gotweb/parse.y

   1 /*
   2  * Copyright (c) 2019, 2020 Tracey Emery <tracey@openbsd.org>
   3  * Copyright (c) 2004, 2005 Esben Norby <norby@openbsd.org>
   4  * Copyright (c) 2004 Ryan McBride <mcbride@openbsd.org>
   5  * Copyright (c) 2002, 2003, 2004 Henning Brauer <henning@openbsd.org>
   6  * Copyright (c) 2001 Markus Friedl.  All rights reserved.
   7  * Copyright (c) 2001 Daniel Hartmeier.  All rights reserved.
   8  * Copyright (c) 2001 Theo de Raadt.  All rights reserved.
   9  *
  10  * Permission to use, copy, modify, and distribute this software for any
  11  * purpose with or without fee is hereby granted, provided that the above
  12  * copyright notice and this permission notice appear in all copies.
  13  *
  14  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
  15  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
  16  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
  17  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
  18  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
  19  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
  20  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
  21  */
  22
  23 %{
  24 #include <sys/types.h>
  25 #include <sys/queue.h>
  26
  27 #include <ctype.h>
  28 #include <err.h>
  29 #include <limits.h>
  30 #include <stdarg.h>
  31 #include <stdio.h>
  32 #include <stdlib.h>
  33 #include <string.h>
  34
  35 #include "got_error.h"
  36 #include "gotweb.h"
  37
  38 TAILQ_HEAD(files, file)          files = TAILQ_HEAD_INITIALIZER(files);
  39 static struct file {
  40         TAILQ_ENTRY(file)        entry;
  41         FILE                    *stream;
  42         char                    *name;
  43         size_t                   ungetpos;
  44         size_t                   ungetsize;
  45         u_char                  *ungetbuf;
  46         int                      eof_reached;
  47         int                      lineno;
  48 } *file, *topfile;
  49 static const struct got_error*  pushfile(struct file**, const char *);
  50 int              popfile(void);
  51 int              yyparse(void);
  52 int              yylex(void);
  53 int              yyerror(const char *, ...)
  54     __attribute__((__format__ (printf, 1, 2)))
  55     __attribute__((__nonnull__ (1)));
  56 int              kw_cmp(const void *, const void *);
  57 int              lookup(char *);
  58 int              igetc(void);
  59 int              lgetc(int);
  60 void             lungetc(int);
  61 int              findeol(void);
  62
  63 TAILQ_HEAD(symhead, sym)         symhead = TAILQ_HEAD_INITIALIZER(symhead);
  64 struct sym {
  65         TAILQ_ENTRY(sym)         entry;
  66         int                      used;
  67         int                      persist;
  68         char                    *nam;
  69         char                    *val;
  70 };
  71
  72 int      symset(const char *, const char *, int);
  73 char    *symget(const char *);
  74
  75 const struct got_error* gerror = NULL;
  76 struct gotweb_config            *gw_conf;
  77
  78 typedef struct {
  79         union {
  80                 int64_t          number;
  81                 char            *string;
  82         } v;
  83         int lineno;
  84 } YYSTYPE;
  85
  86 %}
  87
  88 %token  GOT_WWW_PATH GOT_MAX_REPOS GOT_SITE_NAME GOT_SITE_OWNER GOT_SITE_LINK
  89 %token  GOT_LOGO GOT_LOGO_URL GOT_SHOW_REPO_OWNER GOT_SHOW_REPO_AGE
  90 %token  GOT_SHOW_REPO_DESCRIPTION GOT_MAX_REPOS_DISPLAY GOT_REPOS_PATH
  91 %token  GOT_MAX_COMMITS_DISPLAY ERROR GOT_SHOW_SITE_OWNER
  92 %token  GOT_SHOW_REPO_CLONEURL
  93 %token  <v.string>      STRING
  94 %token  <v.number>      NUMBER
  95 %type   <v.number>      boolean
  96 %%
  97
  98 grammar         : /* empty */
  99                 | grammar '\n'
 100                 | grammar main '\n'
 101                 ;
 102
 103 boolean         : STRING {
 104                         if (strcasecmp($1, "true") == 0 ||
 105                             strcasecmp($1, "on") == 0 ||
 106                             strcasecmp($1, "yes") == 0)
 107                                 $$ = 1;
 108                         else if (strcasecmp($1, "false") == 0 ||
 109                             strcasecmp($1, "off") == 0 ||
 110                             strcasecmp($1, "no") == 0)
