blob: 442ace2dbad40adc5333cbceb31e2eafe852e224 [file] [log] [blame]
//========================================================================
//
// TextOutputDev.cc
//
// Copyright 1997-2003 Glyph & Cog, LLC
//
//========================================================================
//========================================================================
//
// Modified under the Poppler project - http://poppler.freedesktop.org
//
// Copyright (C) 2005-2007 Kristian Høgsberg <krh@redhat.com>
// Copyright (C) 2005 Nickolay V. Shmyrev <nshmyrev@yandex.ru>
// Copyright (C) 2006-2008 Carlos Garcia Campos <carlosgc@gnome.org>
// Copyright (C) 2006, 2007 Ed Catmur <ed@catmur.co.uk>
// Copyright (C) 2006 Jeff Muizelaar <jeff@infidigm.net>
// Copyright (C) 2007, 2008 Adrian Johnson <ajohnson@redneon.com>
// Copyright (C) 2008 Koji Otani <sho@bbr.jp>
// Copyright (C) 2008 Albert Astals Cid <aacid@kde.org>
// Copyright (C) 2008 Pino Toscano <pino@kde.org>
// Copyright (C) 2008 Hib Eris <hib@hiberis.nl>
// Copyright (C) 2009 Ross Moore <ross@maths.mq.edu.au>
// Copyright (C) 2009 Kovid Goyal <kovid@kovidgoyal.net>
//
// To see a description of the changes please see the Changelog file that
// came with your tarball or type make ChangeLog if you are building from git
//
//========================================================================
#include <config.h>
#ifdef USE_GCC_PRAGMAS
#pragma implementation
#endif
#include <stdio.h>
#include <stdlib.h>
#include <stddef.h>
#include <math.h>
#include <ctype.h>
#ifdef _WIN32
#include <fcntl.h> // for O_BINARY
#include <io.h> // for setmode
#endif
#include "goo/gmem.h"
#include "goo/GooString.h"
#include "goo/GooList.h"
#include "poppler-config.h"
#include "Error.h"
#include "GlobalParams.h"
#include "UnicodeMap.h"
#include "UnicodeTypeTable.h"
#include "Link.h"
#include "TextOutputDev.h"
#include "Page.h"
#include "PDFDocEncoding.h"
#ifdef MACOS
// needed for setting type/creator of MacOS files
#include "ICSupport.h"
#endif
//------------------------------------------------------------------------
// parameters
//------------------------------------------------------------------------
// 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
// Upper limit on spacing between characters in a word.
#define maxWideCharSpacing 0.4
// 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
// Max width of underlines (in points).
#define maxUnderlineWidth 3
// Min distance between baseline and underline (in points).
//~ this should be font-size-dependent
#define minUnderlineGap -2
// Max distance between baseline and underline (in points).
//~ this should be font-size-dependent
#define maxUnderlineGap 4
// Max horizontal distance between edge of word and start of underline
// (in points).
