1 //========================================================================
5 // Copyright 1996-2003 Glyph & Cog, LLC
7 //========================================================================
11 #ifdef USE_GCC_PRAGMAS
12 #pragma implementation
23 #include "GlobalParams.h"
25 #include "CharCodeToUnicode.h"
26 #include "FontEncodingTables.h"
27 #include "BuiltinFontTables.h"
28 #include "FoFiType1.h"
29 #include "FoFiType1C.h"
30 #include "FoFiTrueType.h"
33 //------------------------------------------------------------------------
35 struct StdFontMapEntry {
40 // Acrobat 4.0 and earlier substituted Base14-compatible fonts without
41 // providing Widths and a FontDescriptor, so we munge the names into
42 // the proper Base14 names. This table is from implementation note 44
43 // in the PDF 1.4 spec, with some additions based on empirical
45 static StdFontMapEntry stdFontMap[] = {
46 { "Arial", "Helvetica" },
47 { "Arial,Bold", "Helvetica-Bold" },
48 { "Arial,BoldItalic", "Helvetica-BoldOblique" },
49 { "Arial,Italic", "Helvetica-Oblique" },
50 { "Arial-Bold", "Helvetica-Bold" },
51 { "Arial-BoldItalic", "Helvetica-BoldOblique" },
52 { "Arial-BoldItalicMT", "Helvetica-BoldOblique" },
53 { "Arial-BoldMT", "Helvetica-Bold" },
54 { "Arial-Italic", "Helvetica-Oblique" },
55 { "Arial-ItalicMT", "Helvetica-Oblique" },
56 { "ArialMT", "Helvetica" },
57 { "Courier,Bold", "Courier-Bold" },
58 { "Courier,BoldItalic", "Courier-BoldOblique" },
59 { "Courier,Italic", "Courier-Oblique" },
60 { "CourierNew", "Courier" },
61 { "CourierNew,Bold", "Courier-Bold" },
62 { "CourierNew,BoldItalic", "Courier-BoldOblique" },
63 { "CourierNew,Italic", "Courier-Oblique" },
64 { "CourierNew-Bold", "Courier-Bold" },
65 { "CourierNew-BoldItalic", "Courier-BoldOblique" },
66 { "CourierNew-Italic", "Courier-Oblique" },
67 { "CourierNewPS-BoldItalicMT", "Courier-BoldOblique" },
68 { "CourierNewPS-BoldMT", "Courier-Bold" },
69 { "CourierNewPS-ItalicMT", "Courier-Oblique" },
70 { "CourierNewPSMT", "Courier" },
71 { "Helvetica,Bold", "Helvetica-Bold" },
72 { "Helvetica,BoldItalic", "Helvetica-BoldOblique" },
73 { "Helvetica,Italic", "Helvetica-Oblique" },
74 { "Helvetica-BoldItalic", "Helvetica-BoldOblique" },
75 { "Helvetica-Italic", "Helvetica-Oblique" },
76 { "Symbol,Bold", "Symbol" },
77 { "Symbol,BoldItalic", "Symbol" },
78 { "Symbol,Italic", "Symbol" },
79 { "TimesNewRoman", "Times-Roman" },
80 { "TimesNewRoman,Bold", "Times-Bold" },
81 { "TimesNewRoman,BoldItalic", "Times-BoldItalic" },
82 { "TimesNewRoman,Italic", "Times-Italic" },
83 { "TimesNewRoman-Bold", "Times-Bold" },
84 { "TimesNewRoman-BoldItalic", "Times-BoldItalic" },
85 { "TimesNewRoman-Italic", "Times-Italic" },
86 { "TimesNewRomanPS", "Times-Roman" },
87 { "TimesNewRomanPS-Bold", "Times-Bold" },
88 { "TimesNewRomanPS-BoldItalic", "Times-BoldItalic" },
89 { "TimesNewRomanPS-BoldItalicMT", "Times-BoldItalic" },
90 { "TimesNewRomanPS-BoldMT", "Times-Bold" },
91 { "TimesNewRomanPS-Italic", "Times-Italic" },
92 { "TimesNewRomanPS-ItalicMT", "Times-Italic" },
93 { "TimesNewRomanPSMT", "Times-Roman" },
94 { "TimesNewRomanPSMT,Bold", "Times-Bold" },
95 { "TimesNewRomanPSMT,BoldItalic", "Times-BoldItalic" },
96 { "TimesNewRomanPSMT,Italic", "Times-Italic" }
99 //------------------------------------------------------------------------
101 //------------------------------------------------------------------------
103 GfxFont *GfxFont::makeFont(XRef *xref, char *tagA, Ref idA, Dict *fontDict) {
108 // get base font name
110 fontDict->lookup("BaseFont", &obj1);
112 nameA = new GString(obj1.getName());
118 fontDict->lookup("Subtype", &obj1);
119 if (obj1.isName("Type1") || obj1.isName("MMType1")) {
120 font = new Gfx8BitFont(xref, tagA, idA, nameA, fontType1, fontDict);
121 } else if (obj1.isName("Type1C")) {
122 font = new Gfx8BitFont(xref, tagA, idA, nameA, fontType1C, fontDict);
123 } else if (obj1.isName("Type3")) {
124 font = new Gfx8BitFont(xref, tagA, idA, nameA, fontType3, fontDict);
125 } else if (obj1.isName("TrueType")) {
126 font = new Gfx8BitFont(xref, tagA, idA, nameA, fontTrueType, fontDict);
127 } else if (obj1.isName("Type0")) {
128 font = new GfxCIDFont(xref, tagA, idA, nameA, fontDict);
130 error(-1, "Unknown font type: '%s'",
131 obj1.isName() ? obj1.getName() : "???");
132 font = new Gfx8BitFont(xref, tagA, idA, nameA, fontUnknownType, fontDict);
139 GfxFont::GfxFont(char *tagA, Ref idA, GString *nameA) {
141 tag = new GString(tagA);
149 GfxFont::~GfxFont() {
151 if (origName && origName != name) {
165 void GfxFont::readFontDescriptor(XRef *xref, Dict *fontDict) {
166 Object obj1, obj2, obj3, obj4;
170 // assume Times-Roman by default (for substitution purposes)
177 if (fontDict->lookup("FontDescriptor", &obj1)->isDict()) {
