docproc: abstract terminating lines at first space
[cascardo/linux.git] / scripts / docproc.c
1 /*
2  *      docproc is a simple preprocessor for the template files
3  *      used as placeholders for the kernel internal documentation.
4  *      docproc is used for documentation-frontend and
5  *      dependency-generator.
6  *      The two usages have in common that they require
7  *      some knowledge of the .tmpl syntax, therefore they
8  *      are kept together.
9  *
10  *      documentation-frontend
11  *              Scans the template file and call kernel-doc for
12  *              all occurrences of ![EIF]file
13  *              Beforehand each referenced file is scanned for
14  *              any symbols that are exported via these macros:
15  *                      EXPORT_SYMBOL(), EXPORT_SYMBOL_GPL(), &
16  *                      EXPORT_SYMBOL_GPL_FUTURE()
17  *              This is used to create proper -function and
18  *              -nofunction arguments in calls to kernel-doc.
19  *              Usage: docproc doc file.tmpl
20  *
21  *      dependency-generator:
22  *              Scans the template file and list all files
23  *              referenced in a format recognized by make.
24  *              Usage:  docproc depend file.tmpl
25  *              Writes dependency information to stdout
26  *              in the following format:
27  *              file.tmpl src.c src2.c
28  *              The filenames are obtained from the following constructs:
29  *              !Efilename
30  *              !Ifilename
31  *              !Dfilename
32  *              !Ffilename
33  *              !Pfilename
34  *
35  */
36
37 #define _GNU_SOURCE
38 #include <stdio.h>
39 #include <stdlib.h>
40 #include <string.h>
41 #include <ctype.h>
42 #include <unistd.h>
43 #include <limits.h>
44 #include <errno.h>
45 #include <sys/types.h>
46 #include <sys/wait.h>
47
48 /* exitstatus is used to keep track of any failing calls to kernel-doc,
49  * but execution continues. */
50 int exitstatus = 0;
51
52 typedef void DFL(char *);
53 DFL *defaultline;
54
55 typedef void FILEONLY(char * file);
56 FILEONLY *internalfunctions;
57 FILEONLY *externalfunctions;
58 FILEONLY *symbolsonly;
59 FILEONLY *findall;
60
61 typedef void FILELINE(char * file, char * line);
62 FILELINE * singlefunctions;
63 FILELINE * entity_system;
64 FILELINE * docsection;
65
66 #define MAXLINESZ     2048
67 #define MAXFILES      250
68 #define KERNELDOCPATH "scripts/"
69 #define KERNELDOC     "kernel-doc"
70 #define DOCBOOK       "-docbook"
71 #define LIST          "-list"
72 #define FUNCTION      "-function"
73 #define NOFUNCTION    "-nofunction"
74 #define NODOCSECTIONS "-no-doc-sections"
75 #define SHOWNOTFOUND  "-show-not-found"
76
77 static char *srctree, *kernsrctree;
78
79 static char **all_list = NULL;
80 static int all_list_len = 0;
81
82 static void consume_symbol(const char *sym)
83 {
84         int i;
85
86         for (i = 0; i < all_list_len; i++) {
87                 if (!all_list[i])
88                         continue;
89                 if (strcmp(sym, all_list[i]))
90                         continue;
91                 all_list[i] = NULL;
92                 break;
93         }
94 }
95
96 static void usage (void)
97 {
98         fprintf(stderr, "Usage: docproc {doc|depend} file\n");
99         fprintf(stderr, "Input is read from file.tmpl. Output is sent to stdout\n");
100         fprintf(stderr, "doc: frontend when generating kernel documentation\n");
101         fprintf(stderr, "depend: generate list of files referenced within file\n");
102         fprintf(stderr, "Environment variable SRCTREE: absolute path to sources.\n");
103         fprintf(stderr, "                     KBUILD_SRC: absolute path to kernel source tree.\n");
104 }
105
106 /*
107  * Execute kernel-doc with parameters given in svec
108  */
109 static void exec_kernel_doc(char **svec)
110 {
111         pid_t pid;
112         int ret;
113         char real_filename[PATH_MAX + 1];
114         /* Make sure output generated so far are flushed */
115         fflush(stdout);
116         switch (pid=fork()) {
117                 case -1:
118                         perror("fork");
119                         exit(1);
120                 case  0:
121                         memset(real_filename, 0, sizeof(real_filename));
122                         strncat(real_filename, kernsrctree, PATH_MAX);
123                         strncat(real_filename, "/" KERNELDOCPATH KERNELDOC,
124                                         PATH_MAX - strlen(real_filename));
125                         execvp(real_filename, svec);
