Commit | Line | Data |
---|---|---|
645c264b | 1 | static char *sccsid = "@(#)checknr.c 4.5 (Berkeley) %G%"; |
b5da8a99 BJ |
2 | /* |
3 | * checknr: check an nroff/troff input file for matching macro calls. | |
4 | * we also attempt to match size and font changes, but only the embedded | |
5 | * kind. These must end in \s0 and \fP resp. Maybe more sophistication | |
6 | * later but for now think of these restrictions as contributions to | |
7 | * structured typesetting. | |
8 | */ | |
9 | #include <stdio.h> | |
10 | #include <ctype.h> | |
11 | ||
12 | #define MAXSTK 100 /* Stack size */ | |
13 | #define MAXBR 100 /* Max number of bracket pairs known */ | |
14 | #define MAXCMDS 500 /* Max number of commands known */ | |
15 | ||
16 | /* | |
17 | * The stack on which we remember what we've seen so far. | |
18 | */ | |
19 | struct stkstr { | |
20 | int opno; /* number of opening bracket */ | |
21 | int pl; /* '+', '-', ' ' for \s, 1 for \f, 0 for .ft */ | |
22 | int parm; /* parm to size, font, etc */ | |
23 | int lno; /* line number the thing came in in */ | |
24 | } stk[MAXSTK]; | |
25 | int stktop; | |
26 | ||
27 | /* | |
28 | * The kinds of opening and closing brackets. | |
29 | */ | |
30 | struct brstr { | |
31 | char *opbr; | |
32 | char *clbr; | |
33 | } br[MAXBR] = { | |
34 | /* A few bare bones troff commands */ | |
35 | #define SZ 0 | |
36 | "sz", "sz", /* also \s */ | |
37 | #define FT 1 | |
38 | "ft", "ft", /* also \f */ | |
e1062a05 BJ |
39 | /* the -mm package */ |
40 | "AL", "LE", | |
41 | "AS", "AE", | |
42 | "BL", "LE", | |
43 | "BS", "BE", | |
44 | "DF", "DE", | |
45 | "DL", "LE", | |
46 | "DS", "DE", | |
47 | "FS", "FE", | |
48 | "ML", "LE", | |
49 | "NS", "NE", | |
50 | "RL", "LE", | |
51 | "VL", "LE", | |
b5da8a99 BJ |
52 | /* the -ms package */ |
53 | "AB", "AE", | |
b5da8a99 | 54 | "CD", "DE", |
e1062a05 BJ |
55 | "DS", "DE", |
56 | "FS", "FE", | |
b5da8a99 | 57 | "ID", "DE", |
b5da8a99 | 58 | "KF", "KE", |
e1062a05 BJ |
59 | "KS", "KE", |
60 | "LD", "DE", | |
61 | "LG", "NL", | |
b5da8a99 | 62 | "QS", "QE", |
e1062a05 BJ |
63 | "RS", "RE", |
64 | "SM", "NL", | |
b5da8a99 | 65 | /* The -me package */ |
b5da8a99 | 66 | "(b", ")b", |
b5da8a99 BJ |
67 | "(c", ")c", |
68 | "(d", ")d", | |
69 | "(f", ")f", | |
e1062a05 BJ |
70 | "(l", ")l", |
71 | "(q", ")q", | |
b5da8a99 | 72 | "(x", ")x", |
e1062a05 BJ |
73 | "(z", ")z", |
74 | /* Things needed by preprocessors */ | |
75 | "EQ", "EN", | |
76 | "TS", "TE", | |
77 | /* Refer */ | |
78 | "[", "]", | |
b5da8a99 BJ |
79 | 0, 0 |
80 | }; | |
81 | ||
82 | /* | |
e1062a05 | 83 | * All commands known to nroff, plus macro packages. |
b5da8a99 BJ |
84 | * Used so we can complain about unrecognized commands. |
85 | */ | |
86 | char *knowncmds[MAXCMDS] = { | |
e1062a05 BJ |
87 | "$c", "$f", "$h", "$p", "$s", "(b", "(c", "(d", "(f", "(l", "(q", "(t", |