180 if (obj1.dictLookup("Flags", &obj2)->isInt()) {
181 flags = obj2.getInt();
186 obj1.dictLookup("FontName", &obj2);
188 embFontName = new GString(obj2.getName());
192 // look for embedded font file
193 if (obj1.dictLookupNF("FontFile", &obj2)->isRef()) {
194 embFontID = obj2.getRef();
195 if (type != fontType1) {
196 error(-1, "Mismatch between font type and embedded font file");
201 if (embFontID.num == -1 &&
202 obj1.dictLookupNF("FontFile2", &obj2)->isRef()) {
203 embFontID = obj2.getRef();
204 if (type != fontTrueType && type != fontCIDType2) {
205 error(-1, "Mismatch between font type and embedded font file");
206 type = type == fontCIDType0 ? fontCIDType2 : fontTrueType;
210 if (embFontID.num == -1 &&
211 obj1.dictLookupNF("FontFile3", &obj2)->isRef()) {
212 if (obj2.fetch(xref, &obj3)->isStream()) {
213 obj3.streamGetDict()->lookup("Subtype", &obj4);
214 if (obj4.isName("Type1")) {
215 embFontID = obj2.getRef();
216 if (type != fontType1) {
217 error(-1, "Mismatch between font type and embedded font file");
220 } else if (obj4.isName("Type1C")) {
221 embFontID = obj2.getRef();
222 if (type != fontType1 && type != fontType1C) {
223 error(-1, "Mismatch between font type and embedded font file");
226 } else if (obj4.isName("TrueType")) {
227 embFontID = obj2.getRef();
228 if (type != fontTrueType) {
229 error(-1, "Mismatch between font type and embedded font file");
232 } else if (obj4.isName("CIDFontType0C")) {
233 embFontID = obj2.getRef();
234 if (type != fontCIDType0) {
235 error(-1, "Mismatch between font type and embedded font file");
237 type = fontCIDType0C;
239 error(-1, "Unknown embedded font type '%s'",
240 obj4.isName() ? obj4.getName() : "???");
248 // look for MissingWidth
249 obj1.dictLookup("MissingWidth", &obj2);
251 missingWidth = obj2.getNum();
255 // get Ascent and Descent
256 obj1.dictLookup("Ascent", &obj2);
258 t = 0.001 * obj2.getNum();
259 // some broken font descriptors set ascent and descent to 0
265 obj1.dictLookup("Descent", &obj2);
267 t = 0.001 * obj2.getNum();
268 // some broken font descriptors set ascent and descent to 0
272 // some broken font descriptors specify a positive descent
280 if (obj1.dictLookup("FontBBox", &obj2)->isArray()) {
281 for (i = 0; i < 4 && i < obj2.arrayGetLength(); ++i) {
282 if (obj2.arrayGet(i, &obj3)->isNum()) {
283 fontBBox[i] = 0.001 * obj3.getNum();
294 CharCodeToUnicode *GfxFont::readToUnicodeCMap(Dict *fontDict, int nBits,
295 CharCodeToUnicode *ctu) {
300 if (!fontDict->lookup("ToUnicode", &obj1)->isStream()) {
306 while ((c = obj1.streamGetChar()) != EOF) {
312 ctu->mergeCMap(buf, nBits);
314 ctu = CharCodeToUnicode::parseCMap(buf, nBits);
320 void GfxFont::findExtFontFile() {
321 static char *type1Exts[] = { ".pfa", ".pfb", ".ps", "", NULL };
322 static char *ttExts[] = { ".ttf", NULL };
325 if (type == fontType1) {
326 extFontFile = globalParams->findFontFile(name, type1Exts);
327 } else if (type == fontTrueType) {
328 extFontFile = globalParams->findFontFile(name, ttExts);
333 char *GfxFont::readExtFontFile(int *len) {
337 if (!(f = fopen(extFontFile->getCString(), "rb"))) {
338 error(-1, "External font file '%s' vanished", extFontFile->getCString());
341 fseek(f, 0, SEEK_END);
342 *len = (int)ftell(f);
343 fseek(f, 0, SEEK_SET);
344 buf = (char *)gmalloc(*len);
345 if ((int)fread(buf, 1, *len, f) != *len) {
346 error(-1, "Error reading external font file '%s'",
347 extFontFile->getCString());
353 char *GfxFont::readEmbFontFile(XRef *xref, int *len) {
360 obj1.initRef(embFontID.num, embFontID.gen);
361 obj1.fetch(xref, &obj2);
362 if (!obj2.isStream()) {
363 error(-1, "Embedded font file is not a stream");
369 str = obj2.getStream();
374 while ((c = str->getChar()) != EOF) {
377 buf = (char *)grealloc(buf, size);
390 //------------------------------------------------------------------------
392 //------------------------------------------------------------------------
394 Gfx8BitFont::Gfx8BitFont(XRef *xref, char *tagA, Ref idA, GString *nameA,
395 GfxFontType typeA, Dict *fontDict):
396 GfxFont(tagA, idA, nameA)
399 BuiltinFont *builtinFont;
401 GBool baseEncFromFontFile;
409 Unicode toUnicode[256];
410 CharCodeToUnicode *utu, *ctu2;
413 int firstChar, lastChar;
415 Object obj1, obj2, obj3;
421 // do font name substitution for various aliases of the Base 14 font
424 name2 = name->copy();
426 while (i < name2->getLength()) {
427 if (name2->getChar(i) == ' ') {
434 b = sizeof(stdFontMap) / sizeof(StdFontMapEntry);
435 // invariant: stdFontMap[a].altName <= name2 < stdFontMap[b].altName
438 if (name2->cmp(stdFontMap[m].altName) >= 0) {
444 if (!name2->cmp(stdFontMap[a].altName)) {
445 name = new GString(stdFontMap[a].properName);
450 // is it a built-in font?
