3e2dd889 |
1 | /* |
2 | * PostScript Type 1 font file support for Halibut |
3 | */ |
4 | /* |
5 | * Type 1 font file formats are specified by Adobe Technical Note |
6 | * #5040: "Supporting Downloadable PostScript Language Fonts". |
7 | * Halibut supports hexadecimal format (section 3.1) and IBM PC format |
8 | * (section 3.3), commonly called PFA and PFB respectively. |
9 | */ |
10 | |
11 | #include <assert.h> |
12 | #include <limits.h> |
44407fea |
13 | #include <stdio.h> |
3e2dd889 |
14 | #include <stdlib.h> |
44407fea |
15 | #include "halibut.h" |
16 | #include "paper.h" |
17 | |
3e2dd889 |
18 | #define PFB_ASCII 1 |
19 | #define PFB_BINARY 2 |
20 | #define PFB_EOF 3 |
21 | |
22 | typedef struct t1_font_Tag t1_font; |
23 | typedef struct t1_data_Tag t1_data; |
24 | |
25 | struct t1_font_Tag { |
26 | t1_data *data; |
27 | size_t length1; |
28 | size_t length2; |
29 | filepos pos; |
30 | }; |
31 | |
32 | struct t1_data_Tag { |
33 | char type; |
34 | size_t length; |
35 | unsigned char *data; |
36 | t1_data *next; |
37 | }; |
38 | |
39 | typedef struct pfstate_Tag { |
40 | t1_data *data; |
41 | t1_data *curblock; |
42 | size_t offset; |
43 | } pfstate; |
44 | |
45 | static void pf_identify(t1_font *tf); |
46 | |
47 | static t1_data *load_pfb_file(FILE *fp, filepos *pos) { |
48 | t1_data *head = NULL, *tail = NULL; |
49 | int c, i; |
50 | char type; |
51 | |
52 | pos->line = 0; |
53 | for (;;) { |
54 | if (fgetc(fp) != 128) abort(); |
55 | type = fgetc(fp); |
56 | if (type == PFB_EOF) return head; |
57 | if (tail) { |
58 | tail->next = snew(t1_data); |
59 | tail = tail->next; |
60 | } else { |
61 | head = snew(t1_data); |
62 | tail = head; |
63 | } |
64 | tail->type = type; |
65 | tail->length = 0; |
66 | for (i = 0; i < 4; i++) { |
67 | c = fgetc(fp); |
68 | if (c == EOF) abort(); |
69 | tail->length |= c << (8 * i); |
70 | } |
71 | tail->data = snewn(tail->length, unsigned char); |
72 | if (fread(tail->data, 1, tail->length, fp) != tail->length) abort(); |
73 | } |
6c64aedd |
74 | tail->next = NULL; |
3e2dd889 |
75 | } |
76 | |
77 | static t1_data *load_pfa_file(FILE *fp, filepos *pos) { |
78 | t1_data *ret = snew(t1_data); |
79 | size_t off = 0, len, got; |
80 | |
81 | pos->line = 0; |
82 | ret->type = PFB_ASCII; |
83 | len = 32768; |
84 | ret->data = snewn(len, unsigned char); |
85 | for (;;) { |
86 | got = fread(ret->data + off, 1, len - off, fp); |
87 | off += got; |
88 | if (off != len) break; |
89 | len *= 2; |
90 | ret->data = sresize(ret->data, len, unsigned char); |
91 | } |
92 | ret->data = sresize(ret->data, off, unsigned char); |
93 | ret->length = off; |
6c64aedd |
94 | ret->next = NULL; |
3e2dd889 |
95 | return ret; |
96 | } |
c885c2ff |
97 | |
44407fea |
98 | void read_pfa_file(input *in) { |
3e2dd889 |
99 | t1_font *tf = snew(t1_font); |
100 | |
101 | tf->data = load_pfa_file(in->currfp, &in->pos); |
102 | tf->pos = in->pos; |
103 | tf->length1 = tf->length2 = 0; |
104 | fclose(in->currfp); |
105 | pf_identify(tf); |
106 | } |
107 | |
108 | void read_pfb_file(input *in) { |
109 | t1_font *tf = snew(t1_font); |
110 | |
111 | tf->data = load_pfb_file(in->currfp, &in->pos); |
112 | tf->pos = in->pos; |
113 | tf->length1 = tf->length2 = 0; |
114 | fclose(in->currfp); |
115 | pf_identify(tf); |
116 | } |
117 | static char *pf_read_token(pfstate *); |
