always store detected spaces at unicode 32
[swftools.git] / lib / pdf / InfoOutputDev.cc
1 #include "config.h"
2 #include "Object.h"
3 #include "InfoOutputDev.h"
4 #include "SplashOutputDev.h"
5 #ifdef HAVE_POPPLER
6 #include <splash/SplashTypes.h>
7 #include <splash/SplashPath.h>
8 #include <splash/SplashFont.h>
9 #include <splash/SplashFontFile.h>
10 #else
11 #include "SplashTypes.h"
12 #include "SplashPath.h"
13 #include "SplashFont.h"
14 #include "SplashFontFile.h"
15 #endif
16 #include "GfxState.h"
17 #include "../log.h"
18 #include "../types.h"
19 #include "../q.h"
20 #include "../gfxfont.h"
21 #include <math.h>
22 #include <assert.h>
23
24 int config_addspace = 1;
25 int config_fontquality = 10;
26 int config_bigchar = 0;
27
28 InfoOutputDev::InfoOutputDev(XRef*xref) 
29 {
30     num_links = 0;
31     num_jpeg_images = 0;
32     num_ppm_images = 0;
33     num_textfields = 0;
34     num_fonts = 0;
35     num_polygons= 0;
36     currentfont = 0;
37     currentglyph = 0;
38     id2font = new GHash(1);
39     SplashColor white = {255,255,255};
40     splash = new SplashOutputDev(splashModeRGB8,320,0,white,0,0);
41     splash->startDoc(xref);
42 }
43 InfoOutputDev::~InfoOutputDev() 
44 {
45     GHashIter*i;
46     id2font->startIter(&i);
47     GString*key;
48     FontInfo*fontinfo;
49     while(id2font->getNext(&i, &key, (void**)&fontinfo)) {
50         delete fontinfo;
51     }
52     id2font->killIter(&i);
53
54     delete id2font;id2font=0;
55     delete splash;splash=0;
56 }
57
58 void FontInfo::grow(int size)
59 {
60     if(size >= this->num_glyphs) {
61         this->glyphs = (GlyphInfo**)realloc(this->glyphs, sizeof(GlyphInfo*)*(size));
62         this->kerning = (dict_t**)realloc(this->kerning, sizeof(dict_t*)*(size));
63         memset(&this->glyphs[this->num_glyphs], 0, sizeof(SplashPath*)*((size)-this->num_glyphs));
64         memset(&this->kerning[this->num_glyphs], 0, sizeof(dict_t*)*((size)-this->num_glyphs));
65         this->num_glyphs = size;
66     }
67 }
68 FontInfo::FontInfo(char*id)
69 {
70     this->id = strdup(id);
71     this->seen = 0;
72     this->num_glyphs = 0;
73     this->glyphs = 0;
74     this->kerning = 0;
75     this->splash_font = 0;
76     this->lastchar = -1;
77     this->lastx = 0;
78     this->lasty = 0;
79     this->gfxfont = 0;
80     this->space_char = -1;
81     this->ascender = 0;
82     this->descender = 0;
83 }
84 FontInfo::~FontInfo()
85 {
86     if(this->id) {free(this->id);this->id=0;}
87     this->font = 0;
88     int t;
89     for(t=0;t<num_glyphs;t++) {
90         if(glyphs[t]) {
91             delete glyphs[t]->path;glyphs[t]->path = 0;
92             delete glyphs[t];
93             glyphs[t]=0;
94         }
95     }
96     free(glyphs);glyphs=0;
97     if(this->gfxfont)
98         gfxfont_free(this->gfxfont);
99    
100     if(kerning) {
101         for(t=0;t<num_glyphs;t++) {
102             dict_t* d = kerning[t];
103             if(!d) continue;
104             DICT_ITERATE_ITEMS(d,void*,key,mtf_t*,m) {
105                 mtf_destroy(m);
106             }
107             dict_destroy(d);
108         }
109         free(kerning);
110         kerning=0;
111     }
112 }
113
114 static int findSpace(gfxfont_t*font)
115 {
116     int first_space = -1;
117     int t;
118     for(t=0;t<font->num_glyphs;t++) {
119         gfxglyph_t*g = &font->glyphs[t];
120         if(GLYPH_IS_SPACE(g)) {
121             if(g->unicode == 32) return t;
122             if(first_space<0)
123                 first_space = t;
124         }
125     }
126     if(font->num_glyphs>32 && GLYPH_IS_SPACE(&font->glyphs[32])) {
127         return 32;
128     }
129     return first_space;
130 }
131
132 static int addSpace(gfxfont_t*font)
133 {
134     /* first, make sure the new space char is the only char that'll use unicode 32 */
135     int t;
136     for(t=0;t<font->num_glyphs;t++) {
137         if(font->glyphs[t].unicode==32)
138             font->glyphs[t].unicode=0;
139     }
140     font->num_glyphs++;
141     font->glyphs = (gfxglyph_t*)realloc(font->glyphs, sizeof(gfxglyph_t)*font->num_glyphs);
142     gfxglyph_t*g = &font->glyphs[font->num_glyphs-1];