453 for (i = 0; i < nBuiltinFonts; ++i) {
454 if (!name->cmp(builtinFonts[i].name)) {
455 builtinFont = &builtinFonts[i];
461 // default ascent/descent values
463 ascent = 0.001 * builtinFont->ascent;
464 descent = 0.001 * builtinFont->descent;
465 fontBBox[0] = 0.001 * builtinFont->bbox[0];
466 fontBBox[1] = 0.001 * builtinFont->bbox[1];
467 fontBBox[2] = 0.001 * builtinFont->bbox[2];
468 fontBBox[3] = 0.001 * builtinFont->bbox[3];
472 fontBBox[0] = fontBBox[1] = fontBBox[2] = fontBBox[3] = 0;
475 // get info from font descriptor
476 readFontDescriptor(xref, fontDict);
478 // for non-embedded fonts, don't trust the ascent/descent/bbox
479 // values from the font descriptor
480 if (builtinFont && embFontID.num < 0) {
481 ascent = 0.001 * builtinFont->ascent;
482 descent = 0.001 * builtinFont->descent;
483 fontBBox[0] = 0.001 * builtinFont->bbox[0];
484 fontBBox[1] = 0.001 * builtinFont->bbox[1];
485 fontBBox[2] = 0.001 * builtinFont->bbox[2];
486 fontBBox[3] = 0.001 * builtinFont->bbox[3];
489 // look for an external font file
493 fontMat[0] = fontMat[3] = 1;
494 fontMat[1] = fontMat[2] = fontMat[4] = fontMat[5] = 0;
495 if (fontDict->lookup("FontMatrix", &obj1)->isArray()) {
496 for (i = 0; i < 6 && i < obj1.arrayGetLength(); ++i) {
497 if (obj1.arrayGet(i, &obj2)->isNum()) {
498 fontMat[i] = obj2.getNum();
505 // get Type 3 bounding box, font definition, and resources
506 if (type == fontType3) {
507 if (fontDict->lookup("FontBBox", &obj1)->isArray()) {
508 for (i = 0; i < 4 && i < obj1.arrayGetLength(); ++i) {
509 if (obj1.arrayGet(i, &obj2)->isNum()) {
510 fontBBox[i] = obj2.getNum();
516 if (!fontDict->lookup("CharProcs", &charProcs)->isDict()) {
517 error(-1, "Missing or invalid CharProcs dictionary in Type 3 font");
520 if (!fontDict->lookup("Resources", &resources)->isDict()) {
525 //----- build the font encoding -----
527 // Encodings start with a base encoding, which can come from
528 // (in order of priority):
529 // 1. FontDict.Encoding or FontDict.Encoding.BaseEncoding
530 // - MacRoman / MacExpert / WinAnsi / Standard
531 // 2. embedded or external font file
533 // - builtin --> builtin encoding
534 // - TrueType --> WinAnsiEncoding
535 // - others --> StandardEncoding
536 // and then add a list of differences (if any) from
537 // FontDict.Encoding.Differences.
