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 *); |
12 | |
13 | #define QUOTE_INITCTRL 1 /* quote initial . and ' on a line */ |
14 | #define QUOTE_QUOTES 2 /* quote double quotes by doubling them */ |
15 | |
16 | void man_backend(paragraph *sourceform, keywordlist *keywords, |
17 | indexdata *idx) { |
18 | paragraph *p; |
19 | FILE *fp; |
20 | char const *sep; |
21 | |
22 | IGNORE(keywords); /* we don't happen to need this */ |
23 | IGNORE(idx); /* or this */ |
24 | |
25 | /* |
26 | * Determine the output file name, and open the output file |
27 | * |
28 | * FIXME: want configurable output file names here. For the |
29 | * moment, we'll just call it `output.1'. |
30 | */ |
31 | fp = fopen("output.1", "w"); |
32 | if (!fp) { |
33 | error(err_cantopenw, "output.1"); |
34 | return; |
35 | } |
36 | |
37 | /* Do the version ID */ |
38 | for (p = sourceform; p; p = p->next) |
39 | if (p->type == para_VersionID) { |
40 | fprintf(fp, ".\\\" "); |
41 | man_text(fp, p->words, TRUE, 0); |
42 | } |
43 | |
44 | /* FIXME: .TH name-of-program manual-section */ |
45 | fprintf(fp, ".TH FIXME 1\n"); |
46 | |
47 | fprintf(fp, ".UC\n"); |
48 | |
49 | /* Do the preamble and copyright */ |
50 | sep = ""; |
51 | for (p = sourceform; p; p = p->next) |
52 | if (p->type == para_Preamble) { |
53 | fprintf(fp, "%s", sep); |
54 | man_text(fp, p->words, TRUE, 0); |
55 | sep = "\n"; |
56 | } |
57 | for (p = sourceform; p; p = p->next) |
58 | if (p->type == para_Copyright) { |
59 | fprintf(fp, "%s", sep); |
60 | man_text(fp, p->words, TRUE, 0); |
61 | sep = "\n"; |
62 | } |
63 | |
64 | /* |
65 | * FIXME: |
66 | * |
67 | * - figure out precisely what needs to be escaped. |
68 | * * A dot or apostrophe at the start of a line wants to be |
69 | * preceded by `\&', which is a zero-width space. |
70 | * * Literal backslashes always want doubling. |
71 | * * Within double quotes, a double quote needs doubling |
72 | * too. |
73 | * |
74 | * - work out what to do about hyphens / minuses... |
75 | */ |
76 | for (p = sourceform; p; p = p->next) switch (p->type) { |
77 | /* |
78 | * Things we ignore because we've already processed them or |
79 | * aren't going to touch them in this pass. |
80 | */ |
81 | case para_IM: |
82 | case para_BR: |
83 | case para_Biblio: /* only touch BiblioCited */ |
84 | case para_VersionID: |
85 | case para_Copyright: |
86 | case para_Preamble: |
87 | case para_NoCite: |
88 | case para_Title: |
89 | break; |
90 | |
91 | /* |
92 | * Headings. |
93 | */ |
94 | case para_Chapter: |
95 | case para_Appendix: |
96 | case para_UnnumberedChapter: |
97 | case para_Heading: |
98 | case para_Subsect: |
99 | fprintf(fp, ".SH \""); |
100 | /* FIXME: disable this, at _least_ by default */ |
101 | if (p->kwtext) |
102 | man_text(fp, p->kwtext, FALSE, QUOTE_QUOTES); |
103 | fprintf(fp, " "); |
104 | man_text(fp, p->words, FALSE, QUOTE_QUOTES); |
105 | fprintf(fp, "\"\n"); |
106 | break; |
107 | |
108 | /* |
109 | * Code paragraphs. |
110 | */ |
111 | case para_Code: |
112 | fprintf(fp, ".PP\n"); |
113 | man_codepara(fp, p->words); |
114 | break; |
115 | |
116 | /* |
117 | * Normal paragraphs. |
118 | */ |
119 | case para_Normal: |
120 | fprintf(fp, ".PP\n"); |
121 | man_text(fp, p->words, TRUE, 0); |
