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 union 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.vc/"),
 122         INIT_OPT_DUMB_PREFIX("d", "http://www.dict.org"),
 123         INIT_OPT_DUMB_PREFIX("ddg", "http://duckduckgo.com/?t=elinks"),
 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         /* lua5.1 may be a more standard format.  --KON */
 134         INIT_OPT_DUMB_PREFIX("lua", "file:///usr/share/doc/lua5.1-doc/doc/contents.html#index"),
 135         INIT_OPT_DUMB_PREFIX("pycur", "http://www.python.org/doc/current/"),
 136         INIT_OPT_DUMB_PREFIX("pydev", "http://www.python.org/dev/doc/devel/"),
 137         INIT_OPT_DUMB_PREFIX("e2", "http://www.everything2.org/"),
 138         INIT_OPT_DUMB_PREFIX("sd", "http://slashdot.org/"),
 139         INIT_OPT_DUMB_PREFIX("vhtml", "http://validator.w3.org/check?uri=%c"),
 140         INIT_OPT_DUMB_PREFIX("vcss", "http://jigsaw.w3.org/css-validator/validator?uri=%c"),
 141
 142 #define INIT_OPT_SMART_PREFIX(prefix, uri) \
 143         INIT_OPT_STRING("protocol.rewrite.smart", NULL, prefix, 0, uri, NULL)
 144 #define bugzilla_prefix(prefix) (ELINKS_BUGS_URL prefix)
 145
 146         INIT_OPT_SMART_PREFIX("bug", bugzilla_prefix("show_bug.cgi?id=%s")),
 147
 148 #ifdef CONFIG_DEBUG
 149         INIT_OPT_SMART_PREFIX("milestone-bugs", bugzilla_prefix("buglist.cgi?target_milestone=%s")),
 150         INIT_OPT_SMART_PREFIX("search-bugs", bugzilla_prefix("buglist.cgi?short_desc_type=allwordssubstr&short_desc=%s")),
 151 #endif
 152
 153         INIT_OPT_SMART_PREFIX("arc", "http://web.archive.org/web/*/%s"),
 154         INIT_OPT_SMART_PREFIX("cambridge", "http://dictionary.cambridge.org/results.asp?searchword=%s"),
 155         INIT_OPT_SMART_PREFIX("cliki", "http://www.cliki.net/site/search?words=%s"),
 156         INIT_OPT_SMART_PREFIX("d", "http://www.dict.org/bin/Dict?Query=%s&Form=Dict1&Strategy=*&Database=*&submit=Submit+query"),
 157         INIT_OPT_SMART_PREFIX("ddg", "http://duckduckgo.com/?q=%s&t=elinks"),
 158         INIT_OPT_SMART_PREFIX("dmoz", "http://search.dmoz.org/cgi-bin/search?search=%s"),
 159         INIT_OPT_SMART_PREFIX("foldoc", "http://foldoc.org/?%s"),
 160         INIT_OPT_SMART_PREFIX("g", "http://www.google.com/search?q=%s&btnG=Google+Search"),
 161         /* Whose idea was it to use 'gg' for websearches? -- Miciah */
 162         /* INIT_OPT_SMART_PREFIX("gg", "http://groups.google.com/groups?q=%s"), */
 163         INIT_OPT_SMART_PREFIX("gg", "http://www.google.com/search?q=%s&btnG=Google+Search"),
 164         INIT_OPT_SMART_PREFIX("gi", "http://images.google.com/images?q=%s"),
 165         INIT_OPT_SMART_PREFIX("gn", "http://news.google.com/news?q=%s"),
 166         INIT_OPT_SMART_PREFIX("go", "http://www.google.com/search?q=%s&btnG=Google+Search"),
 167         INIT_OPT_SMART_PREFIX("gr", "http://groups.google.com/groups?q=%s"),
 168         INIT_OPT_SMART_PREFIX("google", "http://www.google.com/search?q=%s"),
 169         INIT_OPT_SMART_PREFIX("gwho", "http://www.googlism.com/search/?ism=%s&type=1"),
 170         INIT_OPT_SMART_PREFIX("gwhat", "http://www.googlism.com/search/?ism=%s&type=2"),
 171         INIT_OPT_SMART_PREFIX("gwhere", "http://www.googlism.com/search/?ism=%s&type=3"),
 172         INIT_OPT_SMART_PREFIX("gwhen", "http://www.googlism.com/search/?ism=%s&type=4"),
 173         INIT_OPT_SMART_PREFIX("fm", "http://freshmeat.net/search/?q=%s"),
 174         INIT_OPT_SMART_PREFIX("savannah", "http://savannah.nongnu.org/search/?words=%s&type_of_search=soft&exact=1"),
 175         INIT_OPT_SMART_PREFIX("sf", "http://sourceforge.net/search/?q=%s"),
 176         INIT_OPT_SMART_PREFIX("sfp", "http://sourceforge.net/projects/%s"),
 177         INIT_OPT_SMART_PREFIX("dbug", "http://bugs.debian.org/%s"),
 178         INIT_OPT_SMART_PREFIX("dpkg", "http://packages.debian.org/%s"),
 179         INIT_OPT_SMART_PREFIX("emacs", "http://www.emacswiki.org/cgi-bin/wiki.pl?search=%s"),
 180         INIT_OPT_SMART_PREFIX("lxr", "http://lxr.linux.no/ident?i=%s"),
