src/protocol/rewrite/rewrite.c

   1 /* URI rewriting module */
   2
   3 #ifdef HAVE_CONFIG_H
   4 #include "config.h"
   5 #endif
   6
   7 #include "elinks.h"
   8
   9 #include "config/options.h"
  10 #include "intl/gettext/libintl.h"
  11 #include "main/event.h"
  12 #include "main/module.h"
  13 #include "protocol/rewrite/rewrite.h"
  14 #include "protocol/uri.h"
  15 #include "session/location.h"
  16 #include "session/session.h"
  17 #include "util/string.h"
  18
  19
  20 enum uri_rewrite_type {
  21         URI_REWRITE_DUMB,
  22         URI_REWRITE_SMART,
  23 };
  24
  25
  26 /* TODO: An event hook for follow-url might also feel at home here. --jonas */
  27
  28 enum uri_rewrite_option {
  29         URI_REWRITE_TREE,
  30
  31         URI_REWRITE_ENABLE_DUMB,
  32         URI_REWRITE_ENABLE_SMART,
  33
  34         URI_REWRITE_DUMB_TREE,
  35         URI_REWRITE_DUMB_TEMPLATE,
  36
  37         URI_REWRITE_SMART_TREE,
  38         URI_REWRITE_SMART_TEMPLATE,
  39
  40         URI_REWRITE_OPTIONS,
  41 };
  42
  43 static struct option_info uri_rewrite_options[] = {
  44         INIT_OPT_TREE("protocol", N_("URI rewriting"),
  45                 "rewrite", OPT_SORT,
  46                 N_("Rules for rewriting URIs entered in the goto dialog. "
  47                 "It makes it possible to define a set of prefixes that will "
  48                 "be expanded if they match a string entered in the goto "
  49                 "dialog. The prefixes can be dumb, meaning that they work "
  50                 "only like URI abbreviations, or smart ones, making it "
  51                 "possible to pass arguments to them like search engine "
  52                 "keywords.")),
  53
  54         INIT_OPT_BOOL("protocol.rewrite", N_("Enable dumb prefixes"),
  55                 "enable-dumb", 0, 1,
  56                 N_("Enable dumb prefixes - simple URI abbreviations which "
  57                 "can be written to the Goto URL dialog instead of actual URIs "
  58                 "- i.e. if you write 'elinks' there, you are directed to "
  59                 "http://elinks.cz/.")),
  60
  61         INIT_OPT_BOOL("protocol.rewrite", N_("Enable smart prefixes"),
  62                 "enable-smart", 0, 1,
  63                 N_("Enable smart prefixes - URI templates triggered by "
  64                 "writing given abbreviation to the Goto URL dialog followed "
  65                 "by a list of arguments from which the actual URI is composed "
  66                 "- i.e. 'gg:search keywords' or 'gn search keywords for "
  67                 "news'.")),
  68
  69         INIT_OPT_TREE("protocol.rewrite", N_("Dumb Prefixes"),
  70                 "dumb", OPT_AUTOCREATE | OPT_SORT,
  71                 N_("Dumb prefixes, see enable-dumb description for details.")),
  72
  73         INIT_OPT_STRING("protocol.rewrite.dumb", NULL,
  74                 "_template_", 0, "",
  75                 /* xgettext:no-c-format */
  76                 N_("Replacement URI for this dumbprefix:\n"
  77                 "%c in the string means the current URL\n"
  78                 "%% in the string means '%'")),
  79
  80         INIT_OPT_TREE("protocol.rewrite", N_("Smart Prefixes"),
  81                 "smart", OPT_AUTOCREATE | OPT_SORT,
  82                 N_("Smart prefixes, see enable-smart description for "
