Commit | Line | Data |
---|---|---|
bcb9ffff | 1 | /* |
083e16be KB |
2 | * Copyright (c) 1987 The Regents of the University of California. |
3 | * All rights reserved. | |
4 | * | |
836fe169 | 5 | * %sccs.include.redist.c% |
bcb9ffff KB |
6 | */ |
7 | ||
8 | #ifndef lint | |
6c2ce1d3 | 9 | static char sccsid[] = "@(#)C.c 5.5 (Berkeley) %G%"; |
083e16be | 10 | #endif /* not lint */ |
bcb9ffff KB |
11 | |
12 | #include <stdio.h> | |
6c2ce1d3 KB |
13 | #include <string.h> |
14 | #include "ctags.h" | |
15 | ||
16 | static int func_entry(), str_entry(); | |
17 | static void hash_entry(); | |
bcb9ffff KB |
18 | |
19 | /* | |
20 | * c_entries -- | |
21 | * read .c and .h files and call appropriate routines | |
22 | */ | |
23 | c_entries() | |
24 | { | |
25 | extern int tflag; /* -t: create tags for typedefs */ | |
26 | register int c, /* current character */ | |
27 | level; /* brace level */ | |
28 | register char *sp; /* buffer pointer */ | |
29 | int token, /* if reading a token */ | |
30 | t_def, /* if reading a typedef */ | |
31 | t_level; /* typedef's brace level */ | |
32 | char tok[MAXTOKEN]; /* token buffer */ | |
33 | ||
34 | lineftell = ftell(inf); | |
35 | sp = tok; token = t_def = NO; t_level = -1; level = 0; lineno = 1; | |
36 | while (GETC(!=,EOF)) { | |
37 | ||
38 | switch ((char)c) { | |
39 | /* | |
40 | * Here's where it DOESN'T handle: | |
41 | * foo(a) | |
42 | * { | |
43 | * #ifdef notdef | |
44 | * } | |
45 | * #endif | |
46 | * if (a) | |
47 | * puts("hello, world"); | |
48 | * } | |
49 | */ | |
50 | case '{': | |
51 | ++level; | |
52 | goto endtok; | |
53 | case '}': | |
54 | /* | |
55 | * if level goes below zero, try and fix | |
56 | * it, even though we've already messed up | |
57 | */ | |
58 | if (--level < 0) | |
59 | level = 0; | |
60 | goto endtok; | |
61 | ||
62 | case '\n': | |
63 | SETLINE; | |
64 | /* | |
65 | * the above 3 cases are similar in that they | |
66 | * are special characters that also end tokens. | |
67 | */ | |
68 | endtok: if (sp > tok) { | |
69 | *sp = EOS; | |
70 | token = YES; | |
71 | sp = tok; | |
72 | } | |
73 | else | |
74 | token = NO; | |
75 | continue; | |
76 | ||
77 | /* we ignore quoted strings and comments in their entirety */ | |
78 | case '"': | |
79 | case '\'': | |
80 | (void)skip_key(c); | |
81 | break; | |
82 | ||
83 | /* | |
84 | * comments can be fun; note the state is unchanged after | |
85 | * return, in case we found: | |
86 | * "foo() XX comment XX { int bar; }" | |
87 | */ | |
88 | case '/': | |
89 | if (GETC(==,'*')) { | |
90 | skip_comment(); | |
91 | continue; | |
92 | } | |
93 | (void)ungetc(c,inf); | |
94 | c = '/'; | |
95 | goto storec; | |
96 | ||
97 | /* hash marks flag #define's. */ | |
98 | case '#': | |
99 | if (sp == tok) { | |
100 | hash_entry(); | |
101 | break; | |
102 | } | |
103 | goto storec; | |
104 | ||
105 | /* | |
106 | * if we have a current token, parenthesis on | |
107 | * level zero indicates a function. | |
108 | */ | |
109 | case '(': | |
110 | if (!level && token) { | |
111 | int curline; | |
112 | ||
113 | if (sp != tok) | |
114 | *sp = EOS; | |
115 | /* | |
116 | * grab the line immediately, we may | |
117 | * already be wrong, for example, | |
118 | * foo\n | |
119 | * (arg1, | |
120 | */ | |
121 | getline(); | |
122 | curline = lineno; | |
123 | if (func_entry()) { | |
124 | ++level; | |
125 | pfnote(tok,curline); | |
126 | } | |
127 | break; | |
128 | } | |
129 | goto storec; | |
130 | ||
131 | /* | |
132 | * semi-colons indicate the end of a typedef; if we find a | |
133 | * typedef we search for the next semi-colon of the same | |