88 | "(x", "(z", ")b", ")c", ")d", ")f", ")l", ")q", ")t", ")x", ")z", "++", | |
89 | "+c", "1C", "1c", "2C", "2c", "@(", "@)", "@C", "@D", "@F", "@I", "@M", | |
90 | "@c", "@e", "@f", "@h", "@m", "@n", "@o", "@p", "@r", "@t", "@z", "AB", | |
91 | "AE", "AF", "AI", "AL", "AS", "AT", "AU", "AX", "B", "B1", "B2", "BD", | |
92 | "BE", "BG", "BL", "BS", "BT", "BX", "C1", "C2", "CD", "CM", "CT", "D", | |
7f56a5b5 | 93 | "DA", "DE", "DF", "DL", "DS", "DT", "EC", "EF", "EG", "EH", "EM", "EN", "EQ", |
e1062a05 BJ |
94 | "EX", "FA", "FD", "FE", "FG", "FJ", "FK", "FL", "FN", "FO", "FQ", "FS", |
95 | "FV", "FX", "H", "HC", "HM", "HO", "HU", "I", "ID", "IE", "IH", "IM", | |
96 | "IP", "IZ", "KD", "KE", "KF", "KQ", "KS", "LB", "LC", "LD", "LE", "LG", | |
97 | "LI", "LP", "MC", "ME", "MF", "MH", "ML", "MR", "MT", "ND", "NE", "NH", | |
98 | "NL", "NP", "NS", "OF", "OH", "OK", "OP", "P", "PF", "PH", "PP", "PT", | |
99 | "PY", "QE", "QP", "QS", "R", "RA", "RC", "RE", "RL", "RP", "RQ", "RS", | |
100 | "RT", "S", "S0", "S2", "S3", "SA", "SG", "SH", "SK", "SM", "SP", "SY", | |
101 | "TA", "TB", "TC", "TD", "TE", "TH", "TL", "TM", "TP", "TQ", "TR", "TS", | |
102 | "TX", "UL", "US", "UX", "VL", "WC", "WH", "XD", "XF", "XK", "XP", "[", "[-", | |
103 | "[0", "[1", "[2", "[3", "[4", "[5", "[<", "[>", "[]", "]", "]-", "]<", "]>", | |
104 | "][", "ab", "ac", "ad", "af", "am", "ar", "as", "b", "ba", "bc", "bd", | |
105 | "bi", "bl", "bp", "br", "bx", "c.", "c2", "cc", "ce", "cf", "ch", "cs", | |
106 | "ct", "cu", "da", "de", "di", "dl", "dn", "ds", "dt", "dw", "dy", "ec", | |
107 | "ef", "eh", "el", "em", "eo", "ep", "ev", "ex", "fc", "fi", "fl", "fo", | |
108 | "fp", "ft", "fz", "hc", "he", "hl", "hp", "ht", "hw", "hx", "hy", "i", | |
109 | "ie", "if", "ig", "in", "ip", "it", "ix", "lc", "lg", "li", "ll", "ln", | |
110 | "lo", "lp", "ls", "lt", "m1", "m2", "m3", "m4", "mc", "mk", "mo", "n1", | |
111 | "n2", "na", "ne", "nf", "nh", "nl", "nm", "nn", "np", "nr", "ns", "nx", | |
112 | "of", "oh", "os", "pa", "pc", "pi", "pl", "pm", "pn", "po", "pp", "ps", | |
113 | "q", "r", "rb", "rd", "re", "rm", "rn", "ro", "rr", "rs", "rt", "sb", | |
114 | "sc", "sh", "sk", "so", "sp", "ss", "st", "sv", "sz", "ta", "tc", "th", | |
115 | "ti", "tl", "tm", "tp", "tr", "u", "uf", "uh", "ul", "vs", "wh", "xp", "yr", | |
b5da8a99 BJ |
116 | 0 |
117 | }; | |
118 | ||
119 | int lineno; /* current line number in input file */ | |
120 | char line[256]; /* the current line */ | |
121 | char *cfilename; /* name of current file */ | |
122 | int nfiles; /* number of files to process */ | |
123 | int fflag; /* -f: ignore \f */ | |
124 | int sflag; /* -s: ignore \s */ | |
125 | int ncmds; /* size of knowncmds */ | |
126 | int slot; /* slot in knowncmds found by binsrch */ | |
127 | ||
128 | char *malloc(); | |
129 | ||
130 | main(argc, argv) | |
131 | int argc; | |