 111                                 $$ = 0;
 112                         else {
 113                                 yyerror("invalid boolean value '%s'", $1);
 114                                 free($1);
 115                                 YYERROR;
 116                         }
 117                         free($1);
 118                 }
 119                 ;
 120 main            : GOT_REPOS_PATH STRING {
 121                         gw_conf->got_repos_path = $2;
 122                 }
 123                 | GOT_WWW_PATH STRING {
 124                         gw_conf->got_www_path = $2;
 125                 }
 126                 | GOT_MAX_REPOS NUMBER {
 127                         if ($2 > 0)
 128                                 gw_conf->got_max_repos = $2;
 129                 }
 130                 | GOT_SITE_NAME STRING {
 131                         gw_conf->got_site_name = $2;
 132                 }
 133                 | GOT_SITE_OWNER STRING {
 134                         gw_conf->got_site_owner = $2;
 135                 }
 136                 | GOT_SITE_LINK STRING {
 137                         gw_conf->got_site_link = $2;
 138                 }
 139                 | GOT_LOGO STRING {
 140                         gw_conf->got_logo = $2;
 141                 }
 142                 | GOT_LOGO_URL STRING {
 143                         gw_conf->got_logo_url = $2;
 144                 }
 145                 | GOT_SHOW_SITE_OWNER boolean {
 146                         gw_conf->got_show_site_owner = $2;
 147                 }
 148                 | GOT_SHOW_REPO_OWNER boolean {
 149                         gw_conf->got_show_repo_owner = $2;
 150                 }
 151                 | GOT_SHOW_REPO_AGE boolean {
 152                         gw_conf->got_show_repo_age = $2;
 153                 }
 154                 | GOT_SHOW_REPO_DESCRIPTION boolean {
 155                         gw_conf->got_show_repo_description = $2;
 156                 }
 157                 | GOT_SHOW_REPO_CLONEURL boolean {
 158                         gw_conf->got_show_repo_cloneurl = $2;
 159                 }
 160                 | GOT_MAX_REPOS_DISPLAY NUMBER {
 161                         if ($2 > 0)
 162                                 gw_conf->got_max_repos_display = $2;
 163                 }
 164                 | GOT_MAX_COMMITS_DISPLAY NUMBER {
 165                         if ($2 > 0)
 166                                 gw_conf->got_max_commits_display = $2;
 167                 }
 168                 ;
 169 %%
 170
 171 struct keywords {
 172         const char      *k_name;
 173         int              k_val;
 174 };
 175
 176 int
 177 yyerror(const char *fmt, ...)
 178 {
 179         va_list          ap;
 180         char            *msg;
 181         char            *err = NULL;
 182
 183         va_start(ap, fmt);
 184         if (vasprintf(&msg, fmt, ap) == -1) {
 185                 gerror =  got_error_from_errno("vasprintf");
 186                 return 0;
 187         }
 188         va_end(ap);
 189         if (asprintf(&err, "%s:%d: %s", file->name, yylval.lineno, msg) == -1) {
 190                 gerror = got_error_from_errno("asprintf");
 191                 return(0);
 192         }
 193         gerror = got_error_msg(GOT_ERR_PARSE_CONFIG, err);
 194         free(msg);
 195         return(0);
 196 }
 197
 198 int
 199 kw_cmp(const void *k, const void *e)
 200 {
 201         return (strcmp(k, ((const struct keywords *)e)->k_name));
 202 }
 203
 204 int
 205 lookup(char *s)
 206 {
 207         /* This has to be sorted always. */
 208         static const struct keywords keywords[] = {
 209                 { "got_logo",                   GOT_LOGO },
 210                 { "got_logo_url",               GOT_LOGO_URL },
 211                 { "got_max_commits_display",    GOT_MAX_COMMITS_DISPLAY },
 212                 { "got_max_repos",              GOT_MAX_REPOS },
 213                 { "got_max_repos_display",      GOT_MAX_REPOS_DISPLAY },
 214                 { "got_repos_path",             GOT_REPOS_PATH },
 215                 { "got_show_repo_age",          GOT_SHOW_REPO_AGE },
 216                 { "got_show_repo_cloneurl",     GOT_SHOW_REPO_CLONEURL },