118 | |
119 | /* |
120 | * Read a character from the initial plaintext part of a Type 1 font |
121 | */ |
122 | static int pf_getc(pfstate *pf) { |
123 | if (pf->offset == pf->curblock->length) { |
124 | if (pf->curblock->next == NULL) return EOF; |
125 | pf->curblock = pf->curblock->next; |
126 | pf->offset = 0; |
127 | } |
128 | if (pf->curblock->type != PFB_ASCII) return EOF; |
129 | return pf->curblock->data[pf->offset++]; |
130 | } |
131 | |
132 | static void pf_ungetc(int c, pfstate *pf) { |
133 | assert(pf->offset > 0); |
134 | pf->offset--; |
135 | assert(c == pf->curblock->data[pf->offset]); |
136 | } |
137 | |
138 | static void pf_rewind(pfstate *pf) { |
139 | pf->curblock = pf->data; |
140 | pf->offset = 0; |
141 | } |
142 | |
143 | static void pf_seek(pfstate *pf, size_t off) { |
144 | t1_data *td = pf->data; |
145 | |
146 | while (td->length < off) { |
147 | off -= td->length; |
148 | td = td->next; |
149 | } |
150 | pf->curblock = td; |
151 | pf->offset = off; |
152 | } |
153 | |
154 | static size_t pf_tell(pfstate *pf) { |
155 | t1_data *td = pf->data; |
156 | size_t o = 0; |
157 | |
158 | while (td != pf->curblock) { |
159 | o += td->length; |
160 | td = td->next; |
161 | } |
162 | return o + pf->offset; |
163 | } |
164 | |
165 | static void pf_identify(t1_font *tf) { |
c885c2ff |
166 | rdstringc rsc = { 0, 0, NULL }; |
167 | char *p; |
44407fea |
168 | size_t len; |
169 | char *fontname; |
170 | font_info *fi; |
c885c2ff |
171 | int c; |
3e2dd889 |
172 | pfstate pfs, *pf = &pfs; |
44407fea |
173 | |
3e2dd889 |
174 | pf->data = tf->data; |
175 | pf_rewind(pf); |
c885c2ff |
176 | do { |
3e2dd889 |
177 | c = pf_getc(pf); |
c885c2ff |
178 | if (c == EOF) { |
179 | sfree(rsc.text); |
3e2dd889 |
180 | error(err_pfeof, &tf->pos); |
c885c2ff |
181 | return; |
182 | } |
183 | rdaddc(&rsc, c); |
184 | } while (c != 012 && c != 015); |
185 | p = rsc.text; |
186 | if ((p = strchr(p, ':')) == NULL) { |
187 | sfree(rsc.text); |
3e2dd889 |
188 | error(err_pfhead, &tf->pos); |
44407fea |
189 | return; |
c885c2ff |
190 | } |
191 | p++; |
44407fea |
192 | p += strspn(p, " \t"); |
193 | len = strcspn(p, " \t"); |
194 | fontname = snewn(len + 1, char); |
195 | memcpy(fontname, p, len); |
196 | fontname[len] = 0; |
c885c2ff |
197 | sfree(rsc.text); |
198 | |
44407fea |
199 | for (fi = all_fonts; fi; fi = fi->next) { |
200 | if (strcmp(fi->name, fontname) == 0) { |
3e2dd889 |
201 | fi->fontfile = tf; |
babfe3e2 |
202 | fi->filetype = TYPE1; |
44407fea |
203 | sfree(fontname); |
204 | return; |
205 | } |
206 | } |
3e2dd889 |
207 | error(err_pfnoafm, &tf->pos, fontname); |
44407fea |
208 | sfree(fontname); |
209 | } |
c885c2ff |
210 | |
211 | /* |
212 | * PostScript white space characters; PLRM3 table 3.1 |
213 | */ |
214 | static int pf_isspace(int c) { |
215 | return c == 000 || c == 011 || c == 012 || c == 014 || c == 015 || |
216 | c == ' '; |
217 | } |
218 | |
219 | /* |
220 | * PostScript special characters; PLRM3 page 27 |
221 | */ |
222 | static int pf_isspecial(int c) { |
223 | return c == '(' || c == ')' || c == '<' || c == '>' || c == '[' || |
224 | c == ']' || c == '{' || c == '}' || c == '/' || c == '%'; |
225 | } |
226 | |
3e2dd889 |
227 | static size_t pf_findtoken(t1_font *tf, size_t off, char const *needle) { |
c885c2ff |
228 | char *tok; |
3e2dd889 |
