7136a6c7 |
1 | /* |
2 | * man page backend for Halibut |
3 | */ |
4 | |
5 | #include <stdio.h> |
6 | #include <stdlib.h> |
7 | #include <assert.h> |
8 | #include "halibut.h" |
9 | |
10 | static void man_text(FILE *, word *, int newline, int quote_props); |
11 | static void man_codepara(FILE *, word *); |
4b3c5afb |
12 | static int man_convert(wchar_t *s, int maxlen, |
13 | char **result, int quote_props); |
14 | |
15 | typedef struct { |
16 | wchar_t *th; |
17 | int headnumbers; |
18 | int mindepth; |
19 | } manconfig; |
20 | |
21 | static manconfig man_configure(paragraph *source) { |
22 | manconfig ret; |
23 | |
24 | /* |
25 | * Defaults. |
26 | */ |
27 | ret.th = NULL; |
28 | ret.headnumbers = FALSE; |
29 | ret.mindepth = 0; |
30 | |
31 | for (; source; source = source->next) { |
32 | if (source->type == para_Config) { |
33 | if (!ustricmp(source->keyword, L"man-identity")) { |
34 | wchar_t *wp, *ep; |
35 | |
36 | wp = uadv(source->keyword); |
37 | ep = wp; |
38 | while (*ep) |
39 | ep = uadv(ep); |
40 | ret.th = mknewa(wchar_t, ep - wp + 1); |
41 | memcpy(ret.th, wp, (ep - wp + 1) * sizeof(wchar_t)); |
42 | } else if (!ustricmp(source->keyword, L"man-headnumbers")) { |
43 | ret.headnumbers = utob(uadv(source->keyword)); |
44 | } else if (!ustricmp(source->keyword, L"man-mindepth")) { |
45 | ret.mindepth = utoi(uadv(source->keyword)); |
46 | } |
47 | } |
48 | } |
49 | |
50 | return ret; |
51 | } |
52 | |
53 | static void man_conf_cleanup(manconfig cf) |
54 | { |
55 | sfree(cf.th); |
56 | } |
7136a6c7 |
57 | |
58 | #define QUOTE_INITCTRL 1 /* quote initial . and ' on a line */ |
59 | #define QUOTE_QUOTES 2 /* quote double quotes by doubling them */ |
60 | |
61 | void man_backend(paragraph *sourceform, keywordlist *keywords, |
62 | indexdata *idx) { |
63 | paragraph *p; |
64 | FILE *fp; |
4b3c5afb |
65 | manconfig conf; |
7136a6c7 |
66 | |
67 | IGNORE(keywords); /* we don't happen to need this */ |
68 | IGNORE(idx); /* or this */ |
69 | |
4b3c5afb |
70 | conf = man_configure(sourceform); |
71 | |
7136a6c7 |
72 | /* |
73 | * Determine the output file name, and open the output file |
74 | * |
75 | * FIXME: want configurable output file names here. For the |
76 | * moment, we'll just call it `output.1'. |
77 | */ |
78 | fp = fopen("output.1", "w"); |
79 | if (!fp) { |
80 | error(err_cantopenw, "output.1"); |
81 | return; |
82 | } |
83 | |
84 | /* Do the version ID */ |
85 | for (p = sourceform; p; p = p->next) |
86 | if (p->type == para_VersionID) { |
87 | fprintf(fp, ".\\\" "); |
88 | man_text(fp, p->words, TRUE, 0); |
89 | } |
90 | |
4b3c5afb |
91 | /* .TH name-of-program manual-section */ |
22905f72 |
92 | fprintf(fp, ".TH"); |
93 | if (conf.th && *conf.th) { |
4b3c5afb |
94 | char *c; |
22905f72 |
95 | wchar_t *wp; |
96 | |
97 | for (wp = conf.th; *wp; wp = uadv(wp)) { |
98 | fputs(" \"", fp); |
99 | man_convert(wp, 0, &c, QUOTE_QUOTES); |
100 | fputs(c, fp); |
101 | sfree(c); |
102 | fputc('"', fp); |
4b3c5afb |
103 | } |
104 | } |
22905f72 |
105 | fputc('\n', fp); |
7136a6c7 |
106 | |
107 | fprintf(fp, ".UC\n"); |
108 | |
7136a6c7 |