//~ this should be font-size-dependent
#define underlineSlack 1
// Max distance between edge of text and edge of link border
#define hyperlinkSlack 2
//------------------------------------------------------------------------
// TextUnderline
//------------------------------------------------------------------------
class TextUnderline {
public:
TextUnderline(double x0A, double y0A, double x1A, double y1A)
{ x0 = x0A; y0 = y0A; x1 = x1A; y1 = y1A; horiz = y0 == y1; }
~TextUnderline() {}
double x0, y0, x1, y1;
GBool horiz;
};
//------------------------------------------------------------------------
// TextLink
//------------------------------------------------------------------------
class TextLink {
public:
TextLink(int xMinA, int yMinA, int xMaxA, int yMaxA, Link *linkA)
{ xMin = xMinA; yMin = yMinA; xMax = xMaxA; yMax = yMaxA; link = linkA; }
~TextLink() {}
int xMin, yMin, xMax, yMax;
Link *link;
};
//------------------------------------------------------------------------
// TextFontInfo
//------------------------------------------------------------------------
TextFontInfo::TextFontInfo(GfxState *state) {
gfxFont = state->getFont();
if (gfxFont)
gfxFont->incRefCnt();
#if TEXTOUT_WORD_LIST
fontName = (gfxFont && gfxFont->getOrigName())
? gfxFont->getOrigName()->copy()
: (GooString *)NULL;
flags = gfxFont ? gfxFont->getFlags() : 0;
#endif
}
TextFontInfo::~TextFontInfo() {
if (gfxFont)
gfxFont->decRefCnt();
#if TEXTOUT_WORD_LIST
if (fontName) {
delete fontName;
}
#endif
}
GBool TextFontInfo::matches(GfxState *state) {
return state->getFont() == gfxFont;
}
//------------------------------------------------------------------------
// TextWord
//------------------------------------------------------------------------
TextWord::TextWord(GfxState *state, int rotA, double x0, double y0,
int charPosA, TextFontInfo *fontA, double fontSizeA) {
GfxFont *gfxFont;
double x, y, ascent, descent;
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;
charcode = NULL;
edge = NULL;
len = size = 0;
spaceAfter = gFalse;
next = NULL;
#if TEXTOUT_WORD_LIST
GfxRGB rgb;
if ((state->getRender() & 3) == 1) {
state->getStrokeRGB(&rgb);
} else {
state->getFillRGB(&rgb);
}
colorR = colToDbl(rgb.r);
colorG = colToDbl(rgb.g);
colorB = colToDbl(rgb.b);
#endif
underlined = gFalse;
link = NULL;
}
TextWord::~TextWord() {
gfree(text);
gfree(charcode);
gfree(edge);
}
void TextWord::addChar(GfxState *state, double x, double y,
double dx, double dy, CharCode c, Unicode u) {
if (len == size) {
size += 16;
text = (Unicode *)greallocn(text, size, sizeof(Unicode));
charcode = (Unicode *)greallocn(charcode, size, sizeof(CharCode));
edge = (double *)greallocn(edge, (size + 1), sizeof(double));
}
text[len] = u;
charcode[len] = c;
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 *)greallocn(text, size, sizeof(Unicode));
charcode = (CharCode *)greallocn(charcode, (size + 1), sizeof(CharCode));
edge = (double *)greallocn(edge, (size + 1), sizeof(double));
}
for (i = 0; i < word->len; ++i) {
text[len + i] = word->text[i];
charcode[len + i] = word->charcode[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
GooString *TextWord::getText() {
GooString *s;
UnicodeMap *uMap;
char buf[8];
int n, i;
s = new GooString();
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;
}
void TextWord::getCharBBox(int charIdx, double *xMinA, double *yMinA,
double *xMaxA, double *yMaxA) {
if (charIdx < 0 || charIdx >= len) {
return;
}
switch (rot) {
case 0:
*xMinA = edge[charIdx];
*xMaxA = edge[charIdx + 1];
*yMinA = yMin;
*yMaxA = yMax;
break;
case 1:
*xMinA = xMin;
*xMaxA = xMax;
*yMinA = edge[charIdx];
*yMaxA = edge[charIdx + 1];
break;
case 2:
*xMinA = edge[charIdx + 1];
*xMaxA = edge[charIdx];
*yMinA = yMin;
*yMaxA = yMax;
break;
case 3:
*xMinA = xMin;