  83                 "details.")),
  84
  85         /* TODO: In some rare occations current link URI and referrer might
  86          * also be useful and dare I mention some kind of proxy argument. --jonas */
  87         INIT_OPT_STRING("protocol.rewrite.smart", NULL,
  88                 "_template_", 0, "",
  89                 /* xgettext:no-c-format */
  90                 N_("Replacement URI for this smartprefix:\n"
  91                 "%c in the string means the current URL\n"
  92                 "%s in the string means the whole argument to smartprefix\n"
  93                 "%0,%1,...,%9 means argument 0, 1, ..., 9\n"
  94                 "%% in the string means '%'")),
  95
  96         INIT_OPT_STRING("protocol.rewrite", N_("Default template"),
  97                 "default_template", 0, "",
  98                 /* xgettext:no-c-format */
  99                 N_("Default URI template used when the string entered in "
 100                 "the goto dialog does not appear to be a URI or a filename "
 101                 "(i.e. contains no '.', ':' or '/' characters), and does "
 102                 "not match any defined prefixes. Set the value to \"\" to "
 103                 "disable use of the default template rewrite rule.\n"
 104                 "\n"
 105                 "%c in the template means the current URL,\n"
 106                 "%s in the template means the whole string from the goto\n"
 107                 "   dialog,\n"
 108                 "%0,%1,...,%9 mean the 1st,2nd,...,10th space-delimited part\n"
 109                 "   of %s,\n"
 110                 "%% in the template means '%'.")),
 111
 112 #define INIT_OPT_DUMB_PREFIX(prefix, uri) \
 113         INIT_OPT_STRING("protocol.rewrite.dumb", NULL, prefix, 0, uri, NULL)
 114
 115         INIT_OPT_DUMB_PREFIX("elinks", ELINKS_WEBSITE_URL),
 116         INIT_OPT_DUMB_PREFIX("documentation", ELINKS_DOC_URL),
 117         INIT_OPT_DUMB_PREFIX("bz", ELINKS_BUGS_URL),
 118         INIT_OPT_DUMB_PREFIX("bug", ELINKS_BUGS_URL),
 119
 120         INIT_OPT_DUMB_PREFIX("arc", "http://web.archive.org/web/*/%c"),
 121         INIT_OPT_DUMB_PREFIX("cia", "http://cia.navi.cx/"),
 122         INIT_OPT_DUMB_PREFIX("b", "http://babelfish.altavista.com/babelfish/tr"),
 123         INIT_OPT_DUMB_PREFIX("d", "http://www.dict.org"),
 124         INIT_OPT_DUMB_PREFIX("g", "http://www.google.com/"),
 125         INIT_OPT_DUMB_PREFIX("gg", "http://www.google.com/"),
 126         INIT_OPT_DUMB_PREFIX("go", "http://www.google.com/"),
 127         INIT_OPT_DUMB_PREFIX("fm", "http://freshmeat.net/"),
 128         INIT_OPT_DUMB_PREFIX("sf", "http://www.sourceforge.net/"),
 129         INIT_OPT_DUMB_PREFIX("dbug", "http://bugs.debian.org/"),
 130         INIT_OPT_DUMB_PREFIX("dpkg", "http://packages.debian.org/"),
 131         /* Hm, is this Debian-centric? -- Miciah */
 132         /* Well, does anyone but Debian use lua40 naming convention? --pasky */
 133         INIT_OPT_DUMB_PREFIX("lua", "file:///usr/share/doc/lua40-doc/manual/idx.html"),
 134         INIT_OPT_DUMB_PREFIX("pycur", "http://www.python.org/doc/current/"),
 135         INIT_OPT_DUMB_PREFIX("pydev", "http://www.python.org/dev/doc/devel/"),
 136         INIT_OPT_DUMB_PREFIX("pyhelp", "http://starship.python.net/crew/theller/pyhelp.cgi"),
 137         INIT_OPT_DUMB_PREFIX("pyvault", "http://www.vex.net/parnassus/"),
