//
// TextOutputDev.cc
//
-// Copyright 1997-2002 Glyph & Cog, LLC
+// Copyright 1997-2003 Glyph & Cog, LLC
//
//========================================================================
#include <stddef.h>
#include <math.h>
#include <ctype.h>
-#include "GString.h"
+#ifdef WIN32
+#include <fcntl.h> // for O_BINARY
+#include <io.h> // for setmode
+#endif
#include "gmem.h"
-#include "config.h"
+#include "GString.h"
+#include "GList.h"
+#include "xpdfconfig.h"
#include "Error.h"
#include "GlobalParams.h"
#include "UnicodeMap.h"
+#include "UnicodeTypeTable.h"
#include "GfxState.h"
#include "TextOutputDev.h"
#include "ICSupport.h"
#endif
+//------------------------------------------------------------------------
+// parameters
//------------------------------------------------------------------------
-#define textOutSpace 0.2
-#define textOutColSpace 0.2
+// Each bucket in a text pool includes baselines within a range of
+// this many points.
+#define textPoolStep 4
+
+// Inter-character space width which will cause addChar to start a new
+// word.
+#define minWordBreakSpace 0.1
+
+// Negative inter-character space width, i.e., overlap, which will
+// cause addChar to start a new word.
+#define minDupBreakOverlap 0.2
+
+// Max distance between baselines of two lines within a block, as a
+// fraction of the font size.
+#define maxLineSpacingDelta 1.5
+
+// Max difference in primary font sizes on two lines in the same
+// block. Delta1 is used when examining new lines above and below the
+// current block; delta2 is used when examining text that overlaps the
+// current block; delta3 is used when examining text to the left and
+// right of the current block.
+#define maxBlockFontSizeDelta1 0.05
+#define maxBlockFontSizeDelta2 0.6
+#define maxBlockFontSizeDelta3 0.2
+
+// Max difference in font sizes inside a word.
+#define maxWordFontSizeDelta 0.05
+
+// Maximum distance between baselines of two words on the same line,
+// e.g., distance between subscript or superscript and the primary
+// baseline, as a fraction of the font size.
+#define maxIntraLineDelta 0.5
+
+// Minimum inter-word spacing, as a fraction of the font size. (Only
+// used for raw ordering.)
+#define minWordSpacing 0.15
+
+// Maximum inter-word spacing, as a fraction of the font size.
+#define maxWordSpacing 1.5
+
+// Maximum horizontal spacing which will allow a word to be pulled
+// into a block.
+#define minColSpacing1 0.3
+
+// Minimum spacing between columns, as a fraction of the font size.
+#define minColSpacing2 1.0
+
+// Maximum vertical spacing between blocks within a flow, as a
+// multiple of the font size.
+#define maxBlockSpacing 2.5
+
+// Minimum spacing between characters within a word, as a fraction of
+// the font size.
+#define minCharSpacing -0.2
+
+// Maximum spacing between characters within a word, as a fraction of
+// the font size, when there is no obvious extra-wide character
+// spacing.
+#define maxCharSpacing 0.03
+
+// When extra-wide character spacing is detected, the inter-character
+// space threshold is set to the minimum inter-character space
+// multiplied by this constant.
+#define maxWideCharSpacingMul 1.3
+
+// Max difference in primary,secondary coordinates (as a fraction of
+// the font size) allowed for duplicated text (fake boldface, drop
+// shadows) which is to be discarded.
+#define dupMaxPriDelta 0.1
+#define dupMaxSecDelta 0.2
+//------------------------------------------------------------------------
+// TextFontInfo
//------------------------------------------------------------------------
-struct TextOutColumnEdge {
- double x, y0, y1;
-};
+TextFontInfo::TextFontInfo(GfxState *state) {
+ gfxFont = state->getFont();
+#if TEXTOUT_WORD_LIST
+ fontName = (gfxFont && gfxFont->getOrigName())
+ ? gfxFont->getOrigName()->copy()
+ : (GString *)NULL;
+#endif
+}
+
+TextFontInfo::~TextFontInfo() {
+#if TEXTOUT_WORD_LIST
+ if (fontName) {
+ delete fontName;
+ }
+#endif
+}
+
+GBool TextFontInfo::matches(GfxState *state) {
+ return state->getFont() == gfxFont;
+}
//------------------------------------------------------------------------
-// TextBlock
+// TextWord
//------------------------------------------------------------------------
-class TextBlock {
-public:
+TextWord::TextWord(GfxState *state, int rotA, double x0, double y0,
+ int charPosA, TextFontInfo *fontA, double fontSizeA) {
+ GfxFont *gfxFont;
+ double x, y, ascent, descent;
- TextBlock();
- ~TextBlock();
+ rot = rotA;
+ charPos = charPosA;
+ charLen = 0;
+ font = fontA;
+ fontSize = fontSizeA;
+ state->transform(x0, y0, &x, &y);
+ if ((gfxFont = font->gfxFont)) {
+ ascent = gfxFont->getAscent() * fontSize;
+ descent = gfxFont->getDescent() * fontSize;
+ } else {
+ // this means that the PDF file draws text without a current font,
+ // which should never happen
+ ascent = 0.95 * fontSize;
+ descent = -0.35 * fontSize;
+ }
+ switch (rot) {
+ case 0:
+ yMin = y - ascent;
+ yMax = y - descent;
+ if (yMin == yMax) {
+ // this is a sanity check for a case that shouldn't happen -- but
+ // if it does happen, we want to avoid dividing by zero later
+ yMin = y;
+ yMax = y + 1;
+ }
+ base = y;
+ break;
+ case 1:
+ xMin = x + descent;
+ xMax = x + ascent;
+ if (xMin == xMax) {
+ // this is a sanity check for a case that shouldn't happen -- but
+ // if it does happen, we want to avoid dividing by zero later
+ xMin = x;
+ xMax = x + 1;
+ }
+ base = x;
+ break;
+ case 2:
+ yMin = y + descent;
+ yMax = y + ascent;
+ if (yMin == yMax) {
+ // this is a sanity check for a case that shouldn't happen -- but
+ // if it does happen, we want to avoid dividing by zero later
+ yMin = y;
+ yMax = y + 1;
+ }
+ base = y;
+ break;
+ case 3:
+ xMin = x - ascent;
+ xMax = x - descent;
+ if (xMin == xMax) {
+ // this is a sanity check for a case that shouldn't happen -- but
+ // if it does happen, we want to avoid dividing by zero later
+ xMin = x;
+ xMax = x + 1;
+ }
+ base = x;
+ break;
+ }
+ text = NULL;
+ edge = NULL;
+ len = size = 0;
+ spaceAfter = gFalse;
+ next = NULL;
- double xMin, xMax;
- double yMin, yMax;
- TextString *strings; // list of strings in the block
- TextBlock *next; // next block in line
- TextBlock *xyNext; // next block on xyBlocks list
- Unicode *text; // Unicode text of the block, including
- // spaces between strings
- double *xRight; // right-hand x coord of each char
- int len; // total number of Unicode characters
- int convertedLen; // total number of converted characters
- int *col; // starting column number for each
- // Unicode character
-};
+#if TEXTOUT_WORD_LIST
+ GfxRGB rgb;
-TextBlock::TextBlock() {
- strings = NULL;
- next = NULL;
- xyNext = NULL;
+ if ((state->getRender() & 3) == 1) {
+ state->getStrokeRGB(&rgb);
+ } else {
+ state->getFillRGB(&rgb);
+ }
+ colorR = rgb.r;
+ colorG = rgb.g;
+ colorB = rgb.b;
+#endif
+}
+
+TextWord::~TextWord() {
+ gfree(text);
+ gfree(edge);
+}
+
+void TextWord::addChar(GfxState *state, double x, double y,
+ double dx, double dy, Unicode u) {
+ if (len == size) {
+ size += 16;
+ text = (Unicode *)grealloc(text, size * sizeof(Unicode));
+ edge = (double *)grealloc(edge, (size + 1) * sizeof(double));
+ }
+ text[len] = u;
+ switch (rot) {
+ case 0:
+ if (len == 0) {
+ xMin = x;
+ }
+ edge[len] = x;
+ xMax = edge[len+1] = x + dx;
+ break;
+ case 1:
+ if (len == 0) {
+ yMin = y;
+ }
+ edge[len] = y;
+ yMax = edge[len+1] = y + dy;
+ break;
+ case 2:
+ if (len == 0) {
+ xMax = x;
+ }
+ edge[len] = x;
+ xMin = edge[len+1] = x + dx;
+ break;
+ case 3:
+ if (len == 0) {
+ yMax = y;
+ }
+ edge[len] = y;
+ yMin = edge[len+1] = y + dy;
+ break;
+ }
+ ++len;
+}
+
+void TextWord::merge(TextWord *word) {
+ int i;
+
+ if (word->xMin < xMin) {
+ xMin = word->xMin;
+ }
+ if (word->yMin < yMin) {
+ yMin = word->yMin;
+ }
+ if (word->xMax > xMax) {
+ xMax = word->xMax;
+ }
+ if (word->yMax > yMax) {
+ yMax = word->yMax;
+ }
+ if (len + word->len > size) {
+ size = len + word->len;
+ text = (Unicode *)grealloc(text, size * sizeof(Unicode));
+ edge = (double *)grealloc(edge, (size + 1) * sizeof(double));
+ }
+ for (i = 0; i < word->len; ++i) {
+ text[len + i] = word->text[i];
+ edge[len + i] = word->edge[i];
+ }
+ edge[len + word->len] = word->edge[word->len];
+ len += word->len;
+ charLen += word->charLen;
+}
+
+inline int TextWord::primaryCmp(TextWord *word) {
+ double cmp;
+
+ cmp = 0; // make gcc happy
+ switch (rot) {
+ case 0:
+ cmp = xMin - word->xMin;
+ break;
+ case 1:
+ cmp = yMin - word->yMin;
+ break;
+ case 2:
+ cmp = word->xMax - xMax;
+ break;
+ case 3:
+ cmp = word->yMax - yMax;
+ break;
+ }
+ return cmp < 0 ? -1 : cmp > 0 ? 1 : 0;
+}
+
+double TextWord::primaryDelta(TextWord *word) {
+ double delta;
+
+ delta = 0; // make gcc happy
+ switch (rot) {
+ case 0:
+ delta = word->xMin - xMax;
+ break;
+ case 1:
+ delta = word->yMin - yMax;
+ break;
+ case 2:
+ delta = xMin - word->xMax;
+ break;
+ case 3:
+ delta = yMin - word->yMax;
+ break;
+ }
+ return delta;
+}
+
+int TextWord::cmpYX(const void *p1, const void *p2) {
+ TextWord *word1 = *(TextWord **)p1;
+ TextWord *word2 = *(TextWord **)p2;
+ double cmp;
+
+ cmp = word1->yMin - word2->yMin;
+ if (cmp == 0) {
+ cmp = word1->xMin - word2->xMin;
+ }
+ return cmp < 0 ? -1 : cmp > 0 ? 1 : 0;
+}
+
+#if TEXTOUT_WORD_LIST
+
+GString *TextWord::getText() {
+ GString *s;
+ UnicodeMap *uMap;
+ char buf[8];
+ int n, i;
+
+ s = new GString();
+ if (!(uMap = globalParams->getTextEncoding())) {
+ return s;
+ }
+ for (i = 0; i < len; ++i) {
+ n = uMap->mapUnicode(text[i], buf, sizeof(buf));
+ s->append(buf, n);
+ }
+ uMap->decRefCnt();
+ return s;
+}
+
+#endif // TEXTOUT_WORD_LIST
+
+//------------------------------------------------------------------------
+// TextPool
+//------------------------------------------------------------------------
+
+TextPool::TextPool() {
+ minBaseIdx = 0;
+ maxBaseIdx = -1;
+ pool = NULL;
+ cursor = NULL;
+ cursorBaseIdx = -1;
+}
+
+TextPool::~TextPool() {
+ int baseIdx;
+ TextWord *word, *word2;
+
+ for (baseIdx = minBaseIdx; baseIdx <= maxBaseIdx; ++baseIdx) {
+ for (word = pool[baseIdx - minBaseIdx]; word; word = word2) {
+ word2 = word->next;
+ delete word;
+ }
+ }
+ gfree(pool);
+}
+
+int TextPool::getBaseIdx(double base) {
+ int baseIdx;
+
+ baseIdx = (int)(base / textPoolStep);
+ if (baseIdx < minBaseIdx) {
+ return minBaseIdx;
+ }
+ if (baseIdx > maxBaseIdx) {
+ return maxBaseIdx;
+ }
+ return baseIdx;
+}
+
+void TextPool::addWord(TextWord *word) {
+ TextWord **newPool;
+ int wordBaseIdx, newMinBaseIdx, newMaxBaseIdx, baseIdx;
+ TextWord *w0, *w1;
+
+ // expand the array if needed
+ wordBaseIdx = (int)(word->base / textPoolStep);
+ if (minBaseIdx > maxBaseIdx) {
+ minBaseIdx = wordBaseIdx - 128;
+ maxBaseIdx = wordBaseIdx + 128;
+ pool = (TextWord **)gmalloc((maxBaseIdx - minBaseIdx + 1) *
+ sizeof(TextWord *));
+ for (baseIdx = minBaseIdx; baseIdx <= maxBaseIdx; ++baseIdx) {
+ pool[baseIdx - minBaseIdx] = NULL;
+ }
+ } else if (wordBaseIdx < minBaseIdx) {
+ newMinBaseIdx = wordBaseIdx - 128;
+ newPool = (TextWord **)gmalloc((maxBaseIdx - newMinBaseIdx + 1) *
+ sizeof(TextWord *));
+ for (baseIdx = newMinBaseIdx; baseIdx < minBaseIdx; ++baseIdx) {
+ newPool[baseIdx - newMinBaseIdx] = NULL;
+ }
+ memcpy(&newPool[minBaseIdx - newMinBaseIdx], pool,
+ (maxBaseIdx - minBaseIdx + 1) * sizeof(TextWord *));
+ gfree(pool);
+ pool = newPool;
+ minBaseIdx = newMinBaseIdx;
+ } else if (wordBaseIdx > maxBaseIdx) {
+ newMaxBaseIdx = wordBaseIdx + 128;
+ pool = (TextWord **)grealloc(pool, (newMaxBaseIdx - minBaseIdx + 1) *
+ sizeof(TextWord *));
+ for (baseIdx = maxBaseIdx + 1; baseIdx <= newMaxBaseIdx; ++baseIdx) {
+ pool[baseIdx - minBaseIdx] = NULL;
+ }
+ maxBaseIdx = newMaxBaseIdx;
+ }
+
+ // insert the new word
+ if (cursor && wordBaseIdx == cursorBaseIdx &&
+ word->primaryCmp(cursor) > 0) {
+ w0 = cursor;
+ w1 = cursor->next;
+ } else {
+ w0 = NULL;
+ w1 = pool[wordBaseIdx - minBaseIdx];
+ }
+ for (; w1 && word->primaryCmp(w1) > 0; w0 = w1, w1 = w1->next) ;
+ word->next = w1;
+ if (w0) {
+ w0->next = word;
+ } else {
+ pool[wordBaseIdx - minBaseIdx] = word;
+ }
+ cursor = word;
+ cursorBaseIdx = wordBaseIdx;
+}
+
+//------------------------------------------------------------------------
+// TextLine
+//------------------------------------------------------------------------
+
+TextLine::TextLine(TextBlock *blkA, int rotA, double baseA) {
+ blk = blkA;
+ rot = rotA;
+ xMin = yMin = 0;
+ xMax = yMax = -1;
+ base = baseA;
+ words = lastWord = NULL;
text = NULL;
- xRight = NULL;
+ edge = NULL;
col = NULL;
+ len = 0;
+ convertedLen = 0;
+ hyphenated = gFalse;
+ next = NULL;
}
-TextBlock::~TextBlock() {
- TextString *p1, *p2;
+TextLine::~TextLine() {
+ TextWord *word;
- for (p1 = strings; p1; p1 = p2) {
- p2 = p1->next;
- delete p1;
+ while (words) {
+ word = words;
+ words = words->next;
+ delete word;
}
gfree(text);
- gfree(xRight);
+ gfree(edge);
gfree(col);
}
+void TextLine::addWord(TextWord *word) {
+ if (lastWord) {
+ lastWord->next = word;
+ } else {
+ words = word;
+ }
+ lastWord = word;
+
+ if (xMin > xMax) {
+ xMin = word->xMin;
+ xMax = word->xMax;
+ yMin = word->yMin;
+ yMax = word->yMax;
+ } else {
+ if (word->xMin < xMin) {
+ xMin = word->xMin;
+ }
+ if (word->xMax > xMax) {
+ xMax = word->xMax;
+ }
+ if (word->yMin < yMin) {
+ yMin = word->yMin;
+ }
+ if (word->yMax > yMax) {
+ yMax = word->yMax;
+ }
+ }
+}
+
+double TextLine::primaryDelta(TextLine *line) {
+ double delta;
+
+ delta = 0; // make gcc happy
+ switch (rot) {
+ case 0:
+ delta = line->xMin - xMax;
+ break;
+ case 1:
+ delta = line->yMin - yMax;
+ break;
+ case 2:
+ delta = xMin - line->xMax;
+ break;
+ case 3:
+ delta = yMin - line->yMax;
+ break;
+ }
+ return delta;
+}
+
+int TextLine::primaryCmp(TextLine *line) {
+ double cmp;
+
+ cmp = 0; // make gcc happy
+ switch (rot) {