 217                 { "got_show_repo_description",  GOT_SHOW_REPO_DESCRIPTION },
 218                 { "got_show_repo_owner",        GOT_SHOW_REPO_OWNER },
 219                 { "got_show_site_owner",        GOT_SHOW_SITE_OWNER },
 220                 { "got_site_link",              GOT_SITE_LINK },
 221                 { "got_site_name",              GOT_SITE_NAME },
 222                 { "got_site_owner",             GOT_SITE_OWNER },
 223                 { "got_www_path",               GOT_WWW_PATH },
 224         };
 225         const struct keywords   *p;
 226
 227         p = bsearch(s, keywords, sizeof(keywords)/sizeof(keywords[0]),
 228             sizeof(keywords[0]), kw_cmp);
 229
 230         if (p)
 231                 return (p->k_val);
 232         else
 233                 return (STRING);
 234 }
 235
 236 #define START_EXPAND    1
 237 #define DONE_EXPAND     2
 238
 239 static int      expanding;
 240
 241 int
 242 igetc(void)
 243 {
 244         int     c;
 245
 246         while (1) {
 247                 if (file->ungetpos > 0)
 248                         c = file->ungetbuf[--file->ungetpos];
 249                 else
 250                         c = getc(file->stream);
 251
 252                 if (c == START_EXPAND)
 253                         expanding = 1;
 254                 else if (c == DONE_EXPAND)
 255                         expanding = 0;
 256                 else
 257                         break;
 258         }
 259         return (c);
 260 }
 261
 262 int
 263 lgetc(int quotec)
 264 {
 265         int             c, next;
 266
 267         if (quotec) {
 268                 if ((c = igetc()) == EOF) {
 269                         yyerror("reached end of file while parsing "
 270                             "quoted string");
 271                         if (file == topfile || popfile() == EOF)
 272                                 return (EOF);
 273                         return (quotec);
 274                 }
 275                 return (c);
 276         }
 277
 278         while ((c = igetc()) == '\\') {
 279                 next = igetc();
 280                 if (next != '\n') {
 281                         c = next;
 282                         break;
 283                 }
 284                 yylval.lineno = file->lineno;
 285                 file->lineno++;
 286         }
 287
 288         if (c == EOF) {
 289                 /*
 290                  * Fake EOL when hit EOF for the first time. This gets line
 291                  * count right if last line in included file is syntactically
 292                  * invalid and has no newline.
 293                  */
 294                 if (file->eof_reached == 0) {
 295                         file->eof_reached = 1;
 296                         return ('\n');
 297                 }
 298                 while (c == EOF) {
 299                         if (file == topfile || popfile() == EOF)
 300                                 return (EOF);
 301                         c = igetc();
 302                 }
 303         }
 304         return (c);
 305 }
 306
 307 void
 308 lungetc(int c)
 309 {
 310         if (c == EOF)
 311                 return;
 312
 313         if (file->ungetpos >= file->ungetsize) {
 314                 void *p = reallocarray(file->ungetbuf, file->ungetsize, 2);
 315                 if (p == NULL)
 316                         err(1, "%s", __func__);
 317                 file->ungetbuf = p;
 318                 file->ungetsize *= 2;
 319         }
 320         file->ungetbuf[file->ungetpos++] = c;
 321 }
 322
 323 int
 324 findeol(void)
 325 {
 326         int     c;
 327
 328         /* Skip to either EOF or the first real EOL. */
 329         while (1) {
 330                 c = lgetc(0);
 331                 if (c == '\n') {
 332                         file->lineno++;
 333                         break;
 334                 }
 335                 if (c == EOF)
 336                         break;
 337         }
 338         return (ERROR);
 339 }
 340
 341 int
 342 yylex(void)
 343 {
 344         unsigned char    buf[8096];
 345         unsigned char   *p, *val;
 346         int              quotec, next, c;
 347         int              token;
 348
 349 top:
 350         p = buf;
 351         while ((c = lgetc(0)) == ' ' || c == '\t')