 138         INIT_OPT_DUMB_PREFIX("e2", "http://www.everything2.org/"),
 139         INIT_OPT_DUMB_PREFIX("sd", "http://slashdot.org/"),
 140         INIT_OPT_DUMB_PREFIX("vhtml", "http://validator.w3.org/check?uri=%c"),
 141         INIT_OPT_DUMB_PREFIX("vcss", "http://jigsaw.w3.org/css-validator/validator?uri=%c"),
 142
 143 #define INIT_OPT_SMART_PREFIX(prefix, uri) \
 144         INIT_OPT_STRING("protocol.rewrite.smart", NULL, prefix, 0, uri, NULL)
 145 #define bugzilla_prefix(prefix) (ELINKS_BUGS_URL prefix)
 146
 147         INIT_OPT_SMART_PREFIX("bug", bugzilla_prefix("show_bug.cgi?id=%s")),
 148
 149 #ifdef CONFIG_DEBUG
 150         INIT_OPT_SMART_PREFIX("milestone-bugs", bugzilla_prefix("buglist.cgi?target_milestone=%s")),
 151         INIT_OPT_SMART_PREFIX("search-bugs", bugzilla_prefix("buglist.cgi?short_desc_type=allwordssubstr&short_desc=%s")),
 152 #endif
 153
 154         INIT_OPT_SMART_PREFIX("arc", "http://web.archive.org/web/*/%s"),
 155         INIT_OPT_SMART_PREFIX("bb", "http://babelfish.altavista.com/babelfish/tr?urltext=%s"),
 156         INIT_OPT_SMART_PREFIX("bb_fr_en", "http://babelfish.altavista.com/babelfish/tr?lp=fr_en&submit=1&urltext=%s"),
 157         INIT_OPT_SMART_PREFIX("bb_en_fr", "http://babelfish.altavista.com/babelfish/tr?lp=en_fr&submit=1&urltext=%s"),
 158         INIT_OPT_SMART_PREFIX("cambridge", "http://dictionary.cambridge.org/results.asp?searchword=%s"),
 159         INIT_OPT_SMART_PREFIX("cliki", "http://www.cliki.net/admin/search?words=%s"),
 160         INIT_OPT_SMART_PREFIX("d", "http://www.dict.org/bin/Dict?Query=%s&Form=Dict1&Strategy=*&Database=*&submit=Submit+query"),
 161         INIT_OPT_SMART_PREFIX("dmoz", "http://search.dmoz.org/cgi-bin/search?search=%s"),
 162         INIT_OPT_SMART_PREFIX("foldoc", "http://wombat.doc.ic.ac.uk/foldoc/foldoc.cgi?%s"),
 163         INIT_OPT_SMART_PREFIX("g", "http://www.google.com/search?q=%s&btnG=Google+Search"),
 164         INIT_OPT_SMART_PREFIX("gd", "http://www.google.com/search?q=%s&cat=gwd/Top"),
 165         /* Whose idea was it to use 'gg' for websearches? -- Miciah */
 166         /* INIT_OPT_SMART_PREFIX("gg", "http://groups.google.com/groups?q=%s"), */
 167         INIT_OPT_SMART_PREFIX("gg", "http://www.google.com/search?q=%s&btnG=Google+Search"),
 168         INIT_OPT_SMART_PREFIX("gi", "http://images.google.com/images?q=%s"),
 169         INIT_OPT_SMART_PREFIX("gn", "http://news.google.com/news?q=%s"),
 170         INIT_OPT_SMART_PREFIX("go", "http://www.google.com/search?q=%s&btnG=Google+Search"),
 171         INIT_OPT_SMART_PREFIX("gr", "http://groups.google.com/groups?q=%s"),
 172         INIT_OPT_SMART_PREFIX("google", "http://www.google.com/search?q=%s"),
 173         INIT_OPT_SMART_PREFIX("gwho", "http://www.googlism.com/?ism=%s&name=1"),
 174         INIT_OPT_SMART_PREFIX("gwhat", "http://www.googlism.com/?ism=%s&name=2"),
 175         INIT_OPT_SMART_PREFIX("gwhere", "http://www.googlism.com/?ism=%s&name=3"),
 176         INIT_OPT_SMART_PREFIX("gwhen", "http://www.googlism.com/?ism=%s&name=4"),
 177         INIT_OPT_SMART_PREFIX("fm", "http://freshmeat.net/search/?q=%s"),
 178         INIT_OPT_SMART_PREFIX("savannah", "http://savannah.nongnu.org/search/?words=%s&type_of_search=soft&exact=1"),
 179         INIT_OPT_SMART_PREFIX("sf", "http://sourceforge.net/search/?q=%s"),
