before rendering. + // Since we might want to re-render the line again, we swap them + // back when we're done. If we left them swapped, measurement + // would be broken since it expects the characters in logical order. + if (runIsRtl) { + swapRun(start, limit); + } + if (mCharsValid) { + c.drawText(mChars, start, limit - start, x, y, wp); + } else { + c.drawText(mText, mStart + start, mStart + limit, x, y, wp); + } + if (runIsRtl) { + swapRun(start, limit); + } + } + + /** + * Reverses the order of characters in the chars array between start and + * limit, used by drawTextRun. + * @param start the start of the run to reverse + * @param limit the limit of the run to reverse + */ + private void swapRun(int start, int limit) { + // First we swap all the characters one for one, then we + // do another pass looking for surrogate pairs and swapping them + // back into their logical order. + char[] chars = mChars; + for (int s = start, e = limit - 1; s < e; ++s, --e) { + char ch = chars[s]; chars[s] = chars[e]; chars[e] = ch; + } + + for (int s = start, e = limit - 1; s < e; ++s) { + char c1 = chars[s]; + if (c1 >= 0xdc00 && c1 < 0xe000) { + char c2 = chars[s+1]; + if (c2 >= 0xd800 && c2 < 0xdc00) { + chars[s++] = c2; + chars[s] = c1; + } + } + } + } + + /** + * Returns the ascent of the text at start. This is used for scaling + * emoji. + * + * @param pos the line-relative position + * @return the ascent of the text at start + */ + float ascent(int pos) { + if (mSpanned == null) { + return mPaint.ascent(); + } + + pos += mStart; + MetricAffectingSpan[] spans = mSpanned.getSpans(pos, pos + 1, + MetricAffectingSpan.class); + if (spans.length == 0) { + return mPaint.ascent(); + } + + TextPaint wp = mWorkPaint; + wp.set(mPaint); + for (MetricAffectingSpan span : spans) { + span.updateMeasureState(wp); + } + return wp.ascent(); + } + + /** + * Returns the next tab position. + * + * @param h the (unsigned) offset from the leading margin + * @return the (unsigned) tab position after this offset + */ + float nextTab(float h) { + float nh = Float.MAX_VALUE; + boolean alltabs = false; + + if (mHasTabs && mTabs != null) { + TabStopSpan[] tabs = mTabs; + for (int i = 0; i < tabs.length && tabs[i] != null; ++i) { + int where = tabs[i].getTabStop(); + if (where < nh && where > h) { + nh = where; + } + } + if (nh != Float.MAX_VALUE) { + return nh; + } + } + + return ((int) ((h + TAB_INCREMENT) / TAB_INCREMENT)) * TAB_INCREMENT; + } + + private static final int TAB_INCREMENT = 20; +} diff --git a/core/java/android/text/TextUtils.java b/core/java/android/text/TextUtils.java index 9589bf369e6ea..2d6c7b68fce36 100644 --- a/core/java/android/text/TextUtils.java +++ b/core/java/android/text/TextUtils.java @@ -17,12 +17,11 @@ package android.text; import com.android.internal.R; +import com.android.internal.util.ArrayUtils; -import android.content.res.ColorStateList; import android.content.res.Resources; import android.os.Parcel; import android.os.Parcelable; -import android.text.method.TextKeyListener.Capitalize; import android.text.style.AbsoluteSizeSpan; import android.text.style.AlignmentSpan; import android.text.style.BackgroundColorSpan; @@ -45,10 +44,8 @@ import android.text.style.URLSpan; import android.text.style.UnderlineSpan; import android.util.Printer; -import com.android.internal.util.ArrayUtils; - -import java.util.regex.Pattern; import java.util.Iterator; +import java.util.regex.Pattern; public class TextUtils { private TextUtils() { /* cannot be instantiated */ } @@ -983,7 +980,7 @@ public class TextUtils { /** * Returns the original text if it fits in the specified width * given the properties of the specified Paint, - * or, if it does not fit, a copy with ellipsis character added + * or, if it does not fit, a copy with ellipsis character added * at the specified edge or center. * If preserveLength is specified, the returned copy * will be padded with zero-width spaces to preserve the original @@ -992,7 +989,7 @@ public class TextUtils { * report the start and end of the ellipsized range. */ public static CharSequence ellipsize(CharSequence text, - TextPaint p, + TextPaint paint, float avail, TruncateAt where, boolean preserveLength, EllipsizeCallback callback) { @@ -1003,13 +1000,12 @@ public class TextUtils { int len = text.length(); - // Use Paint.breakText() for the non-Spanned case to avoid having - // to allocate memory and accumulate the character widths ourselves. + MeasuredText mt = MeasuredText.obtain(); + try { + float width = setPara(mt, paint, text, 0, text.length(), + Layout.DIR_REQUEST_DEFAULT_LTR); - if (!(text instanceof Spanned)) { - float wid = p.measureText(text, 0, len); - - if (wid <= avail) { + if (width <= avail) { if (callback != null) { callback.ellipsized(0, 0); } @@ -1017,250 +1013,69 @@ public class TextUtils { return text; } - float ellipsiswid = p.measureText(sEllipsis); + // XXX assumes ellipsis string does not require shaping and + // is unaffected by style + float ellipsiswid = paint.measureText(sEllipsis); + avail -= ellipsiswid; - if (ellipsiswid > avail) { - if (callback != null) { - callback.ellipsized(0, len); - } - - if (preserveLength) { - char[] buf = obtain(len); - for (int i = 0; i < len; i++) { - buf[i] = '\uFEFF'; - } - String ret = new String(buf, 0, len); - recycle(buf); - return ret; - } else { - return ""; - } - } - - if (where == TruncateAt.START) { - int fit = p.breakText(text, 0, len, false, - avail - ellipsiswid, null); - - if (callback != null) { - callback.ellipsized(0, len - fit); - } - - if (preserveLength) { - return blank(text, 0, len - fit); - } else { - return sEllipsis + text.toString().substring(len - fit, len); - } + int left = 0; + int right = len; + if (avail < 0) { + // it all goes + } else if (where == TruncateAt.START) { + right = len - mt.breakText(0, len, false, avail); } else if (where == TruncateAt.END) { - int fit = p.breakText(text, 0, len, true, - avail - ellipsiswid, null); - - if (callback != null) { - callback.ellipsized(fit, len); - } - - if (preserveLength) { - return blank(text, fit, len); - } else { - return text.toString().substring(0, fit) + sEllipsis; - } - } else /* where == TruncateAt.MIDDLE */ { - int right = p.breakText(text, 0, len, false, - (avail - ellipsiswid) / 2, null); - float used = p.measureText(text, len - right, len); - int left = p.breakText(text, 0, len - right, true, - avail - ellipsiswid - used, null); - - if (callback != null) { - callback.ellipsized(left, len - right); - } - - if (preserveLength) { - return blank(text, left, len - right); - } else { - String s = text.toString(); - return s.substring(0, left) + sEllipsis + - s.substring(len - right, len); - } - } - } - - // But do the Spanned cases by hand, because it's such a pain - // to iterate the span transitions backwards and getTextWidths() - // will give us the information we need. - - // getTextWidths() always writes into the start of the array, - // so measure each span into the first half and then copy the - // results into the second half to use later. - - float[] wid = new float[len * 2]; - TextPaint temppaint = new TextPaint(); - Spanned sp = (Spanned) text; - - int next; - for (int i = 0; i < len; i = next) { - next = sp.nextSpanTransition(i, len, MetricAffectingSpan.class); - - Styled.getTextWidths(p, temppaint, sp, i, next, wid, null); - System.arraycopy(wid, 0, wid, len + i, next - i); - } - - float sum = 0; - for (int i = 0; i < len; i++) { - sum += wid[len + i]; - } - - if (sum <= avail) { - if (callback != null) { - callback.ellipsized(0, 0); - } - - return text; - } - - float ellipsiswid = p.measureText(sEllipsis); - - if (ellipsiswid > avail) { - if (callback != null) { - callback.ellipsized(0, len); - } - - if (preserveLength) { - char[] buf = obtain(len); - for (int i = 0; i < len; i++) { - buf[i] = '\uFEFF'; - } - SpannableString ss = new SpannableString(new String(buf, 0, len)); - recycle(buf); - copySpansFrom(sp, 0, len, Object.class, ss, 0); - return ss; + left = mt.breakText(0, len, true, avail); } else { - return ""; - } - } - - if (where == TruncateAt.START) { - sum = 0; - int i; - - for (i = len; i >= 0; i--) { - float w = wid[len + i - 1]; - - if (w + sum + ellipsiswid > avail) { - break; - } - - sum += w; - } - - if (callback != null) { - callback.ellipsized(0, i); - } - - if (preserveLength) { - SpannableString ss = new SpannableString(blank(text, 0, i)); - copySpansFrom(sp, 0, len, Object.class, ss, 0); - return ss; - } else { - SpannableStringBuilder out = new SpannableStringBuilder(sEllipsis); - out.insert(1, text, i, len); - - return out; - } - } else if (where == TruncateAt.END) { - sum = 0; - int i; - - for (i = 0; i < len; i++) { - float w = wid[len + i]; - - if (w + sum + ellipsiswid > avail) { - break; - } - - sum += w; - } - - if (callback != null) { - callback.ellipsized(i, len); - } - - if (preserveLength) { - SpannableString ss = new SpannableString(blank(text, i, len)); - copySpansFrom(sp, 0, len, Object.class, ss, 0); - return ss; - } else { - SpannableStringBuilder out = new SpannableStringBuilder(sEllipsis); - out.insert(0, text, 0, i); - - return out; - } - } else /* where = TruncateAt.MIDDLE */ { - float lsum = 0, rsum = 0; - int left = 0, right = len; - - float ravail = (avail - ellipsiswid) / 2; - for (right = len; right >= 0; right--) { - float w = wid[len + right - 1]; - - if (w + rsum > ravail) { - break; - } - - rsum += w; - } - - float lavail = avail - ellipsiswid - rsum; - for (left = 0; left < right; left++) { - float w = wid[len + left]; - - if (w + lsum > lavail) { - break; - } - - lsum += w; + right = len - mt.breakText(0, len, false, avail / 2); + avail -= mt.measure(right, len); + left = mt.breakText(0, right, true, avail); } if (callback != null) { callback.ellipsized(left, right); } + char[] buf = mt.mChars; + Spanned sp = text instanceof Spanned ? (Spanned) text : null; + + int remaining = len - (right - left); if (preserveLength) { - SpannableString ss = new SpannableString(blank(text, left, right)); + if (remaining > 0) { // else eliminate the ellipsis too + buf[left++] = '\u2026'; + } + for (int i = left; i < right; i++) { + buf[i] = '\uFEFF'; + } + String s = new String(buf, 0, len); + if (sp == null) { + return s; + } + SpannableString ss = new SpannableString(s); copySpansFrom(sp, 0, len, Object.class, ss, 0); return ss; - } else { - SpannableStringBuilder out = new SpannableStringBuilder(sEllipsis); - out.insert(0, text, 0, left); - out.insert(out.length(), text, right, len); - - return out; } - } - } - private static String blank(CharSequence source, int start, int end) { - int len = source.length(); - char[] buf = obtain(len); - - if (start != 0) { - getChars(source, 0, start, buf, 0); - } - if (end != len) { - getChars(source, end, len, buf, end); - } - - if (start != end) { - buf[start] = '\u2026'; - - for (int i = start + 1; i < end; i++) { - buf[i] = '\uFEFF'; + if (remaining == 0) { + return ""; } - } - - String ret = new String(buf, 0, len); - recycle(buf); - return ret; + if (sp == null) { + StringBuilder sb = new StringBuilder(remaining + sEllipsis.length()); + sb.append(buf, 0, left); + sb.append(sEllipsis); + sb.append(buf, right, len - right); + return sb.toString(); + } + + SpannableStringBuilder ssb = new SpannableStringBuilder(); + ssb.append(text, 0, left); + ssb.append(sEllipsis); + ssb.append(text, right, len); + return ssb; + } finally { + MeasuredText.recycle(mt); + } } /** @@ -1278,80 +1093,121 @@ public class TextUtils { TextPaint p, float avail, String oneMore, String more) { - int len = text.length(); - char[] buf = new char[len]; - TextUtils.getChars(text, 0, len, buf, 0); - int commaCount = 0; - for (int i = 0; i < len; i++) { - if (buf[i] == ',') { - commaCount++; - } - } - - float[] wid; - - if (text instanceof Spanned) { - Spanned sp = (Spanned) text; - TextPaint temppaint = new TextPaint(); - wid = new float[len * 2]; - - int next; - for (int i = 0; i < len; i = next) { - next = sp.nextSpanTransition(i, len, MetricAffectingSpan.class); - - Styled.getTextWidths(p, temppaint, sp, i, next, wid, null); - System.arraycopy(wid, 0, wid, len + i, next - i); + MeasuredText mt = MeasuredText.obtain(); + try { + int len = text.length(); + float width = setPara(mt, p, text, 0, len, Layout.DIR_REQUEST_DEFAULT_LTR); + if (width <= avail) { + return text; } - System.arraycopy(wid, len, wid, 0, len); - } else { - wid = new float[len]; - p.getTextWidths(text, 0, len, wid); - } + char[] buf = mt.mChars; - int ok = 0; - int okRemaining = commaCount + 1; - String okFormat = ""; - - int w = 0; - int count = 0; - - for (int i = 0; i < len; i++) { - w += wid[i]; - - if (buf[i] == ',') { - count++; - - int remaining = commaCount - count + 1; - float moreWid; - String format; - - if (remaining == 1) { - format = " " + oneMore; - } else { - format = " " + String.format(more, remaining); - } - - moreWid = p.measureText(format); - - if (w + moreWid <= avail) { - ok = i + 1; - okRemaining = remaining; - okFormat = format; + int commaCount = 0; + for (int i = 0; i < len; i++) { + if (buf[i] == ',') { + commaCount++; } } - } - if (w <= avail) { - return text; - } else { + int remaining = commaCount + 1; + + int ok = 0; + int okRemaining = remaining; + String okFormat = ""; + + int w = 0; + int count = 0; + float[] widths = mt.mWidths; + + int request = mt.mDir == 1 ? Layout.DIR_REQUEST_LTR : + Layout.DIR_REQUEST_RTL; + + MeasuredText tempMt = MeasuredText.obtain(); + for (int i = 0; i < len; i++) { + w += widths[i]; + + if (buf[i] == ',') { + count++; + + String format; + // XXX should not insert spaces, should be part of string + // XXX should use plural rules and not assume English plurals + if (--remaining == 1) { + format = " " + oneMore; + } else { + format = " " + String.format(more, remaining); + } + + // XXX this is probably ok, but need to look at it more + tempMt.setPara(format, 0, format.length(), request); + float moreWid = mt.addStyleRun(p, mt.mLen, null); + + if (w + moreWid <= avail) { + ok = i + 1; + okRemaining = remaining; + okFormat = format; + } + } + } + MeasuredText.recycle(tempMt); + SpannableStringBuilder out = new SpannableStringBuilder(okFormat); out.insert(0, text, 0, ok); return out; + } finally { + MeasuredText.recycle(mt); } } + private static float setPara(MeasuredText mt, TextPaint paint, + CharSequence text, int start, int end, int bidiRequest) { + + mt.setPara(text, start, end, bidiRequest); + + float width; + Spanned sp = text instanceof Spanned ? (Spanned) text : null; + int len = end - start; + if (sp == null) { + width = mt.addStyleRun(paint, len, null); + } else { + width = 0; + int spanEnd; + for (int spanStart = 0; spanStart < len; spanStart = spanEnd) { + spanEnd = sp.nextSpanTransition(spanStart, len, + MetricAffectingSpan.class); + MetricAffectingSpan[] spans = sp.getSpans( + spanStart, spanEnd, MetricAffectingSpan.class); + width += mt.addStyleRun(paint, spans, spanEnd - spanStart, null); + } + } + + return width; + } + + private static final char FIRST_RIGHT_TO_LEFT = '\u0590'; + + /* package */ + static boolean doesNotNeedBidi(CharSequence s, int start, int end) { + for (int i = start; i < end; i++) { + if (s.charAt(i) >= FIRST_RIGHT_TO_LEFT) { + return false; + } + } + return true; + } + + /* package */ + static boolean doesNotNeedBidi(char[] text, int start, int len) { + for (int i = start, e = i + len; i < e; i++) { + if (text[i] >= FIRST_RIGHT_TO_LEFT) { + return false; + } + } + return true; + } + /* package */ static char[] obtain(int len) { char[] buf; @@ -1529,7 +1385,7 @@ public class TextUtils { */ public static final int CAP_MODE_CHARACTERS = InputType.TYPE_TEXT_FLAG_CAP_CHARACTERS; - + /** * Capitalization mode for {@link #getCapsMode}: capitalize the first * character of all words. This value is explicitly defined to be the same as @@ -1537,7 +1393,7 @@ public class TextUtils { */ public static final int CAP_MODE_WORDS = InputType.TYPE_TEXT_FLAG_CAP_WORDS; - + /** * Capitalization mode for {@link #getCapsMode}: capitalize the first * character of each sentence. This value is explicitly defined to be the same as @@ -1545,13 +1401,13 @@ public class TextUtils { */ public static final int CAP_MODE_SENTENCES = InputType.TYPE_TEXT_FLAG_CAP_SENTENCES; - + /** * Determine what caps mode should be in effect at the current offset in * the text. Only the mode bits set in reqModes will be * checked. Note that the caps mode flags here are explicitly defined * to match those in {@link InputType}. - * + * * @param cs The text that should be checked for caps modes. * @param off Location in the text at which to check. * @param reqModes The modes to be checked: may be any combination of @@ -1651,7 +1507,7 @@ public class TextUtils { return mode; } - + private static Object sLock = new Object(); private static char[] sTemp = null; } diff --git a/icu4j/java/android/icu/text/ArabicShaping.java b/icu4j/java/android/icu/text/ArabicShaping.java new file mode 100644 index 0000000000000..13e2175340a3a --- /dev/null +++ b/icu4j/java/android/icu/text/ArabicShaping.java @@ -0,0 +1,1947 @@ +/* +******************************************************************************* +* Copyright (C) 2001-2009, International Business Machines +* Corporation and others. All Rights Reserved. +******************************************************************************* +*/ + +/* + * Ported with minor modifications from ICU4J 4.2's + * com.ibm.icu.text.ArabicShaping class. + */ + +package android.icu.text; + + +/** + * Shape Arabic text on a character basis. + * + *