+ case 0:
+ cmp = xMin - line->xMin;
+ break;
+ case 1:
+ cmp = yMin - line->yMin;
+ break;
+ case 2:
+ cmp = line->xMax - xMax;
+ break;
+ case 3:
+ cmp = line->yMax - yMax;
+ break;
+ }
+ return cmp < 0 ? -1 : cmp > 0 ? 1 : 0;
+}
+
+int TextLine::secondaryCmp(TextLine *line) {
+ double cmp;
+
+ cmp = (rot == 0 || rot == 3) ? base - line->base : line->base - base;
+ return cmp < 0 ? -1 : cmp > 0 ? 1 : 0;
+}
+
+int TextLine::cmpYX(TextLine *line) {
+ int cmp;
+
+ if ((cmp = secondaryCmp(line))) {
+ return cmp;
+ }
+ return primaryCmp(line);
+}
+
+int TextLine::cmpXY(const void *p1, const void *p2) {
+ TextLine *line1 = *(TextLine **)p1;
+ TextLine *line2 = *(TextLine **)p2;
+ int cmp;
+
+ if ((cmp = line1->primaryCmp(line2))) {
+ return cmp;
+ }
+ return line1->secondaryCmp(line2);
+}
+
+void TextLine::coalesce(UnicodeMap *uMap) {
+ TextWord *word0, *word1;
+ double space, delta, minSpace;
+ GBool isUnicode;
+ char buf[8];
+ int i, j;
+
+ if (words->next) {
+
+ // compute the inter-word space threshold
+ if (words->len > 1 || words->next->len > 1) {
+ minSpace = 0;
+ } else {
+ minSpace = words->primaryDelta(words->next);
+ for (word0 = words->next, word1 = word0->next;
+ word1 && minSpace > 0;
+ word0 = word1, word1 = word0->next) {
+ if (word1->len > 1) {
+ minSpace = 0;
+ }
+ delta = word0->primaryDelta(word1);
+ if (delta < minSpace) {
+ minSpace = delta;
+ }
+ }
+ }
+ if (minSpace <= 0) {
+ space = maxCharSpacing * words->fontSize;
+ } else {
+ space = maxWideCharSpacingMul * minSpace;
+ }
+
+ // merge words
+ word0 = words;
+ word1 = words->next;
+ while (word1) {
+ if (word0->primaryDelta(word1) >= space) {
+ word0->spaceAfter = gTrue;
+ word0 = word1;
+ word1 = word1->next;
+ } else if (word0->font == word1->font &&
+ fabs(word0->fontSize - word1->fontSize) <
+ maxWordFontSizeDelta * words->fontSize &&
+ word1->charPos == word0->charPos + word0->charLen) {
+ word0->merge(word1);
+ word0->next = word1->next;
+ delete word1;
+ word1 = word0->next;
+ } else {
+ word0 = word1;
+ word1 = word1->next;
+ }
+ }
+ }
+
+ // build the line text
+ isUnicode = uMap ? uMap->isUnicode() : gFalse;
+ len = 0;
+ for (word1 = words; word1; word1 = word1->next) {
+ len += word1->len;
+ if (word1->spaceAfter) {
+ ++len;
+ }
+ }
+ text = (Unicode *)gmalloc(len * sizeof(Unicode));
+ edge = (double *)gmalloc((len + 1) * sizeof(double));
+ i = 0;
+ for (word1 = words; word1; word1 = word1->next) {
+ for (j = 0; j < word1->len; ++j) {
+ text[i] = word1->text[j];
+ edge[i] = word1->edge[j];
+ ++i;
+ }
+ edge[i] = word1->edge[word1->len];
+ if (word1->spaceAfter) {
+ text[i] = (Unicode)0x0020;
+ ++i;
+ }
+ }
+
+ // compute convertedLen and set up the col array
+ col = (int *)gmalloc((len + 1) * sizeof(int));
+ convertedLen = 0;
+ for (i = 0; i < len; ++i) {
+ col[i] = convertedLen;
+ if (isUnicode) {
+ ++convertedLen;
+ } else if (uMap) {
+ convertedLen += uMap->mapUnicode(text[i], buf, sizeof(buf));
+ }
+ }
+ col[len] = convertedLen;
+
+ // check for hyphen at end of line
+ //~ need to check for other chars used as hyphens
+ hyphenated = text[len - 1] == (Unicode)'-';
+}
+
//------------------------------------------------------------------------
-// TextLine
+// TextLineFrag
//------------------------------------------------------------------------
-class TextLine {
+class TextLineFrag {
public:
- TextLine();
- ~TextLine();
-
- TextBlock *blocks;
- TextLine *next;
+ TextLine *line; // the line object
+ int start, len; // offset and length of this fragment
+ // (in Unicode chars)
+ double xMin, xMax; // bounding box coordinates
double yMin, yMax;
+ double base; // baseline virtual coordinate
+ int col; // first column
+
+ void init(TextLine *lineA, int startA, int lenA);
+ void computeCoords(GBool oneRot);
+
+ static int cmpYXPrimaryRot(const void *p1, const void *p2);
+ static int cmpYXLineRot(const void *p1, const void *p2);
+ static int cmpXYLineRot(const void *p1, const void *p2);
};
-TextLine::TextLine() {
- blocks = NULL;
+void TextLineFrag::init(TextLine *lineA, int startA, int lenA) {
+ line = lineA;
+ start = startA;
+ len = lenA;
+ col = line->col[start];
+}
+
+void TextLineFrag::computeCoords(GBool oneRot) {
+ TextBlock *blk;
+ double d0, d1, d2, d3, d4;
+
+ if (oneRot) {
+
+ switch (line->rot) {
+ case 0:
+ xMin = line->edge[start];
+ xMax = line->edge[start + len];
+ yMin = line->yMin;
+ yMax = line->yMax;
+ break;
+ case 1:
+ xMin = line->xMin;
+ xMax = line->xMax;
+ yMin = line->edge[start];
+ yMax = line->edge[start + len];
+ break;
+ case 2:
+ xMin = line->edge[start + len];
+ xMax = line->edge[start];
+ yMin = line->yMin;
+ yMax = line->yMax;
+ break;
+ case 3:
+ xMin = line->xMin;
+ xMax = line->xMax;
+ yMin = line->edge[start + len];
+ yMax = line->edge[start];
+ break;
+ }
+ base = line->base;
+
+ } else {
+
+ if (line->rot == 0 && line->blk->page->primaryRot == 0) {
+
+ xMin = line->edge[start];
+ xMax = line->edge[start + len];
+ yMin = line->yMin;
+ yMax = line->yMax;
+ base = line->base;
+
+ } else {
+
+ blk = line->blk;
+ d0 = line->edge[start];
+ d1 = line->edge[start + len];
+ d2 = d3 = d4 = 0; // make gcc happy
+
+ switch (line->rot) {
+ case 0:
+ d2 = line->yMin;
+ d3 = line->yMax;
+ d4 = line->base;
+ d0 = (d0 - blk->xMin) / (blk->xMax - blk->xMin);
+ d1 = (d1 - blk->xMin) / (blk->xMax - blk->xMin);
+ d2 = (d2 - blk->yMin) / (blk->yMax - blk->yMin);
+ d3 = (d3 - blk->yMin) / (blk->yMax - blk->yMin);
+ d4 = (d4 - blk->yMin) / (blk->yMax - blk->yMin);
+ break;
+ case 1:
+ d2 = line->xMax;
+ d3 = line->xMin;
+ d4 = line->base;
+ d0 = (d0 - blk->yMin) / (blk->yMax - blk->yMin);
+ d1 = (d1 - blk->yMin) / (blk->yMax - blk->yMin);
+ d2 = (blk->xMax - d2) / (blk->xMax - blk->xMin);
+ d3 = (blk->xMax - d3) / (blk->xMax - blk->xMin);
+ d4 = (blk->xMax - d4) / (blk->xMax - blk->xMin);
+ break;
+ case 2:
+ d2 = line->yMax;
+ d3 = line->yMin;
+ d4 = line->base;
+ d0 = (blk->xMax - d0) / (blk->xMax - blk->xMin);
+ d1 = (blk->xMax - d1) / (blk->xMax - blk->xMin);
+ d2 = (blk->yMax - d2) / (blk->yMax - blk->yMin);
+ d3 = (blk->yMax - d3) / (blk->yMax - blk->yMin);
+ d4 = (blk->yMax - d4) / (blk->yMax - blk->yMin);
+ break;
+ case 3:
+ d2 = line->xMin;
+ d3 = line->xMax;
+ d4 = line->base;
+ d0 = (blk->yMax - d0) / (blk->yMax - blk->yMin);
+ d1 = (blk->yMax - d1) / (blk->yMax - blk->yMin);
+ d2 = (d2 - blk->xMin) / (blk->xMax - blk->xMin);
+ d3 = (d3 - blk->xMin) / (blk->xMax - blk->xMin);
+ d4 = (d4 - blk->xMin) / (blk->xMax - blk->xMin);
+ break;
+ }
+
+ switch (line->blk->page->primaryRot) {
+ case 0:
+ xMin = blk->xMin + d0 * (blk->xMax - blk->xMin);
+ xMax = blk->xMin + d1 * (blk->xMax - blk->xMin);
+ yMin = blk->yMin + d2 * (blk->yMax - blk->yMin);
+ yMax = blk->yMin + d3 * (blk->yMax - blk->yMin);
+ base = blk->yMin + base * (blk->yMax - blk->yMin);
+ break;
+ case 1:
+ xMin = blk->xMax - d3 * (blk->xMax - blk->xMin);
+ xMax = blk->xMax - d2 * (blk->xMax - blk->xMin);
+ yMin = blk->yMin + d0 * (blk->yMax - blk->yMin);
+ yMax = blk->yMin + d1 * (blk->yMax - blk->yMin);
+ base = blk->xMax - d4 * (blk->xMax - blk->xMin);
+ break;
+ case 2:
+ xMin = blk->xMax - d1 * (blk->xMax - blk->xMin);
+ xMax = blk->xMax - d0 * (blk->xMax - blk->xMin);
+ yMin = blk->yMax - d3 * (blk->yMax - blk->yMin);
+ yMax = blk->yMax - d2 * (blk->yMax - blk->yMin);
+ base = blk->yMax - d4 * (blk->yMax - blk->yMin);
+ break;
+ case 3:
+ xMin = blk->xMin + d2 * (blk->xMax - blk->xMin);
+ xMax = blk->xMin + d3 * (blk->xMax - blk->xMin);
+ yMin = blk->yMax - d1 * (blk->yMax - blk->yMin);
+ yMax = blk->yMax - d0 * (blk->yMax - blk->yMin);
+ base = blk->xMin + d4 * (blk->xMax - blk->xMin);
+ break;
+ }
+
+ }
+ }
+}
+
+int TextLineFrag::cmpYXPrimaryRot(const void *p1, const void *p2) {
+ TextLineFrag *frag1 = (TextLineFrag *)p1;
+ TextLineFrag *frag2 = (TextLineFrag *)p2;
+ double cmp;
+
+ cmp = 0; // make gcc happy
+ switch (frag1->line->blk->page->primaryRot) {
+ case 0:
+ if ((cmp = frag1->yMin - frag2->yMin) == 0) {
+ cmp = frag1->xMin - frag2->xMin;
+ }
+ break;
+ case 1:
+ if ((cmp = frag2->xMax - frag1->xMax) == 0) {
+ cmp = frag1->yMin - frag2->yMin;
+ }
+ break;
+ case 2:
+ if ((cmp = frag2->yMin - frag1->yMin) == 0) {
+ cmp = frag2->xMax - frag1->xMax;
+ }
+ break;
+ case 3:
+ if ((cmp = frag1->xMax - frag2->xMax) == 0) {
+ cmp = frag2->yMax - frag1->yMax;
+ }
+ break;
+ }
+ return cmp < 0 ? -1 : cmp > 0 ? 1 : 0;
+}
+
+int TextLineFrag::cmpYXLineRot(const void *p1, const void *p2) {
+ TextLineFrag *frag1 = (TextLineFrag *)p1;
+ TextLineFrag *frag2 = (TextLineFrag *)p2;
+ double cmp;
+
+ cmp = 0; // make gcc happy
+ switch (frag1->line->rot) {
+ case 0:
+ if ((cmp = frag1->yMin - frag2->yMin) == 0) {
+ cmp = frag1->xMin - frag2->xMin;
+ }
+ break;
+ case 1:
+ if ((cmp = frag2->xMax - frag1->xMax) == 0) {
+ cmp = frag1->yMin - frag2->yMin;
+ }
+ break;
+ case 2:
+ if ((cmp = frag2->yMin - frag1->yMin) == 0) {
+ cmp = frag2->xMax - frag1->xMax;
+ }
+ break;
+ case 3:
+ if ((cmp = frag1->xMax - frag2->xMax) == 0) {
+ cmp = frag2->yMax - frag1->yMax;
+ }
+ break;
+ }
+ return cmp < 0 ? -1 : cmp > 0 ? 1 : 0;
+}
+
+int TextLineFrag::cmpXYLineRot(const void *p1, const void *p2) {
+ TextLineFrag *frag1 = (TextLineFrag *)p1;
+ TextLineFrag *frag2 = (TextLineFrag *)p2;
+ double cmp;
+
+ cmp = 0; // make gcc happy
+ switch (frag1->line->rot) {
+ case 0:
+ if ((cmp = frag1->xMin - frag2->xMin) == 0) {
+ cmp = frag1->yMin - frag2->yMin;
+ }
+ break;
+ case 1:
+ if ((cmp = frag1->yMin - frag2->yMin) == 0) {
+ cmp = frag2->xMax - frag1->xMax;
+ }
+ break;
+ case 2:
+ if ((cmp = frag2->xMax - frag1->xMax) == 0) {
+ cmp = frag2->yMin - frag1->yMin;
+ }
+ break;
+ case 3:
+ if ((cmp = frag2->yMax - frag1->yMax) == 0) {
+ cmp = frag1->xMax - frag2->xMax;
+ }
+ break;
+ }
+ return cmp < 0 ? -1 : cmp > 0 ? 1 : 0;
+}
+
+//------------------------------------------------------------------------
+// TextBlock
+//------------------------------------------------------------------------
+
+TextBlock::TextBlock(TextPage *pageA, int rotA) {
+ page = pageA;
+ rot = rotA;
+ xMin = yMin = 0;
+ xMax = yMax = -1;
+ priMin = 0;
+ priMax = page->pageWidth;
+ pool = new TextPool();
+ lines = NULL;
+ curLine = NULL;
+ next = NULL;
+ stackNext = NULL;
+}
+
+TextBlock::~TextBlock() {
+ TextLine *line;
+
+ delete pool;
+ while (lines) {
+ line = lines;
+ lines = lines->next;
+ delete line;
+ }
+}
+
+void TextBlock::addWord(TextWord *word) {
+ pool->addWord(word);
+ if (xMin > xMax) {
+ xMin = word->xMin;
+ xMax = word->xMax;
+ yMin = word->yMin;
+ yMax = word->yMax;
+ } else {
+ if (word->xMin < xMin) {
+ xMin = word->xMin;
+ }
+ if (word->xMax > xMax) {
+ xMax = word->xMax;
+ }
+ if (word->yMin < yMin) {
+ yMin = word->yMin;
+ }
+ if (word->yMax > yMax) {
+ yMax = word->yMax;
+ }
+ }
+}
+
+void TextBlock::coalesce(UnicodeMap *uMap) {
+ TextWord *word0, *word1, *word2, *bestWord0, *bestWord1, *lastWord;
+ TextLine *line, *line0, *line1;
+ int poolMinBaseIdx, startBaseIdx, minBaseIdx, maxBaseIdx;
+ int baseIdx, bestWordBaseIdx, idx0, idx1;
+ double minBase, maxBase;
+ double fontSize, delta, priDelta, secDelta;
+ TextLine **lineArray;
+ GBool found;
+ int col1, col2;
+ int i, j, k;
+
+ // discard duplicated text (fake boldface, drop shadows)
+ for (idx0 = pool->minBaseIdx; idx0 <= pool->maxBaseIdx; ++idx0) {
+ word0 = pool->getPool(idx0);
+ while (word0) {
+ priDelta = dupMaxPriDelta * word0->fontSize;
+ secDelta = dupMaxSecDelta * word0->fontSize;
+ if (rot == 0 || rot == 3) {
+ maxBaseIdx = pool->getBaseIdx(word0->base + secDelta);
+ } else {
+ maxBaseIdx = pool->getBaseIdx(word0->base - secDelta);
+ }
+ found = gFalse;
+ word1 = word2 = NULL; // make gcc happy
+ for (idx1 = idx0; idx1 <= maxBaseIdx; ++idx1) {
+ if (idx1 == idx0) {
+ word1 = word0;
+ word2 = word0->next;
+ } else {
+ word1 = NULL;
+ word2 = pool->getPool(idx1);
+ }
+ for (; word2; word1 = word2, word2 = word2->next) {
+ if (word2->len == word0->len &&
+ !memcmp(word2->text, word0->text,
+ word0->len * sizeof(Unicode))) {
+ switch (rot) {
+ case 0:
+ case 2:
+ found = fabs(word0->xMin - word2->xMin) < priDelta &&
+ fabs(word0->xMax - word2->xMax) < priDelta &&
+ fabs(word0->yMin - word2->yMin) < secDelta &&
+ fabs(word0->yMax - word2->yMax) < secDelta;
+ break;
+ case 1:
+ case 3:
+ found = fabs(word0->xMin - word2->xMin) < secDelta &&
+ fabs(word0->xMax - word2->xMax) < secDelta &&
+ fabs(word0->yMin - word2->yMin) < priDelta &&
+ fabs(word0->yMax - word2->yMax) < priDelta;
+ break;
+ }
+ }
+ if (found) {
+ break;
+ }
+ }
+ if (found) {
+ break;
+ }
+ }
+ if (found) {
+ if (word1) {
+ word1->next = word2->next;
+ } else {
+ pool->setPool(idx1, word2->next);
+ }
+ delete word2;
+ } else {
+ word0 = word0->next;
+ }
+ }
+ }
+
+ // build the lines
+ curLine = NULL;
+ poolMinBaseIdx = pool->minBaseIdx;
+ charCount = 0;
+ nLines = 0;
+ while (1) {
+
+ // find the first non-empty line in the pool
+ for (;
+ poolMinBaseIdx <= pool->maxBaseIdx && !pool->getPool(poolMinBaseIdx);
+ ++poolMinBaseIdx) ;
+ if (poolMinBaseIdx > pool->maxBaseIdx) {
+ break;
+ }
+
+ // look for the left-most word in the first four lines of the
+ // pool -- this avoids starting with a superscript word
+ startBaseIdx = poolMinBaseIdx;
+ for (baseIdx = poolMinBaseIdx + 1;
+ baseIdx < poolMinBaseIdx + 4 && baseIdx <= pool->maxBaseIdx;
+ ++baseIdx) {
+ if (!pool->getPool(baseIdx)) {
+ continue;
+ }
+ if (pool->getPool(baseIdx)->primaryCmp(pool->getPool(startBaseIdx))
+ < 0) {
+ startBaseIdx = baseIdx;
+ }
+ }
+
+ // create a new line
+ word0 = pool->getPool(startBaseIdx);
+ pool->setPool(startBaseIdx, word0->next);
+ word0->next = NULL;
+ line = new TextLine(this, word0->rot, word0->base);
+ line->addWord(word0);
+ lastWord = word0;
+
+ // compute the search range
+ fontSize = word0->fontSize;
+ minBase = word0->base - maxIntraLineDelta * fontSize;