126                         fprintf(stderr, "exec ");
127                         perror(real_filename);
128                         exit(1);
129                 default:
130                         waitpid(pid, &ret ,0);
131         }
132         if (WIFEXITED(ret))
133                 exitstatus |= WEXITSTATUS(ret);
134         else
135                 exitstatus = 0xff;
136 }
137
138 /* Types used to create list of all exported symbols in a number of files */
139 struct symbols
140 {
141         char *name;
142 };
143
144 struct symfile
145 {
146         char *filename;
147         struct symbols *symbollist;
148         int symbolcnt;
149 };
150
151 struct symfile symfilelist[MAXFILES];
152 int symfilecnt = 0;
153
154 static void add_new_symbol(struct symfile *sym, char * symname)
155 {
156         sym->symbollist =
157           realloc(sym->symbollist, (sym->symbolcnt + 1) * sizeof(char *));
158         sym->symbollist[sym->symbolcnt++].name = strdup(symname);
159 }
160
161 /* Add a filename to the list */
162 static struct symfile * add_new_file(char * filename)
163 {
164         symfilelist[symfilecnt++].filename = strdup(filename);
165         return &symfilelist[symfilecnt - 1];
166 }
167
168 /* Check if file already are present in the list */
169 static struct symfile * filename_exist(char * filename)
170 {
171         int i;
172         for (i=0; i < symfilecnt; i++)
173                 if (strcmp(symfilelist[i].filename, filename) == 0)
174                         return &symfilelist[i];
175         return NULL;
176 }
177
178 /*
179  * List all files referenced within the template file.
180  * Files are separated by tabs.
181  */
182 static void adddep(char * file)            { printf("\t%s", file); }
183 static void adddep2(char * file, char * line)     { line = line; adddep(file); }
184 static void noaction(char * line)                  { line = line; }
185 static void noaction2(char * file, char * line)   { file = file; line = line; }
186
187 /* Echo the line without further action */
188 static void printline(char * line)               { printf("%s", line); }
189
190 /*
191  * Find all symbols in filename that are exported with EXPORT_SYMBOL &
192  * EXPORT_SYMBOL_GPL (& EXPORT_SYMBOL_GPL_FUTURE implicitly).
193  * All symbols located are stored in symfilelist.
194  */
195 static void find_export_symbols(char * filename)
196 {
197         FILE * fp;
198         struct symfile *sym;
199         char line[MAXLINESZ];
200         if (filename_exist(filename) == NULL) {
201                 char real_filename[PATH_MAX + 1];
202                 memset(real_filename, 0, sizeof(real_filename));
203                 strncat(real_filename, srctree, PATH_MAX);
204                 strncat(real_filename, "/", PATH_MAX - strlen(real_filename));
205                 strncat(real_filename, filename,
206                                 PATH_MAX - strlen(real_filename));
207                 sym = add_new_file(filename);
208                 fp = fopen(real_filename, "r");
209                 if (fp == NULL) {
210                         fprintf(stderr, "docproc: ");
211                         perror(real_filename);
212                         exit(1);
213                 }
214                 while (fgets(line, MAXLINESZ, fp)) {
215                         char *p;
216                         char *e;
217                         if (((p = strstr(line, "EXPORT_SYMBOL_GPL")) != NULL) ||
218                             ((p = strstr(line, "EXPORT_SYMBOL")) != NULL)) {
219                                 /* Skip EXPORT_SYMBOL{_GPL} */
220                                 while (isalnum(*p) || *p == '_')
221                                         p++;
222                                 /* Remove parentheses & additional whitespace */
223                                 while (isspace(*p))
224                                         p++;
225                                 if (*p != '(')
226                                         continue; /* Syntax error? */
227                                 else
228                                         p++;
229                                 while (isspace(*p))
230                                         p++;
231                                 e = p;
232                                 while (isalnum(*e) || *e == '_')
233                                         e++;
234                                 *e = '\0';
235                                 add_new_symbol(sym, p);
236                         }
237                 }
238                 fclose(fp);
239         }
240 }
241
242 /*
243  * Document all external or internal functions in a file.
