Commit | Line | Data |
---|---|---|
205b7e14 | 1 | static char *sccsid = "@(#)checknr.c 4.2 (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 */ | |
39 | /* the -ms package */ | |
40 | "AB", "AE", | |
41 | "RS", "RE", | |
42 | "LG", "NL", | |
43 | "SM", "NL", | |
44 | "FS", "FE", | |
45 | "DS", "DE", | |
46 | "CD", "DE", | |
47 | "LD", "DE", | |
48 | "ID", "DE", | |
49 | "KS", "KE", | |
50 | "KF", "KE", | |
51 | "QS", "QE", | |
52 | /* Things needed by preprocessors */ | |
53 | "TS", "TE", | |
54 | "EQ", "EN", | |
55 | /* The -me package */ | |
56 | "(l", ")l", | |
57 | "(q", ")q", | |
58 | "(b", ")b", | |
59 | "(z", ")z", | |
60 | "(c", ")c", | |
61 | "(d", ")d", | |
62 | "(f", ")f", | |
63 | "(x", ")x", | |
64 | 0, 0 | |
65 | }; | |
66 | ||
67 | /* | |
68 | * All commands known to nroff, plus ms and me. | |
69 | * Used so we can complain about unrecognized commands. | |
70 | */ | |
71 | char *knowncmds[MAXCMDS] = { | |
72 | "$c", "$f", "$h", "$p", "$s", "(b", "(c", "(d", "(f", "(l", | |
73 | "(q", "(t", "(x", "(z", ")b", ")c", ")d", ")f", ")l", ")q", | |
74 | ")t", ")x", ")z", "++", "+c", "1C", "1c", "2C", "2c", "@(", | |
75 | "@)", "@C", "@D", "@F", "@I", "@M", "@c", "@e", "@f", "@h", | |
76 | "@m", "@n", "@o", "@p", "@r", "@t", "@z", "AB", "AB", "AE", | |
77 | "AE", "AI", "AI", "AT", "AU", "AU", "AX", "B", "B1", "B2", | |
78 | "BD", "BG", "BT", "BX", "C1", "C2", "CD", "CM", "CT", "D", | |
79 | "DA", "DE", "DF", "DS", "EG", "EM", "EN", "EQ", "EQ", "FA", | |
80 | "FE", "FJ", "FK", "FL", "FN", "FO", "FQ", "FS", "FV", "FX", | |
81 | "HO", "I", "ID", "IE", "IH", "IM", "IP", "IZ", "KD", "KE", | |
82 | "KF", "KQ", "KS", "LB", "LD", "LG", "LP", "MC", "ME", "MF", | |
83 | "MH", "MR", "ND", "NH", "NL", "NP", "OK", "PP", "PT", "PY", | |
84 | "QE", "QP", "QS", "R", "RA", "RC", "RE", "RP", "RQ", "RS", | |
85 | "RT", "S0", "S2", "S3", "SG", "SH", "SM", "SY", "TA", "TC", | |
86 | "TD", "TE", "TH", "TL", "TL", "TM", "TQ", "TR", "TS", "TS", | |
87 | "TX", "UL", "US", "UX", "WH", "XD", "XF", "XK", "XP", "[-", | |
88 | "[0", "[1", "[2", "[3", "[4", "[5", "[<", "[>", "[]", "]-", | |
89 | "]<", "]>", "][", "ab", "ac", "ad", "af", "am", "ar", "as", | |
90 | "b", "ba", "bc", "bd", "bi", "bl", "bp", "bp", "br", "bx", | |
91 | "c.", "c2", "cc", "ce", "cf", "ch", "cs", "ct", "cu", "da", | |
92 | "de", "di", "dl", "dn", "ds", "dt", "dw", "dy", "ec", "ef", | |
93 | "eh", "el", "em", "eo", "ep", "ev", "ex", "fc", "fi", "fl", | |
94 | "fo", "fp", "ft", "fz", "hc", "he", "hl", "hp", "ht", "hw", | |
95 | "hx", "hy", "i", "ie", "if", "ig", "in", "ip", "it", "ix", | |
96 | "lc", "lg", "li", "ll", "ll", "ln", "lo", "lp", "ls", "lt", | |
97 | "m1", "m2", "m3", "m4", "mc", "mk", "mo", "n1", "n2", "na", | |
98 | "ne", "nf", "nh", "nl", "nm", "nn", "np", "nr", "ns", "nx", | |
99 | "of", "oh", "os", "pa", "pc", "pi", "pl", "pm", "pn", "po", | |
100 | "po", "pp", "ps", "q", "r", "rb", "rd", "re", "re", "rm", | |
101 | "rn", "ro", "rr", "rs", "rt", "sb", "sc", "sh", "sk", "so", | |