539 // check FontDict for base encoding
540 hasEncoding = gFalse;
541 usesMacRomanEnc = gFalse;
543 baseEncFromFontFile = gFalse;
544 fontDict->lookup("Encoding", &obj1);
546 obj1.dictLookup("BaseEncoding", &obj2);
547 if (obj2.isName("MacRomanEncoding")) {
549 usesMacRomanEnc = gTrue;
550 baseEnc = macRomanEncoding;
551 } else if (obj2.isName("MacExpertEncoding")) {
553 baseEnc = macExpertEncoding;
554 } else if (obj2.isName("WinAnsiEncoding")) {
556 baseEnc = winAnsiEncoding;
559 } else if (obj1.isName("MacRomanEncoding")) {
561 usesMacRomanEnc = gTrue;
562 baseEnc = macRomanEncoding;
563 } else if (obj1.isName("MacExpertEncoding")) {
565 baseEnc = macExpertEncoding;
566 } else if (obj1.isName("WinAnsiEncoding")) {
568 baseEnc = winAnsiEncoding;
571 // check embedded or external font file for base encoding
572 // (only for Type 1 fonts - trying to get an encoding out of a
573 // TrueType font is a losing proposition)
577 if (type == fontType1 && (extFontFile || embFontID.num >= 0)) {
579 ffT1 = FoFiType1::load(extFontFile->getCString());
581 buf = readEmbFontFile(xref, &len);
582 ffT1 = FoFiType1::make(buf, len);
585 if (ffT1->getName()) {
589 embFontName = new GString(ffT1->getName());
592 baseEnc = ffT1->getEncoding();
593 baseEncFromFontFile = gTrue;
596 } else if (type == fontType1C && (extFontFile || embFontID.num >= 0)) {
598 ffT1C = FoFiType1C::load(extFontFile->getCString());
600 buf = readEmbFontFile(xref, &len);
601 ffT1C = FoFiType1C::make(buf, len);
604 if (ffT1C->getName()) {
608 embFontName = new GString(ffT1C->getName());
611 baseEnc = ffT1C->getEncoding();
612 baseEncFromFontFile = gTrue;
620 // get default base encoding
622 if (builtinFont && embFontID.num < 0) {
623 baseEnc = builtinFont->defaultBaseEnc;
625 } else if (type == fontTrueType) {
626 baseEnc = winAnsiEncoding;
628 baseEnc = standardEncoding;
632 // copy the base encoding
633 for (i = 0; i < 256; ++i) {
635 if ((encFree[i] = baseEncFromFontFile) && enc[i]) {
636 enc[i] = copyString(baseEnc[i]);
640 // some Type 1C font files have empty encodings, which can break the
641 // T1C->T1 conversion (since the 'seac' operator depends on having
642 // the accents in the encoding), so we fill in any gaps from
644 if (type == fontType1C && (extFontFile || embFontID.num >= 0) &&
645 baseEncFromFontFile) {
646 for (i = 0; i < 256; ++i) {
647 if (!enc[i] && standardEncoding[i]) {
648 enc[i] = standardEncoding[i];
654 // merge differences into encoding
656 obj1.dictLookup("Differences", &obj2);
657 if (obj2.isArray()) {
660 for (i = 0; i < obj2.arrayGetLength(); ++i) {
661 obj2.arrayGet(i, &obj3);
663 code = obj3.getInt();
664 } else if (obj3.isName()) {
665 if (code >= 0 && code < 256) {
669 enc[code] = copyString(obj3.getName());
670 encFree[code] = gTrue;
674 error(-1, "Wrong type in font encoding resource differences (%s)",
690 //----- build the mapping to Unicode -----
692 // pass 1: use the name-to-Unicode mapping table
693 missing = hex = gFalse;
694 for (code = 0; code < 256; ++code) {
695 if ((charName = enc[code])) {
696 if (!(toUnicode[code] = globalParams->mapNameToUnicode(charName)) &&
697 strcmp(charName, ".notdef")) {
698 // if it wasn't in the name-to-Unicode table, check for a
699 // name that looks like 'Axx' or 'xx', where 'A' is any letter
700 // and 'xx' is two hex digits
701 if ((strlen(charName) == 3 &&
702 isalpha(charName[0]) &&
703 isxdigit(charName[1]) && isxdigit(charName[2]) &&
704 ((charName[1] >= 'a' && charName[1] <= 'f') ||
705 (charName[1] >= 'A' && charName[1] <= 'F') ||
706 (charName[2] >= 'a' && charName[2] <= 'f') ||
707 (charName[2] >= 'A' && charName[2] <= 'F'))) ||
708 (strlen(charName) == 2 &&
709 isxdigit(charName[0]) && isxdigit(charName[1]) &&
710 ((charName[0] >= 'a' && charName[0] <= 'f') ||
711 (charName[0] >= 'A' && charName[0] <= 'F') ||
712 (charName[1] >= 'a' && charName[1] <= 'f') ||
713 (charName[1] >= 'A' && charName[1] <= 'F')))) {
723 // pass 2: try to fill in the missing chars, looking for names of
724 // the form 'Axx', 'xx', 'Ann', 'ABnn', or 'nn', where 'A' and 'B'
725 // are any letters, 'xx' is two hex digits, and 'nn' is 2-4
727 if (missing && globalParams->getMapNumericCharNames()) {
728 for (code = 0; code < 256; ++code) {
729 if ((charName = enc[code]) && !toUnicode[code] &&
730 strcmp(charName, ".notdef")) {
731 n = strlen(charName);
733 if (hex && n == 3 && isalpha(charName[0]) &&
734 isxdigit(charName[1]) && isxdigit(charName[2])) {
735 sscanf(charName+1, "%x", &code2);
736 } else if (hex && n == 2 &&
737 isxdigit(charName[0]) && isxdigit(charName[1])) {
738 sscanf(charName, "%x", &code2);
739 } else if (!hex && n >= 2 && n <= 4 &&
740 isdigit(charName[0]) && isdigit(charName[1])) {
741 code2 = atoi(charName);
742 } else if (n >= 3 && n <= 5 &&
743 isdigit(charName[1]) && isdigit(charName[2])) {