132 | char **argv; | |
133 | { | |
134 | FILE *f; | |
135 | int i; | |
136 | char *cp; | |
205b7e14 | 137 | char b1[4]; |
b5da8a99 | 138 | |
205b7e14 MH |
139 | /* Figure out how many known commands there are */ |
140 | while (knowncmds[ncmds]) | |
141 | ncmds++; | |
b5da8a99 BJ |
142 | while (argc > 1 && argv[1][0] == '-') { |
143 | switch(argv[1][1]) { | |
205b7e14 MH |
144 | |
145 | /* -a: add pairs of macros */ | |
b5da8a99 | 146 | case 'a': |
b5da8a99 | 147 | i = strlen(argv[1]) - 2; |
205b7e14 MH |
148 | if (i % 6 != 0) |
149 | usage(); | |
b5da8a99 BJ |
150 | /* look for empty macro slots */ |
151 | for (i=0; br[i].opbr; i++) | |
152 | ; | |
153 | for (cp=argv[1]+3; cp[-1]; cp += 6) { | |
205b7e14 MH |
154 | br[i].opbr = malloc(3); |
155 | strncpy(br[i].opbr, cp, 2); | |
156 | br[i].clbr = malloc(3); | |
157 | strncpy(br[i].clbr, cp+3, 2); | |
158 | addmac(br[i].opbr); /* knows pairs are also known cmds */ | |
159 | addmac(br[i].clbr); | |
b5da8a99 BJ |
160 | i++; |
161 | } | |
162 | break; | |
205b7e14 MH |
163 | |
164 | /* -c: add known commands */ | |
165 | case 'c': | |
166 | i = strlen(argv[1]) - 2; | |
167 | if (i % 3 != 0) | |
168 | usage(); | |
169 | for (cp=argv[1]+3; cp[-1]; cp += 3) { | |
170 | if (cp[2] && cp[2] != '.') | |
171 | usage(); | |
172 | strncpy(b1, cp, 2); | |
173 | addmac(b1); | |
174 | } | |
175 | break; | |
176 | ||
177 | /* -f: ignore font changes */ | |
b5da8a99 BJ |
178 | case 'f': |
179 | fflag = 1; | |
180 | break; | |
205b7e14 MH |
181 | |
182 | /* -s: ignore size changes */ | |
b5da8a99 BJ |
183 | case 's': |
184 | sflag = 1; | |
185 | break; | |
186 | default: | |
205b7e14 | 187 | usage(); |
b5da8a99 BJ |
188 | } |
189 | argc--; argv++; | |
190 | } | |
191 | ||
192 | nfiles = argc - 1; | |
193 | ||
194 | if (nfiles > 0) { | |
195 | for (i=1; i<argc; i++) { | |
196 | cfilename = argv[i]; | |
197 | f = fopen(cfilename, "r"); | |
198 | if (f == NULL) | |
199 | perror(cfilename); | |
200 | else | |
201 | process(f); | |
202 | } | |
203 | } else { | |
204 | cfilename = "stdin"; | |
205 | process(stdin); | |
206 | } | |
207 | exit(0); | |
208 | } | |
209 | ||
205b7e14 MH |
210 | usage() |
211 | { | |
212 | printf("Usage: checknr -s -f -a.xx.yy.xx.yy... -c.xx.xx.xx...\n"); | |
213 | exit(1); | |
214 | } | |
215 | ||
b5da8a99 BJ |
216 | process(f) |
217 | FILE *f; | |
218 | { | |
219 | register int i, n; | |
220 | char mac[5]; /* The current macro or nroff command */ | |
221 | int pl; | |
222 | ||
223 | stktop = -1; | |
224 | for (lineno = 1; fgets(line, sizeof line, f); lineno++) { | |
225 | if (line[0] == '.') { | |
226 | /* | |
227 | * find and isolate the macro/command name. | |
228 | */ | |
229 | strncpy(mac, line+1, 4); | |
230 | if (isspace(mac[0])) { | |
231 | pe(lineno); | |
232 | printf("Empty command\n"); | |
233 | } else if (isspace(mac[1])) { | |