102 | "sp", "ss", "st", "sv", "sz", "ta", "tc", "th", "ti", "tl", | |
103 | "tm", "tp", "tr", "u", "uf", "uh", "ul", "vs", "wh", "yr", | |
104 | 0 | |
105 | }; | |
106 | ||
107 | int lineno; /* current line number in input file */ | |
108 | char line[256]; /* the current line */ | |
109 | char *cfilename; /* name of current file */ | |
110 | int nfiles; /* number of files to process */ | |
111 | int fflag; /* -f: ignore \f */ | |
112 | int sflag; /* -s: ignore \s */ | |
113 | int ncmds; /* size of knowncmds */ | |
114 | int slot; /* slot in knowncmds found by binsrch */ | |
115 | ||
116 | char *malloc(); | |
117 | ||
118 | main(argc, argv) | |
119 | int argc; | |
120 | char **argv; | |
121 | { | |
122 | FILE *f; | |
123 | int i; | |
124 | char *cp; | |
205b7e14 | 125 | char b1[4]; |
b5da8a99 BJ |
126 | |
127 | if (argc <= 1) | |
205b7e14 MH |
128 | usage(); |
129 | /* Figure out how many known commands there are */ | |
130 | while (knowncmds[ncmds]) | |
131 | ncmds++; | |
b5da8a99 BJ |
132 | while (argc > 1 && argv[1][0] == '-') { |
133 | switch(argv[1][1]) { | |
205b7e14 MH |
134 | |
135 | /* -a: add pairs of macros */ | |
b5da8a99 | 136 | case 'a': |
b5da8a99 | 137 | i = strlen(argv[1]) - 2; |
205b7e14 MH |
138 | if (i % 6 != 0) |
139 | usage(); | |
b5da8a99 BJ |
140 | /* look for empty macro slots */ |
141 | for (i=0; br[i].opbr; i++) | |
142 | ; | |
143 | for (cp=argv[1]+3; cp[-1]; cp += 6) { | |
205b7e14 MH |
144 | br[i].opbr = malloc(3); |
145 | strncpy(br[i].opbr, cp, 2); | |
146 | br[i].clbr = malloc(3); | |
147 | strncpy(br[i].clbr, cp+3, 2); | |
148 | addmac(br[i].opbr); /* knows pairs are also known cmds */ | |
149 | addmac(br[i].clbr); | |
b5da8a99 BJ |
150 | i++; |
151 | } | |
152 | break; | |
205b7e14 MH |
153 | |
154 | /* -c: add known commands */ | |
155 | case 'c': | |
156 | i = strlen(argv[1]) - 2; | |
157 | if (i % 3 != 0) | |
158 | usage(); | |
159 | for (cp=argv[1]+3; cp[-1]; cp += 3) { | |
160 | if (cp[2] && cp[2] != '.') | |
161 | usage(); | |
162 | strncpy(b1, cp, 2); | |
163 | addmac(b1); | |
164 | } | |
165 | break; | |
166 | ||
167 | /* -f: ignore font changes */ | |
b5da8a99 BJ |
168 | case 'f': |
169 | fflag = 1; | |
170 | break; | |
205b7e14 MH |
171 | |
172 | /* -s: ignore size changes */ | |
b5da8a99 BJ |
173 | case 's': |
174 | sflag = 1; | |
175 | break; | |
176 | default: | |
205b7e14 | 177 | usage(); |
b5da8a99 BJ |
178 | } |
179 | argc--; argv++; | |
180 | } | |
181 | ||
182 | nfiles = argc - 1; | |
183 | ||
184 | if (nfiles > 0) { | |
185 | for (i=1; i<argc; i++) { | |
186 | cfilename = argv[i]; | |
187 | f = fopen(cfilename, "r"); | |
188 | if (f == NULL) | |
189 | perror(cfilename); | |
190 | else | |
191 | process(f); | |
192 | } | |
193 | } else { | |
194 | cfilename = "stdin"; | |
195 | process(stdin); | |
196 | } | |
197 | exit(0); | |
198 | } | |
199 | ||
205b7e14 MH |
200 | usage() |
201 | { | |
202 | printf("Usage: checknr -s -f -a.xx.yy.xx.yy... -c.xx.xx.xx...\n"); | |
