Add an error check for correct formatting in Deflate uncompressed
[sgt/halibut] / bk_ps.c
1 /*
2 * PostScript backend for Halibut
3 */
4
5 #include <assert.h>
6 #include <stdarg.h>
7 #include "halibut.h"
8 #include "paper.h"
9
10 /* Ideal number of characters per line, for use in PostScript code */
11 #define PS_WIDTH 79
12 /* Absolute maxiumum characters per line, for use in DSC comments */
13 #define PS_MAXWIDTH 255
14
15 static void ps_comment(FILE *fp, char const *leader, word *words);
16 static void ps_string_len(FILE *fp, int *cc, char const *str, int len);
17 static void ps_string(FILE *fp, int *cc, char const *str);
18
19 paragraph *ps_config_filename(char *filename)
20 {
21 return cmdline_cfg_simple("ps-filename", filename, NULL);
22 }
23
24 void ps_backend(paragraph *sourceform, keywordlist *keywords,
25 indexdata *idx, void *vdoc) {
26 document *doc = (document *)vdoc;
27 int font_index;
28 font_encoding *fe;
29 page_data *page;
30 int pageno;
31 FILE *fp;
32 char *filename;
33 paragraph *p;
34 outline_element *oe;
35 int noe;
36 int cc; /* Character count on current line */
37
38 IGNORE(keywords);
39 IGNORE(idx);
40
41 filename = dupstr("output.ps");
42 for (p = sourceform; p; p = p->next) {
43 if (p->type == para_Config) {
44 if (!ustricmp(p->keyword, L"ps-filename")) {
45 sfree(filename);
46 filename = dupstr(adv(p->origkeyword));
47 }
48 }
49 }
50
51 if (!strcmp(filename, "-"))
52 fp = stdout;
53 else
54 fp = fopen(filename, "w");
55 if (!fp) {
56 error(err_cantopenw, filename);
57 return;
58 }
59
60 fprintf(fp, "%%!PS-Adobe-3.0\n");
61 fprintf(fp, "%%%%Creator: Halibut, %s\n", version);
62 fprintf(fp, "%%%%DocumentData: Clean7Bit\n");
63 fprintf(fp, "%%%%LanguageLevel: 1\n");
64 for (pageno = 0, page = doc->pages; page; page = page->next)
65 pageno++;
66 fprintf(fp, "%%%%Pages: %d\n", pageno);
67 for (p = sourceform; p; p = p->next)
68 if (p->type == para_Title)
69 ps_comment(fp, "%%Title: ", p->words);
70 fprintf(fp, "%%%%DocumentNeededResources:\n");
71 for (fe = doc->fonts->head; fe; fe = fe->next)
72 /* XXX This may request the same font multiple times. */
73 if (!fe->font->info->fontfile)
74 fprintf(fp, "%%%%+ font %s\n", fe->font->info->name);
75 fprintf(fp, "%%%%DocumentSuppliedResources: procset Halibut 0 3\n");
76 for (fe = doc->fonts->head; fe; fe = fe->next)
77 /* XXX This may request the same font multiple times. */
78 if (fe->font->info->fontfile)
79 fprintf(fp, "%%%%+ font %s\n", fe->font->info->name);
80 fprintf(fp, "%%%%EndComments\n");
81
82 fprintf(fp, "%%%%BeginProlog\n");
83 fprintf(fp, "%%%%BeginResource: procset Halibut 0 3\n");
84 /*
85 * Supply a prologue function which allows a reasonably
86 * compressed representation of the text on the pages.
87 *
88 * "t" expects two arguments: a y-coordinate, and then an array.
89 * Elements of the array are processed sequentially as follows:
90 *
91 * - a number is treated as an x-coordinate
92 * - an array is treated as a (font, size) pair
93 * - a string is shown
94 *
95 * "r" takes four arguments, and behaves like "rectfill".