109 | for (p = sourceform; p; p = p->next) switch (p->type) { |
110 | /* |
111 | * Things we ignore because we've already processed them or |
112 | * aren't going to touch them in this pass. |
113 | */ |
114 | case para_IM: |
115 | case para_BR: |
116 | case para_Biblio: /* only touch BiblioCited */ |
117 | case para_VersionID: |
7136a6c7 |
118 | case para_NoCite: |
119 | case para_Title: |
120 | break; |
121 | |
122 | /* |
123 | * Headings. |
124 | */ |
125 | case para_Chapter: |
126 | case para_Appendix: |
127 | case para_UnnumberedChapter: |
128 | case para_Heading: |
129 | case para_Subsect: |
8902e0ed |
130 | |
4b3c5afb |
131 | { |
132 | int depth; |
133 | if (p->type == para_Subsect) |
134 | depth = p->aux + 2; |
135 | else if (p->type == para_Heading) |
136 | depth = 1; |
137 | else |
138 | depth = 0; |
139 | if (depth >= conf.mindepth) { |
140 | fprintf(fp, ".SH \""); |
141 | if (conf.headnumbers && p->kwtext) { |
142 | man_text(fp, p->kwtext, FALSE, QUOTE_QUOTES); |
143 | fprintf(fp, " "); |
144 | } |
145 | man_text(fp, p->words, FALSE, QUOTE_QUOTES); |
146 | fprintf(fp, "\"\n"); |
147 | } |
148 | break; |
149 | } |
7136a6c7 |
150 | |
151 | /* |
152 | * Code paragraphs. |
153 | */ |
154 | case para_Code: |
155 | fprintf(fp, ".PP\n"); |
156 | man_codepara(fp, p->words); |
157 | break; |
158 | |
159 | /* |
160 | * Normal paragraphs. |
161 | */ |
162 | case para_Normal: |
9057a0a8 |
163 | case para_Copyright: |
7136a6c7 |
164 | fprintf(fp, ".PP\n"); |
165 | man_text(fp, p->words, TRUE, 0); |
166 | break; |
167 | |
168 | /* |
169 | * List paragraphs. |
170 | */ |
171 | case para_Description: |
172 | case para_BiblioCited: |
173 | case para_Bullet: |
174 | case para_NumberedList: |
175 | if (p->type == para_Bullet) { |
176 | fprintf(fp, ".IP \"\\fBo\\fP\"\n"); /* FIXME: configurable? */ |
177 | } else if (p->type == para_NumberedList) { |
178 | fprintf(fp, ".IP \""); |
179 | man_text(fp, p->kwtext, FALSE, QUOTE_QUOTES); |
180 | fprintf(fp, "\"\n"); |
181 | } else if (p->type == para_Description) { |
182 | /* |
183 | * Do nothing; the .xP for this paragraph is the .IP |
184 | * which has come before it in the DescribedThing. |
185 | */ |
186 | } else if (p->type == para_BiblioCited) { |
187 | fprintf(fp, ".IP \""); |
188 | man_text(fp, p->kwtext, FALSE, QUOTE_QUOTES); |
189 | fprintf(fp, "\"\n"); |
190 | } |
191 | man_text(fp, p->words, TRUE, 0); |
192 | break; |
193 | |
194 | case para_DescribedThing: |
195 | fprintf(fp, ".IP \""); |
196 | man_text(fp, p->words, FALSE, QUOTE_QUOTES); |
197 | fprintf(fp, "\"\n"); |
198 | break; |
199 | |
200 | case para_Rule: |
201 | /* |
4b3c5afb |
202 | * This isn't terribly good. Anyone who wants to do better |
203 | * should feel free! |
7136a6c7 |
204 | */ |
4b3c5afb |
205 | fprintf(fp, ".PP\n----------------------------------------\n"); |
7136a6c7 |
206 | break; |
207 | |
208 | case para_LcontPush: |
2614b01d |
209 | case para_QuotePush: |
7136a6c7 |
210 | fprintf(fp, ".RS\n"); |
211 | break; |
212 | case para_LcontPop: |
2614b01d |
213 | case para_QuotePop: |
7136a6c7 |
214 | fprintf(fp, ".RE\n"); |
215 | break; |
216 | } |
217 | |
218 | /* |