*xMaxA = xMax;
*yMinA = edge[charIdx + 1];
*yMaxA = edge[charIdx];
break;
}
}
#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 **)gmallocn(maxBaseIdx - minBaseIdx + 1,
sizeof(TextWord *));
for (baseIdx = minBaseIdx; baseIdx <= maxBaseIdx; ++baseIdx) {
pool[baseIdx - minBaseIdx] = NULL;
}
} else if (wordBaseIdx < minBaseIdx) {
newMinBaseIdx = wordBaseIdx - 128;
newPool = (TextWord **)gmallocn(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 **)greallocn(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;
base = baseA;
words = lastWord = NULL;
text = NULL;
edge = NULL;
col = NULL;
len = 0;
convertedLen = 0;
hyphenated = gFalse;
next = NULL;
xMin = yMin = 0;
xMax = yMax = -1;
normalized = NULL;
normalized_len = 0;
normalized_idx = NULL;
}
TextLine::~TextLine() {
TextWord *word;
while (words) {
word = words;
words = words->next;
delete word;
}
gfree(text);
gfree(edge);
gfree(col);
if (normalized) {
gfree(normalized);
gfree(normalized_idx);
}
}
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;
if (space > maxWideCharSpacing * words->fontSize) {
space = maxWideCharSpacing * words->fontSize;
}
}
// 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 &&
word0->underlined == word1->underlined &&
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 *)gmallocn(len, sizeof(Unicode));
edge = (double *)gmallocn(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 *)gmallocn(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)'-';
}
//------------------------------------------------------------------------
// TextLineFrag
//------------------------------------------------------------------------
class TextLineFrag {
public:
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);
static int cmpXYColumnPrimaryRot(const void *p1, const void *p2);
static int cmpXYColumnLineRot(const void *p1, const void *p2);
};
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 (fabs(cmp = frag1->yMin - frag2->yMin) < 0.01) {
cmp = frag1->xMin - frag2->xMin;
}
break;
case 1:
if (fabs(cmp = frag2->xMax - frag1->xMax) < 0.01) {
cmp = frag1->yMin - frag2->yMin;
}
break;
case 2:
if (fabs(cmp = frag2->yMin - frag1->yMin) < 0.01) {
cmp = frag2->xMax - frag1->xMax;
}
break;
case 3:
if (fabs(cmp = frag1->xMax - frag2->xMax) < 0.01) {
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;
}
int TextLineFrag::cmpXYColumnPrimaryRot(const void *p1, const void *p2) {
TextLineFrag *frag1 = (TextLineFrag *)p1;
TextLineFrag *frag2 = (TextLineFrag *)p2;
double cmp;
// if columns overlap, compare y values
if (frag1->col < frag2->col + (frag2->line->col[frag2->start + frag2->len] -
frag2->line->col[frag2->start]) &&
frag2->col < frag1->col + (frag1->line->col[frag1->start + frag1->len] -
frag1->line->col[frag1->start])) {
cmp = 0; // make gcc happy
switch (frag1->line->blk->page->primaryRot) {
case 0: cmp = frag1->yMin - frag2->yMin; break;
case 1: cmp = frag2->xMax - frag1->xMax; break;
case 2: cmp = frag2->yMin - frag1->yMin; break;
case 3: cmp = frag1->xMax - frag2->xMax; break;
}
return cmp < 0 ? -1 : cmp > 0 ? 1 : 0;
}
// otherwise, compare starting column
return frag1->col - frag2->col;
}
int TextLineFrag::cmpXYColumnLineRot(const void *p1, const void *p2) {
TextLineFrag *frag1 = (TextLineFrag *)p1;
TextLineFrag *frag2 = (TextLineFrag *)p2;
double cmp;
// if columns overlap, compare y values
if (frag1->col < frag2->col + (frag2->line->col[frag2->start + frag2->len] -
frag2->line->col[frag2->start]) &&
frag2->col < frag1->col + (frag1->line->col[frag1->start + frag1->len] -
frag1->line->col[frag1->start])) {
cmp = 0; // make gcc happy
switch (frag1->line->rot) {
case 0: cmp = frag1->yMin - frag2->yMin; break;
case 1: cmp = frag2->xMax - frag1->xMax; break;
case 2: cmp = frag2->yMin - frag1->yMin; break;
case 3: cmp = frag1->xMax - frag2->xMax; break;