 180         INIT_OPT_SMART_PREFIX("sfp", "http://sourceforge.net/projects/%s"),
 181         INIT_OPT_SMART_PREFIX("sd", "http://slashdot.org/search.pl?query=%s"),
 182         INIT_OPT_SMART_PREFIX("sdc", "http://slashdot.org/search.pl?query=%s&op=comments"),
 183         INIT_OPT_SMART_PREFIX("sdu", "http://slashdot.org/search.pl?query=%s&op=users"),
 184         INIT_OPT_SMART_PREFIX("sdp", "http://slashdot.org/search.pl?query=%s&op=polls"),
 185         INIT_OPT_SMART_PREFIX("sdj", "http://slashdot.org/search.pl?query=%s&op=journals"),
 186         INIT_OPT_SMART_PREFIX("dbug", "http://bugs.debian.org/%s"),
 187         INIT_OPT_SMART_PREFIX("dpkg", "http://packages.debian.org/%s"),
 188         INIT_OPT_SMART_PREFIX("emacs", "http://www.emacswiki.org/cgi-bin/wiki.pl?search=%s"),
 189         INIT_OPT_SMART_PREFIX("lyrics", "http://music.lycos.com/lyrics/results.asp?QT=L&QW=%s"),
 190         INIT_OPT_SMART_PREFIX("lxr", "http://lxr.linux.no/ident?i=%s"),
 191         INIT_OPT_SMART_PREFIX("onelook", "http://onelook.com/?w=%s&ls=a"),
 192         INIT_OPT_SMART_PREFIX("py", "http://starship.python.net/crew/theller/pyhelp.cgi?keyword=%s&version=current"),
 193         INIT_OPT_SMART_PREFIX("pydev", "http://starship.python.net/crew/theller/pyhelp.cgi?keyword=%s&version=devel"),
 194         INIT_OPT_SMART_PREFIX("pyvault", "http://py.vaults.ca/apyllo.py?find=%s"),
 195         INIT_OPT_SMART_PREFIX("e2", "http://www.everything2.org/?node=%s"),
 196         INIT_OPT_SMART_PREFIX("encz", "http://www.slovnik.cz/bin/ecd?ecd_il=1&ecd_vcb=%s&ecd_trn=translate&ecd_trn_dir=0&ecd_lines=15&ecd_hptxt=0"),
 197         INIT_OPT_SMART_PREFIX("czen", "http://www.slovnik.cz/bin/ecd?ecd_il=1&ecd_vcb=%s&ecd_trn=translate&ecd_trn_dir=1&ecd_lines=15&ecd_hptxt=0"),
 198         INIT_OPT_SMART_PREFIX("dict", "http://dictionary.reference.com/search?q=%s"),
 199         INIT_OPT_SMART_PREFIX("thes", "http://thesaurus.reference.com/search?q=%s"),
 200         INIT_OPT_SMART_PREFIX("a", "http://acronymfinder.com/af-query.asp?String=exact&Acronym=%s"),
 201         INIT_OPT_SMART_PREFIX("imdb", "http://imdb.com/Find?%s"),
 202         INIT_OPT_SMART_PREFIX("mw", "http://www.m-w.com/cgi-bin/dictionary?book=Dictionary&va=%s"),
 203         INIT_OPT_SMART_PREFIX("mwt", "http://www.m-w.com/cgi-bin/thesaurus?book=Thesaurus&va=%s"),
 204         INIT_OPT_SMART_PREFIX("whatis", "http://uptime.netcraft.com/up/graph/?host=%s"),
 205         INIT_OPT_SMART_PREFIX("wiki", "http://en.wikipedia.org/w/wiki.phtml?search=%s"),
 206         INIT_OPT_SMART_PREFIX("wn", "http://www.cogsci.princeton.edu/cgi-bin/webwn1.7.1?stage=1&word=%s"),
 207         /* Search the Free Software Directory */
 208         INIT_OPT_SMART_PREFIX("fsd", "http://directory.fsf.org/search/fsd-search.py?q=%s"),
 209         /* rfc by number */
 210         INIT_OPT_SMART_PREFIX("rfc", "http://www.rfc-editor.org/rfc/rfc%s.txt"),
 211         /* rfc search */
 212         INIT_OPT_SMART_PREFIX("rfcs", "http://www.rfc-editor.org/cgi-bin/rfcsearch.pl?searchwords=%s&format=http&abstract=abson&keywords=keyon&num=25"),
 213         INIT_OPT_SMART_PREFIX("cr", "http://www.rfc-editor.org/cgi-bin/rfcsearch.pl?searchwords=%s&format=http&abstract=abson&keywords=keyon&num=25"),
 214         /* Internet Draft search */