203 | exit(1); | |
204 | } | |
205 | ||
b5da8a99 BJ |
206 | process(f) |
207 | FILE *f; | |
208 | { | |
209 | register int i, n; | |
210 | char mac[5]; /* The current macro or nroff command */ | |
211 | int pl; | |
212 | ||
213 | stktop = -1; | |
214 | for (lineno = 1; fgets(line, sizeof line, f); lineno++) { | |
215 | if (line[0] == '.') { | |
216 | /* | |
217 | * find and isolate the macro/command name. | |
218 | */ | |
219 | strncpy(mac, line+1, 4); | |
220 | if (isspace(mac[0])) { | |
221 | pe(lineno); | |
222 | printf("Empty command\n"); | |
223 | } else if (isspace(mac[1])) { | |
224 | mac[1] = 0; | |
225 | } else if (isspace(mac[2])) { | |
226 | mac[2] = 0; | |
227 | } else if (mac[2] != '\\' || mac[3] != '\"') { | |
228 | pe(lineno); | |
229 | printf("Command too long\n"); | |
230 | } | |
231 | ||
232 | /* | |
233 | * Is it a known command? | |
234 | */ | |
235 | checkknown(mac); | |
236 | ||
237 | /* | |
238 | * Should we add it? | |
239 | */ | |
240 | if (eq(mac, "de")) | |
241 | addcmd(line); | |
242 | ||
243 | chkcmd(line, mac); | |
244 | } | |
245 | ||
246 | /* | |
247 | * At this point we process the line looking | |
248 | * for \s and \f. | |
249 | */ | |
250 | for (i=0; line[i]; i++) | |
251 | if (line[i]=='\\' && (i==0 || line[i-1]!='\\')) { | |
252 | if (!sflag && line[++i]=='s') { | |
253 | pl = line[++i]; | |
254 | if (isdigit(pl)) { | |
255 | n = pl - '0'; | |
256 | pl = ' '; | |
257 | } else | |
258 | n = 0; | |
259 | while (isdigit(line[++i])) | |
260 | n = 10 * n + line[i] - '0'; | |
261 | i--; | |
262 | if (n == 0) { | |
263 | if (stk[stktop].opno == SZ) { | |
264 | stktop--; | |
265 | } else { | |
266 | pe(lineno); | |
267 | printf("unmatched \\s0\n"); | |
268 | } | |
269 | } else { | |
270 | stk[++stktop].opno = SZ; | |
271 | stk[stktop].pl = pl; | |
272 | stk[stktop].parm = n; | |
273 | stk[stktop].lno = lineno; | |
274 | } | |
275 | } else if (!fflag && line[i]=='f') { | |
276 | n = line[++i]; | |
277 | if (n == 'P') { | |
278 | if (stk[stktop].opno == FT) { | |
279 | stktop--; | |
280 | } else { | |
281 | pe(lineno); | |
282 | printf("unmatched \\fP\n"); | |
283 | } | |
284 | } else { | |
285 | stk[++stktop].opno = FT; | |
286 | stk[stktop].pl = 1; | |
287 | stk[stktop].parm = n; | |
288 | stk[stktop].lno = lineno; | |
289 | } | |
290 | } | |
291 | } | |
292 | } | |
293 | /* | |
294 | * We've hit the end and look at all this stuff that hasn't been | |
295 | * matched yet! Complain, complain. | |
296 | */ | |
297 | for (i=stktop; i>=0; i--) { | |
298 | complain(i); | |
299 | } | |
300 | } | |
301 | ||
302 | complain(i) | |
303 | { | |
304 | pe(stk[i].lno); | |
305 | printf("Unmatched "); | |
306 | prop(i); | |
307 | printf("\n"); | |
308 | } | |
309 | ||
310 | prop(i) | |
311 | { | |
312 | if (stk[i].pl == 0) | |
313 | printf(".%s", br[stk[i].opno].opbr); | |
314 | else switch(stk[i].opno) { | |
315 | case SZ: | |
316 | printf("\\s%c%d", stk[i].pl, stk[i].parm); | |
317 | break; | |