744 code2 = atoi(charName+1);
745 } else if (n >= 4 && n <= 6 &&
746 isdigit(charName[2]) && isdigit(charName[3])) {
747 code2 = atoi(charName+2);
749 if (code2 >= 0 && code2 <= 0xff) {
750 toUnicode[code] = (Unicode)code2;
756 // construct the char code -> Unicode mapping object
757 ctu = CharCodeToUnicode::make8BitToUnicode(toUnicode);
759 // merge in a ToUnicode CMap, if there is one -- this overwrites
760 // existing entries in ctu, i.e., the ToUnicode CMap takes
761 // precedence, but the other encoding info is allowed to fill in any
763 readToUnicodeCMap(fontDict, 8, ctu);
765 // look for a Unicode-to-Unicode mapping
766 if (name && (utu = globalParams->getUnicodeToUnicode(name))) {
767 for (i = 0; i < 256; ++i) {
770 ctu2 = CharCodeToUnicode::make8BitToUnicode(toUnicode);
771 for (i = 0; i < 256; ++i) {
772 n = ctu->mapToUnicode((CharCode)i, uBuf, 8);
774 n = utu->mapToUnicode((CharCode)uBuf[0], uBuf, 8);
776 ctu2->setMapping((CharCode)i, uBuf, n);
785 //----- get the character widths -----
787 // initialize all widths
788 for (code = 0; code < 256; ++code) {
789 widths[code] = missingWidth * 0.001;
792 // use widths from font dict, if present
793 fontDict->lookup("FirstChar", &obj1);
794 firstChar = obj1.isInt() ? obj1.getInt() : 0;
796 if (firstChar < 0 || firstChar > 255) {
799 fontDict->lookup("LastChar", &obj1);
800 lastChar = obj1.isInt() ? obj1.getInt() : 255;
802 if (lastChar < 0 || lastChar > 255) {
805 mul = (type == fontType3) ? fontMat[0] : 0.001;
806 fontDict->lookup("Widths", &obj1);
807 if (obj1.isArray()) {
808 flags |= fontFixedWidth;
809 if (obj1.arrayGetLength() < lastChar - firstChar + 1) {
810 lastChar = firstChar + obj1.arrayGetLength() - 1;
812 for (code = firstChar; code <= lastChar; ++code) {
813 obj1.arrayGet(code - firstChar, &obj2);
815 widths[code] = obj2.getNum() * mul;
816 if (widths[code] != widths[firstChar]) {
817 flags &= ~fontFixedWidth;
823 // use widths from built-in font
824 } else if (builtinFont) {
825 // this is a kludge for broken PDF files that encode char 32
827 if (builtinFont->widths->getWidth("space", &w)) {
828 widths[32] = 0.001 * w;
830 for (code = 0; code < 256; ++code) {
831 if (enc[code] && builtinFont->widths->getWidth(enc[code], &w)) {
832 widths[code] = 0.001 * w;
836 // couldn't find widths -- use defaults
838 // this is technically an error -- the Widths entry is required
839 // for all but the Base-14 fonts -- but certain PDF generators
840 // apparently don't include widths for Arial and TimesNewRoman
841 if (isFixedWidth()) {
843 } else if (isSerif()) {
854 builtinFont = builtinFontSubst[i];
855 // this is a kludge for broken PDF files that encode char 32
857 if (builtinFont->widths->getWidth("space", &w)) {
858 widths[32] = 0.001 * w;
860 for (code = 0; code < 256; ++code) {
861 if (enc[code] && builtinFont->widths->getWidth(enc[code], &w)) {
862 widths[code] = 0.001 * w;
871 Gfx8BitFont::~Gfx8BitFont() {
874 for (i = 0; i < 256; ++i) {
875 if (encFree[i] && enc[i]) {
880 if (charProcs.isDict()) {
883 if (resources.isDict()) {
888 int Gfx8BitFont::getNextChar(char *s, int len, CharCode *code,
889 Unicode *u, int uSize, int *uLen,
890 double *dx, double *dy, double *ox, double *oy) {
893 *code = c = (CharCode)(*s & 0xff);
894 *uLen = ctu->mapToUnicode(c, u, uSize);
900 CharCodeToUnicode *Gfx8BitFont::getToUnicode() {
905 Gushort *Gfx8BitFont::getCodeToGIDMap(FoFiTrueType *ff) {
907 int cmapPlatform, cmapEncoding;
908 int unicodeCmap, macRomanCmap, msSymbolCmap, cmap;
909 GBool useMacRoman, useUnicode;
914 map = (Gushort *)gmallocn(256, sizeof(Gushort));
915 for (i = 0; i < 256; ++i) {
919 // To match up with the Adobe-defined behaviour, we choose a cmap
921 // 1. If the PDF font has an encoding:
922 // 1a. If the PDF font specified MacRomanEncoding and the
923 // TrueType font has a Macintosh Roman cmap, use it, and
924 // reverse map the char names through MacRomanEncoding to
926 // 1b. If the TrueType font has a Microsoft Unicode cmap or a
927 // non-Microsoft Unicode cmap, use it, and use the Unicode
928 // indexes, not the char codes.
929 // 1c. If the PDF font is symbolic and the TrueType font has a
930 // Microsoft Symbol cmap, use it, and use char codes
931 // directly (possibly with an offset of 0xf000).
932 // 1d. If the TrueType font has a Macintosh Roman cmap, use it,
934 // 2. If the PDF font does not have an encoding or the PDF font is
936 // 2a. If the TrueType font has a Macintosh Roman cmap, use it,
937 // and use char codes directly (possibly with an offset of
939 // 2b. If the TrueType font has a Microsoft Symbol cmap, use it,
940 // and use char codes directly (possible with an offset of
942 // 3. If none of these rules apply, use the first cmap and hope for
943 // the best (this shouldn't happen).
