modules/search-engine.js

   1 /**
   2  * (C) Copyright 2008 Jeremy Maitin-Shepard
   3  *
   4  * Use, modification, and distribution are subject to the terms specified in the
   5  * COPYING file.
   6 **/
   7
   8 require("utils.js");
   9
  10 var search_engines = new string_hashmap();
  11
  12 function search_engine_parse_error(msg) {
  13     var e = new Error(msg);
  14     e.__proto__ = search_engine_parse_error.prototype;
  15     return e;
  16 }
  17 search_engine_parse_error.prototype.__proto__ = Error.prototype;
  18
  19 function search_engine() {
  20     this.urls = new string_hashmap();
  21 }
  22
  23 function search_engine_url(type, method, template) {
  24     if (!method || !type || !template)
  25         throw search_engine_parse_error("Missing method, type, or template for search engine URL");
  26     method = method.toUpperCase();
  27     type = type.toUpperCase();
  28     if (method != "GET" && method != "POST")
  29         throw search_engine_parse_error("Invalid method");
  30     var template_uri = make_uri(template);
  31     switch (template_uri.scheme) {
  32     case "http":
  33     case "https":
  34         break;
  35     default:
  36         throw search_engine_parse_error("URL template has invalid scheme.");
  37         break;
  38     }
  39     this.type = type;
  40     this.method = method;
  41     this.template = template;
  42     this.params =  [];
  43 }
  44 search_engine_url.prototype.add_param = function search_engine_url__add_param(name, value) {
  45     this.params.push({name: name, value: value});
  46 }
  47
  48 function load_search_engines_in_directory(dir) {
  49     var files = null;
  50     try {
  51         files = dir.directoryEntries.QueryInterface(Ci.nsIDirectoryEnumerator);
  52
  53         while (files.hasMoreElements()) {
  54             var file = files.nextFile;
  55
  56             if (!file.isFile())
  57                 continue;
  58
  59             try {
  60                 load_search_engine_from_file(file);
  61             } catch (e) {
  62                 dumpln("WARNING: Failed to load search engine from file: " + file.path);
  63                 dump_error(e);
  64             }
  65         }
  66     } catch (e) {
  67         // FIXME: maybe have a better error message
  68         dump_error(e);
  69     } finally {
  70         if (files)
  71             files.close();
  72     }
  73 }
  74
  75 function load_search_engine_from_file(file) {
  76     var file_istream = Cc["@mozilla.org/network/file-input-stream;1"].createInstance(Ci.nsIFileInputStream);
  77     file_istream.init(file, MODE_RDONLY, 0644, false);
  78     var dom_parser = Cc["@mozilla.org/xmlextras/domparser;1"].createInstance(Ci.nsIDOMParser);
  79     var doc = dom_parser.parseFromStream(file_istream, "UTF-8", file.fileSize, "text/xml");
  80
  81     var eng = parse_search_engine_from_dom_node(doc.documentElement);
  82     search_engines.put(file.leafName, eng);
  83 }
  84
  85 // Supported OpenSearch parameters
  86 // http://www.opensearch.org/Specifications/OpenSearch/1.1#OpenSearch_URL_template_syntax
  87 const OPENSEARCH_PARAM_USER_DEFINED    = /\{searchTerms\??\}/g;
  88 const OPENSEARCH_PARAM_INPUT_ENCODING  = /\{inputEncoding\??\}/g;
  89 const OPENSEARCH_PARAM_LANGUAGE        = /\{language\??\}/g;
  90 const OPENSEARCH_PARAM_OUTPUT_ENCODING = /\{outputEncoding\??\}/g;
  91
  92 // Default values
  93 const OPENSEARCH_PARAM_LANGUAGE_DEF         = "*";
  94 const OPENSEARCH_PARAM_OUTPUT_ENCODING_DEF  = "UTF-8";
  95 const OPENSEARCH_PARAM_INPUT_ENCODING_DEF   = "UTF-8";
  96
  97 // "Unsupported" OpenSearch parameters. For example, we don't support
  98 // page-based results, so if the engine requires that we send the "page index"
  99 // parameter, we'll always send "1".
 100 const OPENSEARCH_PARAM_COUNT        = /\{count\??\}/g;
 101 const OPENSEARCH_PARAM_START_INDEX  = /\{startIndex\??\}/g;
 102 const OPENSEARCH_PARAM_START_PAGE   = /\{startPage\??\}/g;
 103
 104 // Default values
 105 const OPENSEARCH_PARAM_COUNT_DEF        = "20"; // 20 results
 106 const OPENSEARCH_PARAM_START_INDEX_DEF  = "1";  // start at 1st result
 107 const OPENSEARCH_PARAM_START_PAGE_DEF   = "1";  // 1st page
 108
 109 // Optional parameter
 110 const OPENSEARCH_PARAM_OPTIONAL     = /\{(?:\w+:)?\w+\?\}/g;
 111
 112 // A array of arrays containing parameters that we don't fully support, and
 113 // their default values. We will only send values for these parameters if
 114 // required, since our values are just really arbitrary "guesses" that should
 115 // give us the output we want.