96 */
97 fprintf(fp,
98 "/tdict 4 dict dup begin\n"
99 " /arraytype {aload pop scalefont setfont} bind def\n"
100 " /realtype {1 index moveto} bind def\n"
101 " /integertype /realtype load def\n"
102 " /stringtype {show} bind def\n"
103 "end def\n"
104 "/t { tdict begin {dup type exec} forall end pop } bind def\n"
105 "/r { 4 2 roll moveto 1 index 0 rlineto 0 exch rlineto\n"
106 " neg 0 rlineto closepath fill } bind def\n");
107 /*
108 * pdfmark wrappers
109 *
110 * "p" generates a named destination referencing this page.
111 * "x" generates a link to a named destination.
112 * "u" generates a link to a URI.
113 * "o" generates an outline entry.
114 * "m" generates a general pdfmark.
115 *
116 * They all do nothing if pdfmark is undefined.
117 */
118 fprintf(fp,
119 "/pdfmark where { pop\n"
120 " /p { [ /Dest 3 -1 roll /View [ /XYZ null null null ]\n"
121 " /DEST pdfmark } bind def\n"
122 " /x { [ /Dest 3 -1 roll /Rect 5 -1 roll /Border [0 0 0]\n"
123 " /Subtype /Link /ANN pdfmark } bind def\n"
124 " /u { 2 dict dup /Subtype /URI put dup /URI 4 -1 roll put\n"
125 " [ /Action 3 -1 roll /Rect 5 -1 roll /Border [0 0 0]\n"
126 " /Subtype /Link /ANN pdfmark } bind def\n"
127 " /o { [ /Count 3 -1 roll /Dest 5 -1 roll /Title 7 -1 roll\n"
128 " /OUT pdfmark } bind def\n"
129 " /m /pdfmark load def\n"
130 "}\n");
131 fprintf(fp, "{\n"
132 " /p { pop } bind def\n"
133 " /x { pop pop } bind def\n"
134 " /u /x load def\n"
135 " /o { pop pop pop } bind def\n"
136 " /m /cleartomark load def\n"
137 "} ifelse\n");
138
139 fprintf(fp, "%%%%EndResource\n");
140 fprintf(fp, "%%%%EndProlog\n");
141
142 fprintf(fp, "%%%%BeginSetup\n");
143
144 /*
145 * Assign a destination name to each page for pdfmark purposes.
146 */
147 pageno = 0;
148 for (page = doc->pages; page; page = page->next) {
149 char *buf;
150 pageno++;
151 buf = snewn(12, char);
152 sprintf(buf, "/p%d", pageno);
153 page->spare = buf;
154 }
155
156 /*
157 * This is as good a place as any to put version IDs.
158 */
159 for (p = sourceform; p; p = p->next)
160 if (p->type == para_VersionID)
161 ps_comment(fp, "% ", p->words);
162
163 cc = 0;
164 /*
165 * Request the correct page size. We might want to bracket this
166 * with "%%BeginFeature: *PageSize A4" or similar, and "%%EndFeature",
167 * but that would require us to have a way of getting the name of
168 * the page size given its dimensions.