 352                 ; /* nothing */
 353
 354         yylval.lineno = file->lineno;
 355         if (c == '#')
 356                 while ((c = lgetc(0)) != '\n' && c != EOF)
 357                         ; /* nothing */
 358         if (c == '$' && !expanding) {
 359                 while (1) {
 360                         if ((c = lgetc(0)) == EOF)
 361                                 return (0);
 362
 363                         if (p + 1 >= buf + sizeof(buf) - 1) {
 364                                 yyerror("string too long");
 365                                 return (findeol());
 366                         }
 367                         if (isalnum(c) || c == '_') {
 368                                 *p++ = c;
 369                                 continue;
 370                         }
 371                         *p = '\0';
 372                         lungetc(c);
 373                         break;
 374                 }
 375                 val = symget(buf);
 376                 if (val == NULL) {
 377                         yyerror("macro '%s' not defined", buf);
 378                         return (findeol());
 379                 }
 380                 p = val + strlen(val) - 1;
 381                 lungetc(DONE_EXPAND);
 382                 while (p >= val) {
 383                         lungetc(*p);
 384                         p--;
 385                 }
 386                 lungetc(START_EXPAND);
 387                 goto top;
 388         }
 389
 390         switch (c) {
 391         case '\'':
 392         case '"':
 393                 quotec = c;
 394                 while (1) {
 395                         if ((c = lgetc(quotec)) == EOF)
 396                                 return (0);
 397                         if (c == '\n') {
 398                                 file->lineno++;
 399                                 continue;
 400                         } else if (c == '\\') {
 401                                 if ((next = lgetc(quotec)) == EOF)
 402                                         return (0);
 403                                 if (next == quotec || c == ' ' || c == '\t')
 404                                         c = next;
 405                                 else if (next == '\n') {
 406                                         file->lineno++;
 407                                         continue;
 408                                 } else
 409                                         lungetc(next);
 410                         } else if (c == quotec) {
 411                                 *p = '\0';
 412                                 break;
 413                         } else if (c == '\0') {
 414                                 yyerror("syntax error");
 415                                 return (findeol());
 416                         }
 417                         if (p + 1 >= buf + sizeof(buf) - 1) {
 418                                 yyerror("string too long");
 419                                 return (findeol());
 420                         }
 421                         *p++ = c;
 422                 }
 423                 yylval.v.string = strdup(buf);
 424                 if (yylval.v.string == NULL)
 425                         err(1, "%s", __func__);
 426                 return (STRING);
 427         }
 428
 429 #define allowed_to_end_number(x) \
 430         (isspace(x) || x == ')' || x ==',' || x == '/' || x == '}' || x == '=')
 431
 432         if (c == '-' || isdigit(c)) {
 433                 do {
 434                         *p++ = c;
 435                         if ((unsigned)(p-buf) >= sizeof(buf)) {
 436                                 yyerror("string too long");
 437                                 return (findeol());
 438                         }
 439                 } while ((c = lgetc(0)) != EOF && isdigit(c));
 440                 lungetc(c);
 441                 if (p == buf + 1 && buf[0] == '-')
 442                         goto nodigits;
 443                 if (c == EOF || allowed_to_end_number(c)) {
 444                         const char *errstr = NULL;
 445
 446                         *p = '\0';
 447                         yylval.v.number = strtonum(buf, LLONG_MIN,
 448                             LLONG_MAX, &errstr);
 449                         if (errstr) {