318 | case FT: | |
319 | printf("\\f%c", stk[i].parm); | |
320 | break; | |
321 | default: | |
322 | printf("Bug: stk[%d].opno = %d = .%s, .%s", | |
323 | i, stk[i].opno, br[stk[i].opno].opbr, br[stk[i].opno].clbr); | |
324 | } | |
325 | } | |
326 | ||
327 | chkcmd(line, mac) | |
328 | char *line; | |
329 | char *mac; | |
330 | { | |
331 | register int i, n; | |
332 | ||
333 | /* | |
334 | * Check to see if it matches top of stack. | |
335 | */ | |
336 | if (stktop >= 0 && eq(mac, br[stk[stktop].opno].clbr)) | |
337 | stktop--; /* OK. Pop & forget */ | |
338 | else { | |
339 | /* No. Maybe it's an opener */ | |
340 | for (i=0; br[i].opbr; i++) { | |
341 | if (eq(mac, br[i].opbr)) { | |
342 | /* Found. Push it. */ | |
343 | stktop++; | |
344 | stk[stktop].opno = i; | |
345 | stk[stktop].pl = 0; | |
346 | stk[stktop].parm = 0; | |
347 | stk[stktop].lno = lineno; | |
348 | break; | |
349 | } | |
350 | /* | |
351 | * Maybe it's an unmatched closer. | |
352 | * NOTE: this depends on the fact | |
353 | * that none of the closers can be | |
354 | * openers too. | |
355 | */ | |
356 | if (eq(mac, br[i].clbr)) { | |
357 | nomatch(mac); | |
358 | break; | |
359 | } | |
360 | } | |
361 | } | |
362 | } | |
363 | ||
364 | nomatch(mac) | |
365 | char *mac; | |
366 | { | |
367 | register int i, j; | |
368 | ||
369 | /* | |
370 | * Look for a match further down on stack | |
371 | * If we find one, it suggests that the stuff in | |
372 | * between is supposed to match itself. | |
373 | */ | |
374 | for (j=stktop; j>=0; j--) | |
375 | if (eq(mac,br[stk[j].opno].clbr)) { | |
376 | /* Found. Make a good diagnostic. */ | |
377 | if (j == stktop-2) { | |
378 | /* | |
379 | * Check for special case \fx..\fR and don't | |
380 | * complain. | |
381 | */ | |
382 | if (stk[j+1].opno==FT && stk[j+1].parm!='R' | |
383 | && stk[j+2].opno==FT && stk[j+2].parm=='R') { | |
384 | stktop = j -1; | |
385 | return; | |
386 | } | |
387 | /* | |
388 | * We have two unmatched frobs. Chances are | |
389 | * they were intended to match, so we mention | |
390 | * them together. | |
391 | */ | |
392 | pe(stk[j+1].lno); | |
393 | prop(j+1); | |
394 | printf(" does not match %d: ", stk[j+2].lno); | |
395 | prop(j+2); | |
396 | printf("\n"); | |
397 | } else for (i=j+1; i <= stktop; i++) { | |
398 | complain(i); | |
399 | } | |
400 | stktop = j-1; | |
401 | return; | |
402 | } | |
403 | /* Didn't find one. Throw this away. */ | |
404 | pe(lineno); | |
405 | printf("Unmatched .%s\n", mac); | |
406 | } | |
407 | ||
408 | /* eq: are two strings equal? */ | |
409 | eq(s1, s2) | |
410 | char *s1, *s2; | |
411 | { | |
412 | return (strcmp(s1, s2) == 0); | |
413 | } | |
414 | ||
415 | /* print the first part of an error message, given the line number */ | |
416 | pe(lineno) | |
417 | int lineno; | |
418 | { | |
419 | if (nfiles > 1) | |
420 | printf("%s: ", cfilename); | |
421 | printf("%d: ", lineno); | |
422 | } | |
423 | ||
424 | checkknown(mac) | |
425 | char *mac; | |
426 | { | |