244  * Call kernel-doc with following parameters:
245  * kernel-doc -docbook -nofunction function_name1 filename
246  * Function names are obtained from all the src files
247  * by find_export_symbols.
248  * intfunc uses -nofunction
249  * extfunc uses -function
250  */
251 static void docfunctions(char * filename, char * type)
252 {
253         int i,j;
254         int symcnt = 0;
255         int idx = 0;
256         char **vec;
257
258         for (i=0; i <= symfilecnt; i++)
259                 symcnt += symfilelist[i].symbolcnt;
260         vec = malloc((2 + 2 * symcnt + 3) * sizeof(char *));
261         if (vec == NULL) {
262                 perror("docproc: ");
263                 exit(1);
264         }
265         vec[idx++] = KERNELDOC;
266         vec[idx++] = DOCBOOK;
267         vec[idx++] = NODOCSECTIONS;
268         for (i=0; i < symfilecnt; i++) {
269                 struct symfile * sym = &symfilelist[i];
270                 for (j=0; j < sym->symbolcnt; j++) {
271                         vec[idx++]     = type;
272                         consume_symbol(sym->symbollist[j].name);
273                         vec[idx++] = sym->symbollist[j].name;
274                 }
275         }
276         vec[idx++]     = filename;
277         vec[idx] = NULL;
278         printf("<!-- %s -->\n", filename);
279         exec_kernel_doc(vec);
280         fflush(stdout);
281         free(vec);
282 }
283 static void intfunc(char * filename) {  docfunctions(filename, NOFUNCTION); }
284 static void extfunc(char * filename) { docfunctions(filename, FUNCTION);   }
285
286 /*
287  * Document specific function(s) in a file.
288  * Call kernel-doc with the following parameters:
289  * kernel-doc -docbook -function function1 [-function function2]
290  */
291 static void singfunc(char * filename, char * line)
292 {
293         char *vec[200]; /* Enough for specific functions */
294         int i, idx = 0;
295         int startofsym = 1;
296         vec[idx++] = KERNELDOC;
297         vec[idx++] = DOCBOOK;
298         vec[idx++] = SHOWNOTFOUND;
299
300         /* Split line up in individual parameters preceded by FUNCTION */
301         for (i=0; line[i]; i++) {
302                 if (isspace(line[i])) {
303                         line[i] = '\0';
304                         startofsym = 1;
305                         continue;
306                 }
307                 if (startofsym) {
308                         startofsym = 0;
309                         vec[idx++] = FUNCTION;
310                         vec[idx++] = &line[i];
311                 }
312         }
313         for (i = 0; i < idx; i++) {
314                 if (strcmp(vec[i], FUNCTION))
315                         continue;
316                 consume_symbol(vec[i + 1]);
317         }
318         vec[idx++] = filename;
319         vec[idx] = NULL;
320         exec_kernel_doc(vec);
321 }
322
323 /*
324  * Insert specific documentation section from a file.