ArabicShaping performs basic operations for "shaping" Arabic text. It is most + * useful for use with legacy data formats and legacy display technology + * (simple terminals). All operations are performed on Unicode characters.

+ * + *

Text-based shaping means that some character code points in the text are + * replaced by others depending on the context. It transforms one kind of text + * into another. In comparison, modern displays for Arabic text select + * appropriate, context-dependent font glyphs for each text element, which means + * that they transform text into a glyph vector.

+ * + *

Text transformations are necessary when modern display technology is not + * available or when text needs to be transformed to or from legacy formats that + * use "shaped" characters. Since the Arabic script is cursive, connecting + * adjacent letters to each other, computers select images for each letter based + * on the surrounding letters. This usually results in four images per Arabic + * letter: initial, middle, final, and isolated forms. In Unicode, on the other + * hand, letters are normally stored abstract, and a display system is expected + * to select the necessary glyphs. (This makes searching and other text + * processing easier because the same letter has only one code.) It is possible + * to mimic this with text transformations because there are characters in + * Unicode that are rendered as letters with a specific shape + * (or cursive connectivity). They were included for interoperability with + * legacy systems and codepages, and for unsophisticated display systems.

+ * + *

A second kind of text transformations is supported for Arabic digits: + * For compatibility with legacy codepages that only include European digits, + * it is possible to replace one set of digits by another, changing the + * character code points. These operations can be performed for either + * Arabic-Indic Digits (U+0660...U+0669) or Eastern (Extended) Arabic-Indic + * digits (U+06f0...U+06f9).

+ * + *

Some replacements may result in more or fewer characters (code points). + * By default, this means that the destination buffer may receive text with a + * length different from the source length. Some legacy systems rely on the + * length of the text to be constant. They expect extra spaces to be added + * or consumed either next to the affected character or at the end of the + * text.