234 | mac[1] = 0; | |
235 | } else if (isspace(mac[2])) { | |
236 | mac[2] = 0; | |
e1062a05 | 237 | } else if (mac[0] != '\\' || mac[1] != '\"') { |
b5da8a99 BJ |
238 | pe(lineno); |
239 | printf("Command too long\n"); | |
240 | } | |
241 | ||
242 | /* | |
243 | * Is it a known command? | |
244 | */ | |
245 | checkknown(mac); | |
246 | ||
247 | /* | |
248 | * Should we add it? | |
249 | */ | |
250 | if (eq(mac, "de")) | |
251 | addcmd(line); | |
252 | ||
253 | chkcmd(line, mac); | |
254 | } | |
255 | ||
256 | /* | |
257 | * At this point we process the line looking | |
258 | * for \s and \f. | |
259 | */ | |
260 | for (i=0; line[i]; i++) | |
261 | if (line[i]=='\\' && (i==0 || line[i-1]!='\\')) { | |
262 | if (!sflag && line[++i]=='s') { | |
263 | pl = line[++i]; | |
264 | if (isdigit(pl)) { | |
265 | n = pl - '0'; | |
266 | pl = ' '; | |
267 | } else | |
268 | n = 0; | |
269 | while (isdigit(line[++i])) | |
270 | n = 10 * n + line[i] - '0'; | |
271 | i--; | |
272 | if (n == 0) { | |
273 | if (stk[stktop].opno == SZ) { | |
274 | stktop--; | |
275 | } else { | |
276 | pe(lineno); | |
277 | printf("unmatched \\s0\n"); | |
278 | } | |
279 | } else { | |
280 | stk[++stktop].opno = SZ; | |
281 | stk[stktop].pl = pl; | |
282 | stk[stktop].parm = n; | |
283 | stk[stktop].lno = lineno; | |
284 | } | |
285 | } else if (!fflag && line[i]=='f') { | |
286 | n = line[++i]; | |
287 | if (n == 'P') { | |
288 | if (stk[stktop].opno == FT) { | |
289 | stktop--; | |
290 | } else { | |
291 | pe(lineno); | |
292 | printf("unmatched \\fP\n"); | |
293 | } | |
294 | } else { | |
295 | stk[++stktop].opno = FT; | |
296 | stk[stktop].pl = 1; | |
297 | stk[stktop].parm = n; | |
298 | stk[stktop].lno = lineno; | |
299 | } | |
300 | } | |
301 | } | |
302 | } | |
303 | /* | |
304 | * We've hit the end and look at all this stuff that hasn't been | |
305 | * matched yet! Complain, complain. | |
306 | */ | |
307 | for (i=stktop; i>=0; i--) { | |
308 | complain(i); | |
309 | } | |
310 | } | |
311 | ||
312 | complain(i) | |
313 | { | |
314 | pe(stk[i].lno); | |
315 | printf("Unmatched "); | |
316 | prop(i); | |
317 | printf("\n"); | |
318 | } | |
319 | ||
320 | prop(i) | |
321 | { | |
322 | if (stk[i].pl == 0) | |
323 | printf(".%s", br[stk[i].opno].opbr); | |
324 | else switch(stk[i].opno) { | |
325 | case SZ: | |
326 | printf("\\s%c%d", stk[i].pl, stk[i].parm); | |
327 | break; | |
328 | case FT: | |
329 | printf("\\f%c", stk[i].parm); | |
330 | break; | |
331 | default: | |
332 | printf("Bug: stk[%d].opno = %d = .%s, .%s", | |
333 | i, stk[i].opno, br[stk[i].opno].opbr, br[stk[i].opno].clbr); | |
334 | } | |
335 | } | |
336 | ||
337 | chkcmd(line, mac) | |
338 | char *line; | |
339 | char *mac; | |
340 | { | |
341 | register int i, n; | |
342 | ||
343 | /* | |
344 | * Check to see if it matches top of stack. | |
345 | */ | |