 116 var OPENSEARCH_UNSUPPORTED_PARAMS = [
 117   [OPENSEARCH_PARAM_COUNT, OPENSEARCH_PARAM_COUNT_DEF],
 118   [OPENSEARCH_PARAM_START_INDEX, OPENSEARCH_PARAM_START_INDEX_DEF],
 119   [OPENSEARCH_PARAM_START_PAGE, OPENSEARCH_PARAM_START_PAGE_DEF]
 120 ];
 121
 122
 123 function parse_search_engine_from_dom_node(node) {
 124     var eng = new search_engine();
 125     eng.query_charset = OPENSEARCH_PARAM_INPUT_ENCODING_DEF;
 126
 127     for each (let child in node.childNodes) {
 128         switch (child.localName) {
 129         case "ShortName":
 130             eng.name = child.textContent;
 131             break;
 132         case "Description":
 133             eng.description = child.textContent;
 134             break;
 135         case "Url":
 136             try {
 137                 let type = child.getAttribute("type");
 138                 let method = child.getAttribute("method") || "GET";
 139                 let template = child.getAttribute("template");
 140
 141                 let engine_url = new search_engine_url(type, method, template);
 142                 for each (let p in child.childNodes) {
 143                     if (p.localName == "Param") {
 144                         let name = p.getAttribute("name");
 145                         let value = p.getAttribute("value");
 146                         if (name && value)
 147                             engine_url.add_param(name, value);
 148                     }
 149                 }
 150                 eng.urls.put(type, engine_url);
 151             } catch (e) {
 152                 // Skip this element if parsing fails
 153             }
 154             break;
 155         case "InputEncoding":
 156             eng.query_charset = child.textContent.toUpperCase();
 157             break;
 158         }
 159     }
 160     return eng;
 161 }
 162
 163 search_engine.prototype.supports_response_type = function (type) {
 164     return this.urls.contains(type);
 165 };
 166
 167 /**
 168  * Returns null if the result mime_type isn't supported.  The string
 169  * search_terms will be escaped by this function.
 170  */
 171 search_engine.prototype.get_query_load_spec = function search_engine__get_query_load_spec(search_terms, type) {
 172     if (type == null)
 173         type = "text/html";
 174     var url = this.urls.get(type);
 175     if (!url)
 176         return null;
 177     search_terms = encodeURIComponent(search_terms);
 178     var eng = this;
 179
 180     function substitute(value) {
 181         // Insert the OpenSearch parameters we're confident about
 182         value = value.replace(OPENSEARCH_PARAM_USER_DEFINED, search_terms);
 183         value = value.replace(OPENSEARCH_PARAM_INPUT_ENCODING, eng.query_charset);
 184         value = value.replace(OPENSEARCH_PARAM_LANGUAGE,
 185                               get_locale() || OPENSEARCH_PARAM_LANGUAGE_DEF);
 186         value = value.replace(OPENSEARCH_PARAM_OUTPUT_ENCODING,
 187                               OPENSEARCH_PARAM_OUTPUT_ENCODING_DEF);
 188
 189         // Replace any optional parameters
 190         value = value.replace(OPENSEARCH_PARAM_OPTIONAL, "");
 191
 192         // Insert any remaining required params with our default values
 193         for (let i = 0; i < OPENSEARCH_UNSUPPORTED_PARAMS.length; ++i) {
 194             value = value.replace(OPENSEARCH_UNSUPPORTED_PARAMS[i][0],
 195                                   OPENSEARCH_UNSUPPORTED_PARAMS[i][1]);
 196         }
 197
 198         return value;
 199     }
 200
 201     var url_string = substitute(url.template);
 202
 203     var data = url.params.map(function (p) (p.name + "=" + substitute(p.value))).join("&");
 204
 205     if (url.method == "GET") {
 206         if (data.length > 0) {
 207             if (url_string.indexOf("?") == -1)
 208                 url_string += "?";
 209             else
 210                 url_string += "&";
 211             url_string += data;
 212         }
 213         return load_spec({uri: url_string});
 214     } else {
 215         return load_spec({uri: url_string, raw_post_data: data,
 216                                request_mime_type: "application/x-www-form-urlencoded"});
 217     }
 218 };
 219
 220 search_engine.prototype.__defineGetter__("completer", function () {
 221     const response_type_json = "application/x-suggestions+json";
 222     const response_type_xml = "application/x-suggestions+xml";
 223
 224     var eng = this;
 225     if (this.supports_response_type(response_type_xml)) {
 226         return function (input, pos, conservative) {
 227             if (pos == 0 && conservative)
 228                 yield co_return(undefined);
 229             let str = input.substring(0,pos);
 230             try {
 231                 let lspec = eng.get_query_load_spec(str, response_type_xml);
 232                 let result = yield send_http_request(lspec);
 233                 let doc = result.responseXML;
 234                 let data = [];
 235                 if (doc) {
