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