+ maxBase = word0->base + maxIntraLineDelta * fontSize;
+ minBaseIdx = pool->getBaseIdx(minBase);
+ maxBaseIdx = pool->getBaseIdx(maxBase);
+
+ // find the rest of the words in this line
+ while (1) {
+
+ // find the left-most word whose baseline is in the range for
+ // this line
+ bestWordBaseIdx = 0;
+ bestWord0 = bestWord1 = NULL;
+ for (baseIdx = minBaseIdx; baseIdx <= maxBaseIdx; ++baseIdx) {
+ for (word0 = NULL, word1 = pool->getPool(baseIdx);
+ word1;
+ word0 = word1, word1 = word1->next) {
+ if (word1->base >= minBase &&
+ word1->base <= maxBase &&
+ (delta = lastWord->primaryDelta(word1)) >=
+ minCharSpacing * fontSize) {
+ if (delta < maxWordSpacing * fontSize &&
+ (!bestWord1 || word1->primaryCmp(bestWord1) < 0)) {
+ bestWordBaseIdx = baseIdx;
+ bestWord0 = word0;
+ bestWord1 = word1;
+ }
+ break;
+ }
+ }
+ }
+ if (!bestWord1) {
+ break;
+ }
+
+ // remove it from the pool, and add it to the line
+ if (bestWord0) {
+ bestWord0->next = bestWord1->next;
+ } else {
+ pool->setPool(bestWordBaseIdx, bestWord1->next);
+ }
+ bestWord1->next = NULL;
+ line->addWord(bestWord1);
+ lastWord = bestWord1;
+ }
+
+ // add the line
+ if (curLine && line->cmpYX(curLine) > 0) {
+ line0 = curLine;
+ line1 = curLine->next;
+ } else {
+ line0 = NULL;
+ line1 = lines;
+ }
+ for (;
+ line1 && line->cmpYX(line1) > 0;
+ line0 = line1, line1 = line1->next) ;
+ if (line0) {
+ line0->next = line;
+ } else {
+ lines = line;
+ }
+ line->next = line1;
+ curLine = line;
+ line->coalesce(uMap);
+ charCount += line->len;
+ ++nLines;
+ }
+
+ // sort lines into xy order for column assignment
+ lineArray = (TextLine **)gmalloc(nLines * sizeof(TextLine *));
+ for (line = lines, i = 0; line; line = line->next, ++i) {
+ lineArray[i] = line;
+ }
+ qsort(lineArray, nLines, sizeof(TextLine *), &TextLine::cmpXY);
+
+ // column assignment
+ nColumns = 0;
+ for (i = 0; i < nLines; ++i) {
+ line0 = lineArray[i];
+ col1 = 0;
+ for (j = 0; j < i; ++j) {
+ line1 = lineArray[j];
+ if (line1->primaryDelta(line0) >= 0) {
+ col2 = line1->col[line1->len] + 1;
+ } else {
+ k = 0; // make gcc happy
+ switch (rot) {
+ case 0:
+ for (k = 0;
+ k < line1->len &&
+ line0->xMin >= 0.5 * (line1->edge[k] + line1->edge[k+1]);
+ ++k) ;
+ break;
+ case 1:
+ for (k = 0;
+ k < line1->len &&
+ line0->yMin >= 0.5 * (line1->edge[k] + line1->edge[k+1]);
+ ++k) ;
+ break;
+ case 2:
+ for (k = 0;
+ k < line1->len &&
+ line0->xMax <= 0.5 * (line1->edge[k] + line1->edge[k+1]);
+ ++k) ;
+ break;
+ case 3:
+ for (k = 0;
+ k < line1->len &&
+ line0->yMax <= 0.5 * (line1->edge[k] + line1->edge[k+1]);
+ ++k) ;
+ break;
+ }
+ col2 = line1->col[k];
+ }
+ if (col2 > col1) {
+ col1 = col2;
+ }
+ }
+ for (k = 0; k <= line0->len; ++k) {
+ line0->col[k] += col1;
+ }
+ if (line0->col[line0->len] > nColumns) {
+ nColumns = line0->col[line0->len];
+ }
+ }
+ gfree(lineArray);
+}
+
+void TextBlock::updatePriMinMax(TextBlock *blk) {
+ double newPriMin, newPriMax;
+ GBool gotPriMin, gotPriMax;
+
+ gotPriMin = gotPriMax = gFalse;
+ newPriMin = newPriMax = 0; // make gcc happy
+ switch (page->primaryRot) {
+ case 0:
+ case 2:
+ if (blk->yMin < yMax && blk->yMax > yMin) {
+ if (blk->xMin < xMin) {
+ newPriMin = blk->xMax;
+ gotPriMin = gTrue;
+ }
+ if (blk->xMax > xMax) {
+ newPriMax = blk->xMin;
+ gotPriMax = gTrue;
+ }
+ }
+ break;
+ case 1:
+ case 3:
+ if (blk->xMin < xMax && blk->xMax > xMin) {
+ if (blk->yMin < yMin) {
+ newPriMin = blk->yMax;
+ gotPriMin = gTrue;
+ }
+ if (blk->yMax > yMax) {
+ newPriMax = blk->yMin;
+ gotPriMax = gTrue;
+ }
+ }
+ break;
+ }
+ if (gotPriMin) {
+ if (newPriMin > xMin) {
+ newPriMin = xMin;
+ }
+ if (newPriMin > priMin) {
+ priMin = newPriMin;
+ }
+ }
+ if (gotPriMax) {
+ if (newPriMax < xMax) {
+ newPriMax = xMax;
+ }
+ if (newPriMax < priMax) {
+ priMax = newPriMax;
+ }
+ }
+}
+
+int TextBlock::cmpXYPrimaryRot(const void *p1, const void *p2) {
+ TextBlock *blk1 = *(TextBlock **)p1;
+ TextBlock *blk2 = *(TextBlock **)p2;
+ double cmp;
+
+ cmp = 0; // make gcc happy
+ switch (blk1->page->primaryRot) {
+ case 0:
+ if ((cmp = blk1->xMin - blk2->xMin) == 0) {
+ cmp = blk1->yMin - blk2->yMin;
+ }
+ break;
+ case 1:
+ if ((cmp = blk1->yMin - blk2->yMin) == 0) {
+ cmp = blk2->xMax - blk1->xMax;
+ }
+ break;
+ case 2:
+ if ((cmp = blk2->xMax - blk1->xMax) == 0) {
+ cmp = blk2->yMin - blk1->yMin;
+ }
+ break;
+ case 3:
+ if ((cmp = blk2->yMax - blk1->yMax) == 0) {
+ cmp = blk1->xMax - blk2->xMax;
+ }
+ break;
+ }
+ return cmp < 0 ? -1 : cmp > 0 ? 1 : 0;
+}
+
+int TextBlock::cmpYXPrimaryRot(const void *p1, const void *p2) {
+ TextBlock *blk1 = *(TextBlock **)p1;
+ TextBlock *blk2 = *(TextBlock **)p2;
+ double cmp;
+
+ cmp = 0; // make gcc happy
+ switch (blk1->page->primaryRot) {
+ case 0:
+ if ((cmp = blk1->yMin - blk2->yMin) == 0) {
+ cmp = blk1->xMin - blk2->xMin;
+ }
+ break;
+ case 1:
+ if ((cmp = blk2->xMax - blk1->xMax) == 0) {
+ cmp = blk1->yMin - blk2->yMin;
+ }
+ break;
+ case 2:
+ if ((cmp = blk2->yMin - blk1->yMin) == 0) {
+ cmp = blk2->xMax - blk1->xMax;
+ }
+ break;
+ case 3:
+ if ((cmp = blk1->xMax - blk2->xMax) == 0) {
+ cmp = blk2->yMax - blk1->yMax;
+ }
+ break;
+ }
+ return cmp < 0 ? -1 : cmp > 0 ? 1 : 0;
+}
+
+int TextBlock::primaryCmp(TextBlock *blk) {
+ double cmp;
+
+ cmp = 0; // make gcc happy
+ switch (rot) {
+ case 0:
+ cmp = xMin - blk->xMin;
+ break;
+ case 1:
+ cmp = yMin - blk->yMin;
+ break;
+ case 2:
+ cmp = blk->xMax - xMax;
+ break;
+ case 3:
+ cmp = blk->yMax - yMax;
+ break;
+ }
+ return cmp < 0 ? -1 : cmp > 0 ? 1 : 0;
+}
+
+double TextBlock::secondaryDelta(TextBlock *blk) {
+ double delta;
+
+ delta = 0; // make gcc happy
+ switch (rot) {
+ case 0:
+ delta = blk->yMin - yMax;
+ break;
+ case 1:
+ delta = xMin - blk->xMax;
+ break;
+ case 2:
+ delta = yMin - blk->yMax;
+ break;
+ case 3:
+ delta = blk->xMin - xMax;
+ break;
+ }
+ return delta;
+}
+
+GBool TextBlock::isBelow(TextBlock *blk) {
+ GBool below;
+
+ below = gFalse; // make gcc happy
+ switch (page->primaryRot) {
+ case 0:
+ below = xMin >= blk->priMin && xMax <= blk->priMax &&
+ yMin > blk->yMin;
+ break;
+ case 1:
+ below = yMin >= blk->priMin && yMax <= blk->priMax &&
+ xMax < blk->xMax;
+ break;
+ case 2:
+ below = xMin >= blk->priMin && xMax <= blk->priMax &&
+ yMax < blk->yMax;
+ break;
+ case 3:
+ below = yMin >= blk->priMin && yMax <= blk->priMax &&
+ xMin > blk->xMin;
+ break;
+ }
+
+ return below;
+}
+
+//------------------------------------------------------------------------
+// TextFlow
+//------------------------------------------------------------------------
+
+TextFlow::TextFlow(TextPage *pageA, TextBlock *blk) {
+ page = pageA;
+ xMin = blk->xMin;
+ xMax = blk->xMax;
+ yMin = blk->yMin;
+ yMax = blk->yMax;
+ priMin = blk->priMin;
+ priMax = blk->priMax;
+ blocks = lastBlk = blk;
next = NULL;
}
-TextLine::~TextLine() {
- TextBlock *p1, *p2;
+TextFlow::~TextFlow() {
+ TextBlock *blk;
+
+ while (blocks) {
+ blk = blocks;
+ blocks = blocks->next;
+ delete blk;
+ }
+}
+
+void TextFlow::addBlock(TextBlock *blk) {
+ if (lastBlk) {
+ lastBlk->next = blk;
+ } else {
+ blocks = blk;
+ }
+ lastBlk = blk;
+ if (blk->xMin < xMin) {
+ xMin = blk->xMin;
+ }
+ if (blk->xMax > xMax) {
+ xMax = blk->xMax;
+ }
+ if (blk->yMin < yMin) {
+ yMin = blk->yMin;
+ }
+ if (blk->yMax > yMax) {
+ yMax = blk->yMax;
+ }
+}
+
+GBool TextFlow::blockFits(TextBlock *blk, TextBlock *prevBlk) {
+ GBool fits;
- for (p1 = blocks; p1; p1 = p2) {
- p2 = p1->next;
- delete p1;
+ // lower blocks must use smaller fonts
+ if (blk->lines->words->fontSize > lastBlk->lines->words->fontSize) {
+ return gFalse;
+ }
+
+ fits = gFalse; // make gcc happy
+ switch (page->primaryRot) {
+ case 0:
+ fits = blk->xMin >= priMin && blk->xMax <= priMax;
+ break;
+ case 1:
+ fits = blk->yMin >= priMin && blk->yMax <= priMax;
+ break;
+ case 2:
+ fits = blk->xMin >= priMin && blk->xMax <= priMax;
+ break;
+ case 3:
+ fits = blk->yMin >= priMin && blk->yMax <= priMax;
+ break;
}
+ return fits;
}
+#if TEXTOUT_WORD_LIST
+
//------------------------------------------------------------------------
-// TextString
+// TextWordList
//------------------------------------------------------------------------
-TextString::TextString(GfxState *state, double x0, double y0,
- double fontSize) {
- GfxFont *font;
- double x, y;
+TextWordList::TextWordList(TextPage *text, GBool physLayout) {
+ TextFlow *flow;
+ TextBlock *blk;
+ TextLine *line;
+ TextWord *word;
+ TextWord **wordArray;
+ int nWords, i;
+
+ words = new GList();
+
+ if (text->rawOrder) {
+ for (word = text->rawWords; word; word = word->next) {
+ words->append(word);
+ }
+
+ } else if (physLayout) {
+ // this is inefficient, but it's also the least useful of these
+ // three cases
+ nWords = 0;
+ for (flow = text->flows; flow; flow = flow->next) {
+ for (blk = flow->blocks; blk; blk = blk->next) {
+ for (line = blk->lines; line; line = line->next) {
+ for (word = line->words; word; word = word->next) {
+ ++nWords;
+ }
+ }
+ }
+ }
+ wordArray = (TextWord **)gmalloc(nWords * sizeof(TextWord *));
+ i = 0;
+ for (flow = text->flows; flow; flow = flow->next) {
+ for (blk = flow->blocks; blk; blk = blk->next) {
+ for (line = blk->lines; line; line = line->next) {
+ for (word = line->words; word; word = word->next) {
+ wordArray[i++] = word;
+ }
+ }
+ }
+ }
+ qsort(wordArray, nWords, sizeof(TextWord *), &TextWord::cmpYX);
+ for (i = 0; i < nWords; ++i) {
+ words->append(wordArray[i]);
+ }
+ gfree(wordArray);
- state->transform(x0, y0, &x, &y);
- if ((font = state->getFont())) {
- yMin = y - font->getAscent() * fontSize;
- yMax = y - font->getDescent() * fontSize;
} else {
- // this means that the PDF file draws text without a current font,
- // which should never happen
- yMin = y - 0.95 * fontSize;
- yMax = y + 0.35 * fontSize;
- }
- if (yMin == yMax) {
- // this is a sanity check for a case that shouldn't happen -- but
- // if it does happen, we want to avoid dividing by zero later
- yMin = y;
- yMax = y + 1;
+ for (flow = text->flows; flow; flow = flow->next) {
+ for (blk = flow->blocks; blk; blk = blk->next) {
+ for (line = blk->lines; line; line = line->next) {
+ for (word = line->words; word; word = word->next) {
+ words->append(word);
+ }
+ }
+ }
+ }
}
- marked = gFalse;
- text = NULL;
- xRight = NULL;
- len = size = 0;
- next = NULL;
}
+TextWordList::~TextWordList() {
+ delete words;
+}
-TextString::~TextString() {
- gfree(text);
- gfree(xRight);
+int TextWordList::getLength() {
+ return words->getLength();
}
-void TextString::addChar(GfxState *state, double x, double y,
- double dx, double dy, Unicode u) {
- if (len == size) {
- size += 16;
- text = (Unicode *)grealloc(text, size * sizeof(Unicode));
- xRight = (double *)grealloc(xRight, size * sizeof(double));
+TextWord *TextWordList::get(int idx) {
+ if (idx < 0 || idx >= words->getLength()) {
+ return NULL;
}
- text[len] = u;
- if (len == 0) {
- xMin = x;
- }
- xMax = xRight[len] = x + dx;
- ++len;
+ return (TextWord *)words->get(idx);
}
+#endif // TEXTOUT_WORD_LIST
+
//------------------------------------------------------------------------
// TextPage
//------------------------------------------------------------------------
TextPage::TextPage(GBool rawOrderA) {
+ int rot;
+
rawOrder = rawOrderA;
- curStr = NULL;
- fontSize = 0;
- xyStrings = NULL;
- xyCur1 = xyCur2 = NULL;
- lines = NULL;
+ curWord = NULL;
+ charPos = 0;
+ curFont = NULL;
+ curFontSize = 0;
nest = 0;
nTinyChars = 0;
+ lastCharOverlap = gFalse;
+ if (!rawOrder) {
+ for (rot = 0; rot < 4; ++rot) {
+ pools[rot] = new TextPool();
+ }
+ }
+ flows = NULL;
+ blocks = NULL;
+ rawWords = NULL;
+ rawLastWord = NULL;
+ fonts = new GList();
+ lastFindXMin = lastFindYMin = 0;
+ haveLastFind = gFalse;
}
TextPage::~TextPage() {
+ int rot;
+
+ clear();
+ if (!rawOrder) {
+ for (rot = 0; rot < 4; ++rot) {
+ delete pools[rot];
+ }
+ }
+ delete fonts;
+}
+
+void TextPage::startPage(GfxState *state) {
clear();
+ if (state) {
+ pageWidth = state->getPageWidth();
+ pageHeight = state->getPageHeight();
+ } else {
+ pageWidth = pageHeight = 0;
+ }
+}
+
+void TextPage::endPage() {
+ if (curWord) {
+ endWord();
+ }
+}
+
+void TextPage::clear() {
+ int rot;
+ TextFlow *flow;
+ TextWord *word;
+
+ if (curWord) {
+ delete curWord;
+ curWord = NULL;
+ }
+ if (rawOrder) {
+ while (rawWords) {
+ word = rawWords;
+ rawWords = rawWords->next;
+ delete word;
+ }
+ } else {
+ for (rot = 0; rot < 4; ++rot) {
+ delete pools[rot];
+ }
+ while (flows) {
+ flow = flows;
+ flows = flows->next;
+ delete flow;
+ }
+ gfree(blocks);
+ }
+ deleteGList(fonts, TextFontInfo);
+
+ curWord = NULL;
+ charPos = 0;
+ curFont = NULL;
+ curFontSize = 0;
+ nest = 0;
+ nTinyChars = 0;
+ if (!rawOrder) {
+ for (rot = 0; rot < 4; ++rot) {
+ pools[rot] = new TextPool();
+ }
+ }
+ flows = NULL;
+ blocks = NULL;
+ rawWords = NULL;
+ rawLastWord = NULL;
+ fonts = new GList();
}
void TextPage::updateFont(GfxState *state) {
- GfxFont *font;
+ GfxFont *gfxFont;
double *fm;
char *name;
int code, mCode, letterCode, anyCode;
double w;
+ int i;
+
+ // get the font info object
+ curFont = NULL;
+ for (i = 0; i < fonts->getLength(); ++i) {
+ curFont = (TextFontInfo *)fonts->get(i);
+ if (curFont->matches(state)) {
+ break;
+ }
+ curFont = NULL;
+ }
+ if (!curFont) {
+ curFont = new TextFontInfo(state);
+ fonts->append(curFont);
+ }
// adjust the font size
- fontSize = state->getTransformedFontSize();
- if ((font = state->getFont()) && font->getType() == fontType3) {
+ gfxFont = state->getFont();
+ curFontSize = state->getTransformedFontSize();
+ if (gfxFont && gfxFont->getType() == fontType3) {
// This is a hack which makes it possible to deal with some Type 3
// fonts. The problem is that it's impossible to know what the
// base coordinate system used in the font is without actually
// subset that doesn't contain 'm').