143     memset(g, 0, sizeof(*g));
144     g->unicode = 32;
145     g->advance = fabs(font->ascent + font->descent)*2 / 3;
146     if(font->max_unicode > 32)
147         font->unicode2glyph[32] = font->num_glyphs-1;
148 #if 0
149     g->line = gfxline_makerectangle(0, -font->ascent, g->advance, font->descent);
150 #endif
151     return font->num_glyphs-1;
152 }
153
154 static gfxfont_t* createGfxFont(FontInfo*src)
155 {
156     gfxfont_t*font = (gfxfont_t*)malloc(sizeof(gfxfont_t));
157     memset(font, 0, sizeof(gfxfont_t));
158
159     font->glyphs = (gfxglyph_t*)malloc(sizeof(gfxglyph_t)*src->num_glyphs);
160     memset(font->glyphs, 0, sizeof(gfxglyph_t)*src->num_glyphs);
161     font->id = 0;
162     int t;
163
164     double quality = (INTERNAL_FONT_SIZE * 200 / config_fontquality) / src->max_size;
165     double scale = 1;
166     //printf("%d glyphs\n", font->num_glyphs);
167     font->num_glyphs = 0;
168     font->ascent = fabs(src->ascender);
169     font->descent = fabs(src->descender);
170     
171     for(t=0;t<src->num_glyphs;t++) {
172         if(src->glyphs[t]) {
173             SplashPath*path = src->glyphs[t]->path;
174             int len = path?path->getLength():0;
175             //printf("glyph %d) %08x (%d line segments)\n", t, path, len);
176             gfxglyph_t*glyph = &font->glyphs[font->num_glyphs];
177             src->glyphs[t]->glyphid = font->num_glyphs;
178             glyph->unicode = src->glyphs[t]->unicode;
179             if(glyph->unicode >= font->max_unicode)
180                 font->max_unicode = glyph->unicode+1;
181             gfxdrawer_t drawer;
182             gfxdrawer_target_gfxline(&drawer);
183             int s;
184             int count = 0;
185             double xmax = 0;
186             for(s=0;s<len;s++) {
187                 Guchar f;
188                 double x, y;
189                 path->getPoint(s, &x, &y, &f);
190                 if(!s || x > xmax)
191                     xmax = x;
192                 if(f&splashPathFirst) {
193                     drawer.moveTo(&drawer, x*scale, y*scale);
194                 }
195                 if(f&splashPathCurve) {
196                     double x2,y2;
197                     path->getPoint(++s, &x2, &y2, &f);
198                     if(f&splashPathCurve) {
199                         double x3,y3;
200                         path->getPoint(++s, &x3, &y3, &f);
201                         gfxdraw_cubicTo(&drawer, x*scale, y*scale, x2*scale, y2*scale, x3*scale, y3*scale, quality);
202                     } else {
203                         drawer.splineTo(&drawer, x*scale, y*scale, x2*scale, y2*scale);
204                     }
205                 } else {
206                     drawer.lineTo(&drawer, x*scale, y*scale);
207                 }
208              //   printf("%f %f %s %s\n", x, y, (f&splashPathCurve)?"curve":"",
209              //                           (f&splashPathFirst)?"first":"",
210              //                           (f&splashPathLast)?"last":"");
211             }
212
213             glyph->line = (gfxline_t*)drawer.result(&drawer);
214             if(src->glyphs[t]->advance>0) {
215                 glyph->advance = src->glyphs[t]->advance;
216             } else {
217                 glyph->advance = xmax*scale;
218             }
219             if(config_bigchar) {
220                 double max = src->glyphs[t]->advance_max;
221                 if(max>0 && max > glyph->advance) {
222                     glyph->advance = max;
223                 }
224             }
225
226             font->num_glyphs++;
227         }
228     }
229
230     gfxfont_fix_unicode(font);
231
232     int kerning_size = 0;
233     for(t=0;t<src->num_glyphs;t++) {
234         dict_t* d = src->kerning[t];
235         if(!d) continue;
236         DICT_ITERATE_ITEMS(d,void*,key,mtf_t*,m) {
237             if(m) {
238                 kerning_size++;
239             }
240         }
241     }
242     font->kerning_size = kerning_size;
243     font->kerning = (gfxkerning_t*)malloc(sizeof(gfxkerning_t)*kerning_size);
244     int pos = 0;