325  * Call kernel-doc with the following parameters:
326  * kernel-doc -docbook -function "doc section" filename
327  */
328 static void docsect(char *filename, char *line)
329 {
330         /* kerneldoc -docbook -show-not-found -function "section" file NULL */
331         char *vec[7];
332         char *s;
333
334         for (s = line; *s; s++)
335                 if (*s == '\n')
336                         *s = '\0';
337
338         if (asprintf(&s, "DOC: %s", line) < 0) {
339                 perror("asprintf");
340                 exit(1);
341         }
342         consume_symbol(s);
343         free(s);
344
345         vec[0] = KERNELDOC;
346         vec[1] = DOCBOOK;
347         vec[2] = SHOWNOTFOUND;
348         vec[3] = FUNCTION;
349         vec[4] = line;
350         vec[5] = filename;
351         vec[6] = NULL;
352         exec_kernel_doc(vec);
353 }
354
355 static void find_all_symbols(char *filename)
356 {
357         char *vec[4]; /* kerneldoc -list file NULL */
358         pid_t pid;
359         int ret, i, count, start;
360         char real_filename[PATH_MAX + 1];
361         int pipefd[2];
362         char *data, *str;
363         size_t data_len = 0;
364
365         vec[0] = KERNELDOC;
366         vec[1] = LIST;
367         vec[2] = filename;
368         vec[3] = NULL;
369
370         if (pipe(pipefd)) {
371                 perror("pipe");
372                 exit(1);
373         }
374
375         switch (pid=fork()) {
376                 case -1:
377                         perror("fork");
378                         exit(1);
379                 case  0:
380                         close(pipefd[0]);
381                         dup2(pipefd[1], 1);
382                         memset(real_filename, 0, sizeof(real_filename));
383                         strncat(real_filename, kernsrctree, PATH_MAX);
384                         strncat(real_filename, "/" KERNELDOCPATH KERNELDOC,
385                                         PATH_MAX - strlen(real_filename));
386                         execvp(real_filename, vec);
387                         fprintf(stderr, "exec ");
388                         perror(real_filename);
389                         exit(1);
390                 default:
391                         close(pipefd[1]);
392                         data = malloc(4096);
393                         do {
394                                 while ((ret = read(pipefd[0],
395                                                    data + data_len,
396                                                    4096)) > 0) {
397                                         data_len += ret;
398                                         data = realloc(data, data_len + 4096);
399                                 }
400                         } while (ret == -EAGAIN);
401                         if (ret != 0) {
402                                 perror("read");
403                                 exit(1);
404                         }
405                         waitpid(pid, &ret ,0);
406         }
407         if (WIFEXITED(ret))
408                 exitstatus |= WEXITSTATUS(ret);
409         else
410                 exitstatus = 0xff;
411
412         count = 0;
413         /* poor man's strtok, but with counting */
414         for (i = 0; i < data_len; i++) {
415                 if (data[i] == '\n') {
416                         count++;
417                         data[i] = '\0';
418                 }
419         }
420         start = all_list_len;
421         all_list_len += count;
422         all_list = realloc(all_list, sizeof(char *) * all_list_len);
423         str = data;
424         for (i = 0; i < data_len && start != all_list_len; i++) {
425                 if (data[i] == '\0') {
426                         all_list[start] = str;
427                         str = data + i + 1;
428                         start++;
429                 }
430         }
431 }
432
433 /*
434  * Terminate s at first space, if any. If there was a space, return pointer to
435  * the character after that. Otherwise, return pointer to the terminating NUL.
436  */
437 static char *chomp(char *s)
438 {
439         while (*s && !isspace(*s))
440                 s++;
441
442         if (*s)
443                 *s++ = '\0';
444
445         return s;
446 }
447
448 /* Return pointer to directive content, or NULL if not a directive. */
449 static char *is_directive(char *line)
450 {
451         if (line[0] == '!')