}
return cmp < 0 ? -1 : cmp > 0 ? 1 : 0;
}
// otherwise, compare starting column
return frag1->col - frag2->col;
}
//------------------------------------------------------------------------
// 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 **)gmallocn(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;
}
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;
// 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
//------------------------------------------------------------------------
// TextWordList
//------------------------------------------------------------------------
TextWordList::TextWordList(TextPage *text, GBool physLayout) {
TextFlow *flow;
TextBlock *blk;
TextLine *line;
TextWord *word;
TextWord **wordArray;
int nWords, i;
words = new GooList();
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 **)gmallocn(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);
} else {
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);
}
}
}
}
}
}
TextWordList::~TextWordList() {
delete words;
}
int TextWordList::getLength() {
return words->getLength();
}
TextWord *TextWordList::get(int idx) {
if (idx < 0 || idx >= words->getLength()) {
return NULL;
}
return (TextWord *)words->get(idx);
}
#endif // TEXTOUT_WORD_LIST
//------------------------------------------------------------------------
// TextPage
//------------------------------------------------------------------------
TextPage::TextPage(GBool rawOrderA) {
int rot;
refCnt = 1;
rawOrder = rawOrderA;
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 GooList();
lastFindXMin = lastFindYMin = 0;
haveLastFind = gFalse;
underlines = new GooList();
links = new GooList();
}
TextPage::~TextPage() {
int rot;
clear();
if (!rawOrder) {
for (rot = 0; rot < 4; ++rot) {
delete pools[rot];
}
}
delete fonts;
deleteGooList(underlines, TextUnderline);
deleteGooList(links, TextLink);
}
void TextPage::incRefCnt() {
refCnt++;
}
void TextPage::decRefCnt() {
if (--refCnt == 0)
delete this;
}
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);
}
deleteGooList(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 GooList();
}
void TextPage::updateFont(GfxState *state) {
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
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
// rendering the font. This code tries to guess by looking at the
// width of the character 'm' (which breaks if the font is a
// subset that doesn't contain 'm').
mCode = letterCode = anyCode = -1;
for (code = 0; code < 256; ++code) {
name = ((Gfx8BitFont *)gfxFont)->getCharName(code);
if (name && name[0] == 'm' && name[1] == '\0') {
mCode = code;
}
if (letterCode < 0 && name && name[1] == '\0' &&
((name[0] >= 'A' && name[0] <= 'Z') ||
(name[0] >= 'a' && name[0] <= 'z'))) {
letterCode = code;
}
if (anyCode < 0 && name &&
((Gfx8BitFont *)gfxFont)->getWidth(code) > 0) {
anyCode = code;
}
}
if (mCode >= 0 &&
(w = ((Gfx8BitFont *)gfxFont)->getWidth(mCode)) > 0) {
// 0.6 is a generic average 'm' width -- yes, this is a hack
curFontSize *= w / 0.6;
} else if (letterCode >= 0 &&
(w = ((Gfx8BitFont *)gfxFont)->getWidth(letterCode)) > 0) {
// even more of a hack: 0.5 is a generic letter width
curFontSize *= w / 0.5;
} else if (anyCode >= 0 &&
(w = ((Gfx8BitFont *)gfxFont)->getWidth(anyCode)) > 0) {
// better than nothing: 0.5 is a generic character width
curFontSize *= w / 0.5;
}
fm = gfxFont->getFontMatrix();
if (fm[0] != 0) {
curFontSize *= fabs(fm[3] / fm[0]);
}
}
}
void TextPage::beginWord(GfxState *state, double x0, double y0) {
GfxFont *gfxFont;
double *fontm;
double m[4], m2[4];
int rot;
// This check is needed because Type 3 characters can contain
// text-drawing operations (when TextPage is being used via
// {X,Win}SplashOutputDev rather than TextOutputDev).