245     for(t=0;t<src->num_glyphs;t++) {
246         dict_t* d = src->kerning[t];
247         if(!d) continue;
248         DICT_ITERATE_ITEMS(d,void*,key,mtf_t*,m) {
249             if(m) {
250                 font->kerning[pos].c1 = src->glyphs[t]->glyphid;
251                 font->kerning[pos].c2 = src->glyphs[(int)(ptroff_t)key]->glyphid;
252                 font->kerning[pos].advance = (int)(ptroff_t)m->first->key;
253                 pos++;
254             }
255         }
256     }
257     //int advance = (int)(ptroff_t)m->first->key;
258
259     return font;
260 }
261
262 static float find_average_glyph_advance(gfxfont_t*f)
263 {
264     if(!f->num_glyphs)
265         return 0.0;
266
267     float*values = (float*)malloc(sizeof(float)*f->num_glyphs);
268     int t;
269     for(t=0;t<f->num_glyphs;t++) {
270         values[t] = f->glyphs[t].advance;
271     }
272     float m = medianf(values, f->num_glyphs);
273     free(values);
274     return m;
275 }
276
277 gfxfont_t* FontInfo::getGfxFont()
278 {
279     if(!this->gfxfont) {
280         this->gfxfont = createGfxFont(this);
281         this->gfxfont->id = strdup(this->id);
282         this->space_char = findSpace(this->gfxfont);
283         this->average_advance = find_average_glyph_advance(this->gfxfont);
284
285         if(this->space_char>=0) {
286             msg("<debug> Font %s has space char %d (unicode=%d)", 
287                     this->id, this->space_char, 
288                     this->gfxfont->glyphs[this->space_char].unicode);
289         } else if(config_addspace) {
290             this->space_char = addSpace(this->gfxfont);
291             msg("<debug> Appending space char to font %s, position %d, width %f", this->gfxfont->id, this->space_char, this->gfxfont->glyphs[this->space_char].advance);
292         }
293     }
294     return this->gfxfont;
295 }
296
297 GBool InfoOutputDev::upsideDown() {return gTrue;}
298 GBool InfoOutputDev::useDrawChar() {return gTrue;}
299 GBool InfoOutputDev::interpretType3Chars() {return gTrue;}
300 GBool InfoOutputDev::useTilingPatternFill() {return gTrue;}
301
302 void InfoOutputDev::startPage(int pageNum, GfxState *state, double crop_x1, double crop_y1, double crop_x2, double crop_y2)
303 {
304     double x1,y1,x2,y2;
305     state->transform(crop_x1,crop_y1,&x1,&y1);
306     state->transform(crop_x2,crop_y2,&x2,&y2);
307     if(x2<x1) {double x3=x1;x1=x2;x2=x3;}
308     if(y2<y1) {double y3=y1;y1=y2;y2=y3;}
309     this->x1 = (int)x1;
310     this->y1 = (int)y1;
311     this->x2 = (int)x2;
312     this->y2 = (int)y2;
313     msg("<verbose> Generating info structure for page %d", pageNum);
314 }
315 void InfoOutputDev::endPage()
316 {
317 }
318 void InfoOutputDev::drawLink(Link *link, Catalog *catalog) 
319 {
320     num_links++;
321 }
322    
323 /*  } else if(!strcmp(key,"fontquality")) {
324         this->config_fontquality = atof(value);
325         if(this->config_fontquality<=1)
326             this->config_fontquality=1;
327     } else if(!strcmp(key,"bigchar")) {
328         this->config_bigchar = atoi(value);
329     }
330  */
331
332 double InfoOutputDev::getMaximumFontSize(char*id)
333 {
334     FontInfo*info = (FontInfo*)id2font->lookup(id);
335     if(!info) {
336         msg("<error> Unknown font id: %s", id);
337         return 0.0;
338     }
339     return info->max_size;
340 }
341
342 char*getFontID(GfxFont*font)
343 {
344     Ref*ref = font->getID();
345     GString*gstr = font->getName();
346     char* fname = gstr==0?0:gstr->getCString();
347     char buf[128];
348     if(fname==0) {
349         if(font->getType() == fontType3) {
350             sprintf(buf, "t3font-%d-%d", ref->num, ref->gen);
351         } else {
352             sprintf(buf, "font-%d-%d", ref->num, ref->gen);
353         }
354     } else {
355         sprintf(buf, "%s-%d-%d", fname, ref->num, ref->gen);
356     }
357     return strdup(buf);
358 }
359
360 void InfoOutputDev::updateFont(GfxState *state) 
361 {
362     GfxFont*font = state->getFont();
363     if(!font) {
364         currentfont = 0;
365         return;
366     }
367     if(font->getType() == fontType3) {
368         currentfont = 0;