mCode = letterCode = anyCode = -1;
for (code = 0; code < 256; ++code) {
- name = ((Gfx8BitFont *)font)->getCharName(code);
+ name = ((Gfx8BitFont *)gfxFont)->getCharName(code);
if (name && name[0] == 'm' && name[1] == '\0') {
mCode = code;
}
(name[0] >= 'a' && name[0] <= 'z'))) {
letterCode = code;
}
- if (anyCode < 0 && name && ((Gfx8BitFont *)font)->getWidth(code) > 0) {
+ if (anyCode < 0 && name &&
+ ((Gfx8BitFont *)gfxFont)->getWidth(code) > 0) {
anyCode = code;
}
}
if (mCode >= 0 &&
- (w = ((Gfx8BitFont *)font)->getWidth(mCode)) > 0) {
+ (w = ((Gfx8BitFont *)gfxFont)->getWidth(mCode)) > 0) {
// 0.6 is a generic average 'm' width -- yes, this is a hack
- fontSize *= w / 0.6;
+ curFontSize *= w / 0.6;
} else if (letterCode >= 0 &&
- (w = ((Gfx8BitFont *)font)->getWidth(letterCode)) > 0) {
+ (w = ((Gfx8BitFont *)gfxFont)->getWidth(letterCode)) > 0) {
// even more of a hack: 0.5 is a generic letter width
- fontSize *= w / 0.5;
+ curFontSize *= w / 0.5;
} else if (anyCode >= 0 &&
- (w = ((Gfx8BitFont *)font)->getWidth(anyCode)) > 0) {
+ (w = ((Gfx8BitFont *)gfxFont)->getWidth(anyCode)) > 0) {
// better than nothing: 0.5 is a generic character width
- fontSize *= w / 0.5;
+ curFontSize *= w / 0.5;
}
- fm = font->getFontMatrix();
+ fm = gfxFont->getFontMatrix();
if (fm[0] != 0) {
- fontSize *= fabs(fm[3] / fm[0]);
+ curFontSize *= fabs(fm[3] / fm[0]);
}
}
}
-void TextPage::beginString(GfxState *state, double x0, double y0) {
+void TextPage::beginWord(GfxState *state, double x0, double y0) {
+ double *txtm, *ctm, *fontm;
+ double m[4], m2[4];
+ int rot;
+
// This check is needed because Type 3 characters can contain
- // text-drawing operations.
- if (curStr) {
+ // text-drawing operations (when TextPage is being used via
+ // {X,Win}SplashOutputDev rather than TextOutputDev).
+ if (curWord) {
++nest;
return;
}
- curStr = new TextString(state, x0, y0, fontSize);
+ // compute the rotation
+ txtm = state->getTextMat();
+ ctm = state->getCTM();
+ m[0] = txtm[0] * ctm[0] + txtm[1] * ctm[2];
+ m[1] = txtm[0] * ctm[1] + txtm[1] * ctm[3];
+ m[2] = txtm[2] * ctm[0] + txtm[3] * ctm[2];
+ m[3] = txtm[2] * ctm[1] + txtm[3] * ctm[3];
+ if (state->getFont()->getType() == fontType3) {
+ fontm = state->getFont()->getFontMatrix();
+ m2[0] = fontm[0] * m[0] + fontm[1] * m[2];
+ m2[1] = fontm[0] * m[1] + fontm[1] * m[3];
+ m2[2] = fontm[2] * m[0] + fontm[3] * m[2];
+ m2[3] = fontm[2] * m[1] + fontm[3] * m[3];
+ m[0] = m2[0];
+ m[1] = m2[1];
+ m[2] = m2[2];
+ m[3] = m2[3];
+ }
+ if (fabs(m[0] * m[3]) > fabs(m[1] * m[2])) {
+ rot = (m[3] < 0) ? 0 : 2;
+ } else {
+ rot = (m[2] > 0) ? 1 : 3;
+ }
+
+ curWord = new TextWord(state, rot, x0, y0, charPos, curFont, curFontSize);
}
void TextPage::addChar(GfxState *state, double x, double y,
- double dx, double dy, Unicode *u, int uLen) {
- double x1, y1, w1, h1, dx2, dy2;
- int n, i;
+ double dx, double dy,
+ CharCode c, Unicode *u, int uLen) {
+ double x1, y1, w1, h1, dx2, dy2, base, sp;
+ int i;
+
+ // if the previous char was a space, addChar will have called
+ // endWord, so we need to start a new word
+ if (!curWord) {
+ beginWord(state, x, y);
+ }
+ // throw away chars that aren't inside the page bounds
state->transform(x, y, &x1, &y1);
- if (x1 < 0 || x1 > state->getPageWidth() ||
- y1 < 0 || y1 > state->getPageHeight()) {
+ if (x1 < 0 || x1 > pageWidth ||
+ y1 < 0 || y1 > pageHeight) {
return;
}
- state->textTransformDelta(state->getCharSpace() * state->getHorizScaling(),
- 0, &dx2, &dy2);
+
+ // subtract char and word spacing from the dx,dy values
+ sp = state->getCharSpace();
+ if (c == (CharCode)0x20) {
+ sp += state->getWordSpace();
+ }
+ state->textTransformDelta(sp * state->getHorizScaling(), 0, &dx2, &dy2);
dx -= dx2;
dy -= dy2;
state->transformDelta(dx, dy, &w1, &h1);
+
+ // check the tiny chars limit
if (!globalParams->getTextKeepTinyChars() &&
fabs(w1) < 3 && fabs(h1) < 3) {
- if (++nTinyChars > 20000) {
+ if (++nTinyChars > 50000) {
return;
}
}
- n = curStr->len;
- if (n > 0 && x1 - curStr->xRight[n-1] >
- 0.1 * (curStr->yMax - curStr->yMin)) {
- // large char spacing is sometimes used to move text around
- endString();
- beginString(state, x, y);
- }
- if (uLen == 1 && u[0] == (Unicode)0x20 &&
- w1 > 0.5 * (curStr->yMax - curStr->yMin)) {
- // large word spacing is sometimes used to move text around
+
+ // break words at space character
+ if (uLen == 1 && u[0] == (Unicode)0x20) {
+ ++curWord->charLen;
+ ++charPos;
+ endWord();
return;
}
+
+ // start a new word if:
+ // (1) this character's baseline doesn't match the current word's
+ // baseline, or
+ // (2) there is space between the end of the current word and this
+ // character, or
+ // (3) this character overlaps the previous one (duplicated text), or
+ // (4) the previous character was an overlap (we want each duplicated
+ // characters to be in a word by itself)
+ base = sp = 0; // make gcc happy
+ if (curWord->len > 0) {
+ switch (curWord->rot) {
+ case 0:
+ base = y1;
+ sp = x1 - curWord->xMax;
+ break;
+ case 1:
+ base = x1;
+ sp = y1 - curWord->yMax;
+ break;
+ case 2:
+ base = y1;
+ sp = curWord->xMin - x1;
+ break;
+ case 3:
+ base = x1;
+ sp = curWord->yMin - y1;
+ break;
+ }
+ if (fabs(base - curWord->base) > 0.5 ||
+ sp > minWordBreakSpace * curWord->fontSize ||
+ sp < -minDupBreakOverlap * curWord->fontSize ||
+ lastCharOverlap) {
+ lastCharOverlap = gTrue;
+ endWord();
+ beginWord(state, x, y);
+ } else {
+ lastCharOverlap = gFalse;
+ }
+ } else {
+ lastCharOverlap = gFalse;
+ }
+
+ // page rotation and/or transform matrices can cause text to be
+ // drawn in reverse order -- in this case, swap the begin/end
+ // coordinates and break text into individual chars
+ if ((curWord->rot == 0 && w1 < 0) ||
+ (curWord->rot == 1 && h1 < 0) ||
+ (curWord->rot == 2 && w1 > 0) ||
+ (curWord->rot == 3 && h1 > 0)) {
+ endWord();
+ beginWord(state, x + dx, y + dy);
+ x1 += w1;
+ y1 += h1;
+ w1 = -w1;
+ h1 = -h1;
+ }
+
+ // add the characters to the current word
if (uLen != 0) {
w1 /= uLen;
h1 /= uLen;
}
for (i = 0; i < uLen; ++i) {
- curStr->addChar(state, x1 + i*w1, y1 + i*h1, w1, h1, u[i]);
+ curWord->addChar(state, x1 + i*w1, y1 + i*h1, w1, h1, u[i]);
}
+ ++curWord->charLen;
+ ++charPos;
}
-void TextPage::endString() {
+void TextPage::endWord() {
// This check is needed because Type 3 characters can contain
- // text-drawing operations.
+ // text-drawing operations (when TextPage is being used via
+ // {X,Win}SplashOutputDev rather than TextOutputDev).
if (nest > 0) {
--nest;
return;
}
- addString(curStr);
- curStr = NULL;
+ if (curWord) {
+ addWord(curWord);
+ curWord = NULL;
+ }
}
-void TextPage::addString(TextString *str) {
- TextString *p1, *p2;
-
- // throw away zero-length strings -- they don't have valid xMin/xMax
+void TextPage::addWord(TextWord *word) {
+ // throw away zero-length words -- they don't have valid xMin/xMax
// values, and they're useless anyway
- if (str->len == 0) {
- delete str;
+ if (word->len == 0) {
+ delete word;
return;
}
- // insert string in xy list
if (rawOrder) {
- p1 = xyCur1;
- p2 = NULL;
- } else if ((!xyCur1 || xyBefore(xyCur1, str)) &&
- (!xyCur2 || xyBefore(str, xyCur2))) {
- p1 = xyCur1;
- p2 = xyCur2;
- } else if (xyCur1 && xyBefore(xyCur1, str)) {
- for (p1 = xyCur1, p2 = xyCur2; p2; p1 = p2, p2 = p2->next) {
- if (xyBefore(str, p2)) {
- break;
- }
- }
- xyCur2 = p2;
- } else {
- for (p1 = NULL, p2 = xyStrings; p2; p1 = p2, p2 = p2->next) {
- if (xyBefore(str, p2)) {
- break;
- }
+ if (rawLastWord) {
+ rawLastWord->next = word;
+ } else {
+ rawWords = word;
}
- xyCur2 = p2;
- }
- xyCur1 = str;
- if (p1) {
- p1->next = str;
+ rawLastWord = word;
} else {
- xyStrings = str;
+ pools[word->rot]->addWord(word);
}
- str->next = p2;
}
-void TextPage::coalesce() {
- TextLine *line, *line0;
- TextBlock *yxBlocks, *xyBlocks, *blk, *blk0, *blk1, *blk2;
- TextString *str0, *str1, *str2, *str3, *str4;
- TextString *str1prev, *str2prev, *str3prev;
- TextOutColumnEdge *edges;
+void TextPage::coalesce(GBool physLayout) {
UnicodeMap *uMap;
- GBool isUnicode;
- char buf[8];
- int edgesLength, edgesSize;
- double x, yMin, yMax;
- double space, fit1, fit2, h;
- int col1, col2, d;
- int i, j;
+ TextPool *pool;
+ TextWord *word0, *word1, *word2;
+ TextLine *line;
+ TextBlock *blkList, *blkStack, *blk, *lastBlk, *blk0, *blk1;
+ TextBlock **blkArray;
+ TextFlow *flow, *lastFlow;
+ int rot, poolMinBaseIdx, baseIdx, startBaseIdx;
+ double minBase, maxBase, newMinBase, newMaxBase;
+ double fontSize, colSpace1, colSpace2, lineSpace, intraLineSpace, blkSpace;
+ GBool found;
+ int count[4];
+ int lrCount;
+ int firstBlkIdx, nBlocksLeft;
+ int col1, col2;
+ int i, j, n;
-#if 0 //~ for debugging
- for (str1 = xyStrings; str1; str1 = str1->next) {
- printf("x=%.2f..%.2f y=%.2f..%.2f size=%.2f '",
- str1->xMin, str1->xMax, str1->yMin, str1->yMax,
- (str1->yMax - str1->yMin));
- for (i = 0; i < str1->len; ++i) {
- fputc(str1->text[i] & 0xff, stdout);
+ if (rawOrder) {
+ primaryRot = 0;
+ primaryLR = gTrue;
+ return;
+ }
+
+ uMap = globalParams->getTextEncoding();
+ blkList = NULL;
+ lastBlk = NULL;
+ nBlocks = 0;
+ primaryRot = -1;
+
+#if 0 // for debugging
+ printf("*** initial words ***\n");
+ for (rot = 0; rot < 4; ++rot) {
+ pool = pools[rot];
+ for (baseIdx = pool->minBaseIdx; baseIdx <= pool->maxBaseIdx; ++baseIdx) {
+ for (word0 = pool->getPool(baseIdx); word0; word0 = word0->next) {
+ printf(" word: x=%.2f..%.2f y=%.2f..%.2f base=%.2f fontSize=%.2f '",
+ word0->xMin, word0->xMax, word0->yMin, word0->yMax,
+ word0->base, word0->fontSize);
+ for (i = 0; i < word0->len; ++i) {
+ fputc(word0->text[i] & 0xff, stdout);
+ }
+ printf("'\n");
+ }
}
- printf("'\n");
}
- printf("\n------------------------------------------------------------\n\n");
+ printf("\n");
#endif
- // build the list of column edges
- edges = NULL;
- edgesLength = edgesSize = 0;
- if (!rawOrder) {
- for (str1prev = NULL, str1 = xyStrings;
- str1;
- str1prev = str1, str1 = str1->next) {
- if (str1->marked) {
- continue;
- }
- h = str1->yMax - str1->yMin;
- if (str1prev && (str1->xMin - str1prev->xMax) / h < textOutColSpace) {
- continue;
+ //----- assemble the blocks
+
+ //~ add an outer loop for writing mode (vertical text)
+
+ // build blocks for each rotation value
+ for (rot = 0; rot < 4; ++rot) {
+ pool = pools[rot];
+ poolMinBaseIdx = pool->minBaseIdx;
+ count[rot] = 0;
+
+ // add blocks until no more words are left
+ while (1) {
+
+ // find the first non-empty line in the pool
+ for (;
+ poolMinBaseIdx <= pool->maxBaseIdx &&
+ !pool->getPool(poolMinBaseIdx);
+ ++poolMinBaseIdx) ;
+ if (poolMinBaseIdx > pool->maxBaseIdx) {
+ break;
}
- x = str1->xMin;
- yMin = str1->yMin;
- yMax = str1->yMax;
- for (str2prev = str1, str2 = str1->next;
- str2;
- str2prev = str2, str2 = str2->next) {
- h = str2->yMax - str2->yMin;
- if (!str2->marked &&
- (str2->xMin - str2prev->xMax) / h > textOutColSpace &&
- fabs(str2->xMin - x) < 0.5 &&
- str2->yMin - yMax < 0.3 * h &&
- yMin - str2->yMax < 0.3 * h) {
- break;
+
+ // look for the left-most word in the first four lines of the
+ // pool -- this avoids starting with a superscript word
+ startBaseIdx = poolMinBaseIdx;
+ for (baseIdx = poolMinBaseIdx + 1;
+ baseIdx < poolMinBaseIdx + 4 && baseIdx <= pool->maxBaseIdx;
+ ++baseIdx) {
+ if (!pool->getPool(baseIdx)) {
+ continue;
}
- }
- if (str2) {
- if (str2->yMin < yMin) {
- yMin = str2->yMin;
+ if (pool->getPool(baseIdx)->primaryCmp(pool->getPool(startBaseIdx))
+ < 0) {
+ startBaseIdx = baseIdx;
}
- if (str2->yMax > yMax) {
- yMax = str2->yMax;
+ }
+
+ // create a new block
+ word0 = pool->getPool(startBaseIdx);
+ pool->setPool(startBaseIdx, word0->next);
+ word0->next = NULL;
+ blk = new TextBlock(this, rot);
+ blk->addWord(word0);
+
+ fontSize = word0->fontSize;
+ minBase = maxBase = word0->base;
+ colSpace1 = minColSpacing1 * fontSize;
+ colSpace2 = minColSpacing2 * fontSize;
+ lineSpace = maxLineSpacingDelta * fontSize;
+ intraLineSpace = maxIntraLineDelta * fontSize;
+