229 | pfstate pfs, *pf = &pfs; |
c885c2ff |
230 | |
3e2dd889 |
231 | pf->data = tf->data; |
232 | pf_seek(pf, off); |
233 | for (;;) { |
234 | tok = pf_read_token(pf); |
235 | if (tok == NULL) { |
236 | if (pf->offset == 0 && pf->curblock->type == PFB_BINARY) |
237 | pf->curblock = pf->curblock->next; |
238 | else |
239 | return (size_t)-1; |
240 | } else { |
241 | if (strcmp(tok, needle) == 0) { |
242 | sfree(tok); |
243 | return pf_tell(pf); |
244 | } |
245 | sfree(tok); |
246 | } |
c885c2ff |
247 | } |
3e2dd889 |
248 | } |
249 | |
250 | static size_t pf_length1(t1_font *tf) { |
251 | size_t ret; |
252 | |
253 | ret = pf_findtoken(tf, 0, "eexec"); |
254 | if (ret == (size_t)-1) { |
255 | error(err_pfeof, &tf->pos); |
c885c2ff |
256 | return 0; |
257 | } |
3e2dd889 |
258 | return ret; |
c885c2ff |
259 | } |
260 | |
3e2dd889 |
261 | static size_t pf_length2(t1_font *tf) { |
262 | size_t ret; |
263 | |
264 | if (tf->length1 == 0) |
265 | tf->length1 = pf_length1(tf); |
266 | ret = pf_findtoken(tf, tf->length1, "cleartomark"); |
267 | if (ret == (size_t)-1) { |
268 | error(err_pfeof, &tf->pos); |
269 | return 0; |
270 | } |
271 | return ret - 12 - tf->length1; /* backspace over "cleartomark\n" */ |
272 | } |
273 | |
274 | static void pf_getascii(t1_font *tf, size_t off, size_t len, |
275 | char **bufp, size_t *lenp) { |
276 | t1_data *td = tf->data; |
277 | size_t blk, i; |
278 | char *p; |
279 | |
280 | while (td && off >= td->length) { |
281 | off -= td->length; |
282 | td = td->next; |
c885c2ff |
283 | } |
3e2dd889 |
284 | *bufp = NULL; |
285 | *lenp = 0; |
286 | while (td && len) { |
287 | blk = len < td->length ? len : td->length; |
288 | if (td->type == PFB_ASCII) { |
289 | *bufp = sresize(*bufp, *lenp + blk, char); |
290 | memcpy(*bufp + *lenp, td->data + off, blk); |
291 | *lenp += blk; |
292 | } else { |
293 | *bufp = sresize(*bufp, *lenp + blk * 2 + blk / 39 + 3, char); |
294 | p = *bufp + *lenp; |
295 | for (i = 0; i < blk; i++) { |
296 | if (i % 39 == 0) p += sprintf(p, "\n"); |
297 | p += sprintf(p, "%02x", td->data[off + i]); |
298 | } |
299 | p += sprintf(p, "\n"); |
300 | *lenp = p - *bufp; |
301 | } |
302 | len -= blk; |
303 | td = td->next; |
304 | off = 0; |
305 | } |
306 | } |
307 | |
308 | void pf_writeps(font_info const *fi, FILE *ofp) { |
309 | char *buf; |
310 | size_t len; |
311 | |
312 | pf_getascii(fi->fontfile, 0, INT_MAX, &buf, &len); |
313 | fwrite(buf, 1, len, ofp); |
314 | sfree(buf); |
c885c2ff |
315 | } |
316 | |
317 | static int hexval(char c) { |
318 | if (c >= '0' && c <= '9') return c - '0'; |
319 | if (c >= 'A' && c <= 'F') return c - 'A' + 0xA; |
320 | if (c >= 'a' && c <= 'f') return c - 'a' + 0xa; |
321 | return 0; |
322 | } |
323 | |
3e2dd889 |
324 | static void pf_getbinary(t1_font *tf, size_t off, size_t len, |
325 | char **bufp, size_t *lenp) { |
326 | t1_data *td = tf->data; |
327 | size_t blk, i; |
328 | int havenybble = 0; |
329 | char *p, nybble; |
330 | |
331 | while (td && off >= td->length) { |
332 | off -= td->length; |
333 | td = td->next; |
334 | } |
335 | *bufp = NULL; |
336 | *lenp = 0; |
337 | while (td && len) { |
338 | blk = len < td->length ? len : td->length; |
339 | if (td->type == PFB_BINARY) { |
340 | *bufp = sresize(*bufp, *lenp + blk, char); |
341 | memcpy(*bufp + *lenp, td->data + off, blk); |
342 | *lenp += blk; |