122 | break; |
123 | |
124 | /* |
125 | * List paragraphs. |
126 | */ |
127 | case para_Description: |
128 | case para_BiblioCited: |
129 | case para_Bullet: |
130 | case para_NumberedList: |
131 | if (p->type == para_Bullet) { |
132 | fprintf(fp, ".IP \"\\fBo\\fP\"\n"); /* FIXME: configurable? */ |
133 | } else if (p->type == para_NumberedList) { |
134 | fprintf(fp, ".IP \""); |
135 | man_text(fp, p->kwtext, FALSE, QUOTE_QUOTES); |
136 | fprintf(fp, "\"\n"); |
137 | } else if (p->type == para_Description) { |
138 | /* |
139 | * Do nothing; the .xP for this paragraph is the .IP |
140 | * which has come before it in the DescribedThing. |
141 | */ |
142 | } else if (p->type == para_BiblioCited) { |
143 | fprintf(fp, ".IP \""); |
144 | man_text(fp, p->kwtext, FALSE, QUOTE_QUOTES); |
145 | fprintf(fp, "\"\n"); |
146 | } |
147 | man_text(fp, p->words, TRUE, 0); |
148 | break; |
149 | |
150 | case para_DescribedThing: |
151 | fprintf(fp, ".IP \""); |
152 | man_text(fp, p->words, FALSE, QUOTE_QUOTES); |
153 | fprintf(fp, "\"\n"); |
154 | break; |
155 | |
156 | case para_Rule: |
157 | /* |
158 | * FIXME. |
159 | */ |
160 | break; |
161 | |
162 | case para_LcontPush: |
163 | fprintf(fp, ".RS\n"); |
164 | break; |
165 | case para_LcontPop: |
166 | fprintf(fp, ".RE\n"); |
167 | break; |
168 | } |
169 | |
170 | /* |
171 | * Tidy up. |
172 | */ |
173 | fclose(fp); |
174 | } |
175 | |
176 | /* |
177 | * Convert a wide string into a string of chars. If `result' is |
178 | * non-NULL, mallocs the resulting string and stores a pointer to |
179 | * it in `*result'. If `result' is NULL, merely checks whether all |
180 | * characters in the string are feasible for the output character |
181 | * set. |
182 | * |
183 | * Return is nonzero if all characters are OK. If not all |
184 | * characters are OK but `result' is non-NULL, a result _will_ |
185 | * still be generated! |
186 | * |
187 | * FIXME: Here is probably also a good place to do escaping sorts |
188 | * of things. I know I at least need to escape backslash, and full |
189 | * stops at the starts of words are probably trouble as well. |
190 | */ |
191 | static int man_convert(wchar_t *s, char **result, int quote_props) { |
192 | /* |
193 | * FIXME. Currently this is ISO8859-1 only. |
194 | */ |
195 | int doing = (result != 0); |
196 | int ok = TRUE; |
197 | char *p = NULL; |
198 | int plen = 0, psize = 0; |
199 | |
200 | for (; *s; s++) { |
201 | wchar_t c = *s; |
202 | char outc; |
203 | |
204 | if ((c >= 32 && c <= 126) || |
205 | (c >= 160 && c <= 255)) { |
206 | /* Char is OK. */ |
207 | outc = (char)c; |
208 | } else { |
209 | /* Char is not OK. */ |
210 | ok = FALSE; |
211 | outc = 0xBF; /* approximate the good old DEC `uh?' */ |
212 | } |
213 | if (doing) { |
214 | if (plen+3 >= psize) { |
215 | psize = plen + 256; |
216 | p = resize(p, psize); |
217 | } |
218 | if (plen == 0 && (outc == '.' || outc == '\'') && |
219 | (quote_props & QUOTE_INITCTRL)) { |
220 | /* |
221 | * Control character (. or ') at the start of a |
222 | * line. Quote it by putting \& (troff zero-width |
223 | * space) before it. |
224 | */ |
225 | p[plen++] = '\\'; |
226 | p[plen++] = '&'; |
227 | } else if (outc == '\\') { |
228 | /* |