944 unicodeCmap = macRomanCmap = msSymbolCmap = -1;
945 for (i = 0; i < ff->getNumCmaps(); ++i) {
946 cmapPlatform = ff->getCmapPlatform(i);
947 cmapEncoding = ff->getCmapEncoding(i);
948 if ((cmapPlatform == 3 && cmapEncoding == 1) ||
951 } else if (cmapPlatform == 1 && cmapEncoding == 0) {
953 } else if (cmapPlatform == 3 && cmapEncoding == 0) {
958 useMacRoman = gFalse;
961 if (usesMacRomanEnc && macRomanCmap >= 0) {
964 } else if (unicodeCmap >= 0) {
967 } else if ((flags & fontSymbolic) && msSymbolCmap >= 0) {
969 } else if ((flags & fontSymbolic) && macRomanCmap >= 0) {
971 } else if (macRomanCmap >= 0) {
976 if (macRomanCmap >= 0) {
978 } else if (msSymbolCmap >= 0) {
983 // reverse map the char names through MacRomanEncoding, then map the
984 // char codes through the cmap
986 for (i = 0; i < 256; ++i) {
987 if ((charName = enc[i])) {
988 if ((code = globalParams->getMacRomanCharCode(charName))) {
989 map[i] = ff->mapCodeToGID(cmap, code);
994 // map Unicode through the cmap
995 } else if (useUnicode) {
996 for (i = 0; i < 256; ++i) {
997 if (((charName = enc[i]) &&
998 (u = globalParams->mapNameToUnicode(charName))) ||
999 (n = ctu->mapToUnicode((CharCode)i, &u, 1))) {
1000 map[i] = ff->mapCodeToGID(cmap, u);
1004 // map the char codes through the cmap, possibly with an offset of
1007 for (i = 0; i < 256; ++i) {
1008 if (!(map[i] = ff->mapCodeToGID(cmap, i))) {
1009 map[i] = ff->mapCodeToGID(cmap, 0xf000 + i);
1014 // try the TrueType 'post' table to handle any unmapped characters
1015 for (i = 0; i < 256; ++i) {
1016 if (!map[i] && (charName = enc[i])) {
1017 map[i] = (Gushort)(int)ff->mapNameToGID(charName);
1024 Dict *Gfx8BitFont::getCharProcs() {
1025 return charProcs.isDict() ? charProcs.getDict() : (Dict *)NULL;
1028 Object *Gfx8BitFont::getCharProc(int code, Object *proc) {
1029 if (enc[code] && charProcs.isDict()) {
1030 charProcs.dictLookup(enc[code], proc);
1037 Dict *Gfx8BitFont::getResources() {
1038 return resources.isDict() ? resources.getDict() : (Dict *)NULL;
1041 //------------------------------------------------------------------------
1043 //------------------------------------------------------------------------
1045 static int CDECL cmpWidthExcep(const void *w1, const void *w2) {
1046 return ((GfxFontCIDWidthExcep *)w1)->first -
1047 ((GfxFontCIDWidthExcep *)w2)->first;
1050 static int CDECL cmpWidthExcepV(const void *w1, const void *w2) {
1051 return ((GfxFontCIDWidthExcepV *)w1)->first -
1052 ((GfxFontCIDWidthExcepV *)w2)->first;
1055 GfxCIDFont::GfxCIDFont(XRef *xref, char *tagA, Ref idA, GString *nameA,
1057 GfxFont(tagA, idA, nameA)
1060 GString *collection, *cMapName;
1061 Object desFontDictObj;
1062 Object obj1, obj2, obj3, obj4, obj5, obj6;
1063 CharCodeToUnicode *utu;
1067 int excepsSize, i, j, k, n;
1071 fontBBox[0] = fontBBox[1] = fontBBox[2] = fontBBox[3] = 0;
1074 widths.defWidth = 1.0;
1075 widths.defHeight = -1.0;
1076 widths.defVY = 0.880;
1077 widths.exceps = NULL;
1079 widths.excepsV = NULL;
1080 widths.nExcepsV = 0;
1084 // get the descendant font
1085 if (!fontDict->lookup("DescendantFonts", &obj1)->isArray()) {
1086 error(-1, "Missing DescendantFonts entry in Type 0 font");
1090 if (!obj1.arrayGet(0, &desFontDictObj)->isDict()) {
1091 error(-1, "Bad descendant font in Type 0 font");
1095 desFontDict = desFontDictObj.getDict();
1098 if (!desFontDict->lookup("Subtype", &obj1)) {
1099 error(-1, "Missing Subtype entry in Type 0 descendant font");
1102 if (obj1.isName("CIDFontType0")) {
1103 type = fontCIDType0;
1104 } else if (obj1.isName("CIDFontType2")) {
1105 type = fontCIDType2;
1107 error(-1, "Unknown Type 0 descendant font type '%s'",
1108 obj1.isName() ? obj1.getName() : "???");
1113 // get info from font descriptor
1114 readFontDescriptor(xref, desFontDict);
1116 // look for an external font file
1119 //----- encoding info -----
1122 if (!desFontDict->lookup("CIDSystemInfo", &obj1)->isDict()) {
1123 error(-1, "Missing CIDSystemInfo dictionary in Type 0 descendant font");
1126 obj1.dictLookup("Registry", &obj2);
1127 obj1.dictLookup("Ordering", &obj3);
1128 if (!obj2.isString() || !obj3.isString()) {
1129 error(-1, "Invalid CIDSystemInfo dictionary in Type 0 descendant font");