 215         INIT_OPT_SMART_PREFIX("rfcid", "http://www.rfc-editor.org/cgi-bin/idsearch.pl?searchwords=%s&format=http&abstract=abson&keywords=keyon&num=25"),
 216         INIT_OPT_SMART_PREFIX("id", "http://www.rfc-editor.org/cgi-bin/idsearch.pl?searchwords=%s&format=http&abstract=abson&keywords=keyon&num=25"),
 217         INIT_OPT_SMART_PREFIX("draft", "http://www.rfc-editor.org/cgi-bin/idsearch.pl?searchwords=%s&format=http&abstract=abson&keywords=keyon&num=25"),
 218
 219         NULL_OPTION_INFO,
 220 };
 221
 222 #define get_opt_rewrite(which)  uri_rewrite_options[(which)].option
 223 #define get_dumb_enable()       get_opt_rewrite(URI_REWRITE_ENABLE_DUMB).value.number
 224 #define get_smart_enable()      get_opt_rewrite(URI_REWRITE_ENABLE_SMART).value.number
 225
 226 static inline struct option *
 227 get_prefix_tree(enum uri_rewrite_option tree)
 228 {
 229         assert(tree == URI_REWRITE_DUMB_TREE
 230                || tree == URI_REWRITE_SMART_TREE);
 231         return &get_opt_rewrite(tree);
 232 }
 233
 234 #define MAX_URI_ARGS 10
 235
 236 static unsigned char *
 237 rewrite_uri(unsigned char *url, struct uri *current_uri, unsigned char *arg)
 238 {
 239         struct string n = NULL_STRING;
 240         unsigned char *args[MAX_URI_ARGS];
 241         int argslen[MAX_URI_ARGS];
 242         int argc = 0;
 243         int i;
 244
 245         if (!init_string(&n)) return NULL;
 246
 247         /* Extract space separated list of arguments */
 248         args[argc] = arg;
 249         for (i = 0; ; i++) {
 250                 if (args[argc][i] == ' ') {
 251                         argslen[argc] = i;
 252                         argc++;
 253                         if (argc == MAX_URI_ARGS) break;
 254                         args[argc] = &args[argc - 1][i];
 255                         i = 0;
 256                         for (; *args[argc] == ' '; args[argc]++);
 257                 } else if (!args[argc][i]) {
 258                         argslen[argc] = i;
 259                         argc++;
 260                         break;
 261                 }
 262         }
 263
 264         while (*url) {
 265                 int p;
 266                 int value;
 267
 268                 for (p = 0; url[p] && url[p] != '%'; p++);
 269
 270                 add_bytes_to_string(&n, url, p);
 271                 url += p;
 272
 273                 if (*url != '%') continue;
 274
 275                 url++;
 276                 switch (*url) {
 277                         case 'c':
 278                                 if (!current_uri) break;
 279                                 add_uri_to_string(&n, current_uri, URI_ORIGINAL);
 280                                 break;
 281                         case 's':
 282                                 if (arg) encode_uri_string(&n, arg, -1, 1);
 283                                 break;
 284                         case '%':
 285                                 add_char_to_string(&n, '%');
 286                                 break;
 287                         case '0':
 288                         case '1':
 289                         case '2':
 290                         case '3':
 291                         case '4':
 292                         case '5':
 293                         case '6':
 294                         case '7':
 295                         case '8':