219 | * Tidy up. |
220 | */ |
221 | fclose(fp); |
4b3c5afb |
222 | man_conf_cleanup(conf); |
7136a6c7 |
223 | } |
224 | |
225 | /* |
226 | * Convert a wide string into a string of chars. If `result' is |
227 | * non-NULL, mallocs the resulting string and stores a pointer to |
228 | * it in `*result'. If `result' is NULL, merely checks whether all |
229 | * characters in the string are feasible for the output character |
230 | * set. |
231 | * |
232 | * Return is nonzero if all characters are OK. If not all |
233 | * characters are OK but `result' is non-NULL, a result _will_ |
234 | * still be generated! |
235 | * |
236 | * FIXME: Here is probably also a good place to do escaping sorts |
237 | * of things. I know I at least need to escape backslash, and full |
238 | * stops at the starts of words are probably trouble as well. |
239 | */ |
4b3c5afb |
240 | static int man_convert(wchar_t *s, int maxlen, |
241 | char **result, int quote_props) { |
7136a6c7 |
242 | /* |
243 | * FIXME. Currently this is ISO8859-1 only. |
244 | */ |
245 | int doing = (result != 0); |
246 | int ok = TRUE; |
247 | char *p = NULL; |
248 | int plen = 0, psize = 0; |
249 | |
4b3c5afb |
250 | if (maxlen <= 0) |
251 | maxlen = -1; |
252 | |
253 | for (; *s && maxlen != 0; s++, maxlen--) { |
7136a6c7 |
254 | wchar_t c = *s; |
255 | char outc; |
256 | |
257 | if ((c >= 32 && c <= 126) || |
258 | (c >= 160 && c <= 255)) { |
259 | /* Char is OK. */ |
260 | outc = (char)c; |
261 | } else { |
262 | /* Char is not OK. */ |
263 | ok = FALSE; |
264 | outc = 0xBF; /* approximate the good old DEC `uh?' */ |
265 | } |
266 | if (doing) { |
267 | if (plen+3 >= psize) { |
268 | psize = plen + 256; |
269 | p = resize(p, psize); |
270 | } |
271 | if (plen == 0 && (outc == '.' || outc == '\'') && |
272 | (quote_props & QUOTE_INITCTRL)) { |
273 | /* |
274 | * Control character (. or ') at the start of a |
275 | * line. Quote it by putting \& (troff zero-width |
276 | * space) before it. |
277 | */ |
278 | p[plen++] = '\\'; |
279 | p[plen++] = '&'; |
280 | } else if (outc == '\\') { |
281 | /* |
282 | * Quote backslashes by doubling them, always. |
283 | */ |
284 | p[plen++] = '\\'; |
285 | } else if (outc == '"' && (quote_props & QUOTE_QUOTES)) { |
286 | /* |
287 | * Double quote within double quotes. Quote it by |
288 | * doubling. |
289 | */ |
290 | p[plen++] = '"'; |
291 | } |
292 | p[plen++] = outc; |
293 | } |
294 | } |
295 | if (doing) { |
296 | p = resize(p, plen+1); |
297 | p[plen] = '\0'; |
298 | *result = p; |
299 | } |
300 | return ok; |
301 | } |
302 | |
303 | static void man_rdaddwc(rdstringc *rs, word *text, word *end, |
304 | int quote_props) { |
305 | char *c; |
306 | |
307 | for (; text && text != end; text = text->next) switch (text->type) { |
308 | case word_HyperLink: |
309 | case word_HyperEnd: |
310 | case word_UpperXref: |
311 | case word_LowerXref: |
312 | case word_XrefEnd: |
313 | case word_IndexRef: |
314 | break; |
315 | |
316 | case word_Normal: |
317 | case word_Emph: |
318 | case word_Code: |
319 | case word_WeakCode: |
320 | case word_WhiteSpace: |
321 | case word_EmphSpace: |