 181         INIT_OPT_SMART_PREFIX("onelook", "http://onelook.com/?w=%s&ls=a"),
 182         INIT_OPT_SMART_PREFIX("e2", "http://www.everything2.org/?node=%s"),
 183         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"),
 184         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"),
 185         INIT_OPT_SMART_PREFIX("dict", "http://dictionary.reference.com/search?q=%s"),
 186         INIT_OPT_SMART_PREFIX("thes", "http://thesaurus.reference.com/search?q=%s"),
 187         INIT_OPT_SMART_PREFIX("a", "http://www.acronymfinder.com/af-query.asp?String=exact&Acronym=%s"),
 188         INIT_OPT_SMART_PREFIX("imdb", "http://www.imdb.com/find?q=%s"),
 189         INIT_OPT_SMART_PREFIX("mw", "http://www.m-w.com/cgi-bin/dictionary?book=Dictionary&va=%s"),
 190         INIT_OPT_SMART_PREFIX("mwt", "http://www.m-w.com/cgi-bin/thesaurus?book=Thesaurus&va=%s"),
 191         INIT_OPT_SMART_PREFIX("wiki", "http://en.wikipedia.org/w/wiki.phtml?search=%s"),
 192         INIT_OPT_SMART_PREFIX("wn", "http://wordnetweb.princeton.edu/perl/webwn?s=%s"),
 193         /* Search the Free Software Directory */
 194         INIT_OPT_SMART_PREFIX("fsd", "http://directory.fsf.org/wiki?title=Special%3ASearch&search=%s"),
 195         /* rfc by number */
 196         INIT_OPT_SMART_PREFIX("rfc", "http://www.rfc-editor.org/rfc/rfc%s.txt"),
 197         /* rfc search */
 198         INIT_OPT_SMART_PREFIX("rfcs", "http://www.rfc-editor.org/cgi-bin/rfcsearch.pl?searchwords=%s&filefmt=txt&format=http&abstract=abson&keywords=keyon&num=25"),
 199         INIT_OPT_SMART_PREFIX("cr", "http://www.rfc-editor.org/cgi-bin/rfcsearch.pl?searchwords=%s&filefmt=txt&format=http&abstract=abson&keywords=keyon&num=25"),
 200
 201         NULL_OPTION_INFO,
 202 };
 203
 204 #define get_opt_rewrite(which)  uri_rewrite_options[(which)].option
 205 #define get_dumb_enable()       get_opt_rewrite(URI_REWRITE_ENABLE_DUMB).value.number
 206 #define get_smart_enable()      get_opt_rewrite(URI_REWRITE_ENABLE_SMART).value.number
 207
 208 static inline struct option *
 209 get_prefix_tree(enum uri_rewrite_option tree)
 210 {
 211         assert(tree == URI_REWRITE_DUMB_TREE
 212                || tree == URI_REWRITE_SMART_TREE);
 213         return &get_opt_rewrite(tree);
 214 }
 215
 216 #define MAX_URI_ARGS 10
 217
 218 static unsigned char *
 219 rewrite_uri(unsigned char *url, struct uri *current_uri, unsigned char *arg)
 220 {
 221         struct string n = NULL_STRING;
 222         unsigned char *args[MAX_URI_ARGS];
 223         int argslen[MAX_URI_ARGS];
 224         int argc = 0;
 225         int i;
 226
 227         if (!init_string(&n)) return NULL;
 228
 229         /* Extract space separated list of arguments */
 230         args[argc] = arg;
 231         for (i = 0; ; i++) {
 232                 if (args[argc][i] == ' ') {
 233                         argslen[argc] = i;
 234                         argc++;
 235                         if (argc == MAX_URI_ARGS) break;
 236                         args[argc] = &args[argc - 1][i];
 237                         i = 0;
 238                         for (; *args[argc] == ' '; args[argc]++);
 239                 } else if (!args[argc][i]) {
 240                         argslen[argc] = i;
 241                         argc++;
 242                         break;
 243                 }
 244         }
 245
 246         while (*url) {
 247                 int p;
 248                 int value;
 249
 250                 for (p = 0; url[p] && url[p] != '%'; p++);
 251
 252                 add_bytes_to_string(&n, url, p);
 253                 url += p;
 254
 255                 if (*url != '%') continue;
 256
 257                 url++;
 258                 switch (*url) {
 259                         case 'c':
 260                                 if (!current_uri) break;
 261                                 add_uri_to_string(&n, current_uri, URI_ORIGINAL);
 262                                 break;
 263                         case 's':