+ // add words to the block
+ do {
+ found = gFalse;
+
+ // look for words on the line above the current top edge of
+ // the block
+ newMinBase = minBase;
+ for (baseIdx = pool->getBaseIdx(minBase);
+ baseIdx >= pool->getBaseIdx(minBase - lineSpace);
+ --baseIdx) {
+ word0 = NULL;
+ word1 = pool->getPool(baseIdx);
+ while (word1) {
+ if (word1->base < minBase &&
+ word1->base >= minBase - lineSpace &&
+ ((rot == 0 || rot == 2)
+ ? (word1->xMin < blk->xMax && word1->xMax > blk->xMin)
+ : (word1->yMin < blk->yMax && word1->yMax > blk->yMin)) &&
+ fabs(word1->fontSize - fontSize) <
+ maxBlockFontSizeDelta1 * fontSize) {
+ word2 = word1;
+ if (word0) {
+ word0->next = word1->next;
+ } else {
+ pool->setPool(baseIdx, word1->next);
+ }
+ word1 = word1->next;
+ word2->next = NULL;
+ blk->addWord(word2);
+ found = gTrue;
+ newMinBase = word2->base;
+ } else {
+ word0 = word1;
+ word1 = word1->next;
+ }
+ }
}
- str2->marked = gTrue;
- for (str3prev = str1, str3 = str1->next;
- str3;
- str3prev = str3, str3 = str3->next) {
- h = str3->yMax - str3->yMin;
- if (!str3->marked &&
- (str3->xMin - str3prev->xMax) / h > textOutColSpace &&
- fabs(str3->xMin - x) < 0.5 &&
- str3->yMin - yMax < 0.3 * h &&
- yMin - str3->yMax < 0.3 * h) {
- break;
+ minBase = newMinBase;
+
+ // look for words on the line below the current bottom edge of
+ // the block
+ newMaxBase = maxBase;
+ for (baseIdx = pool->getBaseIdx(maxBase);
+ baseIdx <= pool->getBaseIdx(maxBase + lineSpace);
+ ++baseIdx) {
+ word0 = NULL;
+ word1 = pool->getPool(baseIdx);
+ while (word1) {
+ if (word1->base > maxBase &&
+ word1->base <= maxBase + lineSpace &&
+ ((rot == 0 || rot == 2)
+ ? (word1->xMin < blk->xMax && word1->xMax > blk->xMin)
+ : (word1->yMin < blk->yMax && word1->yMax > blk->yMin)) &&
+ fabs(word1->fontSize - fontSize) <
+ maxBlockFontSizeDelta1 * fontSize) {
+ word2 = word1;
+ if (word0) {
+ word0->next = word1->next;
+ } else {
+ pool->setPool(baseIdx, word1->next);
+ }
+ word1 = word1->next;
+ word2->next = NULL;
+ blk->addWord(word2);
+ found = gTrue;
+ newMaxBase = word2->base;
+ } else {
+ word0 = word1;
+ word1 = word1->next;
+ }
}
}
- if (str3) {
- if (str3->yMin < yMin) {
- yMin = str3->yMin;
+ maxBase = newMaxBase;
+
+ // look for words that are on lines already in the block, and
+ // that overlap the block horizontally
+ for (baseIdx = pool->getBaseIdx(minBase - intraLineSpace);
+ baseIdx <= pool->getBaseIdx(maxBase + intraLineSpace);
+ ++baseIdx) {
+ word0 = NULL;
+ word1 = pool->getPool(baseIdx);
+ while (word1) {
+ if (word1->base >= minBase - intraLineSpace &&
+ word1->base <= maxBase + intraLineSpace &&
+ ((rot == 0 || rot == 2)
+ ? (word1->xMin < blk->xMax + colSpace1 &&
+ word1->xMax > blk->xMin - colSpace1)
+ : (word1->yMin < blk->yMax + colSpace1 &&
+ word1->yMax > blk->yMin - colSpace1)) &&
+ fabs(word1->fontSize - fontSize) <
+ maxBlockFontSizeDelta2 * fontSize) {
+ word2 = word1;
+ if (word0) {
+ word0->next = word1->next;
+ } else {
+ pool->setPool(baseIdx, word1->next);
+ }
+ word1 = word1->next;
+ word2->next = NULL;
+ blk->addWord(word2);
+ found = gTrue;
+ } else {
+ word0 = word1;
+ word1 = word1->next;
+ }
}
- if (str3->yMax > yMax) {
- yMax = str3->yMax;
+ }
+
+ // only check for outlying words (the next two chunks of code)
+ // if we didn't find anything else
+ if (found) {
+ continue;
+ }
+
+ // scan down the left side of the block, looking for words
+ // that are near (but not overlapping) the block; if there are
+ // three or fewer, add them to the block
+ n = 0;
+ for (baseIdx = pool->getBaseIdx(minBase - intraLineSpace);
+ baseIdx <= pool->getBaseIdx(maxBase + intraLineSpace);
+ ++baseIdx) {
+ word1 = pool->getPool(baseIdx);
+ while (word1) {
+ if (word1->base >= minBase - intraLineSpace &&
+ word1->base <= maxBase + intraLineSpace &&
+ ((rot == 0 || rot == 2)
+ ? (word1->xMax <= blk->xMin &&
+ word1->xMax > blk->xMin - colSpace2)
+ : (word1->yMax <= blk->yMin &&
+ word1->yMax > blk->yMin - colSpace2)) &&
+ fabs(word1->fontSize - fontSize) <
+ maxBlockFontSizeDelta3 * fontSize) {
+ ++n;
+ break;
+ }
+ word1 = word1->next;
}
- str3->marked = gTrue;
- do {
- for (str2prev = str1, str2 = str1->next;
- str2;
- str2prev = str2, str2 = str2->next) {
- h = str2->yMax - str2->yMin;
- if (!str2->marked &&
- (str2->xMin - str2prev->xMax) / h > textOutColSpace &&
- fabs(str2->xMin - x) < 0.5 &&
- str2->yMin - yMax < 0.3 * h &&
- yMin - str2->yMax < 0.3 * h) {
- if (str2->yMin < yMin) {
- yMin = str2->yMin;
+ }
+ if (n > 0 && n <= 3) {
+ for (baseIdx = pool->getBaseIdx(minBase - intraLineSpace);
+ baseIdx <= pool->getBaseIdx(maxBase + intraLineSpace);
+ ++baseIdx) {
+ word0 = NULL;
+ word1 = pool->getPool(baseIdx);
+ while (word1) {
+ if (word1->base >= minBase - intraLineSpace &&
+ word1->base <= maxBase + intraLineSpace &&
+ ((rot == 0 || rot == 2)
+ ? (word1->xMax <= blk->xMin &&
+ word1->xMax > blk->xMin - colSpace2)
+ : (word1->yMax <= blk->yMin &&
+ word1->yMax > blk->yMin - colSpace2)) &&
+ fabs(word1->fontSize - fontSize) <
+ maxBlockFontSizeDelta3 * fontSize) {
+ word2 = word1;
+ if (word0) {
+ word0->next = word1->next;
+ } else {
+ pool->setPool(baseIdx, word1->next);
}
- if (str2->yMax > yMax) {
- yMax = str2->yMax;
+ word1 = word1->next;
+ word2->next = NULL;
+ blk->addWord(word2);
+ if (word2->base < minBase) {
+ minBase = word2->base;
+ } else if (word2->base > maxBase) {
+ maxBase = word2->base;
}
- str2->marked = gTrue;
+ found = gTrue;
break;
+ } else {
+ word0 = word1;
+ word1 = word1->next;
}
}
- } while (str2);
- if (edgesLength == edgesSize) {
- edgesSize = edgesSize ? 2 * edgesSize : 16;
- edges = (TextOutColumnEdge *)
- grealloc(edges, edgesSize * sizeof(TextOutColumnEdge));
}
- edges[edgesLength].x = x;
- edges[edgesLength].y0 = yMin;
- edges[edgesLength].y1 = yMax;
- ++edgesLength;
- } else {
- str2->marked = gFalse;
}
- }
- str1->marked = gTrue;
- }
- }
-
-#if 0 //~ for debugging
- printf("column edges:\n");
- for (i = 0; i < edgesLength; ++i) {
- printf("%d: x=%.2f y0=%.2f y1=%.2f\n",
- i, edges[i].x, edges[i].y0, edges[i].y1);
- }
- printf("\n------------------------------------------------------------\n\n");
-#endif
- // build the blocks
- yxBlocks = NULL;
- blk1 = blk2 = NULL;
- while (xyStrings) {
-
- // build the block
- str0 = xyStrings;
- xyStrings = xyStrings->next;
- str0->next = NULL;
- blk = new TextBlock();
- blk->strings = str0;
- blk->xMin = str0->xMin;
- blk->xMax = str0->xMax;
- blk->yMin = str0->yMin;
- blk->yMax = str0->yMax;
- while (xyStrings) {
- str1 = NULL;
- str2 = xyStrings;
- fit1 = coalesceFit(str0, str2);
- if (!rawOrder) {
- // look for best-fitting string
- space = str0->yMax - str0->yMin;
- for (str3 = xyStrings, str4 = xyStrings->next;
- str4 && str4->xMin - str0->xMax <= space;
- str3 = str4, str4 = str4->next) {
- fit2 = coalesceFit(str0, str4);
- if (fit2 < fit1) {
- str1 = str3;
- str2 = str4;
- fit1 = fit2;
+ // scan down the right side of the block, looking for words
+ // that are near (but not overlapping) the block; if there are
+ // three or fewer, add them to the block
+ n = 0;
+ for (baseIdx = pool->getBaseIdx(minBase - intraLineSpace);
+ baseIdx <= pool->getBaseIdx(maxBase + intraLineSpace);
+ ++baseIdx) {
+ word1 = pool->getPool(baseIdx);
+ while (word1) {
+ if (word1->base >= minBase - intraLineSpace &&
+ word1->base <= maxBase + intraLineSpace &&
+ ((rot == 0 || rot == 2)
+ ? (word1->xMin >= blk->xMax &&
+ word1->xMin < blk->xMax + colSpace2)
+ : (word1->yMin >= blk->yMax &&
+ word1->yMin < blk->yMax + colSpace2)) &&
+ fabs(word1->fontSize - fontSize) <
+ maxBlockFontSizeDelta3 * fontSize) {
+ ++n;
+ break;
+ }
+ word1 = word1->next;
}
}
- }
- if (fit1 > 1) {
- // no fit - we're done with this block
- break;
- }
-
- // if we've hit a column edge we're done with this block
- if (fit1 > 0.2) {
- for (i = 0; i < edgesLength; ++i) {
- if (str0->xMax < edges[i].x + 0.5 && edges[i].x - 0.5 < str2->xMin &&
- str0->yMin < edges[i].y1 && str0->yMax > edges[i].y0 &&
- str2->yMin < edges[i].y1 && str2->yMax > edges[i].y0) {
- break;
+ if (n > 0 && n <= 3) {
+ for (baseIdx = pool->getBaseIdx(minBase - intraLineSpace);
+ baseIdx <= pool->getBaseIdx(maxBase + intraLineSpace);
+ ++baseIdx) {
+ word0 = NULL;
+ word1 = pool->getPool(baseIdx);
+ while (word1) {
+ if (word1->base >= minBase - intraLineSpace &&
+ word1->base <= maxBase + intraLineSpace &&
+ ((rot == 0 || rot == 2)
+ ? (word1->xMin >= blk->xMax &&
+ word1->xMin < blk->xMax + colSpace2)
+ : (word1->yMin >= blk->yMax &&
+ word1->yMin < blk->yMax + colSpace2)) &&
+ fabs(word1->fontSize - fontSize) <
+ maxBlockFontSizeDelta3 * fontSize) {
+ word2 = word1;
+ if (word0) {
+ word0->next = word1->next;
+ } else {
+ pool->setPool(baseIdx, word1->next);
+ }
+ word1 = word1->next;
+ word2->next = NULL;
+ blk->addWord(word2);
+ if (word2->base < minBase) {
+ minBase = word2->base;
+ } else if (word2->base > maxBase) {
+ maxBase = word2->base;
+ }
+ found = gTrue;
+ break;
+ } else {
+ word0 = word1;
+ word1 = word1->next;
+ }
+ }
}
}
- if (i < edgesLength) {
- break;
- }
- }
- if (str1) {
- str1->next = str2->next;
+ } while (found);
+
+ //~ need to compute the primary writing mode (horiz/vert) in
+ //~ addition to primary rotation
+
+ // coalesce the block, and add it to the list
+ blk->coalesce(uMap);
+ if (lastBlk) {
+ lastBlk->next = blk;
} else {
- xyStrings = str2->next;
+ blkList = blk;
}
- str0->next = str2;
- str2->next = NULL;
- if (str2->xMax > blk->xMax) {
- blk->xMax = str2->xMax;
+ lastBlk = blk;
+ count[rot] += blk->charCount;
+ if (primaryRot < 0 || count[rot] > count[primaryRot]) {
+ primaryRot = rot;
}
- if (str2->yMin < blk->yMin) {
- blk->yMin = str2->yMin;
- }
- if (str2->yMax > blk->yMax) {
- blk->yMax = str2->yMax;
- }
- str0 = str2;
- }
-
- // insert block on list
- if (!rawOrder) {
- // insert block on list in yx order
- for (blk1 = NULL, blk2 = yxBlocks;
- blk2 && !yxBefore(blk, blk2);
- blk1 = blk2, blk2 = blk2->next) ;
+ ++nBlocks;
}
- blk->next = blk2;
- if (blk1) {
- blk1->next = blk;
- } else {
- yxBlocks = blk;
- }
- blk1 = blk;
}
- gfree(edges);
-
- // the strings are now owned by the lines/blocks tree
- xyStrings = NULL;
+#if 0 // for debugging
+ printf("*** rotation ***\n");
+ for (rot = 0; rot < 4; ++rot) {
+ printf(" %d: %6d\n", rot, count[rot]);
+ }
+ printf(" primary rot = %d\n", primaryRot);
+ printf("\n");
+#endif
- // build the block text
- uMap = globalParams->getTextEncoding();
- isUnicode = uMap ? uMap->isUnicode() : gFalse;
- for (blk = yxBlocks; blk; blk = blk->next) {
- blk->len = 0;
- for (str1 = blk->strings; str1; str1 = str1->next) {
- blk->len += str1->len;
- if (str1->next && str1->next->xMin - str1->xMax >
- textOutSpace * (str1->yMax - str1->yMin)) {
- str1->spaceAfter = gTrue;
- ++blk->len;
- } else {
- str1->spaceAfter = gFalse;
- }
- }
- blk->text = (Unicode *)gmalloc(blk->len * sizeof(Unicode));
- blk->xRight = (double *)gmalloc(blk->len * sizeof(double));
- blk->col = (int *)gmalloc(blk->len * sizeof(int));
- i = 0;
- for (str1 = blk->strings; str1; str1 = str1->next) {
- for (j = 0; j < str1->len; ++j) {
- blk->text[i] = str1->text[j];
- blk->xRight[i] = str1->xRight[j];
- ++i;
- }
- if (str1->spaceAfter) {
- blk->text[i] = (Unicode)0x0020;
- blk->xRight[i] = str1->next->xMin;
- ++i;
- }
- }
- blk->convertedLen = 0;
- for (j = 0; j < blk->len; ++j) {
- blk->col[j] = blk->convertedLen;
- if (isUnicode) {
- ++blk->convertedLen;
- } else if (uMap) {
- blk->convertedLen += uMap->mapUnicode(blk->text[j], buf, sizeof(buf));
+#if 0 // for debugging
+ printf("*** blocks ***\n");
+ for (blk = blkList; blk; blk = blk->next) {
+ printf("block: rot=%d x=%.2f..%.2f y=%.2f..%.2f\n",
+ blk->rot, blk->xMin, blk->xMax, blk->yMin, blk->yMax);
+ for (line = blk->lines; line; line = line->next) {
+ printf(" line: x=%.2f..%.2f y=%.2f..%.2f base=%.2f\n",
+ line->xMin, line->xMax, line->yMin, line->yMax, line->base);
+ for (word0 = line->words; word0; word0 = word0->next) {
+ printf(" word: x=%.2f..%.2f y=%.2f..%.2f base=%.2f fontSize=%.2f space=%d: '",