369         return;
370     }
371     char*id = getFontID(font);
372
373     if(currentfont)
374         currentfont->splash_font = 0;
375
376     currentfont = (FontInfo*)id2font->lookup(id);
377     if(!currentfont) {
378         currentfont = new FontInfo(id);
379         currentfont->font = font;
380         currentfont->max_size = 0;
381         GString* idStr = new GString(id);
382         id2font->add(idStr, (void*)currentfont);
383         num_fonts++;
384     }
385
386     state->setCTM(1.0,0,0,1.0,0,0);
387     splash->updateCTM(state, 0,0,0,0,0,0);
388     state->setTextMat(1.0,0,0,1.0,0,0);
389     state->setFont(font, 1024.0);
390     splash->doUpdateFont(state);
391     currentfont->splash_font = splash->getCurrentFont();
392     if(currentfont->splash_font) {
393         currentfont->ascender = currentfont->splash_font->ascender;
394         currentfont->descender = currentfont->splash_font->descender;
395     } else {
396         currentfont->ascender = currentfont->descender = 0;
397     }
398
399     free(id);
400 }
401
402 void InfoOutputDev::fill(GfxState *state)
403 {
404     num_polygons++;
405 }
406
407 void InfoOutputDev::eoFill(GfxState *state)
408 {
409     num_polygons++;
410 }
411
412 FontInfo* InfoOutputDev::getFont(char*id)
413 {
414     return (FontInfo*)id2font->lookup(id);
415 }
416
417 void InfoOutputDev::drawChar(GfxState *state, double x, double y,
418                       double dx, double dy,
419                       double originX, double originY,
420                       CharCode code, int nBytes, Unicode *u, int uLen)
421 {
422     double m11,m21,m12,m22;
423     state->getFontTransMat(&m11, &m12, &m21, &m22);
424     m11 *= state->getHorizScaling();
425     m21 *= state->getHorizScaling();
426     double lenx = sqrt(m11*m11 + m12*m12);
427     double leny = sqrt(m21*m21 + m22*m22);
428     double len = lenx>leny?lenx:leny;
429     if(!currentfont || !currentfont->splash_font) {
430         return; //error
431     }
432     if(currentfont && currentfont->max_size < len) {
433         currentfont->max_size = len;
434     }
435     
436     num_textfields++;
437
438     currentfont->grow(code+1);
439     GlyphInfo*g = currentfont->glyphs[code];
440     if(!g) {
441         g = currentfont->glyphs[code] = new GlyphInfo();
442         g->advance_max = 0;
443         currentfont->splash_font->last_advance = -1;
444         g->path = currentfont->splash_font->getGlyphPath(code);
445         g->advance = currentfont->splash_font->last_advance;
446         g->unicode = 0;
447     }
448     if(uLen && ((u[0]>=32 && u[0]<g->unicode) || !g->unicode)) {
449         g->unicode = u[0];
450     }
451     if(currentfont->lastchar>=0 && currentfont->lasty == y) {
452         double xshift = (x - currentfont->lastx);
453         if(xshift>=0 && xshift > g->advance_max) {
454             g->advance_max = xshift;
455         }
456         int advance = (int)xshift;
457         if(advance>=0 && advance<g->advance*4 && advance!=currentfont->lastadvance) {
458             int c1 = currentfont->lastchar;
459             int c2 = code;
460             dict_t*d = currentfont->kerning[c1];
461             if(!d) {
462                 d = currentfont->kerning[c1] = dict_new2(&int_type);
463             }
464             mtf_t*k = (mtf_t*)dict_lookup(d, (void*)(ptroff_t)c2);
465             if(!k) {
466                 k = mtf_new(&int_type);
467                 dict_put(d, (void*)(ptroff_t)c2, k);
468             }
469             mtf_increase(k, (void*)(ptroff_t)advance);
470         }
471     }
472
473     currentfont->lastx = x;
474     currentfont->lasty = y;
475     currentfont->lastchar = code;
476     currentfont->lastadvance = (int)(g->advance+0.5);
477 }
478
479 GBool InfoOutputDev::beginType3Char(GfxState *state, double x, double y, double dx, double dy, CharCode code, Unicode *u, int uLen)
480 {
481     GfxFont*font = state->getFont();
482     if(!font)
483         return gTrue;
484     if(font->getType() != fontType3)
485         return gTrue;
486
487     char*id = getFontID(font);
488     currentfont = (FontInfo*)id2font->lookup(id);