 236                     let elems = doc.getElementsByTagName("CompleteSuggestion");
 237                     for (let i = 0; i < elems.length; ++i) {
 238                         let node = elems[i];
 239                         let name = node.firstChild.getAttribute("data");
 240                         let desc = node.lastChild.getAttribute("int");
 241                         if (name && desc)
 242                             data.push([name,desc]);
 243                     }
 244                     delete doc;
 245                     delete elem;
 246                     delete result;
 247                     delete lspec;
 248                     let c = { count: data.length,
 249                               get_string: function (i) data[i][0],
 250                               get_description: function (i) data[i][1] + " results",
 251                               get_input_state: function (i) [data[i][0]]
 252                             };
 253                     yield co_return(c);
 254                 }
 255             } catch (e) {
 256                 yield co_return(null);
 257             }
 258         };
 259     } else if (JSON && this.supports_response_type(response_type_json)) {
 260         return function (input, pos, conservative) {
 261             if (pos == 0 && conservative)
 262                 yield co_return(undefined);
 263             let str = input.substring(0,pos);
 264             try {
 265                 let lspec = eng.get_query_load_spec(str, response_type_json);
 266                 let result = yield send_http_request(lspec);
 267                 let data = JSON.decode(result.responseText);
 268                 delete result;
 269                 delete lspec;
 270
 271                 if (!(data instanceof Array &&
 272                       data.length >= 2 &&
 273                       typeof(data[0]) == "string" &&
 274                       data[0] == str &&
 275                       data[1] instanceof Array &&
 276                       (data[2] == null || (data[2] instanceof Array))))
 277                     yield co_return(null);
 278                 if (data[2] && data[2].length != data[1].length)
 279                     data[2] = null;
 280                 let c = { count: data[1].length,
 281                           get_string: function (i) String(data[1][i]),
 282                           get_description: (data[2] != null ? (function (i) String(data[2][i])) : null),
 283                           get_input_state: function (i) [String(data[1][i])]
 284                         };
 285                 yield co_return(c);
 286             } catch (e) {
 287                 yield co_return(null);
 288             }
 289         };
 290     } else {
 291         return null;
 292     }
 293 });
 294
 295 /**
 296  * Guess the url of a home page to correspond with the search engine.
 297  * Take the text/html url for the search engine, trim off the path and
 298  * any "search." prefix on the domain.
 299  * This works for all the provided search engines.
 300  */
 301 function search_engine_get_homepage(search_engine) {
 302     var url = search_engine.urls.get("text/html");
 303     if (!url)
 304         return null;
 305     url = url_path_trim(url.template);
 306     url = url.replace("//search.", "//");
 307     return url;
 308 }
 309
 310 // Load search engines from default directories
 311 {
 312     let dir = file_locator.get("CurProcD", Ci.nsIFile);
 313     dir.append("search-engines");
 314     if (dir.exists() && dir.isDirectory())
 315         load_search_engines_in_directory(dir);
 316
 317     dir = file_locator.get("ProfD", Ci.nsIFile);
 318     dir.append("search-engines");
 319     if (dir.exists() && dir.isDirectory())
 320         load_search_engines_in_directory(dir);
 321 }
 322
 323 define_keywords("$alternative");
 324 function define_search_engine_webjump(search_engine_name, key) {
 325     keywords(arguments);
 326     var eng = search_engines.get(search_engine_name);
 327     let alternative = arguments.$alternative;
 328
 329     if (key == null)
 330         key = search_engine_name;
 331     if (alternative == null)
 332         alternative = search_engine_get_homepage(eng);
 333
 334     define_webjump(key,
 335                    function (arg) {
 336                        return eng.get_query_load_spec(arg);
 337                    },
 338                    $alternative = alternative,
 339                    $description = eng.description,
 340                    $completer = eng.completer);
 341 }
 342
 343 define_search_engine_webjump("google.xml", "google");
 344 define_search_engine_webjump("mozilla-bugzilla.xml", "bugzilla");
 345 define_search_engine_webjump("wikipedia.xml", "wikipedia");
 346 define_search_engine_webjump("wiktionary.xml", "wiktionary");
 347 define_search_engine_webjump("answers.xml", "answers");
 348 define_search_engine_webjump("yahoo.xml", "yahoo");
 349 define_search_engine_webjump("creativecommons.xml", "creativecommons");
 350 define_search_engine_webjump("eBay.xml", "ebay");