+ word0->xMin, word0->xMax, word0->yMin, word0->yMax,
+ word0->base, word0->fontSize, word0->spaceAfter);
+ for (i = 0; i < word0->len; ++i) {
+ fputc(word0->text[i] & 0xff, stdout);
+ }
+ printf("'\n");
}
}
}
- if (uMap) {
- uMap->decRefCnt();
- }
+ printf("\n");
+#endif
-#if 0 //~ for debugging
- for (blk = yxBlocks; blk; blk = blk->next) {
- printf("[block: x=%.2f..%.2f y=%.2f..%.2f len=%d]\n",
- blk->xMin, blk->xMax, blk->yMin, blk->yMax, blk->len);
- TextString *str;
- for (str = blk->strings; str; str = str->next) {
- printf(" x=%.2f..%.2f y=%.2f..%.2f size=%.2f'",
- str->xMin, str->xMax, str->yMin, str->yMax,
- (str->yMax - str->yMin));
- for (i = 0; i < str->len; ++i) {
- fputc(str->text[i] & 0xff, stdout);
- }
- if (str->spaceAfter) {
- fputc(' ', stdout);
+ // determine the primary direction
+ lrCount = 0;
+ for (blk = blkList; blk; blk = blk->next) {
+ for (line = blk->lines; line; line = line->next) {
+ for (word0 = line->words; word0; word0 = word0->next) {
+ for (i = 0; i < word0->len; ++i) {
+ if (unicodeTypeL(word0->text[i])) {
+ ++lrCount;
+ } else if (unicodeTypeR(word0->text[i])) {
+ --lrCount;
+ }
+ }
}
- printf("'\n");
}
}
- printf("\n------------------------------------------------------------\n\n");
+ primaryLR = lrCount >= 0;
+
+#if 0 // for debugging
+ printf("*** direction ***\n");
+ printf("lrCount = %d\n", lrCount);
+ printf("primaryLR = %d\n", primaryLR);
#endif
- // build the lines
- lines = NULL;
- line0 = NULL;
- while (yxBlocks) {
- blk0 = yxBlocks;
- yxBlocks = yxBlocks->next;
- blk0->next = NULL;
- line = new TextLine();
- line->blocks = blk0;
- line->yMin = blk0->yMin;
- line->yMax = blk0->yMax;
- while (yxBlocks) {
-
- // remove duplicated text (fake boldface, shadowed text)
- h = blk0->yMax - blk0->yMin;
- if (yxBlocks->len == blk0->len &&
- !memcmp(yxBlocks->text, blk0->text,
- yxBlocks->len * sizeof(Unicode)) &&
- fabs(yxBlocks->yMin - blk0->yMin) / h < 0.2 &&
- fabs(yxBlocks->yMax - blk0->yMax) / h < 0.2 &&
- fabs(yxBlocks->xMin - blk0->xMin) / h < 0.2 &&
- fabs(yxBlocks->xMax - blk0->xMax) / h < 0.2) {
- blk1 = yxBlocks;
- yxBlocks = yxBlocks->next;
- delete blk1;
- continue;
- }
+ //----- column assignment
- if (rawOrder && yxBlocks->yMax < blk0->yMin) {
+ // sort blocks into xy order for column assignment
+ blocks = (TextBlock **)gmalloc(nBlocks * sizeof(TextBlock *));
+ for (blk = blkList, i = 0; blk; blk = blk->next, ++i) {
+ blocks[i] = blk;
+ }
+ qsort(blocks, nBlocks, sizeof(TextBlock *), &TextBlock::cmpXYPrimaryRot);
+
+ // column assignment
+ for (i = 0; i < nBlocks; ++i) {
+ blk0 = blocks[i];
+ col1 = 0;
+ for (j = 0; j < i; ++j) {
+ blk1 = blocks[j];
+ col2 = 0; // make gcc happy
+ switch (primaryRot) {
+ case 0:
+ if (blk0->xMin > blk1->xMax) {
+ col2 = blk1->col + blk1->nColumns + 3;
+ } else {
+ col2 = blk1->col + (int)(((blk0->xMin - blk1->xMin) /
+ (blk1->xMax - blk1->xMin)) *
+ blk1->nColumns);
+ }
break;
- }
- if (yxBlocks->yMin > 0.2*blk0->yMin + 0.8*blk0->yMax ||
- yxBlocks->xMin < blk0->xMax) {
+ case 1:
+ if (blk0->yMin > blk1->yMax) {
+ col2 = blk1->col + blk1->nColumns + 3;
+ } else {
+ col2 = blk1->col + (int)(((blk0->yMin - blk1->yMin) /
+ (blk1->yMax - blk1->yMin)) *
+ blk1->nColumns);
+ }
+ break;
+ case 2:
+ if (blk0->xMax < blk1->xMin) {
+ col2 = blk1->col + blk1->nColumns + 3;
+ } else {
+ col2 = blk1->col + (int)(((blk0->xMax - blk1->xMax) /
+ (blk1->xMin - blk1->xMax)) *
+ blk1->nColumns);
+ }
+ break;
+ case 3:
+ if (blk0->yMax < blk1->yMin) {
+ col2 = blk1->col + blk1->nColumns + 3;
+ } else {
+ col2 = blk1->col + (int)(((blk0->yMax - blk1->yMax) /
+ (blk1->yMin - blk1->yMax)) *
+ blk1->nColumns);
+ }
break;
}
- blk1 = yxBlocks;
- yxBlocks = yxBlocks->next;
- blk0->next = blk1;
- blk1->next = NULL;
- if (blk1->yMin < line->yMin) {
- line->yMin = blk1->yMin;
+ if (col2 > col1) {
+ col1 = col2;
}
- if (blk1->yMax > line->yMax) {
- line->yMax = blk1->yMax;
+ }
+ blk0->col = col1;
+ for (line = blk0->lines; line; line = line->next) {
+ for (j = 0; j <= line->len; ++j) {
+ line->col[j] += col1;
}
- blk0 = blk1;
}
- if (line0) {
- line0->next = line;
- } else {
- lines = line;
+ }
+
+#if 0 // for debugging
+ printf("*** blocks, after column assignment ***\n");
+ for (blk = blkList; blk; blk = blk->next) {
+ printf("block: rot=%d x=%.2f..%.2f y=%.2f..%.2f col=%d nCols=%d\n",
+ blk->rot, blk->xMin, blk->xMax, blk->yMin, blk->yMax, blk->col,
+ blk->nColumns);
+ for (line = blk->lines; line; line = line->next) {
+ printf(" line:\n");
+ for (word0 = line->words; word0; word0 = word0->next) {
+ printf(" word: x=%.2f..%.2f y=%.2f..%.2f base=%.2f fontSize=%.2f space=%d: '",
+ word0->xMin, word0->xMax, word0->yMin, word0->yMax,
+ word0->base, word0->fontSize, word0->spaceAfter);
+ for (i = 0; i < word0->len; ++i) {
+ fputc(word0->text[i] & 0xff, stdout);
+ }
+ printf("'\n");
+ }
}
- line->next = NULL;
- line0 = line;
}
+ printf("\n");
+#endif
+ //----- reading order sort
- // sort the blocks into xy order
- xyBlocks = NULL;
- for (line = lines; line; line = line->next) {
- for (blk = line->blocks; blk; blk = blk->next) {
- for (blk1 = NULL, blk2 = xyBlocks;
- blk2 && !xyBefore(blk, blk2);
- blk1 = blk2, blk2 = blk2->xyNext) ;
- blk->xyNext = blk2;
- if (blk1) {
- blk1->xyNext = blk;
- } else {
- xyBlocks = blk;
+ // sort blocks into yx order (in preparation for reading order sort)
+ qsort(blocks, nBlocks, sizeof(TextBlock *), &TextBlock::cmpYXPrimaryRot);
+
+ // compute space on left and right sides of each block
+ for (i = 0; i < nBlocks; ++i) {
+ blk0 = blocks[i];
+ for (j = 0; j < nBlocks; ++j) {
+ blk1 = blocks[j];
+ if (blk1 != blk0) {
+ blk0->updatePriMinMax(blk1);
}
}
}
-#if 0 //~ for debugging
- for (blk = xyBlocks; blk; blk = blk->xyNext) {
- printf("[block: x=%.2f..%.2f y=%.2f..%.2f len=%d]\n",
- blk->xMin, blk->xMax, blk->yMin, blk->yMax, blk->len);
- TextString *str;
- for (str = blk->strings; str; str = str->next) {
- printf(" x=%.2f..%.2f y=%.2f..%.2f size=%.2f '",
- str->xMin, str->xMax, str->yMin, str->yMax,
- (str->yMax - str->yMin));
- for (i = 0; i < str->len; ++i) {
- fputc(str->text[i] & 0xff, stdout);
+#if 0 // for debugging
+ printf("*** blocks, after yx sort ***\n");
+ for (i = 0; i < nBlocks; ++i) {
+ blk = blocks[i];
+ printf("block: rot=%d x=%.2f..%.2f y=%.2f..%.2f space=%.2f..%.2f\n",
+ blk->rot, blk->xMin, blk->xMax, blk->yMin, blk->yMax,
+ blk->priMin, blk->priMax);
+ for (line = blk->lines; line; line = line->next) {
+ printf(" line:\n");
+ for (word0 = line->words; word0; word0 = word0->next) {
+ printf(" word: x=%.2f..%.2f y=%.2f..%.2f base=%.2f fontSize=%.2f space=%d: '",
+ word0->xMin, word0->xMax, word0->yMin, word0->yMax,
+ word0->base, word0->fontSize, word0->spaceAfter);
+ for (j = 0; j < word0->len; ++j) {
+ fputc(word0->text[j] & 0xff, stdout);
+ }
+ printf("'\n");
}
- printf("'\n");
}
}
- printf("\n------------------------------------------------------------\n\n");
+ printf("\n");
#endif
- // do column assignment
- for (blk1 = xyBlocks; blk1; blk1 = blk1->xyNext) {
- col1 = 0;
- for (blk2 = xyBlocks; blk2 != blk1; blk2 = blk2->xyNext) {
- if (blk1->xMin >= blk2->xMax) {
- d = (int)((blk1->xMin - blk2->xMax) /
- (0.4 * (blk1->yMax - blk1->yMin)));
- if (d > 4) {
- d = 4;
- }
- col2 = blk2->col[0] + blk2->convertedLen + d;
- if (col2 > col1) {
- col1 = col2;
+ // build the flows
+ //~ this needs to be adjusted for writing mode (vertical text)
+ //~ this also needs to account for right-to-left column ordering
+ blkArray = (TextBlock **)gmalloc(nBlocks * sizeof(TextBlock *));
+ memcpy(blkArray, blocks, nBlocks * sizeof(TextBlock *));
+ flows = lastFlow = NULL;
+ firstBlkIdx = 0;
+ nBlocksLeft = nBlocks;
+ while (nBlocksLeft > 0) {
+
+ // find the upper-left-most block
+ for (; !blkArray[firstBlkIdx]; ++firstBlkIdx) ;
+ i = firstBlkIdx;
+ blk = blkArray[i];
+ for (j = firstBlkIdx + 1; j < nBlocks; ++j) {
+ blk1 = blkArray[j];
+ if (blk1) {
+ if (blk && blk->secondaryDelta(blk1) > 0) {
+ break;
}
- } else if (blk1->xMin > blk2->xMin) {
- for (i = 0; i < blk2->len && blk1->xMin >= blk2->xRight[i]; ++i) ;
- col2 = blk2->col[i];
- if (col2 > col1) {
- col1 = col2;
+ if (blk1->primaryCmp(blk) < 0) {
+ i = j;
+ blk = blk1;
}
}
}
- for (j = 0; j < blk1->len; ++j) {
- blk1->col[j] += col1;
+ blkArray[i] = NULL;
+ --nBlocksLeft;
+ blk->next = NULL;
+
+ // create a new flow, starting with the upper-left-most block
+ flow = new TextFlow(this, blk);
+ if (lastFlow) {
+ lastFlow->next = flow;
+ } else {
+ flows = flow;
}
- }
-
-#if 0 //~ for debugging
- for (line = lines; line; line = line->next) {
- printf("[line]\n");
- for (blk = line->blocks; blk; blk = blk->next) {
- printf("[block: col=%d, len=%d]\n", blk->col[0], blk->len);
- TextString *str;
- for (str = blk->strings; str; str = str->next) {
- printf(" x=%.2f..%.2f y=%.2f..%.2f size=%.2f '",
- str->xMin, str->xMax, str->yMin, str->yMax,
- (str->yMax - str->yMin));
- for (i = 0; i < str->len; ++i) {
- fputc(str->text[i] & 0xff, stdout);
+ lastFlow = flow;
+ fontSize = blk->lines->words->fontSize;
+
+ // push the upper-left-most block on the stack
+ blk->stackNext = NULL;
+ blkStack = blk;
+
+ // find the other blocks in this flow
+ while (blkStack) {
+
+ // find the upper-left-most block under (but within
+ // maxBlockSpacing of) the top block on the stack
+ blkSpace = maxBlockSpacing * blkStack->lines->words->fontSize;
+ blk = NULL;
+ i = -1;
+ for (j = firstBlkIdx; j < nBlocks; ++j) {
+ blk1 = blkArray[j];
+ if (blk1) {
+ if (blkStack->secondaryDelta(blk1) > blkSpace) {
+ break;
+ }
+ if (blk && blk->secondaryDelta(blk1) > 0) {
+ break;
+ }
+ if (blk1->isBelow(blkStack) &&
+ (!blk || blk1->primaryCmp(blk) < 0)) {
+ i = j;
+ blk = blk1;
+ }
}
- if (str->spaceAfter) {
- printf(" [space]\n");
+ }
+
+ // if a suitable block was found, add it to the flow and push it
+ // onto the stack
+ if (blk && flow->blockFits(blk, blkStack)) {
+ blkArray[i] = NULL;
+ --nBlocksLeft;
+ blk->next = NULL;
+ flow->addBlock(blk);
+ fontSize = blk->lines->words->fontSize;
+ blk->stackNext = blkStack;
+ blkStack = blk;
+
+ // otherwise (if there is no block under the top block or the
+ // block is not suitable), pop the stack
+ } else {
+ blkStack = blkStack->stackNext;
+ }
+ }
+ }
+ gfree(blkArray);
+
+#if 0 // for debugging
+ printf("*** flows ***\n");
+ for (flow = flows; flow; flow = flow->next) {
+ printf("flow: x=%.2f..%.2f y=%.2f..%.2f pri:%.2f..%.2f\n",
+ flow->xMin, flow->xMax, flow->yMin, flow->yMax,
+ flow->priMin, flow->priMax);
+ for (blk = flow->blocks; blk; blk = blk->next) {
+ printf(" block: rot=%d x=%.2f..%.2f y=%.2f..%.2f pri=%.2f..%.2f\n",
+ blk->rot, blk->xMin, blk->xMax, blk->yMin, blk->yMax,
+ blk->priMin, blk->priMax);
+ for (line = blk->lines; line; line = line->next) {
+ printf(" line:\n");
+ for (word0 = line->words; word0; word0 = word0->next) {
+ printf(" word: x=%.2f..%.2f y=%.2f..%.2f base=%.2f fontSize=%.2f space=%d: '",
+ word0->xMin, word0->xMax, word0->yMin, word0->yMax,
+ word0->base, word0->fontSize, word0->spaceAfter);
+ for (i = 0; i < word0->len; ++i) {
+ fputc(word0->text[i] & 0xff, stdout);
+ }
+ printf("'\n");
}
- printf("'\n");
}
}
}
- printf("\n------------------------------------------------------------\n\n");
+ printf("\n");
#endif
-}
+ if (uMap) {
+ uMap->decRefCnt();
+ }
+}
GBool TextPage::findText(Unicode *s, int len,
- GBool top, GBool bottom,
+ GBool startAtTop, GBool stopAtBottom,
+ GBool startAtLast, GBool stopAtLast,
double *xMin, double *yMin,
double *xMax, double *yMax) {
- TextLine *line;
TextBlock *blk;
+ TextLine *line;
Unicode *p;
Unicode u1, u2;
- int m, i, j;
- double x0, x1, x;
+ int m, i, j, k;
+ double xStart, yStart, xStop, yStop;
+ double xMin0, yMin0, xMax0, yMax0;
+ double xMin1, yMin1, xMax1, yMax1;
+ GBool found;
- // scan all blocks on page
- for (line = lines; line; line = line->next) {
- for (blk = line->blocks; blk; blk = blk->next) {
+ //~ needs to handle right-to-left text
- // check: above top limit?