if (curWord) {
++nest;
return;
}
// compute the rotation
state->getFontTransMat(&m[0], &m[1], &m[2], &m[3]);
gfxFont = state->getFont();
if (gfxFont && gfxFont->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,
CharCode c, int nBytes, Unicode *u, int uLen) {
double x1, y1, w1, h1, dx2, dy2, base, sp, delta;
GBool overlap;
int i;
// 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);
// throw away chars that aren't inside the page bounds
// (and also do a sanity check on the character size)
state->transform(x, y, &x1, &y1);
if (x1 + w1 < 0 || x1 > pageWidth ||
y1 + h1 < 0 || y1 > pageHeight ||
w1 > pageWidth || h1 > pageHeight) {
charPos += nBytes;
return;
}
// check the tiny chars limit
if (!globalParams->getTextKeepTinyChars() &&
fabs(w1) < 3 && fabs(h1) < 3) {
if (++nTinyChars > 50000) {
charPos += nBytes;
return;
}
}
// break words at space character
if (uLen == 1 && u[0] == (Unicode)0x20) {
if (curWord) {
++curWord->charLen;
}
charPos += nBytes;
endWord();
return;
}
// start a new word if:
// (1) this character doesn't fall in the right place relative to
// the end of the previous word (this places upper and lower
// constraints on the position deltas along both the primary
// and secondary axes), or
// (2) this character overlaps the previous one (duplicated text), or
// (3) the previous character was an overlap (we want each duplicated
// character to be in a word by itself at this stage),
// (4) the font size has changed
if (curWord && curWord->len > 0) {
base = sp = delta = 0; // make gcc happy
switch (curWord->rot) {
case 0:
base = y1;
sp = x1 - curWord->xMax;
delta = x1 - curWord->edge[curWord->len - 1];
break;
case 1:
base = x1;
sp = y1 - curWord->yMax;
delta = y1 - curWord->edge[curWord->len - 1];
break;
case 2:
base = y1;
sp = curWord->xMin - x1;
delta = curWord->edge[curWord->len - 1] - x1;
break;
case 3:
base = x1;
sp = curWord->yMin - y1;
delta = curWord->edge[curWord->len - 1] - y1;
break;
}
overlap = fabs(delta) < dupMaxPriDelta * curWord->fontSize &&
fabs(base - curWord->base) < dupMaxSecDelta * curWord->fontSize;
if (overlap || lastCharOverlap ||
sp < -minDupBreakOverlap * curWord->fontSize ||
sp > minWordBreakSpace * curWord->fontSize ||
fabs(base - curWord->base) > 0.5 ||
curFontSize != curWord->fontSize) {
endWord();
}
lastCharOverlap = overlap;
} else {
lastCharOverlap = gFalse;
}
if (uLen != 0) {
// start a new word if needed
if (!curWord) {
beginWord(state, x, y);
}
// 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
w1 /= uLen;
h1 /= uLen;
for (i = 0; i < uLen; ++i) {
if (u[i] >= 0xd800 && u[i] < 0xdc00) { /* surrogate pair */
if (i + 1 < uLen && u[i+1] >= 0xdc00 && u[i+1] < 0xe000) {
/* next code is a low surrogate */
Unicode uu = (((u[i] & 0x3ff) << 10) | (u[i+1] & 0x3ff)) + 0x10000;
i++;
curWord->addChar(state, x1 + i*w1, y1 + i*h1, w1, h1, c, uu);
} else {
/* missing low surrogate
replace it with REPLACEMENT CHARACTER (U+FFFD) */
curWord->addChar(state, x1 + i*w1, y1 + i*h1, w1, h1, c, 0xfffd);
}
} else if (u[i] >= 0xdc00 && u[i] < 0xe000) {
/* invalid low surrogate
replace it with REPLACEMENT CHARACTER (U+FFFD) */
curWord->addChar(state, x1 + i*w1, y1 + i*h1, w1, h1, c, 0xfffd);
} else {
curWord->addChar(state, x1 + i*w1, y1 + i*h1, w1, h1, c, u[i]);
}
}
}
if (curWord) {
curWord->charLen += nBytes;
}
charPos += nBytes;
}
void TextPage::endWord() {
// This check is needed because Type 3 characters can contain
// text-drawing operations (when TextPage is being used via
// {X,Win}SplashOutputDev rather than TextOutputDev).