489     if(!currentfont) {
490         currentfont = new FontInfo(id);
491         currentfont->font = font;
492         GString* idStr = new GString(id);
493         id2font->add(idStr, (void*)currentfont);
494         num_fonts++;
495     }
496     currentfont = currentfont;
497     free(id);
498
499     currentfont->grow(code+1);
500     if(!currentfont->glyphs[code]) {
501         currentglyph = currentfont->glyphs[code] = new GlyphInfo();
502         currentglyph->unicode = uLen?u[0]:0;
503         currentglyph->path = new SplashPath();
504         currentglyph->x1=0;
505         currentglyph->y1=0;
506         currentglyph->x2=dx;
507         currentglyph->y2=dy;
508         currentglyph->advance=dx;
509         return gFalse;
510     } else {
511         return gTrue;
512     }
513 }
514
515 void InfoOutputDev::type3D0(GfxState *state, double wx, double wy)
516 {
517     currentglyph->x1=0;
518     currentglyph->y1=0;
519     currentglyph->x2=wx;
520     currentglyph->y2=wy;
521 }
522
523 void InfoOutputDev::type3D1(GfxState *state, double wx, double wy, double llx, double lly, double urx, double ury)
524 {
525     if(-lly>currentfont->descender)
526         currentfont->descender = -lly;
527     if(ury>currentfont->ascender)
528         currentfont->ascender = ury;
529
530     currentglyph->x1=llx;
531     currentglyph->y1=lly;
532     currentglyph->x2=urx;
533     currentglyph->y2=ury;
534 }
535
536 void InfoOutputDev::endType3Char(GfxState *state)
537 {
538     double x1 = currentglyph->x1;
539     double y1 = currentglyph->y1;
540     double x2 = currentglyph->x2;
541     double y2 = currentglyph->y2;
542     currentglyph->path->moveTo(x1,y1);
543     currentglyph->path->lineTo(x2,y1);
544     currentglyph->path->lineTo(x2,y2);
545     currentglyph->path->lineTo(x1,y2);
546     currentglyph->path->close();
547 }
548
549 void InfoOutputDev::drawImageMask(GfxState *state, Object *ref, Stream *str,
550                            int width, int height, GBool invert,
551                            GBool inlineImg) 
552 {
553     if(str->getKind()==strDCT) num_jpeg_images++; else num_ppm_images++;
554
555     OutputDev::drawImageMask(state,ref,str,width,height,invert,inlineImg);
556 }
557 void InfoOutputDev::drawImage(GfxState *state, Object *ref, Stream *str,
558                        int width, int height, GfxImageColorMap *colorMap,
559                        int *maskColors, GBool inlineImg)
560 {
561     if(str->getKind()==strDCT) num_jpeg_images++; else num_ppm_images++;
562
563     OutputDev::drawImage(state,ref,str,width,height,colorMap,maskColors,inlineImg);
564 }
565 void InfoOutputDev::drawMaskedImage(GfxState *state, Object *ref, Stream *str,
566                                 int width, int height,
567                                 GfxImageColorMap *colorMap,
568                                 Stream *maskStr,
569                                 int maskWidth, int maskHeight,
570                                 GBool maskInvert) 
571 {
572     if(str->getKind()==strDCT) num_jpeg_images++; else num_ppm_images++;
573
574     OutputDev::drawMaskedImage(state,ref,str,width,height,colorMap,maskStr,maskWidth,maskHeight,maskInvert);
575 }
576
577 void InfoOutputDev::drawSoftMaskedImage(GfxState *state, Object *ref, Stream *str,
578                                     int width, int height,
579                                     GfxImageColorMap *colorMap,
580                                     Stream *maskStr,
581                                     int maskWidth, int maskHeight,
582                                     GfxImageColorMap *maskColorMap) 
583 {
584     if(str->getKind()==strDCT) num_jpeg_images++; else num_ppm_images++;
585
586     OutputDev::drawSoftMaskedImage(state,ref,str,width,height,colorMap,maskStr,maskWidth,maskHeight,maskColorMap);
587 }
588     
589 void InfoOutputDev::dumpfonts(gfxdevice_t*dev)
590 {
591     GHashIter*i;
592     GString*key;
593     FontInfo*font;
594     id2font->startIter(&i);
595     while(id2font->getNext(&i, &key, (void**)&font)) {
596         dev->addfont(dev, font->getGfxFont());
597     }
598 }