+ * @stable ICU 2.0 + * + * @hide + */ +public class ArabicShaping { + private final int options; + private boolean isLogical; // convenience + private boolean spacesRelativeToTextBeginEnd; + private char tailChar; + + public static final ArabicShaping SHAPER = new ArabicShaping( + ArabicShaping.TEXT_DIRECTION_LOGICAL | + ArabicShaping.LENGTH_FIXED_SPACES_NEAR | + ArabicShaping.LETTERS_SHAPE | + ArabicShaping.DIGITS_NOOP); + + /** + * Convert a range of text in the source array, putting the result + * into a range of text in the destination array, and return the number + * of characters written. + * + * @param source An array containing the input text + * @param sourceStart The start of the range of text to convert + * @param sourceLength The length of the range of text to convert + * @param dest The destination array that will receive the result. + * It may be NULL only if destSize is 0. + * @param destStart The start of the range of the destination buffer to use. + * @param destSize The size (capacity) of the destination buffer. + * If destSize is 0, then no output is produced, + * but the necessary buffer size is returned ("preflighting"). This + * does not validate the text against the options, for example, + * if letters are being unshaped, and spaces are being consumed + * following lamalef, this will not detect a lamalef without a + * corresponding space. An error will be thrown when the actual + * conversion is attempted. + * @return The number of chars written to the destination buffer. + * If an error occurs, then no output was written, or it may be + * incomplete. + * @throws ArabicShapingException if the text cannot be converted according to the options. + * @stable ICU 2.0 + */ + public int shape(char[] source, int sourceStart, int sourceLength, + char[] dest, int destStart, int destSize) throws ArabicShapingException { + if (source == null) { + throw new IllegalArgumentException("source can not be null"); + } + if (sourceStart < 0 || sourceLength < 0 || sourceStart + sourceLength > source.length) { + throw new IllegalArgumentException("bad source start (" + sourceStart + + ") or length (" + sourceLength + + ") for buffer of length " + source.length); + } + if (dest == null && destSize != 0) { + throw new IllegalArgumentException("null dest requires destSize == 0"); + } + if ((destSize != 0) && + (destStart < 0 || destSize < 0 || destStart + destSize > dest.length)) { + throw new IllegalArgumentException("bad dest start (" + destStart + + ") or size (" + destSize + + ") for buffer of length " + dest.length); + } + /* Validate input options */ + if ( ((options&TASHKEEL_MASK) > 0) && + !(((options & TASHKEEL_MASK)==TASHKEEL_BEGIN) || + ((options & TASHKEEL_MASK)==TASHKEEL_END ) || + ((options & TASHKEEL_MASK)==TASHKEEL_RESIZE )|| + ((options & TASHKEEL_MASK)==TASHKEEL_REPLACE_BY_TATWEEL)) ){ + throw new IllegalArgumentException("Wrong Tashkeel argument"); + } + + ///CLOVER:OFF + //According to Steven Loomis, the code is unreachable when you OR all the constants within the if statements + if(((options&LAMALEF_MASK) > 0)&& + !(((options & LAMALEF_MASK)==LAMALEF_BEGIN) || + ((options & LAMALEF_MASK)==LAMALEF_END ) || + ((options & LAMALEF_MASK)==LAMALEF_RESIZE )|| + ((options & LAMALEF_MASK)==LAMALEF_AUTO) || + ((options & LAMALEF_MASK)==LAMALEF_NEAR))){ + throw new IllegalArgumentException("Wrong Lam Alef argument"); + } + ///CLOVER:ON + + /* Validate Tashkeel (Tashkeel replacement options should be enabled in shaping mode only)*/ + if(((options&TASHKEEL_MASK) > 0) && (options&LETTERS_MASK) == LETTERS_UNSHAPE) { + throw new IllegalArgumentException("Tashkeel replacement should not be enabled in deshaping mode "); + } + return internalShape(source, sourceStart, sourceLength, dest, destStart, destSize); + } + + /** + * Convert a range of text in place. This may only be used if the Length option + * does not grow or shrink the text. + * + * @param source An array containing the input text + * @param start The start of the range of text to convert + * @param length The length of the range of text to convert + * @throws ArabicShapingException if the text cannot be converted according to the options. + * @stable ICU 2.0 + */ + public void shape(char[] source, int start, int length) throws ArabicShapingException { + if ((options & LAMALEF_MASK) == LAMALEF_RESIZE) { + throw new ArabicShapingException("Cannot shape in place with length option resize."); + } + shape(source, start, length, source, start, length); + } + + /** + * Convert a string, returning the new string. + * + * @param text the string to convert + * @return the converted string + * @throws ArabicShapingException if the string cannot be converted according to the options. + * @stable ICU 2.0 + */ + public String shape(String text) throws ArabicShapingException { + char[] src = text.toCharArray(); + char[] dest = src; + if (((options & LAMALEF_MASK) == LAMALEF_RESIZE) && + ((options & LETTERS_MASK) == LETTERS_UNSHAPE)) { + + dest = new char[src.length * 2]; // max + } + int len = shape(src, 0, src.length, dest, 0, dest.length); + + return new String(dest, 0, len); + } + + /** + * Construct ArabicShaping using the options flags. + * The flags are as follows:
+ * 'LENGTH' flags control whether the text can change size, and if not, + * how to maintain the size of the text when LamAlef ligatures are + * formed or broken.
+ * 'TEXT_DIRECTION' flags control whether the text is read and written + * in visual order or in logical order.
+ * 'LETTERS_SHAPE' flags control whether conversion is to or from + * presentation forms.
+ * 'DIGITS' flags control whether digits are shaped, and whether from + * European to Arabic-Indic or vice-versa.
+ * 'DIGIT_TYPE' flags control whether standard or extended Arabic-Indic + * digits are used when performing digit conversion. + * @stable ICU 2.0 + */ + public ArabicShaping(int options) { + this.options = options; + if ((options & DIGITS_MASK) > 0x80) { + throw new IllegalArgumentException("bad DIGITS options"); + } + + isLogical = ( (options & TEXT_DIRECTION_MASK) == TEXT_DIRECTION_LOGICAL ); + /* Validate options */ + spacesRelativeToTextBeginEnd = ( (options & SPACES_RELATIVE_TO_TEXT_MASK) == SPACES_RELATIVE_TO_TEXT_BEGIN_END ); + if ( (options&SHAPE_TAIL_TYPE_MASK) == SHAPE_TAIL_NEW_UNICODE){ + tailChar = NEW_TAIL_CHAR; + } else { + tailChar = OLD_TAIL_CHAR; + } + } + + /* Seen Tail options */ + /** + * Memory option: the result must have the same length as the source. + * Shaping mode: The SEEN family character will expand into two characters using space near + * the SEEN family character(i.e. the space after the character). + * if there are no spaces found, ArabicShapingException will be thrown + * + * De-shaping mode: Any Seen character followed by Tail character will be + * replaced by one cell Seen and a space will replace the Tail. + * Affects: Seen options + */ + public static final int SEEN_TWOCELL_NEAR = 0x200000; + + /** Bit mask for Seen memory options. */ + public static final int SEEN_MASK = 0x700000; + + /* YehHamza options */ + /** + * Memory option: the result must have the same length as the source. + * Shaping mode: The YEHHAMZA character will expand into two characters using space near it + * (i.e. the space after the character) + * if there are no spaces found, ArabicShapingException will be thrown + * + * De-shaping mode: Any Yeh (final or isolated) character followed by Hamza character will be + * replaced by one cell YehHamza and space will replace the Hamza. + * Affects: YehHamza options + */ + public static final int YEHHAMZA_TWOCELL_NEAR = 0x1000000; + + + /** Bit mask for YehHamza memory options. */ + public static final int YEHHAMZA_MASK = 0x3800000; + + /* New Tashkeel options */ + /** + * Memory option: the result must have the same length as the source. + * Shaping mode: Tashkeel characters will be replaced by spaces. + * Spaces will be placed at beginning of the buffer + * + * De-shaping mode: N/A + * Affects: Tashkeel options + */ + public static final int TASHKEEL_BEGIN = 0x40000; + + /** + * Memory option: the result must have the same length as the source. + * Shaping mode: Tashkeel characters will be replaced by spaces. + * Spaces will be placed at end of the buffer + * + * De-shaping mode: N/A + * Affects: Tashkeel options + */ + public static final int TASHKEEL_END = 0x60000; + + /** + * Memory option: allow the result to have a different length than the source. + * Shaping mode: Tashkeel characters will be removed, buffer length will shrink. + * De-shaping mode: N/A + * + * Affects: Tashkeel options + */ + public static final int TASHKEEL_RESIZE = 0x80000; + + /** + * Memory option: the result must have the same length as the source. + * Shaping mode: Tashkeel characters will be replaced by Tatweel if it is connected to adjacent + * characters (i.e. shaped on Tatweel) or replaced by space if it is not connected. + * + * De-shaping mode: N/A + * Affects: YehHamza options + */ + public static final int TASHKEEL_REPLACE_BY_TATWEEL = 0xC0000; + + /** Bit mask for Tashkeel replacement with Space or Tatweel memory options. */ + public static final int TASHKEEL_MASK = 0xE0000; + + /* Space location Control options */ + /** + * This option effects the meaning of BEGIN and END options. if this option is not used the default + * for BEGIN and END will be as following: + * The Default (for both Visual LTR, Visual RTL and Logical Text) + * 1. BEGIN always refers to the start address of physical memory. + * 2. END always refers to the end address of physical memory. + * + * If this option is used it will swap the meaning of BEGIN and END only for Visual LTR text. + * + * The affect on BEGIN and END Memory Options will be as following: + * A. BEGIN For Visual LTR text: This will be the beginning (right side) of the visual text + * (corresponding to the physical memory address end, same as END in default behavior) + * B. BEGIN For Logical text: Same as BEGIN in default behavior. + * C. END For Visual LTR text: This will be the end (left side) of the visual text. (corresponding to + * the physical memory address beginning, same as BEGIN in default behavior) + * D. END For Logical text: Same as END in default behavior. + * Affects: All LamAlef BEGIN, END and AUTO options. + */ + public static final int SPACES_RELATIVE_TO_TEXT_BEGIN_END = 0x4000000; + + /** Bit mask for swapping BEGIN and END for Visual LTR text */ + public static final int SPACES_RELATIVE_TO_TEXT_MASK = 0x4000000; + + /** + * If this option is used, shaping will use the new Unicode code point for TAIL (i.e. 0xFE73). + * If this option is not specified (Default), old unofficial Unicode TAIL code point is used (i.e. 0x200B) + * De-shaping will not use this option as it will always search for both the new Unicode code point for the + * TAIL (i.e. 0xFE73) or the old unofficial Unicode TAIL code point (i.e. 0x200B) and de-shape the + * Seen-Family letter accordingly. + * + * Shaping Mode: Only shaping. + * De-shaping Mode: N/A. + * Affects: All Seen options + */ + public static final int SHAPE_TAIL_NEW_UNICODE = 0x8000000; + + /** Bit mask for new Unicode Tail option */ + public static final int SHAPE_TAIL_TYPE_MASK = 0x8000000; + + /** + * Memory option: allow the result to have a different length than the source. + * @stable ICU 2.0 + */ + public static final int LENGTH_GROW_SHRINK = 0; + + /** + * Memory option: allow the result to have a different length than the source. + * Affects: LamAlef options + * This option is an alias to LENGTH_GROW_SHRINK + */ + public static final int LAMALEF_RESIZE = 0; + + /** + * Memory option: the result must have the same length as the source. + * If more room is necessary, then try to consume spaces next to modified characters. + * @stable ICU 2.0 + */ + public static final int LENGTH_FIXED_SPACES_NEAR = 1; + + /** + * Memory option: the result must have the same length as the source. + * If more room is necessary, then try to consume spaces next to modified characters. + * Affects: LamAlef options + * This option is an alias to LENGTH_FIXED_SPACES_NEAR + */ + public static final int LAMALEF_NEAR = 1 ; + + /** + * Memory option: the result must have the same length as the source. + * If more room is necessary, then try to consume spaces at the end of the text. + * @stable ICU 2.0 + */ + public static final int LENGTH_FIXED_SPACES_AT_END = 2; + + + /** + * Memory option: the result must have the same length as the source. + * If more room is necessary, then try to consume spaces at the end of the text. + * Affects: LamAlef options + * This option is an alias to LENGTH_FIXED_SPACES_AT_END + */ + public static final int LAMALEF_END = 2; + + /** + * Memory option: the result must have the same length as the source. + * If more room is necessary, then try to consume spaces at the beginning of the text. + * @stable ICU 2.0 + */ + public static final int LENGTH_FIXED_SPACES_AT_BEGINNING = 3; + + /** + * Memory option: the result must have the same length as the source. + * If more room is necessary, then try to consume spaces at the beginning of the text. + * Affects: LamAlef options + * This option is an alias to LENGTH_FIXED_SPACES_AT_BEGINNING + */ + public static final int LAMALEF_BEGIN = 3; + + /** + * Memory option: the result must have the same length as the source. + * Shaping Mode: For each LAMALEF character found, expand LAMALEF using space at end. + * If there is no space at end, use spaces at beginning of the buffer. If there + * is no space at beginning of the buffer, use spaces at the near (i.e. the space + * after the LAMALEF character). + * + * Deshaping Mode: Perform the same function as the flag equals LAMALEF_END. + * Affects: LamAlef options + */ + public static final int LAMALEF_AUTO = 0x10000; + + /** + * Bit mask for memory options. + * @stable ICU 2.0 + */ + public static final int LENGTH_MASK = 0x10003; + + /** Bit mask for LamAlef memory options. */ + + public static final int LAMALEF_MASK = 0x10003; + + /** + * Direction indicator: the source is in logical (keyboard) order. + * @stable ICU 2.0 + */ + public static final int TEXT_DIRECTION_LOGICAL = 0; + + /** + * Direction indicator:the source is in visual RTL order, + * the rightmost displayed character stored first. + * This option is an alias to U_SHAPE_TEXT_DIRECTION_LOGICAL + */ + public static final int TEXT_DIRECTION_VISUAL_RTL = 0; + + /** + * Direction indicator: the source is in visual (display) order, that is, + * the leftmost displayed character is stored first. + * @stable ICU 2.0 + */ + public static final int TEXT_DIRECTION_VISUAL_LTR = 4; + + /** + * Bit mask for direction indicators. + * @stable ICU 2.0 + */ + public static final int TEXT_DIRECTION_MASK = 4; + + + /** + * Letter shaping option: do not perform letter shaping. + * @stable ICU 2.0 + */ + public static final int LETTERS_NOOP = 0; + + /** + * Letter shaping option: replace normative letter characters in the U+0600 (Arabic) block, + * by shaped ones in the U+FE70 (Presentation Forms B) block. Performs Lam-Alef ligature + * substitution. + * @stable ICU 2.0 + */ + public static final int LETTERS_SHAPE = 8; + + /** + * Letter shaping option: replace shaped letter characters in the U+FE70 (Presentation Forms B) block + * by normative ones in the U+0600 (Arabic) block. Converts Lam-Alef ligatures to pairs of Lam and + * Alef characters, consuming spaces if required. + * @stable ICU 2.0 + */ + public static final int LETTERS_UNSHAPE = 0x10; + + /** + * Letter shaping option: replace normative letter characters in the U+0600 (Arabic) block, + * except for the TASHKEEL characters at U+064B...U+0652, by shaped ones in the U+Fe70 + * (Presentation Forms B) block. The TASHKEEL characters will always be converted to + * the isolated forms rather than to their correct shape. + * @stable ICU 2.0 + */ + public static final int LETTERS_SHAPE_TASHKEEL_ISOLATED = 0x18; + + /** + * Bit mask for letter shaping options. + * @stable ICU 2.0 + */ + public static final int LETTERS_MASK = 0x18; + + + /** + * Digit shaping option: do not perform digit shaping. + * @stable ICU 2.0 + */ + public static final int DIGITS_NOOP = 0; + + /** + * Digit shaping option: Replace European digits (U+0030...U+0039) by Arabic-Indic digits. + * @stable ICU 2.0 + */ + public static final int DIGITS_EN2AN = 0x20; + + /** + * Digit shaping option: Replace Arabic-Indic digits by European digits (U+0030...U+0039). + * @stable ICU 2.0 + */ + public static final int DIGITS_AN2EN = 0x40; + + /** + * Digit shaping option: + * Replace European digits (U+0030...U+0039) by Arabic-Indic digits + * if the most recent strongly directional character + * is an Arabic letter (its Bidi direction value is RIGHT_TO_LEFT_ARABIC). + * The initial state at the start of the text is assumed to be not an Arabic, + * letter, so European digits at the start of the text will not change. + * Compare to DIGITS_ALEN2AN_INIT_AL. + * @stable ICU 2.0 + */ + public static final int DIGITS_EN2AN_INIT_LR = 0x60; + + /** + * Digit shaping option: + * Replace European digits (U+0030...U+0039) by Arabic-Indic digits + * if the most recent strongly directional character + * is an Arabic letter (its Bidi direction value is RIGHT_TO_LEFT_ARABIC). + * The initial state at the start of the text is assumed to be an Arabic, + * letter, so European digits at the start of the text will change. + * Compare to DIGITS_ALEN2AN_INT_LR. + * @stable ICU 2.0 + */ + public static final int DIGITS_EN2AN_INIT_AL = 0x80; + + /** Not a valid option value. */ + //private static final int DIGITS_RESERVED = 0xa0; + + /** + * Bit mask for digit shaping options. + * @stable ICU 2.0 + */ + public static final int DIGITS_MASK = 0xe0; + + /** + * Digit type option: Use Arabic-Indic digits (U+0660...U+0669). + * @stable ICU 2.0 + */ + public static final int DIGIT_TYPE_AN = 0; + + /** + * Digit type option: Use Eastern (Extended) Arabic-Indic digits (U+06f0...U+06f9). + * @stable ICU 2.0 + */ + public static final int DIGIT_TYPE_AN_EXTENDED = 0x100; + + /** + * Bit mask for digit type options. + * @stable ICU 2.0 + */ + public static final int DIGIT_TYPE_MASK = 0x0100; // 0x3f00? + + /** + * some constants + */ + private static final char HAMZAFE_CHAR = '\ufe80'; + private static final char HAMZA06_CHAR = '\u0621'; + private static final char YEH_HAMZA_CHAR = '\u0626'; + private static final char YEH_HAMZAFE_CHAR = '\uFE89'; + private static final char LAMALEF_SPACE_SUB = '\uffff'; + private static final char TASHKEEL_SPACE_SUB = '\ufffe'; + private static final char LAM_CHAR = '\u0644'; + private static final char SPACE_CHAR = '\u0020'; + private static final char SPACE_CHAR_FOR_LAMALEF = '\ufeff'; // XXX: tweak for TextLine use + private static final char SHADDA_CHAR = '\uFE7C'; + private static final char TATWEEL_CHAR = '\u0640'; + private static final char SHADDA_TATWEEL_CHAR = '\uFE7D'; + private static final char NEW_TAIL_CHAR = '\uFE73'; + private static final char OLD_TAIL_CHAR = '\u200B'; + private static final int SHAPE_MODE = 0; + private static final int DESHAPE_MODE = 1; + + /** + * @stable ICU 2.0 + */ + public boolean equals(Object rhs) { + return rhs != null && + rhs.getClass() == ArabicShaping.class && + options == ((ArabicShaping)rhs).options; + } + + /** + * @stable ICU 2.0 + */ + ///CLOVER:OFF + public int hashCode() { + return options; + } + + /** + * @stable ICU 2.0 + */ + public String toString() { + StringBuffer buf = new StringBuffer(super.toString()); + buf.append('['); + + switch (options & LAMALEF_MASK) { + case LAMALEF_RESIZE: buf.append("LamAlef resize"); break; + case LAMALEF_NEAR: buf.append("LamAlef spaces at near"); break; + case LAMALEF_BEGIN: buf.append("LamAlef spaces at begin"); break; + case LAMALEF_END: buf.append("LamAlef spaces at end"); break; + case LAMALEF_AUTO: buf.append("lamAlef auto"); break; + } + switch (options & TEXT_DIRECTION_MASK) { + case TEXT_DIRECTION_LOGICAL: buf.append(", logical"); break; + case TEXT_DIRECTION_VISUAL_LTR: buf.append(", visual"); break; + } + switch (options & LETTERS_MASK) { + case LETTERS_NOOP: buf.append(", no letter shaping"); break; + case LETTERS_SHAPE: buf.append(", shape letters"); break; + case LETTERS_SHAPE_TASHKEEL_ISOLATED: buf.append(", shape letters tashkeel isolated"); break; + case LETTERS_UNSHAPE: buf.append(", unshape letters"); break; + } + switch (options & SEEN_MASK) { + case SEEN_TWOCELL_NEAR: buf.append(", Seen at near"); break; + } + switch (options & YEHHAMZA_MASK) { + case YEHHAMZA_TWOCELL_NEAR: buf.append(", Yeh Hamza at near"); break; + } + switch (options & TASHKEEL_MASK) { + case TASHKEEL_BEGIN: buf.append(", Tashkeel at begin"); break; + case TASHKEEL_END: buf.append(", Tashkeel at end"); break; + case TASHKEEL_REPLACE_BY_TATWEEL: buf.append(", Tashkeel replace with tatweel"); break; + case TASHKEEL_RESIZE: buf.append(", Tashkeel resize"); break; + } + + switch (options & DIGITS_MASK) { + case DIGITS_NOOP: buf.append(", no digit shaping"); break; + case DIGITS_EN2AN: buf.append(", shape digits to AN"); break; + case DIGITS_AN2EN: buf.append(", shape digits to EN"); break; + case DIGITS_EN2AN_INIT_LR: buf.append(", shape digits to AN contextually: default EN"); break; + case DIGITS_EN2AN_INIT_AL: buf.append(", shape digits to AN contextually: default AL"); break; + } + switch (options & DIGIT_TYPE_MASK) { + case DIGIT_TYPE_AN: buf.append(", standard Arabic-Indic digits"); break; + case DIGIT_TYPE_AN_EXTENDED: buf.append(", extended Arabic-Indic digits"); break; + } + buf.append("]"); + + return buf.toString(); + } + ///CLOVER:ON + + // + // ported api + // + + private static final int IRRELEVANT = 4; + private static final int LAMTYPE = 16; + private static final int ALEFTYPE = 32; + + private static final int LINKR = 1; + private static final int LINKL = 2; + private static final int LINK_MASK = 3; + + private static final int irrelevantPos[] = { + 0x0, 0x2, 0x4, 0x6, 0x8, 0xA, 0xC, 0xE + }; + +/* + private static final char convertLamAlef[] = { + '\u0622', // FEF5 + '\u0622', // FEF6 + '\u0623', // FEF7 + '\u0623', // FEF8 + '\u0625', // FEF9 + '\u0625', // FEFA + '\u0627', // FEFB + '\u0627' // FEFC + }; +*/ + + private static final int tailFamilyIsolatedFinal[] = { + /* FEB1 */ 1, + /* FEB2 */ 1, + /* FEB3 */ 0, + /* FEB4 */ 0, + /* FEB5 */ 1, + /* FEB6 */ 1, + /* FEB7 */ 0, + /* FEB8 */ 0, + /* FEB9 */ 1, + /* FEBA */ 1, + /* FEBB */ 0, + /* FEBC */ 0, + /* FEBD */ 1, + /* FEBE */ 1 + }; + + private static final int tashkeelMedial[] = { + /* FE70 */ 0, + /* FE71 */ 1, + /* FE72 */ 0, + /* FE73 */ 0, + /* FE74 */ 0, + /* FE75 */ 0, + /* FE76 */ 0, + /* FE77 */ 1, + /* FE78 */ 0, + /* FE79 */ 1, + /* FE7A */ 0, + /* FE7B */ 1, + /* FE7C */ 0, + /* FE7D */ 1, + /* FE7E */ 0, + /* FE7F */ 1 + }; + + private static final char yehHamzaToYeh[] = + { + /* isolated*/ 0xFEEF, + /* final */ 0xFEF0 + }; + + private static final char convertNormalizedLamAlef[] = { + '\u0622', // 065C + '\u0623', // 065D + '\u0625', // 065E + '\u0627', // 065F + }; + + private static final int[] araLink = { + 1 + 32 + 256 * 