if (nest > 0) {
--nest;
return;
}
if (curWord) {
addWord(curWord);
curWord = NULL;
}
}
void TextPage::addWord(TextWord *word) {
// throw away zero-length words -- they don't have valid xMin/xMax
// values, and they're useless anyway
if (word->len == 0) {
delete word;
return;
}
if (rawOrder) {
if (rawLastWord) {
rawLastWord->next = word;
} else {
rawWords = word;
}
rawLastWord = word;
} else {
pools[word->rot]->addWord(word);
}
}
void TextPage::addUnderline(double x0, double y0, double x1, double y1) {
underlines->append(new TextUnderline(x0, y0, x1, y1));
}
void TextPage::addLink(int xMin, int yMin, int xMax, int yMax, Link *link) {
links->append(new TextLink(xMin, yMin, xMax, yMax, link));
}
void TextPage::coalesce(GBool physLayout, GBool doHTML) {
UnicodeMap *uMap;
TextPool *pool;
TextWord *word0, *word1, *word2;
TextLine *line;
TextBlock *blkList, *blkStack, *blk, *lastBlk, *blk0, *blk1;
TextBlock **blkArray;
TextFlow *flow, *lastFlow;
TextUnderline *underline;
TextLink *link;
int rot, poolMinBaseIdx, baseIdx, startBaseIdx, endBaseIdx;
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 (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 rot=%d link=%p '",
word0->xMin, word0->xMax, word0->yMin, word0->yMax,
word0->base, word0->fontSize, rot*90, word0->link);
for (i = 0; i < word0->len; ++i) {
fputc(word0->text[i] & 0xff, stdout);
}
printf("'\n");
}
}
}
printf("\n");
#endif
#if 0 //~ for debugging
for (i = 0; i < underlines->getLength(); ++i) {
underline = (TextUnderline *)underlines->get(i);
printf("underline: x=%g..%g y=%g..%g horiz=%d\n",
underline->x0, underline->x1, underline->y0, underline->y1,
underline->horiz);
}
#endif
if (doHTML) {
//----- handle underlining
for (i = 0; i < underlines->getLength(); ++i) {
underline = (TextUnderline *)underlines->get(i);
if (underline->horiz) {
// rot = 0
if (pools[0]->minBaseIdx <= pools[0]->maxBaseIdx) {
startBaseIdx = pools[0]->getBaseIdx(underline->y0 + minUnderlineGap);
endBaseIdx = pools[0]->getBaseIdx(underline->y0 + maxUnderlineGap);
for (j = startBaseIdx; j <= endBaseIdx; ++j) {
for (word0 = pools[0]->getPool(j); word0; word0 = word0->next) {
//~ need to check the y value against the word baseline
if (underline->x0 < word0->xMin + underlineSlack &&
word0->xMax - underlineSlack < underline->x1) {
word0->underlined = gTrue;
}
}
}
}
// rot = 2
if (pools[2]->minBaseIdx <= pools[2]->maxBaseIdx) {
startBaseIdx = pools[2]->getBaseIdx(underline->y0 - maxUnderlineGap);
endBaseIdx = pools[2]->getBaseIdx(underline->y0 - minUnderlineGap);
for (j = startBaseIdx; j <= endBaseIdx; ++j) {
for (word0 = pools[2]->getPool(j); word0; word0 = word0->next) {
if (underline->x0 < word0->xMin + underlineSlack &&
word0->xMax - underlineSlack < underline->x1) {
word0->underlined = gTrue;
}
}
}
}
} else {
// rot = 1
if (pools[1]->minBaseIdx <= pools[1]->maxBaseIdx) {
startBaseIdx = pools[1]->getBaseIdx(underline->x0 - maxUnderlineGap);
endBaseIdx = pools[1]->getBaseIdx(underline->x0 - minUnderlineGap);