169 */
170 ps_token(fp, &cc, "/setpagedevice where {\n");
171 ps_token(fp, &cc, " pop 2 dict dup /PageSize [%g %g] put setpagedevice\n",
172 doc->paper_width / FUNITS_PER_PT,
173 doc->paper_height / FUNITS_PER_PT);
174 ps_token(fp, &cc, "} if\n");
175
176 ps_token(fp, &cc, "[/PageMode/UseOutlines/DOCVIEW m\n");
177 noe = doc->n_outline_elements;
178 for (oe = doc->outline_elements; noe; oe++, noe--) {
179 char *title;
180 int titlelen, count, i;
181
182 title = pdf_outline_convert(oe->pdata->outline_title, &titlelen);
183 if (oe->level == 0) {
184 ps_token(fp, &cc, "[/Title");
185 ps_string_len(fp, &cc, title, titlelen);
186 ps_token(fp, &cc, "/DOCINFO m\n");
187 }
188
189 count = 0;
190 for (i = 1; i < noe && oe[i].level > oe->level; i++)
191 if (oe[i].level == oe->level + 1)
192 count++;
193 if (oe->level > 0) count = -count;
194
195 ps_string_len(fp, &cc, title, titlelen);
196 sfree(title);
197 ps_token(fp, &cc, "%s %d o\n",
198 (char *)oe->pdata->first->page->spare, count);
199 }
200
201 for (fe = doc->fonts->head; fe; fe = fe->next) {
202 /* XXX This may request the same font multiple times. */
203 if (fe->font->info->fontfile) {
204 fprintf(fp, "%%%%BeginResource: font %s\n", fe->font->info->name);
205 if (fe->font->info->filetype == TYPE1)
206 pf_writeps(fe->font->info, fp);
207 else
208 sfnt_writeps(fe->font->info, fp);
209 fprintf(fp, "%%%%EndResource\n");
210 } else {
211 fprintf(fp, "%%%%IncludeResource: font %s\n",
212 fe->font->info->name);
213 }
214 }
215
216 /*
217 * Re-encode the fonts.
218 */
219 font_index = 0;
220 for (fe = doc->fonts->head; fe; fe = fe->next) {
221 char fname[40];
222 int i;
223
224 sprintf(fname, "f%d", font_index++);
225 fe->name = dupstr(fname);
226
227 ps_token(fp, &cc, "/%s findfont dup length dict begin\n",
228 fe->font->info->name);
229 ps_token(fp, &cc, "{1 index /FID ne {def} {pop pop} ifelse} forall\n");
230 ps_token(fp, &cc, "/Encoding [\n");
231 for (i = 0; i < 256; i++)
232 ps_token(fp, &cc, "/%s", glyph_extern(fe->vector[i]));
233 ps_token(fp, &cc, "] def\n");
234 ps_token(fp, &cc, "currentdict end\n");
235 ps_token(fp, &cc, "/fontname-%s exch definefont /%s exch def\n",
236 fe->name, fe->name);
237 }
238 fprintf(fp, "%%%%EndSetup\n");
239
240 /*
241 * Output the text and graphics.
242 */
243 pageno = 0;
244 for (page = doc->pages; page; page = page->next) {
245 text_fragment *frag, *frag_end;
246 rect *r;
247 xref *xr;
248 font_encoding *fe;
249 int fs;
250
251 pageno++;
252 fprintf(fp, "%%%%Page: %d %d\n", pageno, pageno);
253 cc = 0;
254 ps_token(fp, &cc, "save %s p\n", (char *)page->spare);
255
256 for (xr = page->first_xref; xr; xr = xr->next) {
257 ps_token(fp, &cc, "[%g %g %g %g]",
258 xr->lx/FUNITS_PER_PT, xr->by/FUNITS_PER_PT,
259 xr->rx/FUNITS_PER_PT, xr->ty/FUNITS_PER_PT);
260 if (xr->dest.type == PAGE) {
261 ps_token(fp, &cc, "%s x\n", (char *)xr->dest.page->spare);
262 } else {
263 ps_string(fp, &cc, xr->dest.url);
264 ps_token(fp, &cc, "u\n");
265 }
266 }
267
268 for (r = page->first_rect; r; r = r->next) {
269 ps_token(fp, &cc, "%g %g %g %g r\n",
270 r->x / FUNITS_PER_PT, r->y / FUNITS_PER_PT,
271 r->w / FUNITS_PER_PT, r->h / FUNITS_PER_PT);
272 }
273
274 frag = page->first_text;
275 fe = NULL;
276 fs = -1;
277 while (frag) {
278 /*
279 * Collect all the adjacent text fragments with the
280 * same y-coordinate.