346 | if (stktop >= 0 && eq(mac, br[stk[stktop].opno].clbr)) | |
347 | stktop--; /* OK. Pop & forget */ | |
348 | else { | |
349 | /* No. Maybe it's an opener */ | |
350 | for (i=0; br[i].opbr; i++) { | |
351 | if (eq(mac, br[i].opbr)) { | |
352 | /* Found. Push it. */ | |
353 | stktop++; | |
354 | stk[stktop].opno = i; | |
355 | stk[stktop].pl = 0; | |
356 | stk[stktop].parm = 0; | |
357 | stk[stktop].lno = lineno; | |
358 | break; | |
359 | } | |
360 | /* | |
361 | * Maybe it's an unmatched closer. | |
362 | * NOTE: this depends on the fact | |
363 | * that none of the closers can be | |
364 | * openers too. | |
365 | */ | |
366 | if (eq(mac, br[i].clbr)) { | |
367 | nomatch(mac); | |
368 | break; | |
369 | } | |
370 | } | |
371 | } | |
372 | } | |
373 | ||
374 | nomatch(mac) | |
375 | char *mac; | |
376 | { | |
377 | register int i, j; | |
378 | ||
379 | /* | |
380 | * Look for a match further down on stack | |
381 | * If we find one, it suggests that the stuff in | |
382 | * between is supposed to match itself. | |
383 | */ | |
384 | for (j=stktop; j>=0; j--) | |
385 | if (eq(mac,br[stk[j].opno].clbr)) { | |
386 | /* Found. Make a good diagnostic. */ | |
387 | if (j == stktop-2) { | |
388 | /* | |
389 | * Check for special case \fx..\fR and don't | |
390 | * complain. | |
391 | */ | |
392 | if (stk[j+1].opno==FT && stk[j+1].parm!='R' | |
393 | && stk[j+2].opno==FT && stk[j+2].parm=='R') { | |
394 | stktop = j -1; | |
395 | return; | |
396 | } | |
397 | /* | |
398 | * We have two unmatched frobs. Chances are | |
399 | * they were intended to match, so we mention | |
400 | * them together. | |
401 | */ | |
402 | pe(stk[j+1].lno); | |
403 | prop(j+1); | |
404 | printf(" does not match %d: ", stk[j+2].lno); | |
405 | prop(j+2); | |
406 | printf("\n"); | |
407 | } else for (i=j+1; i <= stktop; i++) { | |
408 | complain(i); | |
409 | } | |
410 | stktop = j-1; | |
411 | return; | |
412 | } | |
413 | /* Didn't find one. Throw this away. */ | |
414 | pe(lineno); | |
415 | printf("Unmatched .%s\n", mac); | |
416 | } | |
417 | ||
418 | /* eq: are two strings equal? */ | |
419 | eq(s1, s2) | |
420 | char *s1, *s2; | |
421 | { | |
422 | return (strcmp(s1, s2) == 0); | |
423 | } | |
424 | ||
425 | /* print the first part of an error message, given the line number */ | |
426 | pe(lineno) | |
427 | int lineno; | |
428 | { | |
429 | if (nfiles > 1) | |
430 | printf("%s: ", cfilename); | |
431 | printf("%d: ", lineno); | |
432 | } | |
433 | ||
434 | checkknown(mac) | |
435 | char *mac; | |
436 | { | |
b5da8a99 BJ |
437 | |
438 | if (eq(mac, ".")) | |
439 | return; | |
440 | if (binsrch(mac) >= 0) | |
441 | return; | |
e1062a05 BJ |
442 | if (mac[0] == '\\' && mac[1] == '"') /* comments */ |
443 | return; | |
b5da8a99 BJ |
444 | |
445 | pe(lineno); | |
446 | printf("Unknown command: .%s\n", mac); | |
447 | } | |
448 | ||
449 | /* | |