for (j = startBaseIdx; j <= endBaseIdx; ++j) {
for (word0 = pools[1]->getPool(j); word0; word0 = word0->next) {
if (underline->y0 < word0->yMin + underlineSlack &&
word0->yMax - underlineSlack < underline->y1) {
word0->underlined = gTrue;
}
}
}
}
// rot = 3
if (pools[3]->minBaseIdx <= pools[3]->maxBaseIdx) {
startBaseIdx = pools[3]->getBaseIdx(underline->x0 + minUnderlineGap);
endBaseIdx = pools[3]->getBaseIdx(underline->x0 + maxUnderlineGap);
for (j = startBaseIdx; j <= endBaseIdx; ++j) {
for (word0 = pools[3]->getPool(j); word0; word0 = word0->next) {
if (underline->y0 < word0->yMin + underlineSlack &&
word0->yMax - underlineSlack < underline->y1) {
word0->underlined = gTrue;
}
}
}
}
}
}
//----- handle links
for (i = 0; i < links->getLength(); ++i) {
link = (TextLink *)links->get(i);
// rot = 0
if (pools[0]->minBaseIdx <= pools[0]->maxBaseIdx) {
startBaseIdx = pools[0]->getBaseIdx(link->yMin);
endBaseIdx = pools[0]->getBaseIdx(link->yMax);
for (j = startBaseIdx; j <= endBaseIdx; ++j) {
for (word0 = pools[0]->getPool(j); word0; word0 = word0->next) {
if (link->xMin < word0->xMin + hyperlinkSlack &&
word0->xMax - hyperlinkSlack < link->xMax &&
link->yMin < word0->yMin + hyperlinkSlack &&
word0->yMax - hyperlinkSlack < link->yMax) {
word0->link = link->link;
}
}
}
}
// rot = 2
if (pools[2]->minBaseIdx <= pools[2]->maxBaseIdx) {
startBaseIdx = pools[2]->getBaseIdx(link->yMin);
endBaseIdx = pools[2]->getBaseIdx(link->yMax);
for (j = startBaseIdx; j <= endBaseIdx; ++j) {
for (word0 = pools[2]->getPool(j); word0; word0 = word0->next) {
if (link->xMin < word0->xMin + hyperlinkSlack &&
word0->xMax - hyperlinkSlack < link->xMax &&
link->yMin < word0->yMin + hyperlinkSlack &&
word0->yMax - hyperlinkSlack < link->yMax) {
word0->link = link->link;
}
}
}
}
// rot = 1
if (pools[1]->minBaseIdx <= pools[1]->maxBaseIdx) {
startBaseIdx = pools[1]->getBaseIdx(link->xMin);
endBaseIdx = pools[1]->getBaseIdx(link->xMax);
for (j = startBaseIdx; j <= endBaseIdx; ++j) {
for (word0 = pools[1]->getPool(j); word0; word0 = word0->next) {
if (link->yMin < word0->yMin + hyperlinkSlack &&
word0->yMax - hyperlinkSlack < link->yMax &&
link->xMin < word0->xMin + hyperlinkSlack &&
word0->xMax - hyperlinkSlack < link->xMax) {
word0->link = link->link;
}
}
}
}
// rot = 3
if (pools[3]->minBaseIdx <= pools[3]->maxBaseIdx) {
startBaseIdx = pools[3]->getBaseIdx(link->xMin);
endBaseIdx = pools[3]->getBaseIdx(link->xMax);
for (j = startBaseIdx; j <= endBaseIdx; ++j) {
for (word0 = pools[3]->getPool(j); word0; word0 = word0->next) {
if (link->yMin < word0->yMin + hyperlinkSlack &&
word0->yMax - hyperlinkSlack < link->yMax &&
link->xMin < word0->xMin + hyperlinkSlack &&
word0->xMax - hyperlinkSlack < link->xMax) {
word0->link = link->link;
}
}
}
}
}
}
//----- 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;
}
// 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 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;
}
}
}
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);