452                 return line + 1;
453
454         return NULL;
455 }
456
457 /*
458  * Parse file, calling action specific functions for:
459  * 1) Lines containing !E
460  * 2) Lines containing !I
461  * 3) Lines containing !D
462  * 4) Lines containing !F
463  * 5) Lines containing !P
464  * 6) Lines containing !C
465  * 7) Default lines - lines not matching the above
466  */
467 static void parse_file(FILE *infile)
468 {
469         char line[MAXLINESZ];
470         char *p, *s;
471         while (fgets(line, MAXLINESZ, infile)) {
472                 p = is_directive(line);
473                 if (!p) {
474                         defaultline(line);
475                         continue;
476                 }
477
478                 switch (*p++) {
479                 case 'E':
480                         chomp(p);
481                         externalfunctions(p);
482                         break;
483                 case 'I':
484                         chomp(p);
485                         internalfunctions(p);
486                         break;
487                 case 'D':
488                         chomp(p);
489                         symbolsonly(p);
490                         break;
491                 case 'F':
492                         /* filename */
493                         s = chomp(p);
494                         /* function names */
495                         while (isspace(*s))
496                                 s++;
497                         singlefunctions(p, s);
498                         break;
499                 case 'P':
500                         /* filename */
501                         s = chomp(p);
502                         /* DOC: section name */
503                         while (isspace(*s))
504                                 s++;
505                         docsection(p, s);
506                         break;
507                 case 'C':
508                         chomp(p);
509                         if (findall)
510                                 findall(p);
511                         break;
512                 default:
513                         defaultline(line);
514                 }
515         }
516         fflush(stdout);
517 }
518
519
520 int main(int argc, char *argv[])
521 {
522         const char *subcommand, *filename;
523         FILE * infile;
524         int i;
525
526         srctree = getenv("SRCTREE");
527         if (!srctree)
528                 srctree = getcwd(NULL, 0);
529         kernsrctree = getenv("KBUILD_SRC");
530         if (!kernsrctree || !*kernsrctree)
531                 kernsrctree = srctree;
532         if (argc != 3) {
533                 usage();
534                 exit(1);
535         }
536
537         subcommand = argv[1];
538         filename = argv[2];
539
540         /* Open file, exit on error */
541         infile = fopen(filename, "r");
542         if (infile == NULL) {
543                 fprintf(stderr, "docproc: ");
544                 perror(filename);
545                 exit(2);
546         }
547
548         if (strcmp("doc", subcommand) == 0) {
549                 /* Need to do this in two passes.
550                  * First pass is used to collect all symbols exported
551                  * in the various files;
552                  * Second pass generate the documentation.
553                  * This is required because some functions are declared
554                  * and exported in different files :-((
555                  */
556                 /* Collect symbols */
557                 defaultline       = noaction;
558                 internalfunctions = find_export_symbols;
559                 externalfunctions = find_export_symbols;
560                 symbolsonly       = find_export_symbols;
561                 singlefunctions   = noaction2;
562                 docsection        = noaction2;
563                 findall           = find_all_symbols;
564                 parse_file(infile);
565
566                 /* Rewind to start from beginning of file again */
567                 fseek(infile, 0, SEEK_SET);
568                 defaultline       = printline;
569                 internalfunctions = intfunc;
570                 externalfunctions = extfunc;
571                 symbolsonly       = printline;
572                 singlefunctions   = singfunc;
573                 docsection        = docsect;
574                 findall           = NULL;
575
576                 parse_file(infile);
577
578                 for (i = 0; i < all_list_len; i++) {
579                         if (!all_list[i])
580                                 continue;
581                         fprintf(stderr, "Warning: didn't use docs for %s\n",
582                                 all_list[i]);
583                 }
584         } else if (strcmp("depend", subcommand) == 0) {
585                 /* Create first part of dependency chain
586                  * file.tmpl */
587                 printf("%s\t", filename);
588                 defaultline       = noaction;
589                 internalfunctions = adddep;
590                 externalfunctions = adddep;
591                 symbolsonly       = adddep;
592                 singlefunctions   = adddep2;
593                 docsection        = adddep2;
594                 findall           = adddep;
595                 parse_file(infile);
596                 printf("\n");
597         } else {
598                 fprintf(stderr, "Unknown option: %s\n", subcommand);
599                 exit(1);
600         }
601         fclose(infile);
602         fflush(stdout);
603         return exitstatus;
604 }