b5da8a99 BJ |
427 | |
428 | if (eq(mac, ".")) | |
429 | return; | |
430 | if (binsrch(mac) >= 0) | |
431 | return; | |
432 | ||
433 | pe(lineno); | |
434 | printf("Unknown command: .%s\n", mac); | |
435 | } | |
436 | ||
437 | /* | |
438 | * We have a .de xx line in "line". Add xx to the list of known commands. | |
439 | */ | |
440 | addcmd(line) | |
441 | char *line; | |
442 | { | |
443 | char *mac; | |
b5da8a99 BJ |
444 | |
445 | /* grab the macro being defined */ | |
446 | mac = line+4; | |
447 | while (isspace(*mac)) | |
448 | mac++; | |
449 | if (*mac == 0) { | |
450 | pe(lineno); | |
451 | printf("illegal define: %s\n", line); | |
452 | return; | |
453 | } | |
454 | mac[2] = 0; | |
455 | if (isspace(mac[1]) || mac[1] == '\\') | |
456 | mac[1] = 0; | |
457 | if (ncmds >= MAXCMDS) { | |
458 | printf("Only %d known commands allowed\n", MAXCMDS); | |
459 | exit(1); | |
460 | } | |
205b7e14 MH |
461 | addmac(mac); |
462 | } | |
463 | ||
464 | /* | |
465 | * Add mac to the list. We should really have some kind of tree | |
466 | * structure here but this is a quick-and-dirty job and I just don't | |
467 | * have time to mess with it. (I wonder if this will come back to haunt | |
468 | * me someday?) Anyway, I claim that .de is fairly rare in user | |
469 | * nroff programs, and the register loop below is pretty fast. | |
470 | */ | |
471 | addmac(mac) | |
472 | char *mac; | |
473 | { | |
474 | register char **src, **dest, **loc; | |
475 | ||
b5da8a99 BJ |
476 | binsrch(mac); /* it's OK to redefine something */ |
477 | /* binsrch sets slot as a side effect */ | |
205b7e14 MH |
478 | #ifdef DEBUG |
479 | printf("binsrch(%s) -> %d\n", mac, slot); | |
480 | #endif | |
b5da8a99 BJ |
481 | loc = &knowncmds[slot]; |
482 | src = &knowncmds[ncmds-1]; | |
483 | dest = src+1; | |
484 | while (dest > loc) | |
485 | *dest-- = *src--; | |
486 | *loc = malloc(3); | |
487 | strcpy(*loc, mac); | |
488 | ncmds++; | |
205b7e14 MH |
489 | #ifdef DEBUG |
490 | 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); | |
491 | #endif | |
b5da8a99 BJ |
492 | } |
493 | ||
494 | /* | |
495 | * Do a binary search in knowncmds for mac. | |
496 | * If found, return the index. If not, return -1. | |
497 | */ | |
498 | binsrch(mac) | |
499 | char *mac; | |
500 | { | |
501 | register char *p; /* pointer to current cmd in list */ | |
502 | register int d; /* difference if any */ | |
503 | register int mid; /* mid point in binary search */ | |
504 | register int top, bot; /* boundaries of bin search, inclusive */ | |
505 | ||
506 | top = ncmds-1; | |
507 | bot = 0; | |
508 | while (top >= bot) { | |
509 | mid = (top+bot)/2; | |
510 | p = knowncmds[mid]; | |
511 | d = p[0] - mac[0]; | |
512 | if (d == 0) | |
513 | d = p[1] - mac[1]; | |
514 | if (d == 0) | |
515 | return mid; | |
516 | if (d < 0) | |
517 | bot = mid + 1; | |
518 | else | |
519 | top = mid - 1; | |
520 | } | |
521 | slot = bot; /* place it would have gone */ | |
522 | return -1; | |
523 | } |