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