 264                                 if (arg) encode_uri_string(&n, arg, -1, 1);
 265                                 break;
 266                         case '%':
 267                                 add_char_to_string(&n, '%');
 268                                 break;
 269                         case '0':
 270                         case '1':
 271                         case '2':
 272                         case '3':
 273                         case '4':
 274                         case '5':
 275                         case '6':
 276                         case '7':
 277                         case '8':
 278                         case '9':
 279                                 value = *url - '0';
 280                                 if (value >= argc) break;
 281                                 encode_uri_string(&n, args[value],
 282                                                   argslen[value], 1);
 283                                 break;
 284                         default:
 285                                 add_bytes_to_string(&n, url - 1, 2);
 286                                 break;
 287                 }
 288                 if (*url) url++;
 289         }
 290
 291         return n.source;
 292 }
 293
 294 static unsigned char *
 295 get_uri_rewrite_prefix(enum uri_rewrite_type type, unsigned char *url)
 296 {
 297         enum uri_rewrite_option tree = type == URI_REWRITE_DUMB
 298                         ? URI_REWRITE_DUMB_TREE : URI_REWRITE_SMART_TREE;
 299         struct option *prefix_tree = get_prefix_tree(tree);
 300         struct option *opt = get_opt_rec_real(prefix_tree, url);
 301         unsigned char *exp = opt ? opt->value.string : NULL;
 302
 303         return (exp && *exp) ? exp : NULL;
 304 }
 305
 306 static enum evhook_status
 307 goto_url_hook(va_list ap, void *data)
 308 {
 309         unsigned char **url = va_arg(ap, unsigned char **);
 310         struct session *ses = va_arg(ap, struct session *);
 311         unsigned char *uu = NULL;
 312         unsigned char *arg = "";
 313         unsigned char *argstart = *url + strcspn(*url, " :");
 314
 315         if (get_smart_enable() && *argstart) {
 316                 unsigned char bucket = *argstart;
 317
 318                 *argstart = '\0';
 319                 uu = get_uri_rewrite_prefix(URI_REWRITE_SMART, *url);
 320                 *argstart = bucket;
 321                 arg = argstart + 1;
 322         }
 323
 324         if (get_dumb_enable() && !uu && !*argstart)
 325                 uu = get_uri_rewrite_prefix(URI_REWRITE_DUMB, *url);
 326
 327         if (!uu
 328             && !strchr(*url, ':')
 329             && !strchr(*url, '.')
 330             && !strchr(*url, '/')) {
 331                 uu = get_opt_str("protocol.rewrite.default_template", NULL);
 332                 if (uu && *uu) {
 333                         arg = *url;
 334                 } else {
 335                         uu = NULL;
 336                 }
 337         }
 338
 339         if (uu) {
 340                 struct uri *uri = ses && have_location(ses)
 341                                 ? cur_loc(ses)->vs.uri : NULL;
 342
 343                 uu = rewrite_uri(uu, uri, arg);
 344                 if (uu) {
 345                         mem_free(*url);
 346                         *url = uu;
 347                 }
 348         }
 349
 350         return EVENT_HOOK_STATUS_NEXT;
 351 }
 352
 353 struct event_hook_info uri_rewrite_hooks[] = {
 354         { "goto-url", -1, goto_url_hook },
 355
 356         NULL_EVENT_HOOK_INFO
 357 };
 358
 359 struct module uri_rewrite_module = struct_module(
 360         /* name: */             N_("URI rewrite"),
 361         /* options: */          uri_rewrite_options,
 362         /* hooks: */            uri_rewrite_hooks,
 363         /* submodules: */       NULL,
 364         /* data: */             NULL,
 365         /* init: */             NULL,
 366         /* done: */             NULL
 367 );