134 | * level as the typedef. Ignoring "structs", they are | |
135 | * tricky, since you can find: | |
136 | * | |
137 | * "typedef long time_t;" | |
138 | * "typedef unsigned int u_int;" | |
139 | * "typedef unsigned int u_int [10];" | |
140 | * | |
141 | * If looking at a typedef, we save a copy of the last token | |
142 | * found. Then, when we find the ';' we take the current | |
143 | * token if it starts with a valid token name, else we take | |
144 | * the one we saved. There's probably some reasonable | |
145 | * alternative to this... | |
146 | */ | |
147 | case ';': | |
148 | if (t_def && level == t_level) { | |
149 | t_def = NO; | |
150 | getline(); | |
151 | if (sp != tok) | |
152 | *sp = EOS; | |
153 | pfnote(tok,lineno); | |
154 | break; | |
155 | } | |
156 | goto storec; | |
157 | ||
158 | /* | |
159 | * store characters until one that can't be part of a token | |
160 | * comes along; check the current token against certain | |
161 | * reserved words. | |
162 | */ | |
163 | default: | |
164 | storec: if (!intoken(c)) { | |
165 | if (sp == tok) | |
166 | break; | |
167 | *sp = EOS; | |
168 | if (tflag) { | |
169 | /* no typedefs inside typedefs */ | |
170 | if (!t_def && !bcmp(tok,"typedef",8)) { | |
171 | t_def = YES; | |
172 | t_level = level; | |
173 | break; | |
174 | } | |
175 | /* catch "typedef struct" */ | |
176 | if ((!t_def || t_level < level) | |
177 | && (!bcmp(tok,"struct",7) | |
178 | || !bcmp(tok,"union",6) | |
179 | || !bcmp(tok,"enum",5))) { | |
180 | /* | |
181 | * get line immediately; | |
182 | * may change before '{' | |
183 | */ | |
184 | getline(); | |
185 | if (str_entry(c)) | |
186 | ++level; | |
187 | break; | |
188 | } | |
189 | } | |
190 | sp = tok; | |
191 | } | |
192 | else if (sp != tok || begtoken(c)) { | |
193 | *sp++ = c; | |
194 | token = YES; | |
195 | } | |
196 | continue; | |
197 | } | |
198 | sp = tok; | |
199 | token = NO; | |
200 | } | |
201 | } | |
202 | ||
203 | /* | |
204 | * func_entry -- | |
205 | * handle a function reference | |
206 | */ | |
207 | static | |
208 | func_entry() | |
209 | { | |
210 | register int c; /* current character */ | |
211 | ||
212 | /* | |
213 | * we assume that the character after a function's right paren | |
214 | * is a token character if it's a function and a non-token | |
215 | * character if it's a declaration. Comments don't count... | |
216 | */ | |
217 | (void)skip_key((int)')'); | |
218 | for (;;) { | |
219 | while (GETC(!=,EOF) && iswhite(c)) | |
220 | if (c == (int)'\n') | |
221 | SETLINE; | |
222 | if (intoken(c) || c == (int)'{') | |
223 | break; | |
224 | if (c == (int)'/' && GETC(==,'*')) | |
225 | skip_comment(); | |
226 | else { /* don't ever "read" '/' */ | |
227 | (void)ungetc(c,inf); | |
228 | return(NO); | |
229 | } | |
230 | } | |
231 | if (c != (int)'{') | |
232 | (void)skip_key((int)'{'); | |
233 | return(YES); | |
234 | } | |
235 | ||
236 | /* | |
237 | * hash_entry -- | |
238 | * handle a line starting with a '#' | |
239 | */ | |
6c2ce1d3 | 240 | static void |
bcb9ffff KB |
241 | hash_entry() |
242 | { | |
243 | extern int dflag; /* -d: non-macro defines */ | |
244 | register int c, /* character read */ | |
245 | curline; /* line started on */ | |
246 | register char *sp; /* buffer pointer */ | |
247 | char tok[MAXTOKEN]; /* storage buffer */ | |
248 | ||
249 | curline = lineno; | |
250 | for (sp = tok;;) { /* get next token */ | |
251 | if (GETC(==,EOF)) | |
252 | return; | |
253 | if (iswhite(c)) | |
254 | break; | |
255 | *sp++ = c; | |
256 | } | |
257 | *sp = EOS; | |