 450                                 yyerror("\"%s\" invalid number: %s",
 451                                     buf, errstr);
 452                                 return (findeol());
 453                         }
 454                         return (NUMBER);
 455                 } else {
 456 nodigits:
 457                         while (p > buf + 1)
 458                                 lungetc(*--p);
 459                         c = *--p;
 460                         if (c == '-')
 461                                 return (c);
 462                 }
 463         }
 464
 465 #define allowed_in_string(x) \
 466         (isalnum(x) || (ispunct(x) && x != '(' && x != ')' && \
 467         x != '{' && x != '}' && \
 468         x != '!' && x != '=' && x != '#' && \
 469         x != ','))
 470
 471         if (isalnum(c) || c == ':' || c == '_') {
 472                 do {
 473                         *p++ = c;
 474                         if ((unsigned)(p-buf) >= sizeof(buf)) {
 475                                 yyerror("string too long");
 476                                 return (findeol());
 477                         }
 478                 } while ((c = lgetc(0)) != EOF && (allowed_in_string(c)));
 479                 lungetc(c);
 480                 *p = '\0';
 481                 if ((token = lookup(buf)) == STRING)
 482                         if ((yylval.v.string = strdup(buf)) == NULL)
 483                                 err(1, "%s", __func__);
 484                 return (token);
 485         }
 486         if (c == '\n') {
 487                 yylval.lineno = file->lineno;
 488                 file->lineno++;
 489         }
 490         if (c == EOF)
 491                 return (0);
 492         return (c);
 493 }
 494
 495 static const struct got_error*
 496 pushfile(struct file **nfile, const char *name)
 497 {
 498         const struct got_error* error = NULL;
 499
 500         if (((*nfile) = calloc(1, sizeof(struct file))) == NULL)
 501                 return got_error_from_errno2(__func__, "calloc");
 502         if (((*nfile)->name = strdup(name)) == NULL) {
 503                 free(nfile);
 504                 return got_error_from_errno2(__func__, "strdup");
 505         }
 506         if (((*nfile)->stream = fopen((*nfile)->name, "r")) == NULL) {
 507                 char *msg = NULL;
 508                 if (asprintf(&msg, "%s", (*nfile)->name) == -1)
 509                         return got_error_from_errno("asprintf");
 510                 error = got_error_msg(GOT_ERR_NO_CONFIG_FILE, msg);
 511                 free((*nfile)->name);
 512                 free((*nfile));
 513                 free(msg);
 514                 return error;
 515         }
 516         (*nfile)->lineno = TAILQ_EMPTY(&files) ? 1 : 0;
 517         (*nfile)->ungetsize = 16;
 518         (*nfile)->ungetbuf = malloc((*nfile)->ungetsize);
 519         if ((*nfile)->ungetbuf == NULL) {
 520                 fclose((*nfile)->stream);
 521                 free((*nfile)->name);
 522                 free((*nfile));
 523                 return got_error_from_errno2(__func__, "malloc");
 524         }
 525         TAILQ_INSERT_TAIL(&files, (*nfile), entry);
 526         return error;
 527 }
 528
 529 int
 530 popfile(void)
 531 {
 532         struct file     *prev = NULL;
 533
 534         TAILQ_REMOVE(&files, file, entry);
 535         fclose(file->stream);
 536         free(file->name);
 537         free(file->ungetbuf);
 538         free(file);
 539         file = prev;
 540         return (file ? 0 : EOF);
 541 }
 542
 543 const struct got_error*
 544 parse_gotweb_config(struct gotweb_config **gconf, const char *filename)
 545 {
 546         gw_conf = malloc(sizeof(struct gotweb_config));
 547         if (gw_conf == NULL) {
 548                 gerror = got_error_from_errno("malloc");
 549                 goto done;
 550         }
 551         gw_conf->got_repos_path = strdup(D_GOTPATH);
 552         if (gw_conf->got_repos_path == NULL) {
 553                 gerror = got_error_from_errno("strdup");
 554                 goto done;
 555         }
 556         gw_conf->got_www_path = strdup(D_GOTWWW);
 557         if (gw_conf->got_www_path == NULL) {
 558                 gerror = got_error_from_errno("strdup");