322 | case word_CodeSpace: |
323 | case word_WkCodeSpace: |
324 | case word_Quote: |
325 | case word_EmphQuote: |
326 | case word_CodeQuote: |
327 | case word_WkCodeQuote: |
328 | assert(text->type != word_CodeQuote && |
329 | text->type != word_WkCodeQuote); |
330 | if (towordstyle(text->type) == word_Emph && |
331 | (attraux(text->aux) == attr_First || |
332 | attraux(text->aux) == attr_Only)) |
333 | rdaddsc(rs, "\\fI"); |
4b3c5afb |
334 | else if ((towordstyle(text->type) == word_Code || |
335 | towordstyle(text->type) == word_WeakCode) && |
7136a6c7 |
336 | (attraux(text->aux) == attr_First || |
337 | attraux(text->aux) == attr_Only)) |
338 | rdaddsc(rs, "\\fB"); |
339 | if (removeattr(text->type) == word_Normal) { |
340 | if (rs->pos > 0) |
341 | quote_props &= ~QUOTE_INITCTRL; /* not at start any more */ |
4b3c5afb |
342 | if (man_convert(text->text, 0, &c, quote_props)) |
7136a6c7 |
343 | rdaddsc(rs, c); |
344 | else |
345 | man_rdaddwc(rs, text->alt, NULL, quote_props); |
346 | sfree(c); |
347 | } else if (removeattr(text->type) == word_WhiteSpace) { |
348 | rdaddc(rs, ' '); |
349 | } else if (removeattr(text->type) == word_Quote) { |
4b3c5afb |
350 | rdaddc(rs, '"'); |
351 | if (quote_props & QUOTE_QUOTES) |
352 | rdaddc(rs, '"'); |
7136a6c7 |
353 | } |
354 | if (towordstyle(text->type) == word_Emph && |
355 | (attraux(text->aux) == attr_Last || |
356 | attraux(text->aux) == attr_Only)) |
357 | rdaddsc(rs, "\\fP"); |
4b3c5afb |
358 | else if ((towordstyle(text->type) == word_Code || |
359 | towordstyle(text->type) == word_WeakCode) && |
7136a6c7 |
360 | (attraux(text->aux) == attr_Last || |
361 | attraux(text->aux) == attr_Only)) |
362 | rdaddsc(rs, "\\fP"); |
363 | break; |
364 | } |
365 | } |
366 | |
367 | static void man_text(FILE *fp, word *text, int newline, int quote_props) { |
368 | rdstringc t = { 0, 0, NULL }; |
369 | |
370 | man_rdaddwc(&t, text, NULL, quote_props | QUOTE_INITCTRL); |
371 | fprintf(fp, "%s", t.text); |
372 | sfree(t.text); |
373 | if (newline) |
374 | fputc('\n', fp); |
375 | } |
376 | |
377 | static void man_codepara(FILE *fp, word *text) { |
378 | fprintf(fp, ".nf\n"); |
379 | for (; text; text = text->next) if (text->type == word_WeakCode) { |
380 | char *c; |
4b3c5afb |
381 | wchar_t *t, *e; |
382 | int quote_props = QUOTE_INITCTRL; |
383 | |
384 | t = text->text; |
385 | if (text->next && text->next->type == word_Emph) { |
386 | e = text->next->text; |
387 | text = text->next; |
388 | } else |
389 | e = NULL; |
390 | |
391 | while (e && *e && *t) { |
392 | int n; |
393 | int ec = *e; |
394 | |
395 | for (n = 0; t[n] && e[n] && e[n] == ec; n++); |
396 | if (ec == 'i') |
397 | fprintf(fp, "\\fI"); |
398 | else if (ec == 'b') |
399 | fprintf(fp, "\\fB"); |
400 | man_convert(t, n, &c, quote_props); |
401 | quote_props &= ~QUOTE_INITCTRL; |
402 | fprintf(fp, "%s", c); |
403 | sfree(c); |
404 | if (ec == 'i' || ec == 'b') |
405 | fprintf(fp, "\\fP"); |
406 | t += n; |
407 | e += n; |
408 | } |
409 | man_convert(t, 0, &c, quote_props); |
7136a6c7 |
410 | fprintf(fp, "%s\n", c); |
411 | sfree(c); |
412 | } |
413 | fprintf(fp, ".fi\n"); |
414 | } |