 296                         case '9':
 297                                 value = *url - '0';
 298                                 if (value >= argc) break;
 299                                 encode_uri_string(&n, args[value],
 300                                                   argslen[value], 1);
 301                                 break;
 302                         default:
 303                                 add_bytes_to_string(&n, url - 1, 2);
 304                                 break;
 305                 }
 306                 if (*url) url++;
 307         }
 308
 309         return n.source;
 310 }
 311
 312 static unsigned char *
 313 get_uri_rewrite_prefix(enum uri_rewrite_type type, unsigned char *url)
 314 {
 315         enum uri_rewrite_option tree = type == URI_REWRITE_DUMB
 316                         ? URI_REWRITE_DUMB_TREE : URI_REWRITE_SMART_TREE;
 317         struct option *prefix_tree = get_prefix_tree(tree);
 318         struct option *opt = get_opt_rec_real(prefix_tree, url);
 319         unsigned char *exp = opt ? opt->value.string : NULL;
 320
 321         return (exp && *exp) ? exp : NULL;
 322 }
 323
 324 static enum evhook_status
 325 goto_url_hook(va_list ap, void *data)
 326 {
 327         unsigned char **url = va_arg(ap, unsigned char **);
 328         struct session *ses = va_arg(ap, struct session *);
 329         unsigned char *uu = NULL;
 330         unsigned char *arg = "";
 331         unsigned char *argstart = *url + strcspn(*url, " :");
 332
 333         if (get_smart_enable() && *argstart) {
 334                 unsigned char bucket = *argstart;
 335
 336                 *argstart = '\0';
 337                 uu = get_uri_rewrite_prefix(URI_REWRITE_SMART, *url);
 338                 *argstart = bucket;
 339                 arg = argstart + 1;
 340         }
 341
 342         if (get_dumb_enable() && !uu && !*argstart)
 343                 uu = get_uri_rewrite_prefix(URI_REWRITE_DUMB, *url);
 344
 345         if (!uu
 346             && !strchr(*url, ':')
 347             && !strchr(*url, '.')
 348             && !strchr(*url, '/')) {
 349                 uu = get_opt_str("protocol.rewrite.default_template", NULL);
 350                 if (uu && *uu) {
 351                         arg = *url;
 352                 } else {
 353                         uu = NULL;
 354                 }
 355         }
 356
 357         if (uu) {
 358                 struct uri *uri = ses && have_location(ses)
 359                                 ? cur_loc(ses)->vs.uri : NULL;
 360
 361                 uu = rewrite_uri(uu, uri, arg);
 362                 if (uu) {
 363                         mem_free(*url);
 364                         *url = uu;
 365                 }
 366         }
 367
 368         return EVENT_HOOK_STATUS_NEXT;
 369 }
 370
 371 struct event_hook_info uri_rewrite_hooks[] = {
 372         { "goto-url", -1, goto_url_hook },
 373
 374         NULL_EVENT_HOOK_INFO
 375 };
 376
 377 struct module uri_rewrite_module = struct_module(
 378         /* name: */             N_("URI rewrite"),
 379         /* options: */          uri_rewrite_options,
 380         /* hooks: */            uri_rewrite_hooks,
 381         /* submodules: */       NULL,
 382         /* data: */             NULL,
 383         /* init: */             NULL,
 384         /* done: */             NULL
 385 );