258 | if (bcmp(tok,"define",6)) /* only interested in #define's */ | |
259 | goto skip; | |
260 | for (;;) { /* this doesn't handle "#define \n" */ | |
261 | if (GETC(==,EOF)) | |
262 | return; | |
263 | if (!iswhite(c)) | |
264 | break; | |
265 | } | |
266 | for (sp = tok;;) { /* get next token */ | |
267 | *sp++ = c; | |
268 | if (GETC(==,EOF)) | |
269 | return; | |
270 | /* | |
271 | * this is where it DOESN'T handle | |
272 | * "#define \n" | |
273 | */ | |
274 | if (!intoken(c)) | |
275 | break; | |
276 | } | |
277 | *sp = EOS; | |
278 | if (dflag || c == (int)'(') { /* only want macros */ | |
279 | getline(); | |
280 | pfnote(tok,curline); | |
281 | } | |
282 | skip: if (c == (int)'\n') { /* get rid of rest of define */ | |
283 | SETLINE | |
284 | if (*(sp - 1) != '\\') | |
285 | return; | |
286 | } | |
287 | (void)skip_key((int)'\n'); | |
288 | } | |
289 | ||
290 | /* | |
291 | * str_entry -- | |
292 | * handle a struct, union or enum entry | |
293 | */ | |
294 | static | |
295 | str_entry(c) | |
296 | register int c; /* current character */ | |
297 | { | |
298 | register char *sp; /* buffer pointer */ | |
299 | int curline; /* line started on */ | |
300 | char tok[BUFSIZ]; /* storage buffer */ | |
301 | ||
302 | curline = lineno; | |
303 | while (iswhite(c)) | |
304 | if (GETC(==,EOF)) | |
305 | return(NO); | |
306 | if (c == (int)'{') /* it was "struct {" */ | |
307 | return(YES); | |
308 | for (sp = tok;;) { /* get next token */ | |
309 | *sp++ = c; | |
310 | if (GETC(==,EOF)) | |
311 | return(NO); | |
312 | if (!intoken(c)) | |
313 | break; | |
314 | } | |
315 | switch ((char)c) { | |
316 | case '{': /* it was "struct foo{" */ | |
317 | --sp; | |
318 | break; | |
319 | case '\n': /* it was "struct foo\n" */ | |
320 | SETLINE; | |
321 | /*FALLTHROUGH*/ | |
322 | default: /* probably "struct foo " */ | |
323 | while (GETC(!=,EOF)) | |
324 | if (!iswhite(c)) | |
325 | break; | |
4c5197ec KB |
326 | if (c != (int)'{') { |
327 | (void)ungetc(c, inf); | |
bcb9ffff | 328 | return(NO); |
4c5197ec | 329 | } |
bcb9ffff KB |
330 | } |
331 | *sp = EOS; | |
332 | pfnote(tok,curline); | |
333 | return(YES); | |
334 | } | |
335 | ||
336 | /* | |
337 | * skip_comment -- | |
338 | * skip over comment | |
339 | */ | |
340 | skip_comment() | |
341 | { | |
342 | register int c, /* character read */ | |
343 | star; /* '*' flag */ | |
344 | ||
345 | for (star = 0;GETC(!=,EOF);) | |
346 | switch((char)c) { | |
347 | /* comments don't nest, nor can they be escaped. */ | |
348 | case '*': | |
349 | star = YES; | |
350 | break; | |
351 | case '/': | |
352 | if (star) | |
353 | return; | |
354 | break; | |
355 | case '\n': | |
356 | SETLINE; | |
357 | /*FALLTHROUGH*/ | |
358 | default: | |
359 | star = NO; | |
360 | } | |
361 | } | |
362 | ||
363 | /* | |
364 | * skip_key -- | |
365 | * skip to next char "key" | |
366 | */ | |
367 | skip_key(key) | |
368 | register int key; | |
369 | { | |
370 | register int c, | |
371 | skip, | |
372 | retval; | |
373 | ||
374 | for (skip = retval = NO;GETC(!=,EOF);) | |
375 | switch((char)c) { | |
376 | case '\\': /* a backslash escapes anything */ | |
377 | skip = !skip; /* we toggle in case it's "\\" */ | |
378 | break; | |
379 | case ';': /* special case for yacc; if one */ | |
380 | case '|': /* of these chars occurs, we may */ | |
381 | retval = YES; /* have moved out of the rule */ | |
382 | break; /* not used by C */ | |
383 | case '\n': | |
384 | SETLINE; | |
385 | /*FALLTHROUGH*/ | |
386 | default: | |
387 | if (c == key && !skip) | |
388 | return(retval); | |
389 | skip = NO; | |
390 | } | |
391 | return(retval); | |
392 | } |