343 | } else { |
344 | *bufp = sresize(*bufp, *lenp + blk / 2 + 1, char); |
345 | p = *bufp + *lenp; |
346 | for (i = 0; i < blk; i++) { |
347 | if (pf_isspace(td->data[off + i])) continue; |
348 | if (!havenybble) |
349 | nybble = hexval(td->data[off+i]); |
350 | else |
351 | *p++ = (nybble << 4) | hexval(td->data[off+i]); |
352 | havenybble = !havenybble; |
353 | } |
354 | *lenp = p - *bufp; |
355 | } |
356 | len -= blk; |
357 | td = td->next; |
358 | off = 0; |
359 | } |
360 | } |
361 | |
362 | |
363 | /* |
364 | * Return the initial, unencrypted, part of a font. |
365 | */ |
366 | void pf_part1(font_info *fi, char **bufp, size_t *lenp) { |
367 | t1_font *tf = fi->fontfile; |
368 | |
369 | if (tf->length1 == 0) |
370 | tf->length1 = pf_length1(tf); |
371 | pf_getascii(tf, 0, tf->length1, bufp, lenp); |
372 | } |
373 | |
c885c2ff |
374 | /* |
375 | * Return the middle, encrypted, part of a font. |
376 | */ |
377 | void pf_part2(font_info *fi, char **bufp, size_t *lenp) { |
3e2dd889 |
378 | t1_font *tf = fi->fontfile; |
c885c2ff |
379 | |
3e2dd889 |
380 | if (tf->length2 == 0) |
381 | tf->length2 = pf_length2(tf); |
382 | pf_getbinary(tf, tf->length1, tf->length2, bufp, lenp); |
383 | if (*lenp >= 256) |
384 | *lenp -= 256; |
c885c2ff |
385 | } |
386 | |
3e2dd889 |
387 | static char *pf_read_litstring(pfstate *pf) { |
c885c2ff |
388 | rdstringc rsc = { 0, 0, NULL }; |
389 | int depth = 1; |
390 | int c; |
391 | |
392 | rdaddc(&rsc, '('); |
393 | do { |
3e2dd889 |
394 | c = pf_getc(pf); |
c885c2ff |
395 | switch (c) { |
396 | case '(': |
397 | depth++; break; |
398 | case ')': |
399 | depth--; break; |
400 | case '\\': |
401 | rdaddc(&rsc, '\\'); |
3e2dd889 |
402 | c = pf_getc(pf); |
c885c2ff |
403 | break; |
404 | } |
405 | if (c != EOF) rdaddc(&rsc, c); |
406 | } while (depth > 0 && c != EOF); |
407 | return rsc.text; |
408 | } |
409 | |
3e2dd889 |
410 | static char *pf_read_hexstring(pfstate *pf) { |
c885c2ff |
411 | rdstringc rsc = { 0, 0, NULL }; |
412 | int c; |
413 | |
414 | rdaddc(&rsc, '<'); |
415 | do { |
3e2dd889 |
416 | c = pf_getc(pf); |
c885c2ff |
417 | if (c != EOF) rdaddc(&rsc, c); |
418 | } while (c != '>' && c != EOF); |
419 | return rsc.text; |
420 | } |
421 | |
3e2dd889 |
422 | static char *pf_read_word(pfstate *pf, int c) { |
c885c2ff |
423 | rdstringc rsc = { 0, 0, NULL }; |
424 | |
425 | rdaddc(&rsc, c); |
426 | if (c == '{' || c == '}' || c == '[' || c == ']') |
427 | return rsc.text; |
428 | for (;;) { |
3e2dd889 |
429 | c = pf_getc(pf); |
c885c2ff |
430 | if (pf_isspecial(c) || pf_isspace(c) || c == EOF) break; |
431 | rdaddc(&rsc, c); |
432 | } |
3e2dd889 |
433 | if (pf_isspecial(c)) pf_ungetc(c, pf); |
c885c2ff |
434 | return rsc.text; |
435 | } |
436 | |
3e2dd889 |
437 | static char *pf_read_token(pfstate *pf) { |
c885c2ff |
438 | int c; |
439 | |
440 | do { |
3e2dd889 |
441 | c = pf_getc(pf); |
c885c2ff |
442 | } while (pf_isspace(c)); |
443 | if (c == EOF) return NULL; |
444 | if (c == '%') { |
445 | do { |
3e2dd889 |
446 | c = pf_getc(pf); |
c885c2ff |
447 | } while (c != 012 && c != 015); |
3e2dd889 |
448 | return pf_read_token(pf); |
c885c2ff |
449 | } |
3e2dd889 |
450 | if (c == '(') return pf_read_litstring(pf); |
451 | if (c == '<') return pf_read_hexstring(pf); |
452 | return pf_read_word(pf, c); |
c885c2ff |
453 | } |