- if (!top && (blk->yMax < *yMin ||
- (blk->yMin < *yMin && blk->xMax <= *xMin))) {
- continue;
- }
+ if (rawOrder) {
+ return gFalse;
+ }
- // check: below bottom limit?
- if (!bottom && (blk->yMin > *yMax ||
- (blk->yMax > *yMax && blk->xMin >= *xMax))) {
- return gFalse;
- }
+ xStart = yStart = xStop = yStop = 0;
+ if (startAtLast && haveLastFind) {
+ xStart = lastFindXMin;
+ yStart = lastFindYMin;
+ } else if (!startAtTop) {
+ xStart = *xMin;
+ yStart = *yMin;
+ }
+ if (stopAtLast && haveLastFind) {
+ xStop = lastFindXMin;
+ yStop = lastFindYMin;
+ } else if (!stopAtBottom) {
+ xStop = *xMax;
+ yStop = *yMax;
+ }
- // search each position in this block
- m = blk->len;
- for (i = 0, p = blk->text; i <= m - len; ++i, ++p) {
+ found = gFalse;
+ xMin0 = xMax0 = yMin0 = yMax0 = 0; // make gcc happy
+ xMin1 = xMax1 = yMin1 = yMax1 = 0; // make gcc happy
- x0 = (i == 0) ? blk->xMin : blk->xRight[i-1];
- x1 = blk->xRight[i];
- x = 0.5 * (x0 + x1);
+ for (i = 0; i < nBlocks; ++i) {
+ blk = blocks[i];
- // check: above top limit?
- if (!top && blk->yMin < *yMin) {
- if (x < *xMin) {
- continue;
- }
- }
+ // check: is the block above the top limit?
+ if (!startAtTop && blk->yMax < yStart) {
+ continue;
+ }
- // check: below bottom limit?
- if (!bottom && blk->yMax > *yMax) {
- if (x > *xMax) {
- return gFalse;
- }
- }
+ // check: is the block below the bottom limit?
+ if (!stopAtBottom && blk->yMin > yStop) {
+ break;
+ }
+
+ for (line = blk->lines; line; line = line->next) {
+
+ // check: is the line above the top limit?
+ if (!startAtTop && line->yMin < yStart) {
+ continue;
+ }
+
+ // check: is the line below the bottom limit?
+ if (!stopAtBottom && line->yMin > yStop) {
+ continue;
+ }
+
+ // search each position in this line
+ m = line->len;
+ for (j = 0, p = line->text; j <= m - len; ++j, ++p) {
// compare the strings
- for (j = 0; j < len; ++j) {
+ for (k = 0; k < len; ++k) {
#if 1 //~ this lowercases Latin A-Z only -- this will eventually be
- //~ extended to handle other character sets
- if (p[j] >= 0x41 && p[j] <= 0x5a) {
- u1 = p[j] + 0x20;
+ //~ extended to handle other character sets
+ if (p[k] >= 0x41 && p[k] <= 0x5a) {
+ u1 = p[k] + 0x20;
} else {
- u1 = p[j];
+ u1 = p[k];
}
- if (s[j] >= 0x41 && s[j] <= 0x5a) {
- u2 = s[j] + 0x20;
+ if (s[k] >= 0x41 && s[k] <= 0x5a) {
+ u2 = s[k] + 0x20;
} else {
- u2 = s[j];
+ u2 = s[k];
}
#endif
if (u1 != u2) {
}
// found it
- if (j == len) {
- *xMin = x0;
- *xMax = blk->xRight[i + len - 1];
- *yMin = blk->yMin;
- *yMax = blk->yMax;
- return gTrue;
+ if (k == len) {
+ switch (line->rot) {
+ case 0:
+ xMin1 = line->edge[j];
+ xMax1 = line->edge[j + len];
+ yMin1 = line->yMin;
+ yMax1 = line->yMax;
+ break;
+ case 1:
+ xMin1 = line->xMin;
+ xMax1 = line->xMax;
+ yMin1 = line->edge[j];
+ yMax1 = line->edge[j + len];
+ break;
+ case 2:
+ xMin1 = line->edge[j + len];
+ xMax1 = line->edge[j];
+ yMin1 = line->yMin;
+ yMax1 = line->yMax;
+ break;
+ case 3:
+ xMin1 = line->xMin;
+ xMax1 = line->xMax;
+ yMin1 = line->edge[j + len];
+ yMax1 = line->edge[j];
+ break;
+ }
+ if ((startAtTop ||
+ yMin1 > yStart || (yMin1 == yStart && xMin1 > xStart)) &&
+ (stopAtBottom ||
+ yMin1 < yStop || (yMin1 == yStop && xMin1 < yStop))) {
+ if (!found || yMin1 < yMin0 || (yMin1 == yMin0 && xMin1 < xMin0)) {
+ xMin0 = xMin1;
+ xMax0 = xMax1;
+ yMin0 = yMin1;
+ yMax0 = yMax1;
+ found = gTrue;
+ }
+ }
}
}
}
}
+ if (found) {
+ *xMin = xMin0;
+ *xMax = xMax0;
+ *yMin = yMin0;
+ *yMax = yMax0;
+ lastFindXMin = xMin0;
+ lastFindYMin = yMin0;
+ haveLastFind = gTrue;
+ return gTrue;
+ }
+
return gFalse;
}
GString *s;
UnicodeMap *uMap;
GBool isUnicode;
- char space[8], eol[16], buf[8];
- int spaceLen, eolLen, len;
- TextLine *line;
TextBlock *blk;
- double x0, x1, y;
- int firstCol, col, i;
- GBool multiLine;
+ TextLine *line;
+ TextLineFrag *frags;
+ int nFrags, fragsSize;
+ TextLineFrag *frag;
+ char space[8], eol[16];
+ int spaceLen, eolLen;
+ int lastRot;
+ double x, y;
+ int col, idx0, idx1, i, j;
+ GBool multiLine, oneRot;
s = new GString();
+ if (rawOrder) {
+ return s;
+ }
+
// get the output encoding
if (!(uMap = globalParams->getTextEncoding())) {
return s;
break;
}
- // find the leftmost column
- multiLine = gFalse;
- firstCol = -1;
- for (line = lines; line; line = line->next) {
- if (line->yMin > yMax) {
- break;
- }
- if (line->yMax < yMin) {
- continue;
- }
-
- for (blk = line->blocks; blk && blk->xMax < xMin; blk = blk->next) ;
- if (!blk || blk->xMin > xMax) {
- continue;
- }
-
- y = 0.5 * (blk->yMin + blk->yMax);
- if (y < yMin || y > yMax) {
- continue;
- }
-
- if (firstCol >= 0) {
- multiLine = gTrue;
- }
-
- i = 0;
- while (1) {
- x0 = (i==0) ? blk->xMin : blk->xRight[i-1];
- x1 = blk->xRight[i];
- if (0.5 * (x0 + x1) > xMin) {
- break;
+ //~ writing mode (horiz/vert)
+
+ // collect the line fragments that are in the rectangle
+ fragsSize = 256;
+ frags = (TextLineFrag *)gmalloc(fragsSize * sizeof(TextLineFrag));
+ nFrags = 0;
+ lastRot = -1;
+ oneRot = gTrue;
+ for (i = 0; i < nBlocks; ++i) {
+ blk = blocks[i];
+ if (xMin < blk->xMax && blk->xMin < xMax &&
+ yMin < blk->yMax && blk->yMin < yMax) {
+ for (line = blk->lines; line; line = line->next) {
+ if (xMin < line->xMax && line->xMin < xMax &&
+ yMin < line->yMax && line->yMin < yMax) {
+ idx0 = idx1 = -1;
+ switch (line->rot) {
+ case 0:
+ y = 0.5 * (line->yMin + line->yMax);
+ if (yMin < y && y < yMax) {
+ j = 0;
+ while (j < line->len) {
+ if (0.5 * (line->edge[j] + line->edge[j+1]) > xMin) {
+ idx0 = j;
+ break;
+ }
+ ++j;
+ }
+ j = line->len - 1;
+ while (j >= 0) {
+ if (0.5 * (line->edge[j] + line->edge[j+1]) < xMax) {
+ idx1 = j;
+ break;
+ }
+ --j;
+ }
+ }
+ break;
+ case 1:
+ x = 0.5 * (line->xMin + line->xMax);
+ if (xMin < x && x < xMax) {
+ j = 0;
+ while (j < line->len) {
+ if (0.5 * (line->edge[j] + line->edge[j+1]) > yMin) {
+ idx0 = j;
+ break;
+ }
+ ++j;
+ }
+ j = line->len - 1;
+ while (j >= 0) {
+ if (0.5 * (line->edge[j] + line->edge[j+1]) < yMax) {
+ idx1 = j;
+ break;
+ }
+ --j;
+ }
+ }
+ break;
+ case 2:
+ y = 0.5 * (line->yMin + line->yMax);
+ if (yMin < y && y < yMax) {
+ j = 0;
+ while (j < line->len) {
+ if (0.5 * (line->edge[j] + line->edge[j+1]) < xMax) {
+ idx0 = j;
+ break;
+ }
+ ++j;
+ }
+ j = line->len - 1;
+ while (j >= 0) {
+ if (0.5 * (line->edge[j] + line->edge[j+1]) > xMin) {
+ idx1 = j;
+ break;
+ }
+ --j;
+ }
+ }
+ break;
+ case 3:
+ x = 0.5 * (line->xMin + line->xMax);
+ if (xMin < x && x < xMax) {
+ j = 0;
+ while (j < line->len) {
+ if (0.5 * (line->edge[j] + line->edge[j+1]) < yMax) {
+ idx0 = j;
+ break;
+ }
+ ++j;
+ }
+ j = line->len - 1;
+ while (j >= 0) {
+ if (0.5 * (line->edge[j] + line->edge[j+1]) > yMin) {
+ idx1 = j;
+ break;
+ }
+ --j;
+ }
+ }
+ break;
+ }
+ if (idx0 >= 0 && idx1 >= 0) {
+ if (nFrags == fragsSize) {
+ fragsSize *= 2;
+ frags = (TextLineFrag *)
+ grealloc(frags, fragsSize * sizeof(TextLineFrag));
+ }
+ frags[nFrags].init(line, idx0, idx1 - idx0 + 1);
+ ++nFrags;
+ if (lastRot >= 0 && line->rot != lastRot) {
+ oneRot = gFalse;
+ }
+ lastRot = line->rot;
+ }
+ }
}
- ++i;
- }
- col = blk->col[i];
-
- if (firstCol < 0 || col < firstCol) {
- firstCol = col;
}
}
- // extract the text
- for (line = lines; line; line = line->next) {
- if (line->yMin > yMax) {
- break;
- }
- if (line->yMax < yMin) {
- continue;
- }
+ // sort the fragments and generate the string
+ if (nFrags > 0) {
- for (blk = line->blocks; blk && blk->xMax < xMin; blk = blk->next) ;
- if (!blk || blk->xMin > xMax) {
- continue;
+ for (i = 0; i < nFrags; ++i) {
+ frags[i].computeCoords(oneRot);
}
+ assignColumns(frags, nFrags, oneRot);
- y = 0.5 * (blk->yMin + blk->yMax);
- if (y < yMin || y > yMax) {
- continue;
+ // if all lines in the region have the same rotation, use it;
+ // otherwise, use the page's primary rotation
+ if (oneRot) {
+ qsort(frags, nFrags, sizeof(TextLineFrag),
+ &TextLineFrag::cmpYXLineRot);
+ } else {
+ qsort(frags, nFrags, sizeof(TextLineFrag),
+ &TextLineFrag::cmpYXPrimaryRot);
}
- i = 0;
- while (1) {
- x0 = (i==0) ? blk->xMin : blk->xRight[i-1];
- x1 = blk->xRight[i];
- if (0.5 * (x0 + x1) > xMin) {
- break;
+ col = 0;
+ multiLine = gFalse;
+ for (i = 0; i < nFrags; ++i) {
+ frag = &frags[i];
+
+ // insert a return
+ if (frag->col < col ||
+ (i > 0 && fabs(frag->base - frags[i-1].base) >
+ maxIntraLineDelta * frags[i-1].line->words->fontSize)) {
+ s->append(eol, eolLen);
+ col = 0;
+ multiLine = gTrue;
}
- ++i;
- }
-
- col = firstCol;
- do {
-
- // line this block up with the correct column
- for (; col < blk->col[i]; ++col) {
+ // column alignment
+ for (; col < frag->col; ++col) {
s->append(space, spaceLen);
}
- // print the block
- for (; i < blk->len; ++i) {
-
- x0 = (i==0) ? blk->xMin : blk->xRight[i-1];
- x1 = blk->xRight[i];
- if (0.5 * (x0 + x1) > xMax) {
- break;
- }
-
- len = uMap->mapUnicode(blk->text[i], buf, sizeof(buf));
- s->append(buf, len);
- col += isUnicode ? 1 : len;
- }
- if (i < blk->len) {
- break;
- }
-
- // next block
- blk = blk->next;
- i = 0;
-
- } while (blk && blk->xMin < xMax);
+ // get the fragment text
+ col += dumpFragment(frag->line->text + frag->start, frag->len, uMap, s);
+ }
if (multiLine) {
s->append(eol, eolLen);
}
}
+ gfree(frags);
uMap->decRefCnt();
return s;
}
-void TextPage::dump(void *outputStream, TextOutputFunc outputFunc) {
- UnicodeMap *uMap;
- char space[8], eol[16], eop[8], buf[8];
- int spaceLen, eolLen, eopLen, len;
+GBool TextPage::findCharRange(int pos, int length,
+ double *xMin, double *yMin,
+ double *xMax, double *yMax) {
+ TextBlock *blk;
TextLine *line;
+ TextWord *word;
+ double xMin0, xMax0, yMin0, yMax0;
+ double xMin1, xMax1, yMin1, yMax1;
+ GBool first;
+ int i, j0, j1;
+
+ if (rawOrder) {
+ return gFalse;
+ }
+
+ //~ this doesn't correctly handle:
+ //~ - ranges split across multiple lines (the highlighted region
+ //~ is the bounding box of all the parts of the range)
+ //~ - cases where characters don't convert one-to-one into Unicode
+ first = gTrue;
+ xMin0 = xMax0 = yMin0 = yMax0 = 0; // make gcc happy
+ xMin1 = xMax1 = yMin1 = yMax1 = 0; // make gcc happy
+ for (i = 0; i < nBlocks; ++i) {
+ blk = blocks[i];
+ for (line = blk->lines; line; line = line->next) {
+ for (word = line->words; word; word = word->next) {
+ if (pos < word->charPos + word->charLen &&
+ word->charPos < pos + length) {
+ j0 = pos - word->charPos;
+ if (j0 < 0) {
+ j0 = 0;
+ }
+ j1 = pos + length - 1 - word->charPos;
+ if (j1 >= word->len) {
+ j1 = word->len - 1;
+ }
+ switch (line->rot) {
+ case 0:
+ xMin1 = word->edge[j0];
+ xMax1 = word->edge[j1 + 1];
+ yMin1 = word->yMin;
+ yMax1 = word->yMax;
+ break;
+ case 1:
+ xMin1 = word->xMin;
+ xMax1 = word->xMax;
+ yMin1 = word->edge[j0];
+ yMax1 = word->edge[j1 + 1];
+ break;
+ case 2:
+ xMin1 = word->edge[j1 + 1];
+ xMax1 = word->edge[j0];
+ yMin1 = word->yMin;
+ yMax1 = word->yMax;
+ break;
+ case 3:
+ xMin1 = word->xMin;
+ xMax1 = word->xMax;
+ yMin1 = word->edge[j1 + 1];
+ yMax1 = word->edge[j0];
+ break;
+ }
+ if (first || xMin1 < xMin0) {
+ xMin0 = xMin1;
+ }
+ if (first || xMax1 > xMax0) {
+ xMax0 = xMax1;
+ }
+ if (first || yMin1 < yMin0) {
+ yMin0 = yMin1;
+ }
+ if (first || yMax1 > yMax0) {
+ yMax0 = yMax1;
+ }
+ first = gFalse;
+ }
+ }
+ }
+ }
+ if (!first) {
+ *xMin = xMin0;
+ *xMax = xMax0;
+ *yMin = yMin0;
+ *yMax = yMax0;
+ return gTrue;
+ }
+ return gFalse;
+}
+
+void TextPage::dump(void *outputStream, TextOutputFunc outputFunc,
+ GBool physLayout) {
+ UnicodeMap *uMap;
+ TextFlow *flow;
TextBlock *blk;
- int col, d, i;
+ TextLine *line;
+ TextLineFrag *frags;
+ TextWord *word;
+ int nFrags, fragsSize;
+ TextLineFrag *frag;
+ char space[8], eol[16], eop[8];
+ int spaceLen, eolLen, eopLen;
+ GBool pageBreaks;
+ GString *s;
+ int col, i, d, n;
// get the output encoding
if (!(uMap = globalParams->getTextEncoding())) {
break;
}
eopLen = uMap->mapUnicode(0x0c, eop, sizeof(eop));
+ pageBreaks = globalParams->getTextPageBreaks();
- // output
- for (line = lines; line; line = line->next) {
- col = 0;
- for (blk = line->blocks; blk; blk = blk->next) {
+ //~ writing mode (horiz/vert)
- // line this block up with the correct column
- if (rawOrder && col == 0) {
- col = blk->col[0];
- } else {
- for (; col < blk->col[0]; ++col) {
+ // output the page in raw (content stream) order
+ if (rawOrder) {
+
+ for (word = rawWords; word; word = word->next) {
+ s = new GString();
+ dumpFragment(word->text, word->len, uMap, s);
+ (*outputFunc)(outputStream, s->getCString(), s->getLength());
+ delete s;
+ if (word->next &&
+ fabs(word->next->base - word->base) <
+ maxIntraLineDelta * word->fontSize) {
+ if (word->next->xMin > word->xMax + minWordSpacing * word->fontSize) {
(*outputFunc)(outputStream, space, spaceLen);
}
+ } else {
+ (*outputFunc)(outputStream, eol, eolLen);
}
+ }
- // print the block
- for (i = 0; i < blk->len; ++i) {
- len = uMap->mapUnicode(blk->text[i], buf, sizeof(buf));
- (*outputFunc)(outputStream, buf, len);
+ // output the page, maintaining the original physical layout
+ } else if (physLayout) {
+
+ // collect the line fragments for the page and sort them
+ fragsSize = 256;
+ frags = (TextLineFrag *)gmalloc(fragsSize * sizeof(TextLineFrag));
+ nFrags = 0;
+ for (i = 0; i < nBlocks; ++i) {
+ blk = blocks[i];
+ for (line = blk->lines; line; line = line->next) {
+ if (nFrags == fragsSize) {
+ fragsSize *= 2;
+ frags = (TextLineFrag *)grealloc(frags,
+ fragsSize * sizeof(TextLineFrag));
+ }
+ frags[nFrags].init(line, 0, line->len);
+ frags[nFrags].computeCoords(gTrue);
+ ++nFrags;
}
- col += blk->convertedLen;
}
+ qsort(frags, nFrags, sizeof(TextLineFrag), &TextLineFrag::cmpYXPrimaryRot);
- // print a return
- (*outputFunc)(outputStream, eol, eolLen);
+ // generate output
+ col = 0;
+ for (i = 0; i < nFrags; ++i) {
+ frag = &frags[i];
- // print extra vertical space if necessary
- if (line->next) {
- d = (int)((line->next->yMin - line->yMax) /
- (line->blocks->strings->yMax - lines->blocks->strings->yMin)
- + 0.5);
- // various things (weird font matrices) can result in bogus
- // values here, so do a sanity check
- if (rawOrder && d > 2) {
- d = 2;
- } else if (!rawOrder && d > 5) {
- d = 5;
+ // column alignment
+ for (; col < frag->col; ++col) {
+ (*outputFunc)(outputStream, space, spaceLen);
}
- for (; d > 0; --d) {
- (*outputFunc)(outputStream, eol, eolLen);
+
+ // print the line
+ s = new GString();
+ col += dumpFragment(frag->line->text + frag->start, frag->len, uMap, s);
+ (*outputFunc)(outputStream, s->getCString(), s->getLength());
+ delete s;
+
+ // print one or more returns if necessary
+ if (i == nFrags - 1 ||
+ frags[i+1].col < col ||
+ fabs(frags[i+1].base - frag->base) >
+ maxIntraLineDelta * frag->line->words->fontSize) {
+ if (i < nFrags - 1) {
+ d = (int)((frags[i+1].base - frag->base) /
+ frag->line->words->fontSize);
+ if (d < 1) {
+ d = 1;
+ } else if (d > 5) {
+ d = 5;
+ }
+ } else {
+ d = 1;
+ }
+ for (; d > 0; --d) {
+ (*outputFunc)(outputStream, eol, eolLen);
+ }
+ col = 0;
+ }
+ }
+
+ gfree(frags);
+
+ // output the page, "undoing" the layout
+ } else {
+ for (flow = flows; flow; flow = flow->next) {
+ for (blk = flow->blocks; blk; blk = blk->next) {
+ for (line = blk->lines; line; line = line->next) {
+ n = line->len;
+ if (line->hyphenated && (line->next || blk->next)) {
+ --n;
+ }
+ s = new GString();
+ dumpFragment(line->text, n, uMap, s);
+ (*outputFunc)(outputStream, s->getCString(), s->getLength());
+ delete s;
+ if (!line->hyphenated) {
+ if (line->next) {
+ (*outputFunc)(outputStream, space, spaceLen);
+ } else if (blk->next) {
+ //~ this is a bit of a kludge - we should really do a more
+ //~ intelligent determination of paragraphs
+ if (blk->next->lines->words->fontSize ==
+ blk->lines->words->fontSize) {
+ (*outputFunc)(outputStream, space, spaceLen);
+ } else {
+ (*outputFunc)(outputStream, eol, eolLen);
+ }
+ }
+ }
+ }
}
+ (*outputFunc)(outputStream, eol, eolLen);
+ (*outputFunc)(outputStream, eol, eolLen);
}
}
// end of page
- (*outputFunc)(outputStream, eol, eolLen);
- (*outputFunc)(outputStream, eop, eopLen);
- (*outputFunc)(outputStream, eol, eolLen);
+ if (pageBreaks) {
+ (*outputFunc)(outputStream, eop, eopLen);
+ (*outputFunc)(outputStream, eol, eolLen);
+ }
uMap->decRefCnt();
}
-// Returns true if <str1> should be inserted before <str2> in xy
-// order.