229 | * Quote backslashes by doubling them, always. |
230 | */ |
231 | p[plen++] = '\\'; |
232 | } else if (outc == '"' && (quote_props & QUOTE_QUOTES)) { |
233 | /* |
234 | * Double quote within double quotes. Quote it by |
235 | * doubling. |
236 | */ |
237 | p[plen++] = '"'; |
238 | } |
239 | p[plen++] = outc; |
240 | } |
241 | } |
242 | if (doing) { |
243 | p = resize(p, plen+1); |
244 | p[plen] = '\0'; |
245 | *result = p; |
246 | } |
247 | return ok; |
248 | } |
249 | |
250 | static void man_rdaddwc(rdstringc *rs, word *text, word *end, |
251 | int quote_props) { |
252 | char *c; |
253 | |
254 | for (; text && text != end; text = text->next) switch (text->type) { |
255 | case word_HyperLink: |
256 | case word_HyperEnd: |
257 | case word_UpperXref: |
258 | case word_LowerXref: |
259 | case word_XrefEnd: |
260 | case word_IndexRef: |
261 | break; |
262 | |
263 | case word_Normal: |
264 | case word_Emph: |
265 | case word_Code: |
266 | case word_WeakCode: |
267 | case word_WhiteSpace: |
268 | case word_EmphSpace: |
269 | case word_CodeSpace: |
270 | case word_WkCodeSpace: |
271 | case word_Quote: |
272 | case word_EmphQuote: |
273 | case word_CodeQuote: |
274 | case word_WkCodeQuote: |
275 | assert(text->type != word_CodeQuote && |
276 | text->type != word_WkCodeQuote); |
277 | if (towordstyle(text->type) == word_Emph && |
278 | (attraux(text->aux) == attr_First || |
279 | attraux(text->aux) == attr_Only)) |
280 | rdaddsc(rs, "\\fI"); |
281 | else if (towordstyle(text->type) == word_Code && |
282 | (attraux(text->aux) == attr_First || |
283 | attraux(text->aux) == attr_Only)) |
284 | rdaddsc(rs, "\\fB"); |
285 | if (removeattr(text->type) == word_Normal) { |
286 | if (rs->pos > 0) |
287 | quote_props &= ~QUOTE_INITCTRL; /* not at start any more */ |
288 | if (man_convert(text->text, &c, quote_props)) |
289 | rdaddsc(rs, c); |
290 | else |
291 | man_rdaddwc(rs, text->alt, NULL, quote_props); |
292 | sfree(c); |
293 | } else if (removeattr(text->type) == word_WhiteSpace) { |
294 | rdaddc(rs, ' '); |
295 | } else if (removeattr(text->type) == word_Quote) { |
296 | rdaddc(rs, quoteaux(text->aux) == quote_Open ? '`' : '\''); |
297 | /* FIXME: configurability */ |
298 | } |
299 | if (towordstyle(text->type) == word_Emph && |
300 | (attraux(text->aux) == attr_Last || |
301 | attraux(text->aux) == attr_Only)) |
302 | rdaddsc(rs, "\\fP"); |
303 | else if (towordstyle(text->type) == word_Code && |
304 | (attraux(text->aux) == attr_Last || |
305 | attraux(text->aux) == attr_Only)) |
306 | rdaddsc(rs, "\\fP"); |
307 | break; |
308 | } |
309 | } |
310 | |
311 | static void man_text(FILE *fp, word *text, int newline, int quote_props) { |
312 | rdstringc t = { 0, 0, NULL }; |
313 | |
314 | man_rdaddwc(&t, text, NULL, quote_props | QUOTE_INITCTRL); |
315 | fprintf(fp, "%s", t.text); |
316 | sfree(t.text); |
317 | if (newline) |
318 | fputc('\n', fp); |
319 | } |
320 | |
321 | static void man_codepara(FILE *fp, word *text) { |
322 | fprintf(fp, ".nf\n"); |
323 | for (; text; text = text->next) if (text->type == word_WeakCode) { |
324 | char *c; |
325 | man_convert(text->text, &c, QUOTE_INITCTRL); |
326 | fprintf(fp, "%s\n", c); |
327 | sfree(c); |
328 | } |
329 | fprintf(fp, ".fi\n"); |
330 | } |