1132 collection = obj2.getString()->copy()->append('-')->append(obj3.getString());
1137 // look for a ToUnicode CMap
1138 if (!(ctu = readToUnicodeCMap(fontDict, 16, NULL))) {
1140 // the "Adobe-Identity" and "Adobe-UCS" collections don't have
1141 // cidToUnicode files
1142 if (collection->cmp("Adobe-Identity") &&
1143 collection->cmp("Adobe-UCS")) {
1145 // look for a user-supplied .cidToUnicode file
1146 if (!(ctu = globalParams->getCIDToUnicode(collection))) {
1147 error(-1, "Unknown character collection '%s'",
1148 collection->getCString());
1149 // fall-through, assuming the Identity mapping -- this appears
1150 // to match Adobe's behavior
1155 // look for a Unicode-to-Unicode mapping
1156 if (name && (utu = globalParams->getUnicodeToUnicode(name))) {
1158 for (c = 0; c < ctu->getLength(); ++c) {
1159 n = ctu->mapToUnicode(c, uBuf, 8);
1161 n = utu->mapToUnicode((CharCode)uBuf[0], uBuf, 8);
1163 ctu->setMapping(c, uBuf, n);
1173 // encoding (i.e., CMap)
1174 //~ need to handle a CMap stream here
1175 //~ also need to deal with the UseCMap entry in the stream dict
1176 if (!fontDict->lookup("Encoding", &obj1)->isName()) {
1177 error(-1, "Missing or invalid Encoding entry in Type 0 font");
1181 cMapName = new GString(obj1.getName());
1183 if (!(cMap = globalParams->getCMap(collection, cMapName))) {
1184 error(-1, "Unknown CMap '%s' for character collection '%s'",
1185 cMapName->getCString(), collection->getCString());
1193 // CIDToGIDMap (for embedded TrueType fonts)
1194 if (type == fontCIDType2) {
1195 desFontDict->lookup("CIDToGIDMap", &obj1);
1196 if (obj1.isStream()) {
1199 cidToGID = (Gushort *)gmallocn(i, sizeof(Gushort));
1201 while ((c1 = obj1.streamGetChar()) != EOF &&
1202 (c2 = obj1.streamGetChar()) != EOF) {
1203 if (cidToGIDLen == i) {
1205 cidToGID = (Gushort *)greallocn(cidToGID, i, sizeof(Gushort));
1207 cidToGID[cidToGIDLen++] = (Gushort)((c1 << 8) + c2);
1209 } else if (!obj1.isName("Identity") && !obj1.isNull()) {
1210 error(-1, "Invalid CIDToGIDMap entry in CID font");
1215 //----- character metrics -----
1217 // default char width
1218 if (desFontDict->lookup("DW", &obj1)->isInt()) {
1219 widths.defWidth = obj1.getInt() * 0.001;
1223 // char width exceptions
1224 if (desFontDict->lookup("W", &obj1)->isArray()) {
1227 while (i + 1 < obj1.arrayGetLength()) {
1228 obj1.arrayGet(i, &obj2);
1229 obj1.arrayGet(i + 1, &obj3);
1230 if (obj2.isInt() && obj3.isInt() && i + 2 < obj1.arrayGetLength()) {
1231 if (obj1.arrayGet(i + 2, &obj4)->isNum()) {
1232 if (widths.nExceps == excepsSize) {
1234 widths.exceps = (GfxFontCIDWidthExcep *)
1235 greallocn(widths.exceps,
1236 excepsSize, sizeof(GfxFontCIDWidthExcep));
1238 widths.exceps[widths.nExceps].first = obj2.getInt();
1239 widths.exceps[widths.nExceps].last = obj3.getInt();
1240 widths.exceps[widths.nExceps].width = obj4.getNum() * 0.001;
1243 error(-1, "Bad widths array in Type 0 font");
1247 } else if (obj2.isInt() && obj3.isArray()) {
1248 if (widths.nExceps + obj3.arrayGetLength() > excepsSize) {
1249 excepsSize = (widths.nExceps + obj3.arrayGetLength() + 15) & ~15;
1250 widths.exceps = (GfxFontCIDWidthExcep *)
1251 greallocn(widths.exceps,
1252 excepsSize, sizeof(GfxFontCIDWidthExcep));
1255 for (k = 0; k < obj3.arrayGetLength(); ++k) {
1256 if (obj3.arrayGet(k, &obj4)->isNum()) {
1257 widths.exceps[widths.nExceps].first = j;
1258 widths.exceps[widths.nExceps].last = j;
1259 widths.exceps[widths.nExceps].width = obj4.getNum() * 0.001;
1263 error(-1, "Bad widths array in Type 0 font");
1269 error(-1, "Bad widths array in Type 0 font");
1275 qsort(widths.exceps, widths.nExceps, sizeof(GfxFontCIDWidthExcep),
1280 // default metrics for vertical font
1281 if (desFontDict->lookup("DW2", &obj1)->isArray() &&
1282 obj1.arrayGetLength() == 2) {
1283 if (obj1.arrayGet(0, &obj2)->isNum()) {
1284 widths.defVY = obj2.getNum() * 0.001;
1287 if (obj1.arrayGet(1, &obj2)->isNum()) {
1288 widths.defHeight = obj2.getNum() * 0.001;
1294 // char metric exceptions for vertical font
1295 if (desFontDict->lookup("W2", &obj1)->isArray()) {
1298 while (i + 1 < obj1.arrayGetLength()) {
1299 obj1.arrayGet(i, &obj2);