 559                 goto done;
 560         }
 561         gw_conf->got_site_name = strdup(D_SITENAME);
 562         if (gw_conf->got_site_name == NULL) {
 563                 gerror = got_error_from_errno("strdup");
 564                 goto done;
 565         }
 566         gw_conf->got_site_owner = strdup(D_SITEOWNER);
 567         if (gw_conf->got_site_owner == NULL) {
 568                 gerror = got_error_from_errno("strdup");
 569                 goto done;
 570         }
 571         gw_conf->got_site_link = strdup(D_SITELINK);
 572         if (gw_conf->got_site_link == NULL) {
 573                 gerror = got_error_from_errno("strdup");
 574                 goto done;
 575         }
 576         gw_conf->got_logo = strdup(D_GOTLOGO);
 577         if (gw_conf->got_logo == NULL) {
 578                 gerror = got_error_from_errno("strdup");
 579                 goto done;
 580         }
 581         gw_conf->got_logo_url = strdup(D_GOTURL);
 582         if (gw_conf->got_logo_url == NULL) {
 583                 gerror = got_error_from_errno("strdup");
 584                 goto done;
 585         }
 586         gw_conf->got_show_site_owner = D_SHOWSOWNER;
 587         gw_conf->got_show_repo_owner = D_SHOWROWNER;
 588         gw_conf->got_show_repo_age = D_SHOWAGE;
 589         gw_conf->got_show_repo_description = D_SHOWDESC;
 590         gw_conf->got_show_repo_cloneurl = D_SHOWURL;
 591         gw_conf->got_max_repos = D_MAXREPO;
 592         gw_conf->got_max_repos_display = D_MAXREPODISP;
 593         gw_conf->got_max_commits_display = D_MAXCOMMITDISP;
 594
 595         /*
 596          * We don't require that the gotweb config file exists
 597          * So reset gerror if it doesn't exist and goto done.
 598          */
 599         gerror = pushfile(&file, filename);
 600         if (gerror && gerror->code == GOT_ERR_NO_CONFIG_FILE) {
 601                 gerror = NULL;
 602                 goto done;
 603         } else if (gerror)
 604                 return gerror;
 605         topfile = file;
 606
 607         yyparse();
 608         popfile();
 609 done:
 610         *gconf = gw_conf;
 611         return gerror;
 612 }
 613
 614 int
 615 symset(const char *nam, const char *val, int persist)
 616 {
 617         struct sym      *sym;
 618
 619         TAILQ_FOREACH(sym, &symhead, entry) {
 620                 if (strcmp(nam, sym->nam) == 0)
 621                         break;
 622         }
 623
 624         if (sym != NULL) {
 625                 if (sym->persist == 1)
 626                         return (0);
 627                 else {
 628                         free(sym->nam);
 629                         free(sym->val);
 630                         TAILQ_REMOVE(&symhead, sym, entry);
 631                         free(sym);
 632                 }
 633         }
 634         if ((sym = calloc(1, sizeof(*sym))) == NULL)
 635                 return (-1);
 636
 637         sym->nam = strdup(nam);
 638         if (sym->nam == NULL) {
 639                 free(sym);
 640                 return (-1);
 641         }
 642         sym->val = strdup(val);
 643         if (sym->val == NULL) {
 644                 free(sym->nam);
 645                 free(sym);
 646                 return (-1);
 647         }
 648         sym->used = 0;
 649         sym->persist = persist;
 650         TAILQ_INSERT_TAIL(&symhead, sym, entry);
 651         return (0);
 652 }
 653
 654 int
 655 cmdline_symset(char *s)
 656 {
 657         char    *sym, *val;
 658         int     ret;
 659         size_t  len;
 660
 661         if ((val = strrchr(s, '=')) == NULL)
 662                 return (-1);
 663
 664         len = strlen(s) - strlen(val) + 1;
 665         if ((sym = malloc(len)) == NULL)
 666                 errx(1, "cmdline_symset: malloc");
 667
 668         strlcpy(sym, s, len);
 669
 670         ret = symset(sym, val + 1, 1);
 671         free(sym);
 672
 673         return (ret);
 674 }
 675
 676 char *
 677 symget(const char *nam)
 678 {
 679         struct sym      *sym;
 680
 681         TAILQ_FOREACH(sym, &symhead, entry) {
 682                 if (strcmp(nam, sym->nam) == 0) {
 683                         sym->used = 1;
 684                         return (sym->val);
 685                 }
 686         }
 687         return (NULL);
 688 }