450 | * We have a .de xx line in "line". Add xx to the list of known commands. | |
451 | */ | |
452 | addcmd(line) | |
453 | char *line; | |
454 | { | |
455 | char *mac; | |
b5da8a99 BJ |
456 | |
457 | /* grab the macro being defined */ | |
458 | mac = line+4; | |
459 | while (isspace(*mac)) | |
460 | mac++; | |
461 | if (*mac == 0) { | |
462 | pe(lineno); | |
463 | printf("illegal define: %s\n", line); | |
464 | return; | |
465 | } | |
466 | mac[2] = 0; | |
467 | if (isspace(mac[1]) || mac[1] == '\\') | |
468 | mac[1] = 0; | |
469 | if (ncmds >= MAXCMDS) { | |
470 | printf("Only %d known commands allowed\n", MAXCMDS); | |
471 | exit(1); | |
472 | } | |
205b7e14 MH |
473 | addmac(mac); |
474 | } | |
475 | ||
476 | /* | |
477 | * Add mac to the list. We should really have some kind of tree | |
478 | * structure here but this is a quick-and-dirty job and I just don't | |
479 | * have time to mess with it. (I wonder if this will come back to haunt | |
480 | * me someday?) Anyway, I claim that .de is fairly rare in user | |
481 | * nroff programs, and the register loop below is pretty fast. | |
482 | */ | |
483 | addmac(mac) | |
484 | char *mac; | |
485 | { | |
486 | register char **src, **dest, **loc; | |
487 | ||
b5da8a99 BJ |
488 | binsrch(mac); /* it's OK to redefine something */ |
489 | /* binsrch sets slot as a side effect */ | |
205b7e14 MH |
490 | #ifdef DEBUG |
491 | printf("binsrch(%s) -> %d\n", mac, slot); | |
492 | #endif | |
b5da8a99 BJ |
493 | loc = &knowncmds[slot]; |
494 | src = &knowncmds[ncmds-1]; | |
495 | dest = src+1; | |
496 | while (dest > loc) | |
497 | *dest-- = *src--; | |
498 | *loc = malloc(3); | |
499 | strcpy(*loc, mac); | |
500 | ncmds++; | |
205b7e14 MH |
501 | #ifdef DEBUG |
502 | printf("after: %s %s %s %s %s, %d cmds\n", knowncmds[slot-2], knowncmds[slot-1], knowncmds[slot], knowncmds[slot+1], knowncmds[slot+2], ncmds); | |
503 | #endif | |
b5da8a99 BJ |
504 | } |
505 | ||
506 | /* | |
507 | * Do a binary search in knowncmds for mac. | |
508 | * If found, return the index. If not, return -1. | |
509 | */ | |
510 | binsrch(mac) | |
511 | char *mac; | |
512 | { | |
513 | register char *p; /* pointer to current cmd in list */ | |
514 | register int d; /* difference if any */ | |
515 | register int mid; /* mid point in binary search */ | |
516 | register int top, bot; /* boundaries of bin search, inclusive */ | |
517 | ||
518 | top = ncmds-1; | |
519 | bot = 0; | |
520 | while (top >= bot) { | |
521 | mid = (top+bot)/2; | |
522 | p = knowncmds[mid]; | |
523 | d = p[0] - mac[0]; | |
524 | if (d == 0) | |
525 | d = p[1] - mac[1]; | |
526 | if (d == 0) | |
527 | return mid; | |
528 | if (d < 0) | |
529 | bot = mid + 1; | |
530 | else | |
531 | top = mid - 1; | |
532 | } | |
533 | slot = bot; /* place it would have gone */ | |
534 | return -1; | |
535 | } | |
e1062a05 BJ |
536 | |
537 |