0x11, /*0x0622*/ + 1 + 32 + 256 * 0x13, /*0x0623*/ + 1 + 256 * 0x15, /*0x0624*/ + 1 + 32 + 256 * 0x17, /*0x0625*/ + 1 + 2 + 256 * 0x19, /*0x0626*/ + 1 + 32 + 256 * 0x1D, /*0x0627*/ + 1 + 2 + 256 * 0x1F, /*0x0628*/ + 1 + 256 * 0x23, /*0x0629*/ + 1 + 2 + 256 * 0x25, /*0x062A*/ + 1 + 2 + 256 * 0x29, /*0x062B*/ + 1 + 2 + 256 * 0x2D, /*0x062C*/ + 1 + 2 + 256 * 0x31, /*0x062D*/ + 1 + 2 + 256 * 0x35, /*0x062E*/ + 1 + 256 * 0x39, /*0x062F*/ + 1 + 256 * 0x3B, /*0x0630*/ + 1 + 256 * 0x3D, /*0x0631*/ + 1 + 256 * 0x3F, /*0x0632*/ + 1 + 2 + 256 * 0x41, /*0x0633*/ + 1 + 2 + 256 * 0x45, /*0x0634*/ + 1 + 2 + 256 * 0x49, /*0x0635*/ + 1 + 2 + 256 * 0x4D, /*0x0636*/ + 1 + 2 + 256 * 0x51, /*0x0637*/ + 1 + 2 + 256 * 0x55, /*0x0638*/ + 1 + 2 + 256 * 0x59, /*0x0639*/ + 1 + 2 + 256 * 0x5D, /*0x063A*/ + 0, 0, 0, 0, 0, /*0x063B-0x063F*/ + 1 + 2, /*0x0640*/ + 1 + 2 + 256 * 0x61, /*0x0641*/ + 1 + 2 + 256 * 0x65, /*0x0642*/ + 1 + 2 + 256 * 0x69, /*0x0643*/ + 1 + 2 + 16 + 256 * 0x6D, /*0x0644*/ + 1 + 2 + 256 * 0x71, /*0x0645*/ + 1 + 2 + 256 * 0x75, /*0x0646*/ + 1 + 2 + 256 * 0x79, /*0x0647*/ + 1 + 256 * 0x7D, /*0x0648*/ + 1 + 256 * 0x7F, /*0x0649*/ + 1 + 2 + 256 * 0x81, /*0x064A*/ + 4, 4, 4, 4, /*0x064B-0x064E*/ + 4, 4, 4, 4, /*0x064F-0x0652*/ + 4, 4, 4, 0, 0, /*0x0653-0x0657*/ + 0, 0, 0, 0, /*0x0658-0x065B*/ + 1 + 256 * 0x85, /*0x065C*/ + 1 + 256 * 0x87, /*0x065D*/ + 1 + 256 * 0x89, /*0x065E*/ + 1 + 256 * 0x8B, /*0x065F*/ + 0, 0, 0, 0, 0, /*0x0660-0x0664*/ + 0, 0, 0, 0, 0, /*0x0665-0x0669*/ + 0, 0, 0, 0, 0, 0, /*0x066A-0x066F*/ + 4, /*0x0670*/ + 0, /*0x0671*/ + 1 + 32, /*0x0672*/ + 1 + 32, /*0x0673*/ + 0, /*0x0674*/ + 1 + 32, /*0x0675*/ + 1, 1, /*0x0676-0x0677*/ + 1+2, 1+2, 1+2, 1+2, 1+2, 1+2, /*0x0678-0x067D*/ + 1+2, 1+2, 1+2, 1+2, 1+2, 1+2, /*0x067E-0x0683*/ + 1+2, 1+2, 1+2, 1+2, /*0x0684-0x0687*/ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /*0x0688-0x0691*/ + 1, 1, 1, 1, 1, 1, 1, 1, /*0x0692-0x0699*/ + 1+2, 1+2, 1+2, 1+2, 1+2, 1+2, /*0x069A-0x06A3*/ + 1+2, 1+2, 1+2, 1+2, /*0x069A-0x06A3*/ + 1+2, 1+2, 1+2, 1+2, 1+2, 1+2, /*0x06A4-0x06AD*/ + 1+2, 1+2, 1+2, 1+2, /*0x06A4-0x06AD*/ + 1+2, 1+2, 1+2, 1+2, 1+2, 1+2, /*0x06AE-0x06B7*/ + 1+2, 1+2, 1+2, 1+2, /*0x06AE-0x06B7*/ + 1+2, 1+2, 1+2, 1+2, 1+2, 1+2, /*0x06B8-0x06BF*/ + 1+2, 1+2, /*0x06B8-0x06BF*/ + 1, /*0x06C0*/ + 1+2, /*0x06C1*/ + 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /*0x06C2-0x06CB*/ + 1+2, /*0x06CC*/ + 1, /*0x06CD*/ + 1+2, 1+2, 1+2, 1+2, /*0x06CE-0x06D1*/ + 1, 1 /*0x06D2-0x06D3*/ + }; + + private static final int[] presLink = { + 1 + 2, /*0xFE70*/ + 1 + 2, /*0xFE71*/ + 1 + 2, 0, 1+ 2, 0, 1+ 2, /*0xFE72-0xFE76*/ + 1 + 2, /*0xFE77*/ + 1+ 2, 1 + 2, 1+2, 1 + 2, /*0xFE78-0xFE81*/ + 1+ 2, 1 + 2, 1+2, 1 + 2, /*0xFE82-0xFE85*/ + 0, 0 + 32, 1 + 32, 0 + 32, /*0xFE86-0xFE89*/ + 1 + 32, 0, 1, 0 + 32, /*0xFE8A-0xFE8D*/ + 1 + 32, 0, 2, 1 + 2, /*0xFE8E-0xFE91*/ + 1, 0 + 32, 1 + 32, 0, /*0xFE92-0xFE95*/ + 2, 1 + 2, 1, 0, /*0xFE96-0xFE99*/ + 1, 0, 2, 1 + 2, /*0xFE9A-0xFE9D*/ + 1, 0, 2, 1 + 2, /*0xFE9E-0xFEA1*/ + 1, 0, 2, 1 + 2, /*0xFEA2-0xFEA5*/ + 1, 0, 2, 1 + 2, /*0xFEA6-0xFEA9*/ + 1, 0, 2, 1 + 2, /*0xFEAA-0xFEAD*/ + 1, 0, 1, 0, /*0xFEAE-0xFEB1*/ + 1, 0, 1, 0, /*0xFEB2-0xFEB5*/ + 1, 0, 2, 1+2, /*0xFEB6-0xFEB9*/ + 1, 0, 2, 1+2, /*0xFEBA-0xFEBD*/ + 1, 0, 2, 1+2, /*0xFEBE-0xFEC1*/ + 1, 0, 2, 1+2, /*0xFEC2-0xFEC5*/ + 1, 0, 2, 1+2, /*0xFEC6-0xFEC9*/ + 1, 0, 2, 1+2, /*0xFECA-0xFECD*/ + 1, 0, 2, 1+2, /*0xFECE-0xFED1*/ + 1, 0, 2, 1+2, /*0xFED2-0xFED5*/ + 1, 0, 2, 1+2, /*0xFED6-0xFED9*/ + 1, 0, 2, 1+2, /*0xFEDA-0xFEDD*/ + 1, 0, 2, 1+2, /*0xFEDE-0xFEE1*/ + 1, 0 + 16, 2 + 16, 1 + 2 +16, /*0xFEE2-0xFEE5*/ + 1 + 16, 0, 2, 1+2, /*0xFEE6-0xFEE9*/ + 1, 0, 2, 1+2, /*0xFEEA-0xFEED*/ + 1, 0, 2, 1+2, /*0xFEEE-0xFEF1*/ + 1, 0, 1, 0, /*0xFEF2-0xFEF5*/ + 1, 0, 2, 1+2, /*0xFEF6-0xFEF9*/ + 1, 0, 1, 0, /*0xFEFA-0xFEFD*/ + 1, 0, 1, 0, + 1 + }; + + private static int[] convertFEto06 = { + /***********0******1******2******3******4******5******6******7******8******9******A******B******C******D******E******F***/ + /*FE7*/ 0x64B, 0x64B, 0x64C, 0x64C, 0x64D, 0x64D, 0x64E, 0x64E, 0x64F, 0x64F, 0x650, 0x650, 0x651, 0x651, 0x652, 0x652, + /*FE8*/ 0x621, 0x622, 0x622, 0x623, 0x623, 0x624, 0x624, 0x625, 0x625, 0x626, 0x626, 0x626, 0x626, 0x627, 0x627, 0x628, + /*FE9*/ 0x628, 0x628, 0x628, 0x629, 0x629, 0x62A, 0x62A, 0x62A, 0x62A, 0x62B, 0x62B, 0x62B, 0x62B, 0x62C, 0x62C, 0x62C, + /*FEA*/ 0x62C, 0x62D, 0x62D, 0x62D, 0x62D, 0x62E, 0x62E, 0x62E, 0x62E, 0x62F, 0x62F, 0x630, 0x630, 0x631, 0x631, 0x632, + /*FEB*/ 0x632, 0x633, 0x633, 0x633, 0x633, 0x634, 0x634, 0x634, 0x634, 0x635, 0x635, 0x635, 0x635, 0x636, 0x636, 0x636, + /*FEC*/ 0x636, 0x637, 0x637, 0x637, 0x637, 0x638, 0x638, 0x638, 0x638, 0x639, 0x639, 0x639, 0x639, 0x63A, 0x63A, 0x63A, + /*FED*/ 0x63A, 0x641, 0x641, 0x641, 0x641, 0x642, 0x642, 0x642, 0x642, 0x643, 0x643, 0x643, 0x643, 0x644, 0x644, 0x644, + /*FEE*/ 0x644, 0x645, 0x645, 0x645, 0x645, 0x646, 0x646, 0x646, 0x646, 0x647, 0x647, 0x647, 0x647, 0x648, 0x648, 0x649, + /*FEF*/ 0x649, 0x64A, 0x64A, 0x64A, 0x64A, 0x65C, 0x65C, 0x65D, 0x65D, 0x65E, 0x65E, 0x65F, 0x65F + }; + + private static final int shapeTable[][][] = { + { {0,0,0,0}, {0,0,0,0}, {0,1,0,3}, {0,1,0,1} }, + { {0,0,2,2}, {0,0,1,2}, {0,1,1,2}, {0,1,1,3} }, + { {0,0,0,0}, {0,0,0,0}, {0,1,0,3}, {0,1,0,3} }, + { {0,0,1,2}, {0,0,1,2}, {0,1,1,2}, {0,1,1,3} } + }; + + /* + * This function shapes European digits to Arabic-Indic digits + * in-place, writing over the input characters. Data is in visual + * order. + */ + private void shapeToArabicDigitsWithContext(char[] dest, + int start, + int length, + char digitBase, + boolean lastStrongWasAL) { + digitBase -= '0'; // move common adjustment out of loop + + for(int i = start + length; --i >= start;) { + char ch = dest[i]; + switch (Character.getDirectionality(ch)) { + case Character.DIRECTIONALITY_LEFT_TO_RIGHT: + case Character.DIRECTIONALITY_RIGHT_TO_LEFT: + lastStrongWasAL = false; + break; + case Character.DIRECTIONALITY_RIGHT_TO_LEFT_ARABIC: + lastStrongWasAL = true; + break; + case Character.DIRECTIONALITY_EUROPEAN_NUMBER: + if (lastStrongWasAL && ch <= '\u0039') { + dest[i] = (char)(ch + digitBase); + } + break; + default: + break; + } + } + } + + /* + * Name : invertBuffer + * Function: This function inverts the buffer, it's used + * in case the user specifies the buffer to be + * TEXT_DIRECTION_LOGICAL + */ + private static void invertBuffer(char[] buffer, + int start, + int length) { + + for(int i = start, j = start + length - 1; i < j; i++, --j) { + char temp = buffer[i]; + buffer[i] = buffer[j]; + buffer[j] = temp; + } + } + + /* + * Name : changeLamAlef + * Function: Converts the Alef characters into an equivalent + * LamAlef location in the 0x06xx Range, this is an + * intermediate stage in the operation of the program + * later it'll be converted into the 0xFExx LamAlefs + * in the shaping function. + */ + private static char changeLamAlef(char ch) { + switch(ch) { + case '\u0622': return '\u065C'; + case '\u0623': return '\u065D'; + case '\u0625': return '\u065E'; + case '\u0627': return '\u065F'; + default: return '\u0000'; // not a lamalef + } + } + + /* + * Name : specialChar + * Function: Special Arabic characters need special handling in the shapeUnicode + * function, this function returns 1 or 2 for these special characters + */ + private static int specialChar(char ch) { + if ((ch > '\u0621' && ch < '\u0626') || + (ch == '\u0627') || + (ch > '\u062E' && ch < '\u0633') || + (ch > '\u0647' && ch < '\u064A') || + (ch == '\u0629')) { + return 1; + } else if (ch >= '\u064B' && ch<= '\u0652') { + return 2; + } else if (ch >= 0x0653 && ch <= 0x0655 || + ch == 0x0670 || + ch >= 0xFE70 && ch <= 0xFE7F) { + return 3; + } else { + return 0; + } + } + + /* + * Name : getLink + * Function: Resolves the link between the characters as + * Arabic characters have four forms : + * Isolated, Initial, Middle and Final Form + */ + private static int getLink(char ch) { + if (ch >= '\u0622' && ch <= '\u06D3') { + return araLink[ch - '\u0622']; + } else if (ch == '\u200D') { + return 3; + } else if (ch >= '\u206D' && ch <= '\u206F') { + return 4; + } else if (ch >= '\uFE70' && ch <= '\uFEFC') { + return presLink[ch - '\uFE70']; + } else { + return 0; + } + } + + /* + * Name : countSpaces + * Function: Counts the number of spaces + * at each end of the logical buffer + */ + private static int countSpacesLeft(char[] dest, + int start, + int count) { + for (int i = start, e = start + count; i < e; ++i) { + if (dest[i] != SPACE_CHAR) { + return i - start; + } + } + return count; + } + + private static int countSpacesRight(char[] dest, + int start, + int count) { + + for (int i = start + count; --i >= start;) { + if (dest[i] != SPACE_CHAR) { + return start + count - 1 - i; + } + } + return count; + } + + /* + * Name : isTashkeelChar + * Function: Returns true for Tashkeel characters else return false + */ + private static boolean isTashkeelChar(char ch) { + return ( ch >='\u064B' && ch <= '\u0652' ); + } + + /* + *Name : isSeenTailFamilyChar + *Function : returns 1 if the character is a seen family isolated character + * in the FE range otherwise returns 0 + */ + + private static int isSeenTailFamilyChar(char ch) { + if (ch >= 0xfeb1 && ch < 0xfebf){ + return tailFamilyIsolatedFinal [ch - 0xFEB1]; + } else { + return 0; + } + } + + /* Name : isSeenFamilyChar + * Function : returns 1 if the character is a seen family character in the Unicode + * 06 range otherwise returns 0 + */ + + private static int isSeenFamilyChar(char ch){ + if (ch >= 0x633 && ch <= 0x636){ + return 1; + }else { + return 0; + } + } + + /* + *Name : isTailChar + *Function : returns true if the character matches one of the tail characters + * (0xfe73 or 0x200b) otherwise returns false + */ + + private static boolean isTailChar(char ch) { + if(ch == OLD_TAIL_CHAR || ch == NEW_TAIL_CHAR){ + return true; + }else{ + return false; + } + } + + /* + *Name : isAlefMaksouraChar + *Function : returns true if the character is a Alef Maksoura Final or isolated + * otherwise returns false + */ + private static boolean isAlefMaksouraChar(char ch) { + return ( (ch == 0xFEEF) || ( ch == 0xFEF0) || (ch == 0x0649)); + } + + /* + * Name : isYehHamzaChar + * Function : returns true if the character is a yehHamza isolated or yehhamza + * final is found otherwise returns false + */ + private static boolean isYehHamzaChar(char ch) { + if((ch==0xFE89)||(ch==0xFE8A)){ + return true; + }else{ + return false; + } + } + + /* + *Name : isTashkeelCharFE + *Function : Returns true for Tashkeel characters in FE range else return false + */ + + private static boolean isTashkeelCharFE(char ch) { + return ( ch!