1300 obj1.arrayGet(i+ 1, &obj3);
1301 if (obj2.isInt() && obj3.isInt() && i + 4 < obj1.arrayGetLength()) {
1302 if (obj1.arrayGet(i + 2, &obj4)->isNum() &&
1303 obj1.arrayGet(i + 3, &obj5)->isNum() &&
1304 obj1.arrayGet(i + 4, &obj6)->isNum()) {
1305 if (widths.nExcepsV == excepsSize) {
1307 widths.excepsV = (GfxFontCIDWidthExcepV *)
1308 greallocn(widths.excepsV,
1309 excepsSize, sizeof(GfxFontCIDWidthExcepV));
1311 widths.excepsV[widths.nExcepsV].first = obj2.getInt();
1312 widths.excepsV[widths.nExcepsV].last = obj3.getInt();
1313 widths.excepsV[widths.nExcepsV].height = obj4.getNum() * 0.001;
1314 widths.excepsV[widths.nExcepsV].vx = obj5.getNum() * 0.001;
1315 widths.excepsV[widths.nExcepsV].vy = obj6.getNum() * 0.001;
1318 error(-1, "Bad widths (W2) array in Type 0 font");
1324 } else if (obj2.isInt() && obj3.isArray()) {
1325 if (widths.nExcepsV + obj3.arrayGetLength() / 3 > excepsSize) {
1327 (widths.nExcepsV + obj3.arrayGetLength() / 3 + 15) & ~15;
1328 widths.excepsV = (GfxFontCIDWidthExcepV *)
1329 greallocn(widths.excepsV,
1330 excepsSize, sizeof(GfxFontCIDWidthExcepV));
1333 for (k = 0; k < obj3.arrayGetLength(); k += 3) {
1334 if (obj3.arrayGet(k, &obj4)->isNum() &&
1335 obj3.arrayGet(k+1, &obj5)->isNum() &&
1336 obj3.arrayGet(k+2, &obj6)->isNum()) {
1337 widths.excepsV[widths.nExceps].first = j;
1338 widths.excepsV[widths.nExceps].last = j;
1339 widths.excepsV[widths.nExceps].height = obj4.getNum() * 0.001;
1340 widths.excepsV[widths.nExceps].vx = obj5.getNum() * 0.001;
1341 widths.excepsV[widths.nExceps].vy = obj6.getNum() * 0.001;
1345 error(-1, "Bad widths (W2) array in Type 0 font");
1353 error(-1, "Bad widths (W2) array in Type 0 font");
1359 qsort(widths.excepsV, widths.nExcepsV, sizeof(GfxFontCIDWidthExcepV),
1364 desFontDictObj.free();
1374 desFontDictObj.free();
1378 GfxCIDFont::~GfxCIDFont() {
1385 gfree(widths.exceps);
1386 gfree(widths.excepsV);
1392 int GfxCIDFont::getNextChar(char *s, int len, CharCode *code,
1393 Unicode *u, int uSize, int *uLen,
1394 double *dx, double *dy, double *ox, double *oy) {
1396 double w, h, vx, vy;
1406 *code = (CharCode)(cid = cMap->getCID(s, len, &n));
1408 *uLen = ctu->mapToUnicode(cid, u, uSize);
1414 if (cMap->getWMode() == 0) {
1415 w = widths.defWidth;
1417 if (widths.nExceps > 0 && cid >= widths.exceps[0].first) {
1420 // invariant: widths.exceps[a].first <= cid < widths.exceps[b].first
1423 if (widths.exceps[m].first <= cid) {
1429 if (cid <= widths.exceps[a].last) {
1430 w = widths.exceps[a].width;
1437 h = widths.defHeight;
1438 vx = widths.defWidth / 2;
1440 if (widths.nExcepsV > 0 && cid >= widths.excepsV[0].first) {
1442 b = widths.nExcepsV;
1443 // invariant: widths.excepsV[a].first <= cid < widths.excepsV[b].first
1446 if (widths.excepsV[m].last <= cid) {
1452 if (cid <= widths.excepsV[a].last) {
1453 h = widths.excepsV[a].height;
1454 vx = widths.excepsV[a].vx;
1455 vy = widths.excepsV[a].vy;
1468 int GfxCIDFont::getWMode() {
1469 return cMap ? cMap->getWMode() : 0;
1472 CharCodeToUnicode *GfxCIDFont::getToUnicode() {
1479 GString *GfxCIDFont::getCollection() {
1480 return cMap ? cMap->getCollection() : (GString *)NULL;
1483 //------------------------------------------------------------------------
1485 //------------------------------------------------------------------------
1487 GfxFontDict::GfxFontDict(XRef *xref, Ref *fontDictRef, Dict *fontDict) {
1492 numFonts = fontDict->getLength();
1493 fonts = (GfxFont **)gmallocn(numFonts, sizeof(GfxFont *));
1494 for (i = 0; i < numFonts; ++i) {
1495 fontDict->getValNF(i, &obj1);
1496 obj1.fetch(xref, &obj2);
1497 if (obj2.isDict()) {
1501 // no indirect reference for this font, so invent a unique one
1502 // (legal generation numbers are five digits, so any 6-digit
1503 // number would be safe)
1506 r.gen = 100000 + fontDictRef->num;
1511 fonts[i] = GfxFont::makeFont(xref, fontDict->getKey(i),
1513 if (fonts[i] && !fonts[i]->isOk()) {
1518 error(-1, "font resource is not a dictionary");
1526 GfxFontDict::~GfxFontDict() {
1529 for (i = 0; i < numFonts; ++i) {
1537 GfxFont *GfxFontDict::lookup(char *tag) {
1540 for (i = 0; i < numFonts; ++i) {
1541 if (fonts[i] && fonts[i]->matches(tag)) {