281 */
282 for (frag_end = frag;
283 frag_end && frag_end->y == frag->y;
284 frag_end = frag_end->next);
285
286 ps_token(fp, &cc, "%g[", frag->y / FUNITS_PER_PT);
287
288 while (frag && frag != frag_end) {
289
290 if (frag->fe != fe || frag->fontsize != fs)
291 ps_token(fp, &cc, "[%s %d]",
292 frag->fe->name, frag->fontsize);
293 fe = frag->fe;
294 fs = frag->fontsize;
295
296 ps_token(fp, &cc, "%g", frag->x/FUNITS_PER_PT);
297 ps_string(fp, &cc, frag->text);
298
299 frag = frag->next;
300 }
301
302 ps_token(fp, &cc, "]t\n");
303 }
304
305 ps_token(fp, &cc, "restore showpage\n");
306 }
307
308 fprintf(fp, "%%%%EOF\n");
309
310 if (fp != stdout)
311 fclose(fp);
312
313 sfree(filename);
314 }
315
316 static void ps_comment(FILE *fp, char const *leader, word *words) {
317 int cc = 0;
318
319 cc += fprintf(fp, "%s", leader);
320
321 for (; words; words = words->next) {
322 char *text;
323 int type;
324
325 switch (words->type) {
326 case word_HyperLink:
327 case word_HyperEnd:
328 case word_UpperXref:
329 case word_LowerXref:
330 case word_XrefEnd:
331 case word_IndexRef:
332 continue;
333 }
334
335 type = removeattr(words->type);
336
337 switch (type) {
338 case word_Normal:
339 text = utoa_dup(words->text, CS_ASCII);
340 break;
341 case word_WhiteSpace:
342 text = dupstr(" ");
343 break;
344 case word_Quote:
345 text = dupstr("'");
346 break;
347 }
348
349 if (cc + strlen(text) > PS_MAXWIDTH)
350 text[PS_MAXWIDTH - cc] = 0;
351 cc += fprintf(fp, "%s", text);
352 sfree(text);
353 }
354
355 fprintf(fp, "\n");
356 }
357
358 void ps_token(FILE *fp, int *cc, char const *fmt, ...) {
359 va_list ap;
360
361 va_start(ap, fmt);
362 if (*cc >= PS_WIDTH - 10) {
363 fprintf(fp, "\n");
364 *cc = 0;
365 }
366 *cc += vfprintf(fp, fmt, ap);
367 /* Assume that \n only occurs at the end of a string */
368 if (fmt[strlen(fmt) - 1] == '\n')
369 *cc = 0;
370 }
371
372 static void ps_string_len(FILE *fp, int *cc, char const *str, int len) {
373 char const *c;
374 int score = 0;
375
376 for (c = str; c < str+len; c++) {
377 if (*c < ' ' || *c > '~')
378 score += 2;
379 else if (*c == '(' || *c == ')' || *c == '\\')
380 score += 0;
381 else
382 score -= 1;
383 }
384 if (score > 0) {
385 ps_token(fp, cc, "<");
386 for (c = str; c < str+len; c++) {
387 ps_token(fp, cc, "%02X", 0xFF & (int)*c);
388 }
389 ps_token(fp, cc, ">");
390 } else {
391 *cc += fprintf(fp, "(");
392 for (c = str; c < str+len; c++) {
393 if (*cc >= PS_WIDTH - 4) {
394 fprintf(fp, "\\\n");
395 *cc = 0;
396 }
397 if (*c < ' ' || *c > '~') {
398 *cc += fprintf(fp, "\\%03o", 0xFF & (int)*c);
399 } else {
400 if (*c == '(' || *c == ')' || *c == '\\') {
401 fputc('\\', fp);
402 (*cc)++;
403 }
404 fputc(*c, fp);
405 (*cc)++;
406 }
407 }
408 *cc += fprintf(fp, ")");
409 }
410 }
411
412 static void ps_string(FILE *fp, int *cc, char const *str) {
413 ps_string_len(fp, cc, str, strlen(str));
414 }