=0xFE75 &&(ch>=0xFE70 && ch<= 0xFE7F) ); + } + + /* + * Name: isTashkeelOnTatweelChar + * Function: Checks if the Tashkeel Character is on Tatweel or not,if the + * Tashkeel on tatweel (FE range), it returns 1 else if the + * Tashkeel with shadda on tatweel (FC range)return 2 otherwise + * returns 0 + */ + private static int isTashkeelOnTatweelChar(char ch){ + if (ch >= 0xfe70 && ch <= 0xfe7f && ch != NEW_TAIL_CHAR && ch != 0xFE75 && ch != SHADDA_TATWEEL_CHAR) + { + return tashkeelMedial [ch - 0xFE70]; + } else if( (ch >= 0xfcf2 && ch <= 0xfcf4) || (ch == SHADDA_TATWEEL_CHAR)) { + return 2; + } else { + return 0; + } + } + + /* + * Name: isIsolatedTashkeelChar + * Function: Checks if the Tashkeel Character is in the isolated form + * (i.e. Unicode FE range) returns 1 else if the Tashkeel + * with shadda is in the isolated form (i.e. Unicode FC range) + * returns 1 otherwise returns 0 + */ + private static int isIsolatedTashkeelChar(char ch){ + if (ch >= 0xfe70 && ch <= 0xfe7f && ch != NEW_TAIL_CHAR && ch != 0xFE75){ + return (1 - tashkeelMedial [ch - 0xFE70]); + } else if(ch >= 0xfc5e && ch <= 0xfc63){ + return 1; + } else{ + return 0; + } + } + + /* + * Name : isAlefChar + * Function: Returns 1 for Alef characters else return 0 + */ + private static boolean isAlefChar(char ch) { + return ch == '\u0622' || ch == '\u0623' || ch == '\u0625' || ch == '\u0627'; + } + + /* + * Name : isLamAlefChar + * Function: Returns true for LamAlef characters else return false + */ + private static boolean isLamAlefChar(char ch) { + return ch >= '\uFEF5' && ch <= '\uFEFC'; + } + + private static boolean isNormalizedLamAlefChar(char ch) { + return ch >= '\u065C' && ch <= '\u065F'; + } + + /* + * Name : calculateSize + * Function: This function calculates the destSize to be used in preflighting + * when the destSize is equal to 0 + */ + private int calculateSize(char[] source, + int sourceStart, + int sourceLength) { + + int destSize = sourceLength; + + switch (options & LETTERS_MASK) { + case LETTERS_SHAPE: + case LETTERS_SHAPE_TASHKEEL_ISOLATED: + if (isLogical) { + for (int i = sourceStart, e = sourceStart + sourceLength - 1; i < e; ++i) { + if ((source[i] == LAM_CHAR && isAlefChar(source[i+1])) || isTashkeelCharFE(source[i])){ + --destSize; + } + } + } else { // visual + for(int i = sourceStart + 1, e = sourceStart + sourceLength; i < e; ++i) { + if ((source[i] == LAM_CHAR && isAlefChar(source[i-1])) || isTashkeelCharFE(source[i])) { + --destSize; + } + } + } + break; + + case LETTERS_UNSHAPE: + for(int i = sourceStart, e = sourceStart + sourceLength; i < e; ++i) { + if (isLamAlefChar(source[i])) { + destSize++; + } + } + break; + + default: + break; + } + + return destSize; + } + + + /* + * Name : countSpaceSub + * Function: Counts number of times the subChar appears in the array + */ + public static int countSpaceSub(char [] dest,int length, char subChar){ + int i = 0; + int count = 0; + while (i < length) { + if (dest[i] == subChar) { + count++; + } + i++; + } + return count; + } + + /* + * Name : shiftArray + * Function: Shifts characters to replace space sub characters + */ + public static void shiftArray(char [] dest,int start, int e, char subChar){ + int w = e; + int r = e; + while (--r >= start) { + char ch = dest[r]; + if (ch != subChar) { + --w; + if (w != r) { + dest[w] = ch; + } + } + } + } + + /* + * Name : flipArray + * Function: inverts array, so that start becomes end and vice versa + */ + public static int flipArray(char [] dest, int start, int e, int w){ + int r; + if (w > start) { + // shift, assume small buffer size so don't use arraycopy + r = w; + w = start; + while (r < e) { + dest[w++] = dest[r++]; + } + } else { + w = e; + } + return w; + } + + /* + * Name : handleTashkeelWithTatweel + * Function : Replaces Tashkeel as following: + * Case 1 :if the Tashkeel on tatweel, replace it with Tatweel. + * Case 2 :if the Tashkeel aggregated with Shadda on Tatweel, replace + * it with Shadda on Tatweel. + * Case 3: if the Tashkeel is isolated replace it with Space. + * + */ + private static int handleTashkeelWithTatweel(char[] dest, int sourceLength) { + int i; + for(i = 0; i < sourceLength; i++){ + if((isTashkeelOnTatweelChar(dest[i]) == 1)){ + dest[i] = TATWEEL_CHAR; + }else if((isTashkeelOnTatweelChar(dest[i]) == 2)){ + dest[i] = SHADDA_TATWEEL_CHAR; + }else if((isIsolatedTashkeelChar(dest[i])==1) && dest[i] != SHADDA_CHAR){ + dest[i] = SPACE_CHAR; + } + } + return sourceLength; + } + + /* + *Name : handleGeneratedSpaces + *Function : The shapeUnicode function converts Lam + Alef into LamAlef + space, + * and Tashkeel to space. + * handleGeneratedSpaces function puts these generated spaces + * according to the options the user specifies. LamAlef and Tashkeel + * spaces can be replaced at begin, at end, at near or decrease the + * buffer size. + * + * There is also Auto option for LamAlef and tashkeel, which will put + * the spaces at end of the buffer (or end of text if the user used + * the option SPACES_RELATIVE_TO_TEXT_BEGIN_END). + * + * If the text type was visual_LTR and the option + * SPACES_RELATIVE_TO_TEXT_BEGIN_END was selected the END + * option will place the space at the beginning of the buffer and + * BEGIN will place the space at the end of the buffer. + */ + private int handleGeneratedSpaces(char[] dest, + int start, + int length) { + + int lenOptionsLamAlef = options & LAMALEF_MASK; + int lenOptionsTashkeel = options & TASHKEEL_MASK; + boolean lamAlefOn = false; + boolean tashkeelOn = false; + + if (!isLogical & !spacesRelativeToTextBeginEnd) { + switch (lenOptionsLamAlef) { + case LAMALEF_BEGIN: lenOptionsLamAlef = LAMALEF_END; break; + case LAMALEF_END: lenOptionsLamAlef = LAMALEF_BEGIN; break; + default: break; + } + switch (lenOptionsTashkeel){ + case TASHKEEL_BEGIN: lenOptionsTashkeel = TASHKEEL_END; break; + case TASHKEEL_END: lenOptionsTashkeel = TASHKEEL_BEGIN; break; + default: break; + } + } + + + if (lenOptionsLamAlef == LAMALEF_NEAR) { + for (int i = start, e = i + length; i < e; ++i) { + if (dest[i] == LAMALEF_SPACE_SUB) { + dest[i] = SPACE_CHAR_FOR_LAMALEF; + } + } + + } else { + + final int e = start + length; + int wL = countSpaceSub(dest, length, LAMALEF_SPACE_SUB); + int wT = countSpaceSub(dest, length, TASHKEEL_SPACE_SUB); + + if (lenOptionsLamAlef == LAMALEF_END){ + lamAlefOn = true; + } + if (lenOptionsTashkeel == TASHKEEL_END){ + tashkeelOn = true; + } + + + if (lamAlefOn && (lenOptionsLamAlef == LAMALEF_END)) { + shiftArray(dest, start, e, LAMALEF_SPACE_SUB); + while (wL > start) { + dest[--wL] = SPACE_CHAR; + } + } + + if (tashkeelOn && (lenOptionsTashkeel == TASHKEEL_END)){ + shiftArray(dest, start, e, TASHKEEL_SPACE_SUB); + while (wT > start) { + dest[--wT] = SPACE_CHAR; + } + } + + lamAlefOn = false; + tashkeelOn = false; + + if (lenOptionsLamAlef == LAMALEF_RESIZE){ + lamAlefOn = true; + } + if (lenOptionsTashkeel == TASHKEEL_RESIZE){ + tashkeelOn = true; + } + + if (lamAlefOn && (lenOptionsLamAlef == LAMALEF_RESIZE)){ + shiftArray(dest, start, e, LAMALEF_SPACE_SUB); + wL = flipArray(dest,start,e, wL); + length = wL - start; + } + if (tashkeelOn && (lenOptionsTashkeel == TASHKEEL_RESIZE)) { + shiftArray(dest, start, e, TASHKEEL_SPACE_SUB); + wT = flipArray(dest,start,e, wT); + length = wT - start; + } + + lamAlefOn = false; + tashkeelOn = false; + + if ((lenOptionsLamAlef == LAMALEF_BEGIN) || + (lenOptionsLamAlef == LAMALEF_AUTO)){ + lamAlefOn = true; + } + if (lenOptionsTashkeel == TASHKEEL_BEGIN){ + tashkeelOn = true; + } + + if (lamAlefOn && ((lenOptionsLamAlef == LAMALEF_BEGIN)|| + (lenOptionsLamAlef == LAMALEF_AUTO))) { // spaces at beginning + shiftArray(dest, start, e, LAMALEF_SPACE_SUB); + wL = flipArray(dest,start,e, wL); + while (wL < e) { + dest[wL++] = SPACE_CHAR; + } + } + if(tashkeelOn && (lenOptionsTashkeel == TASHKEEL_BEGIN)){ + shiftArray(dest, start, e, TASHKEEL_SPACE_SUB); + wT = flipArray(dest,start,e, wT); + while (wT < e) { + dest[wT++] = SPACE_CHAR; + } + } + } + + return length; + } + + + /* + *Name :expandCompositCharAtBegin + *Function :Expands the LamAlef character to Lam and Alef consuming the required + * space from beginning of the buffer. If the text type was visual_LTR + * and the option SPACES_RELATIVE_TO_TEXT_BEGIN_END was selected + * the spaces will be located at end of buffer. + * If there are no spaces to expand the LamAlef, an exception is thrown. +*/ + private boolean expandCompositCharAtBegin(char[] dest,int start, int length, + int lacount) { + boolean spaceNotFound = false; + + if (lacount > countSpacesRight(dest, start, length)) { + spaceNotFound = true; + return spaceNotFound; + } + for (int r = start + length - lacount, w = start + length; --r >= start;) { + char ch = dest[r]; + if (isNormalizedLamAlefChar(ch)) { + dest[--w] = LAM_CHAR; + dest[--w] = convertNormalizedLamAlef[ch - '\u065C']; + } else { + dest[--w] = ch; + } + } + return spaceNotFound; + + } + + /* + *Name : expandCompositCharAtEnd + *Function : Expands the LamAlef character to Lam and Alef consuming the + * required space from end of the buffer. If the text type was + * Visual LTR and the option SPACES_RELATIVE_TO_TEXT_BEGIN_END + * was used, the spaces will be consumed from begin of buffer. If + * there are no spaces to expand the LamAlef, an exception is thrown. + */ + + private boolean expandCompositCharAtEnd(char[] dest,int start, int length, + int lacount){ + boolean spaceNotFound = false; + + if (lacount > countSpacesLeft(dest, start, length)) { + spaceNotFound = true; + return spaceNotFound; + } + for (int r = start + lacount, w = start, e = start + length; r < e; ++r) { + char ch = dest[r]; + if (isNormalizedLamAlefChar(ch)) { + dest[w++] = convertNormalizedLamAlef[ch - '\u065C']; + dest[w++] = LAM_CHAR; + } else { + dest[w++] = ch; + } + } + return spaceNotFound; + } + + /* + *Name : expandCompositCharAtNear + *Function : Expands the LamAlef character into Lam + Alef, YehHamza character + * into Yeh + Hamza, SeenFamily character into SeenFamily character + * + Tail, while consuming the space next to the character. + */ + + private boolean expandCompositCharAtNear(char[] dest,int start, int length, + int yehHamzaOption, int seenTailOption, int lamAlefOption){ + + boolean spaceNotFound = false; + + + + if (isNormalizedLamAlefChar(dest[start])) { + spaceNotFound = true; + return spaceNotFound; + } + for (int i = start + length; --i >=start;) { + char ch = dest[i]; + if (lamAlefOption == 1 && isNormalizedLamAlefChar(ch)) { + if (i>start &&dest[i-1] == SPACE_CHAR) { + dest[i] = LAM_CHAR; + dest[--i] = convertNormalizedLamAlef[ch - '\u065C']; + } else { + spaceNotFound = true; + return spaceNotFound; + } + }else if(seenTailOption == 1 && isSeenTailFamilyChar(ch) == 1){ + if(i>start &&dest[i-1] == SPACE_CHAR){ + dest[i-1] = tailChar; + } else{ + spaceNotFound = true; + return spaceNotFound; + } + }else if(yehHamzaOption == 1 && isYehHamzaChar(ch)){ + + if(i>start &&dest[i-1] == SPACE_CHAR){ + dest[i] = yehHamzaToYeh[ch - YEH_HAMZAFE_CHAR]; + dest[i-1] = HAMZAFE_CHAR; + }else{ + spaceNotFound = true; + return spaceNotFound; + } + + + } + } + return false; + + } + + /* + * Name : expandCompositChar + * Function: LamAlef needs special handling as the LamAlef is + * one character while expanding it will give two + * characters Lam + Alef, so we need to expand the LamAlef + * in near or far spaces according to the options the user + * specifies or increase the buffer size. + * Dest has enough room for the expansion if we are growing. + * lamalef are normalized to the 'special characters' + */ + private int expandCompositChar(char[] dest, + int start, + int length, + int lacount, + int shapingMode) throws ArabicShapingException { + + int lenOptionsLamAlef = options & LAMALEF_MASK; + int lenOptionsSeen = options & SEEN_MASK; + int lenOptionsYehHamza = options & YEHHAMZA_MASK; + boolean spaceNotFound = false; + + if (!isLogical && !spacesRelativeToTextBeginEnd) { + switch (lenOptionsLamAlef) { + case LAMALEF_BEGIN: lenOptionsLamAlef = LAMALEF_END; break; + case LAMALEF_END: lenOptionsLamAlef = LAMALEF_BEGIN; break; + default: break; + } + } + + if(shapingMode == 1){ + if(lenOptionsLamAlef == LAMALEF_AUTO){ + if(isLogical){ + spaceNotFound = expandCompositCharAtEnd(dest, start, length, lacount); + if(spaceNotFound){ + spaceNotFound = expandCompositCharAtBegin(dest, start, length, lacount); + } + if(spaceNotFound){ + spaceNotFound = expandCompositCharAtNear(dest, start, length,0,0,1); + } + if(spaceNotFound){ + throw new ArabicShapingException("No spacefor lamalef"); + } + }else{ + spaceNotFound = expandCompositCharAtBegin(dest, start, length, lacount); + if(spaceNotFound){ + spaceNotFound = expandCompositCharAtEnd(dest, start, length, lacount); + } + if(spaceNotFound){ + spaceNotFound = expandCompositCharAtNear(dest, start, length,0,0,1); + } + if(spaceNotFound){ + throw new ArabicShapingException("No spacefor lamalef"); + } + } + }else if(lenOptionsLamAlef == LAMALEF_END){ + spaceNotFound = expandCompositCharAtEnd(dest, start, length, lacount); + if(spaceNotFound){ + throw new ArabicShapingException("No spacefor lamalef"); + } + }else if(lenOptionsLamAlef == LAMALEF_BEGIN){ + spaceNotFound = expandCompositCharAtBegin(dest, start, length, lacount); + if(spaceNotFound){ + throw new ArabicShapingException("No spacefor lamalef"); + } + }else if(lenOptionsLamAlef == LAMALEF_NEAR){ + spaceNotFound = expandCompositCharAtNear(dest, start, length,0,0,1); + if(spaceNotFound){ + throw new ArabicShapingException("No spacefor lamalef"); + } + }else if(lenOptionsLamAlef == LAMALEF_RESIZE){ + for (int r = start + length, w = r + lacount; --r >= start;) { + char ch = dest[r]; + if (isNormalizedLamAlefChar(ch)) { + dest[--w] = '\u0644'; + dest[--w] = convertNormalizedLamAlef[ch - '\u065C']; + } else { + dest[--w] = ch; + } + } + length += lacount; + } + }else{ + if(lenOptionsSeen == SEEN_TWOCELL_NEAR){ + spaceNotFound = expandCompositCharAtNear(dest, start, length,0,1,0); + if(spaceNotFound){ + throw new ArabicShapingException("No space for Seen tail expansion"); + } + } + if(lenOptionsYehHamza == YEHHAMZA_TWOCELL_NEAR){ + spaceNotFound = expandCompositCharAtNear(dest, start, length,1,0,0); + if(spaceNotFound){ + throw new ArabicShapingException("No space for YehHamza expansion"); + } + } + } + return length; + } + + + /* Convert the input buffer from FExx Range into 06xx Range + * to put all characters into the 06xx range + * even the lamalef is converted to the special region in + * the 06xx range. Return the number of lamalef chars found. + */ + private int normalize(char[] dest, int start, int length) { + int lacount = 0; + for (int i = start, e = i + length; i < e; ++i) { + char ch = dest[i]; + if (ch >= '\uFE70' && ch <= '\uFEFC') { + if (isLamAlefChar(ch)) { + ++lacount; + } + dest[i] = (char)convertFEto06[ch - '\uFE70']; + } + } + return lacount; + } + + /* + * Name : deshapeNormalize + * Function: Convert the input buffer from FExx Range into 06xx Range + * even the lamalef is converted to the special region in the 06xx range. + * According to the options the user enters, all seen family characters + * followed by a tail character are merged to seen tail family character and + * any yeh followed by a hamza character are merged to yehhamza character. + * Method returns the number of lamalef chars found. + */ + private int deshapeNormalize(char[] dest, int start, int length) { + int lacount = 0; + int yehHamzaComposeEnabled = 0; + int seenComposeEnabled = 0; + + yehHamzaComposeEnabled = ((options&YEHHAMZA_MASK) == YEHHAMZA_TWOCELL_NEAR) ? 1 : 0; + seenComposeEnabled = ((options&SEEN_MASK) == SEEN_TWOCELL_NEAR)? 1 : 0; + + for (int i = start, e = i + length; i < e; ++i) { + char ch = dest[i]; + + if( (yehHamzaComposeEnabled == 1) && ((ch == HAMZA06_CHAR) || (ch == HAMZAFE_CHAR)) + && (i < (length - 1)) && isAlefMaksouraChar(dest[i+1] )) { + dest[i] = SPACE_CHAR; + dest[i+1] = YEH_HAMZA_CHAR; + } else if ( (seenComposeEnabled == 1) && (isTailChar(ch)) && (i< (length - 1)) + && (isSeenTailFamilyChar(dest[i+1])==1) ) { + dest[i] = SPACE_CHAR; + } + else if (ch >= '\uFE70' && ch <= '\uFEFC') { + if (isLamAlefChar(ch)) { + ++lacount; + } + dest[i] = (char)convertFEto06[ch - '\uFE70']; + } + } + return lacount; + } + + /* + * Name : shapeUnicode + * Function: Converts an Arabic Unicode buffer in 06xx Range into a shaped + * arabic Unicode buffer in FExx Range + */ + private int shapeUnicode(char[] dest, + int start, + int length, + int destSize, + int tashkeelFlag)throws ArabicShapingException { + + int lamalef_count = normalize(dest, start, length); + + // resolve the link between the characters. + // Arabic characters have four forms: Isolated, Initial, Medial and Final. + // Tashkeel characters have two, isolated or medial, and sometimes only isolated. + // tashkeelFlag == 0: shape normally, 1: shape isolated, 2: don't shape + + boolean lamalef_found = false, seenfam_found = false; + boolean yehhamza_found = false, tashkeel_found = false; + int i = start + length - 1; + int currLink = getLink(dest[i]); + int nextLink = 0; + int prevLink = 0; + int lastLink = 0; + //int prevPos = i; + int lastPos = i; + int nx = -2; + int nw = 0; + + while (i >= 0) { + // If high byte of currLink > 0 then there might be more than one shape + if ((currLink & '\uFF00') > 0 || isTashkeelChar(dest[i])) { + nw = i - 1; + nx = -2; + while (nx < 0) { // we need to know about next char + if (nw == -1) { + nextLink = 0; + nx = Integer.MAX_VALUE; + } else { + nextLink = getLink(dest[nw]); + if ((nextLink & IRRELEVANT) == 0) { + nx = nw; + } else { + --nw; + } + } + } + + if (((currLink & ALEFTYPE) > 0) && ((lastLink & LAMTYPE) > 0)) { + lamalef_found = true; + char wLamalef = changeLamAlef(dest[i]); // get from 0x065C-0x065f + if (wLamalef != '\u0000') { + // replace alef by marker, it will be removed later + dest[i] = '\uffff'; + dest[lastPos] = wLamalef; + i = lastPos; + } + + lastLink = prevLink; + currLink = getLink(wLamalef); // requires '\u0000', unfortunately + } + if ((i > 0) && (dest[i-1] == SPACE_CHAR)) + { + if ( isSeenFamilyChar(dest[i]) == 1){ + seenfam_found = true; + } else if (dest[i] == YEH_HAMZA_CHAR) { + yehhamza_found = true; + } + } + else if(i==0){ + if ( isSeenFamilyChar(dest[i]) == 1){ + seenfam_found = true; + } else if (dest[i] == YEH_HAMZA_CHAR) { + yehhamza_found = true; + } + } + + + // get the proper shape according to link ability of neighbors + // and of character; depends on the order of the shapes + // (isolated, initial, middle, final) in the compatibility area + + int flag = specialChar(dest[i]); + + int shape = shapeTable[nextLink & LINK_MASK] + [lastLink & LINK_MASK] + [currLink & LINK_MASK]; + + if (flag == 1) { + shape &= 0x1; + } else if (flag == 2) { + if (tashkeelFlag == 0 && + ((lastLink & LINKL) != 0) && + ((nextLink & LINKR) != 0) && + dest[i] != '\u064C' && + dest[i] != '\u064D' && + !((nextLink & ALEFTYPE) == ALEFTYPE && + (lastLink & LAMTYPE) == LAMTYPE)) { + + shape = 1; + } else { + shape = 0; + } + } + if (flag == 2) { + if (tashkeelFlag == 2) { + dest[i] = TASHKEEL_SPACE_SUB; + tashkeel_found = true; + } + else{ + dest[i] = (char)('\uFE70' + irrelevantPos[dest[i] - '\u064B'] + shape); + } + // else leave tashkeel alone + } else { + dest[i] = (char)('\uFE70' + (currLink >> 8) + shape); + } + } + + // move one notch forward + if ((currLink & IRRELEVANT) == 0) { + prevLink = lastLink; + lastLink = currLink; + //prevPos = lastPos; + lastPos = i; + } + + --i; + if (i == nx) { + currLink = nextLink; + nx = -2; + } else if (i != -1) { + currLink = getLink(dest[i]); + } + } + + // If we found a lam/alef pair in the buffer + // call handleGeneratedSpaces to remove the spaces that were added + + destSize = length; + if (lamalef_found || tashkeel_found) { + destSize = handleGeneratedSpaces(dest, start, length); + } + if (seenfam_found || yehhamza_found){ + destSize = expandCompositChar(dest, start, destSize, lamalef_count, SHAPE_MODE); + } + return destSize; + } + + /* + * Name : deShapeUnicode + * Function: Converts an Arabic Unicode buffer in FExx Range into unshaped + * arabic Unicode buffer in 06xx Range + */ + private int deShapeUnicode(char[] dest, + int start, + int length, + int destSize) throws ArabicShapingException { + + int lamalef_count = deshapeNormalize(dest, start, length); + + // If there was a lamalef in the buffer call expandLamAlef + if (lamalef_count != 0) { + // need to adjust dest to fit expanded buffer... !!! + destSize = expandCompositChar(dest, start, length, lamalef_count,DESHAPE_MODE); + } else { + destSize = length; + } + + return destSize; + } + + private int internalShape(char[] source, + int sourceStart, + int sourceLength, + char[] dest, + int destStart, + int destSize) throws ArabicShapingException { + + if (sourceLength == 0) { + return 0; + } + + if (destSize == 0) { + if (((options & LETTERS_MASK) != LETTERS_NOOP) && + ((options & LAMALEF_MASK) == LAMALEF_RESIZE)) { + + return calculateSize(source, sourceStart, sourceLength); + } else { + return sourceLength; // by definition + } + } + + // always use temp buffer + char[] temp = new char[sourceLength * 2]; // all lamalefs requiring expansion + System.arraycopy(source, sourceStart, temp, 0, sourceLength); + + if (isLogical) { + invertBuffer(temp, 0, sourceLength); + } + + int outputSize = sourceLength; + + switch (options & LETTERS_MASK) { + case LETTERS_SHAPE_TASHKEEL_ISOLATED: + outputSize = shapeUnicode(temp, 0, sourceLength, destSize, 1); + break; + + case LETTERS_SHAPE: + if( ((options&TASHKEEL_MASK)> 0) && + ((options&TASHKEEL_MASK) !=TASHKEEL_REPLACE_BY_TATWEEL)) { + /* Call the shaping function with tashkeel flag == 2 for removal of tashkeel */ + outputSize = shapeUnicode(temp, 0, sourceLength, destSize, 2); + }else { + //default Call the shaping function with tashkeel flag == 1 */ + outputSize = shapeUnicode(temp, 0, sourceLength, destSize, 0); + + /*After shaping text check if user wants to remove tashkeel and replace it with tatweel*/ + if( (options&TASHKEEL_MASK) == TASHKEEL_REPLACE_BY_TATWEEL){ + outputSize = handleTashkeelWithTatweel(temp,sourceLength); + } + } + break; + + case LETTERS_UNSHAPE: + outputSize = deShapeUnicode(temp, 0, sourceLength, destSize); + break; + + default: + break; + } + + if (outputSize > destSize) { + throw new ArabicShapingException("not enough room for result data"); + } + + if ((options & DIGITS_MASK) != DIGITS_NOOP) { + char digitBase = '\u0030'; // European digits + switch (options & DIGIT_TYPE_MASK) { + case DIGIT_TYPE_AN: + digitBase = '\u0660'; // Arabic-Indic digits + break; + + case DIGIT_TYPE_AN_EXTENDED: + digitBase = '\u06f0'; // Eastern Arabic-Indic digits (Persian and Urdu) + break; + + default: + break; + } + + switch (options & DIGITS_MASK) { + case DIGITS_EN2AN: + { + int digitDelta = digitBase - '\u0030'; + for (int i = 0; i < outputSize; ++i) { + char ch = temp[i]; + if (ch <= '\u0039' && ch >= '\u0030') { + temp[i] += digitDelta; + } + } + } + break; + + case DIGITS_AN2EN: + { + char digitTop = (char)(digitBase + 9); + int digitDelta = '\u0030' - digitBase; + for (int i = 0; i < outputSize; ++i) { + char ch = temp[i]; + if (ch <= digitTop && ch >= digitBase) { + temp[i] += digitDelta; + } + } + } + break; + + case DIGITS_EN2AN_INIT_LR: + shapeToArabicDigitsWithContext(temp, 0, outputSize, digitBase, false); + break; + + case DIGITS_EN2AN_INIT_AL: + shapeToArabicDigitsWithContext(temp, 0, outputSize, digitBase, true); + break; + + default: + break; + } + } + + if (isLogical) { + invertBuffer(temp, 0, outputSize); + } + + System.arraycopy(temp, 0, dest, destStart, outputSize); + + return outputSize; + } + + private static class ArabicShapingException extends RuntimeException { + ArabicShapingException(String msg) { + super(msg); + } + } +} diff --git a/icu4j/license.html b/icu4j/license.html new file mode 100644 index 0000000000000..b905ddf41e8ea --- /dev/null +++ b/icu4j/license.html @@ -0,0 +1,51 @@ + + + + +ICU License - ICU 1.8.1 and later + + + +