-GBool TextPage::xyBefore(TextString *str1, TextString *str2) {
- return str1->xMin < str2->xMin ||
- (str1->xMin == str2->xMin && str1->yMin < str2->yMin);
-}
+void TextPage::assignColumns(TextLineFrag *frags, int nFrags, GBool oneRot) {
+ TextLineFrag *frag0, *frag1;
+ int rot, col1, col2, i, j, k;
+
+ // all text in the region has the same rotation -- recompute the
+ // column numbers based only on the text in the region
+ if (oneRot) {
+ qsort(frags, nFrags, sizeof(TextLineFrag), &TextLineFrag::cmpXYLineRot);
+ rot = frags[0].line->rot;
+ for (i = 0; i < nFrags; ++i) {
+ frag0 = &frags[i];
+ col1 = 0;
+ for (j = 0; j < i; ++j) {
+ frag1 = &frags[j];
+ col2 = 0; // make gcc happy
+ switch (rot) {
+ case 0:
+ if (frag0->xMin >= frag1->xMax) {
+ col2 = frag1->col + (frag1->line->col[frag1->start + frag1->len] -
+ frag1->line->col[frag1->start]) + 1;
+ } else {
+ for (k = frag1->start;
+ k < frag1->start + frag1->len &&
+ frag0->xMin >= 0.5 * (frag1->line->edge[k] +
+ frag1->line->edge[k+1]);
+ ++k) ;
+ col2 = frag1->col +
+ frag1->line->col[k] - frag1->line->col[frag1->start];
+ }
+ break;
+ case 1:
+ if (frag0->yMin >= frag1->yMax) {
+ col2 = frag1->col + (frag1->line->col[frag1->start + frag1->len] -
+ frag1->line->col[frag1->start]) + 1;
+ } else {
+ for (k = frag1->start;
+ k < frag1->start + frag1->len &&
+ frag0->yMin >= 0.5 * (frag1->line->edge[k] +
+ frag1->line->edge[k+1]);
+ ++k) ;
+ col2 = frag1->col +
+ frag1->line->col[k] - frag1->line->col[frag1->start];
+ }
+ break;
+ case 2:
+ if (frag0->xMax <= frag1->xMin) {
+ col2 = frag1->col + (frag1->line->col[frag1->start + frag1->len] -
+ frag1->line->col[frag1->start]) + 1;
+ } else {
+ for (k = frag1->start;
+ k < frag1->start + frag1->len &&
+ frag0->xMax <= 0.5 * (frag1->line->edge[k] +
+ frag1->line->edge[k+1]);
+ ++k) ;
+ col2 = frag1->col +
+ frag1->line->col[k] - frag1->line->col[frag1->start];
+ }
+ break;
+ case 3:
+ if (frag0->yMax <= frag1->yMin) {
+ col2 = frag1->col + (frag1->line->col[frag1->start + frag1->len] -
+ frag1->line->col[frag1->start]) + 1;
+ } else {
+ for (k = frag1->start;
+ k < frag1->start + frag1->len &&
+ frag0->yMax <= 0.5 * (frag1->line->edge[k] +
+ frag1->line->edge[k+1]);
+ ++k) ;
+ col2 = frag1->col +
+ frag1->line->col[k] - frag1->line->col[frag1->start];
+ }
+ break;
+ }
+ if (col2 > col1) {
+ col1 = col2;
+ }
+ }
+ frag0->col = col1;
+ }
-// Returns true if <blk1> should be inserted before <blk2> in xy
-// order.
-GBool TextPage::xyBefore(TextBlock *blk1, TextBlock *blk2) {
- return blk1->xMin < blk2->xMin ||
- (blk1->xMin == blk2->xMin && blk1->yMin < blk2->yMin);
+ // the region includes text at different rotations -- use the
+ // globally assigned column numbers, offset by the minimum column
+ // number (i.e., shift everything over to column 0)
+ } else {
+ col1 = frags[0].col;
+ for (i = 1; i < nFrags; ++i) {
+ if (frags[i].col < col1) {
+ col1 = frags[i].col;
+ }
+ }
+ for (i = 0; i < nFrags; ++i) {
+ frags[i].col -= col1;
+ }
+ }
}
-// Returns true if <blk1> should be inserted before <blk2> in yx
-// order, allowing a little slack for vertically overlapping text.
-GBool TextPage::yxBefore(TextBlock *blk1, TextBlock *blk2) {
- double h1, h2, overlap;
+int TextPage::dumpFragment(Unicode *text, int len, UnicodeMap *uMap,
+ GString *s) {
+ char lre[8], rle[8], popdf[8], buf[8];
+ int lreLen, rleLen, popdfLen, n;
+ int nCols, i, j, k;
- h1 = blk1->yMax - blk1->yMin;
- h2 = blk2->yMax - blk2->yMin;
- overlap = ((blk1->yMax < blk2->yMax ? blk1->yMax : blk2->yMax) -
- (blk1->yMin > blk2->yMin ? blk1->yMin : blk2->yMin)) /
- (h1 < h2 ? h1 : h2);
- if (overlap > 0.6) {
- return blk1->xMin < blk2->xMin;
- }
- return blk1->yMin < blk2->yMin;
-}
+ nCols = 0;
-double TextPage::coalesceFit(TextString *str1, TextString *str2) {
- double h1, h2, w1, w2, r, overlap, spacing;
+ if (uMap->isUnicode()) {
- h1 = str1->yMax - str1->yMin;
- h2 = str2->yMax - str2->yMin;
- w1 = str1->xMax - str1->xMin;
- w2 = str2->xMax - str2->xMin;
- r = h1 / h2;
- if (r < (1.0 / 3.0) || r > 3) {
- return 10;
- }
- overlap = ((str1->yMax < str2->yMax ? str1->yMax : str2->yMax) -
- (str1->yMin > str2->yMin ? str1->yMin : str2->yMin)) /
- (h1 < h2 ? h1 : h2);
- if (overlap < 0.5) {
- return 10;
- }
- spacing = (str2->xMin - str1->xMax) / (h1 > h2 ? h1 : h2);
- if (spacing < -0.5) {
- return 10;
- }
- // separate text that overlaps - duplicated text (so that fake
- // boldface and shadowed text can be cleanly removed)
- if ((str2->xMin - str1->xMax) / (w1 < w2 ? w1 : w2) < -0.7) {
- return 10;
- }
- return spacing;
-}
+ lreLen = uMap->mapUnicode(0x202a, lre, sizeof(lre));
+ rleLen = uMap->mapUnicode(0x202b, rle, sizeof(rle));
+ popdfLen = uMap->mapUnicode(0x202c, popdf, sizeof(popdf));
-void TextPage::clear() {
- TextLine *p1, *p2;
- TextString *s1, *s2;
+ if (primaryLR) {
+
+ i = 0;
+ while (i < len) {
+ // output a left-to-right section
+ for (j = i; j < len && !unicodeTypeR(text[j]); ++j) ;
+ for (k = i; k < j; ++k) {
+ n = uMap->mapUnicode(text[k], buf, sizeof(buf));
+ s->append(buf, n);
+ ++nCols;
+ }
+ i = j;
+ // output a right-to-left section
+ for (j = i; j < len && !unicodeTypeL(text[j]); ++j) ;
+ if (j > i) {
+ s->append(rle, rleLen);
+ for (k = j - 1; k >= i; --k) {
+ n = uMap->mapUnicode(text[k], buf, sizeof(buf));
+ s->append(buf, n);
+ ++nCols;
+ }
+ s->append(popdf, popdfLen);
+ i = j;
+ }
+ }
+
+ } else {
+
+ s->append(rle, rleLen);
+ i = len - 1;
+ while (i >= 0) {
+ // output a right-to-left section
+ for (j = i; j >= 0 && !unicodeTypeL(text[j]); --j) ;
+ for (k = i; k > j; --k) {
+ n = uMap->mapUnicode(text[k], buf, sizeof(buf));
+ s->append(buf, n);
+ ++nCols;
+ }
+ i = j;
+ // output a left-to-right section
+ for (j = i; j >= 0 && !unicodeTypeR(text[j]); --j) ;
+ if (j < i) {
+ s->append(lre, lreLen);
+ for (k = j + 1; k <= i; ++k) {
+ n = uMap->mapUnicode(text[k], buf, sizeof(buf));
+ s->append(buf, n);
+ ++nCols;
+ }
+ s->append(popdf, popdfLen);
+ i = j;
+ }
+ }
+ s->append(popdf, popdfLen);
- if (curStr) {
- delete curStr;
- curStr = NULL;
- }
- if (lines) {
- for (p1 = lines; p1; p1 = p2) {
- p2 = p1->next;
- delete p1;
}
- } else if (xyStrings) {
- for (s1 = xyStrings; s1; s1 = s2) {
- s2 = s1->next;
- delete s1;
+
+ } else {
+ for (i = 0; i < len; ++i) {
+ n = uMap->mapUnicode(text[i], buf, sizeof(buf));
+ s->append(buf, n);
+ nCols += n;
}
}
- xyStrings = NULL;
- xyCur1 = xyCur2 = NULL;
- lines = NULL;
- nest = 0;
- nTinyChars = 0;
+
+ return nCols;
}
+#if TEXTOUT_WORD_LIST
+TextWordList *TextPage::makeWordList(GBool physLayout) {
+ return new TextWordList(this, physLayout);
+}
+#endif
+
//------------------------------------------------------------------------
// TextOutputDev
//------------------------------------------------------------------------
fwrite(text, 1, len, (FILE *)stream);
}
-TextOutputDev::TextOutputDev(char *fileName, GBool rawOrderA, GBool append) {
+TextOutputDev::TextOutputDev(char *fileName, GBool physLayoutA,
+ GBool rawOrderA, GBool append) {
text = NULL;
+ physLayout = physLayoutA;
rawOrder = rawOrderA;
ok = gTrue;
if (fileName) {
if (!strcmp(fileName, "-")) {
outputStream = stdout;
+#ifdef WIN32
+ // keep DOS from munging the end-of-line characters
+ setmode(fileno(stdout), O_BINARY);
+#endif
} else if ((outputStream = fopen(fileName, append ? "ab" : "wb"))) {
needClose = gTrue;
} else {
}
// set up text object
- text = new TextPage(rawOrder);
+ text = new TextPage(rawOrderA);
}
TextOutputDev::TextOutputDev(TextOutputFunc func, void *stream,
- GBool rawOrderA) {
+ GBool physLayoutA, GBool rawOrderA) {
outputFunc = func;
outputStream = stream;
needClose = gFalse;
+ physLayout = physLayoutA;
rawOrder = rawOrderA;
- text = new TextPage(rawOrder);
+ text = new TextPage(rawOrderA);
ok = gTrue;
}
}
void TextOutputDev::startPage(int pageNum, GfxState *state) {
- text->clear();
+ text->startPage(state);
}
void TextOutputDev::endPage() {
- text->coalesce();
+ text->endPage();
+ text->coalesce(physLayout);
if (outputStream) {
- text->dump(outputStream, outputFunc);
+ text->dump(outputStream, outputFunc, physLayout);
}
}
}
void TextOutputDev::beginString(GfxState *state, GString *s) {
- text->beginString(state, state->getCurX(), state->getCurY());
}
void TextOutputDev::endString(GfxState *state) {
- text->endString();
}
void TextOutputDev::drawChar(GfxState *state, double x, double y,
double dx, double dy,
double originX, double originY,
CharCode c, Unicode *u, int uLen) {
- text->addChar(state, x, y, dx, dy, u, uLen);
+ text->addChar(state, x, y, dx, dy, c, u, uLen);
}
GBool TextOutputDev::findText(Unicode *s, int len,
- GBool top, GBool bottom,
+ GBool startAtTop, GBool stopAtBottom,
+ GBool startAtLast, GBool stopAtLast,
double *xMin, double *yMin,
double *xMax, double *yMax) {
- return text->findText(s, len, top, bottom, xMin, yMin, xMax, yMax);
+ return text->findText(s, len, startAtTop, stopAtBottom,
+ startAtLast, stopAtLast, xMin, yMin, xMax, yMax);
}
GString *TextOutputDev::getText(double xMin, double yMin,
return text->getText(xMin, yMin, xMax, yMax);
}
+GBool TextOutputDev::findCharRange(int pos, int length,
+ double *xMin, double *yMin,
+ double *xMax, double *yMax) {
+ return text->findCharRange(pos, length, xMin, yMin, xMax, yMax);
+}
+
+#if TEXTOUT_WORD_LIST
+TextWordList *TextOutputDev::makeWordList() {
+ return text->makeWordList(physLayout);
+}
+#endif