ICU License - ICU 1.8.1 and later

+ +

COPYRIGHT AND PERMISSION NOTICE

+ +

+Copyright (c) 1995-2006 International Business Machines Corporation and others +

+

+All rights reserved. +

+

+Permission is hereby granted, free of charge, to any person obtaining a copy +of this software and associated documentation files (the "Software"), +to deal in the Software without restriction, including without limitation +the rights to use, copy, modify, merge, publish, distribute, and/or sell +copies of the Software, and to permit persons +to whom the Software is furnished to do so, provided that the above +copyright notice(s) and this permission notice appear in all copies +of the Software and that both the above copyright notice(s) and this +permission notice appear in supporting documentation. +

+

+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, +INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A +PARTICULAR PURPOSE AND NONINFRINGEMENT OF THIRD PARTY RIGHTS. IN NO EVENT SHALL +THE COPYRIGHT HOLDER OR HOLDERS INCLUDED IN THIS NOTICE BE LIABLE FOR ANY CLAIM, +OR ANY SPECIAL INDIRECT OR CONSEQUENTIAL DAMAGES, OR ANY DAMAGES WHATSOEVER +RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, +NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE +USE OR PERFORMANCE OF THIS SOFTWARE. +

+

+Except as contained in this notice, the name of a copyright holder shall not be +used in advertising or otherwise to promote the sale, use or other dealings in +this Software without prior written authorization of the copyright holder. +

+ +
+

+All trademarks and registered trademarks mentioned herein are the property of their respective owners. +

+ +