1 |
|
---|
2 | var langs_with_pos = [ "en","de","pt","da","nl", "sv" ];
|
---|
3 |
|
---|
4 | var langs_without_pos = [
|
---|
5 | "af","ar","bg","bn","cs","el","es","et","fa","fi","fr","he","hi","hr","hu",
|
---|
6 | "id","it","ja","kn","ko","lt","lv","mk","ml","mr","ne","no","pa","pl",
|
---|
7 | "ro","ru","sk","sl","so","sq","sv","sw","ta","te","th","tl","tr",
|
---|
8 | "uk","ur","vi","zh-cn","zh-tw" ];
|
---|
9 |
|
---|
10 | var num_rows = 20;
|
---|
11 |
|
---|
12 | function lang_pos_toggle(event) {
|
---|
13 | var $this =$(this);
|
---|
14 | var checked_state = $this.prop("checked");
|
---|
15 |
|
---|
16 | var id = $this.attr("id");
|
---|
17 | var split_id = id.split("-");
|
---|
18 | var related_id = split_id[0] + "-pos-choice";
|
---|
19 |
|
---|
20 | var disable_state = !checked_state;
|
---|
21 | $('#'+related_id + " *").prop('disabled',disable_state);
|
---|
22 | }
|
---|
23 |
|
---|
24 | function ajax_error(jqXHR, textStatus, errorThrown) {
|
---|
25 | alert('An error occurred... Look at the console (F12 or Ctrl+Shift+I, Console tab) for more information!');
|
---|
26 |
|
---|
27 | console.log('jqXHR:' + jqXHR);
|
---|
28 | console.log('textStatus:' + textStatus);
|
---|
29 | console.log('errorThrown:' + errorThrown);
|
---|
30 | }
|
---|
31 |
|
---|
32 |
|
---|
33 | function add_titles(json_data) {
|
---|
34 | $.each( json_data, function( htid_with_prefix, htid_val ) {
|
---|
35 | var htid = htid_with_prefix.replace(/^htid:/,"");
|
---|
36 |
|
---|
37 | $.each(htid_val.records, function( internalid, metadata ) {
|
---|
38 | var title = metadata.titles[0];
|
---|
39 | $("[name='"+htid+"']").each(function() {$(this).html(title)});
|
---|
40 | console.log(htid + ", title = " + metadata.titles[0]);
|
---|
41 | });
|
---|
42 | });
|
---|
43 | }
|
---|
44 |
|
---|
45 | function show_new_results(delta)
|
---|
46 | {
|
---|
47 | $('.search-in-progress').css("cursor","wait");
|
---|
48 |
|
---|
49 | var start = parseInt(store_search_args.start)
|
---|
50 |
|
---|
51 | store_search_args.start = start + parseInt(delta);
|
---|
52 |
|
---|
53 | $.ajax({
|
---|
54 | type: 'GET',
|
---|
55 | url: store_search_action,
|
---|
56 | data: store_search_args,
|
---|
57 | dataType: 'json',
|
---|
58 | success: show_results,
|
---|
59 | error: ajax_error
|
---|
60 | });
|
---|
61 | }
|
---|
62 |
|
---|
63 | function generate_item(line,id,id_pages)
|
---|
64 | {
|
---|
65 | var css_class = (line%2 == 0) ? 'class="evenline"' : 'class="oddline"';
|
---|
66 |
|
---|
67 | var html_item = "";
|
---|
68 |
|
---|
69 | var id_pages_len = id_pages.length;
|
---|
70 |
|
---|
71 | for (var pi=0; pi<id_pages_len; pi++) {
|
---|
72 | var page = id_pages[pi];
|
---|
73 |
|
---|
74 | var seqnum = page;
|
---|
75 | var babel_url = "https://babel.hathitrust.org/cgi/pt?id="+id+";view=1up;seq="+seqnum;
|
---|
76 |
|
---|
77 | if (id_pages_len>1) {
|
---|
78 |
|
---|
79 | if (pi==0) {
|
---|
80 | html_item += '<p ' + css_class + '>';
|
---|
81 | html_item += '<span style="font-style: italic;" name="'
|
---|
82 | +id+'"><span style="cursor: progress;">Loading ...</span></span><br>';
|
---|
83 |
|
---|
84 | html_item += id + ': <a target="_blank" href="' + babel_url + '">seq ' + seqnum + '</a> ';
|
---|
85 | }
|
---|
86 | else {
|
---|
87 | html_item += ', <a target="_blank" href="' + babel_url + '">seq ' + seqnum + '</a> ';
|
---|
88 | }
|
---|
89 | }
|
---|
90 | else {
|
---|
91 | html_item += '<p ' + css_class + '>';
|
---|
92 | html_item += ' <span style="font-style: italic;" name="'
|
---|
93 | +id+'"><span style="cursor: progress;">Loading ...</span></span><br>';
|
---|
94 | html_item += '<a target="_blank" href="' + babel_url + '">' + id + ', seq ' + seqnum + '</a>';
|
---|
95 | html_item += '</p>';
|
---|
96 | }
|
---|
97 |
|
---|
98 | }
|
---|
99 |
|
---|
100 | if (id_pages_len>1) {
|
---|
101 | html_item += "</p>";
|
---|
102 | }
|
---|
103 |
|
---|
104 | return html_item;
|
---|
105 | }
|
---|
106 |
|
---|
107 |
|
---|
108 |
|
---|
109 | function show_results(jsonData) {
|
---|
110 | var response = jsonData.response;
|
---|
111 | var num_found = response.numFound;
|
---|
112 | var docs = response.docs;
|
---|
113 | var num_docs = docs.length;
|
---|
114 |
|
---|
115 | $('.search-in-progress').css("cursor","auto");
|
---|
116 |
|
---|
117 | var $search_results = $('#search-results');
|
---|
118 |
|
---|
119 | if (num_docs>0) {
|
---|
120 | $search_results.html("<p>Results: " + num_found + " pages matched</p>");
|
---|
121 | var from = parseInt(store_search_args.start);
|
---|
122 | var to = from + num_rows;
|
---|
123 | $search_results.append("<p>Showing matches: "+from+"-" + to + "</p>");
|
---|
124 | }
|
---|
125 | else {
|
---|
126 | $search_results.html("<p>No pages matched your query</p>");
|
---|
127 | }
|
---|
128 |
|
---|
129 | // Example form of URL
|
---|
130 | // https://babel.hathitrust.org/cgi/pt?id=hvd.hnnssu;view=1up;seq=11
|
---|
131 |
|
---|
132 | var htids = [];
|
---|
133 |
|
---|
134 | var prev_id = null;
|
---|
135 | var prev_pages = [];
|
---|
136 |
|
---|
137 | var i=0;
|
---|
138 | var line_num = 0;
|
---|
139 | while (i<num_docs) {
|
---|
140 | var doc = docs[i];
|
---|
141 | var id_and_page = doc.id.split(".page-");
|
---|
142 | var id = id_and_page[0];
|
---|
143 | var seqnum = parseInt(id_and_page[1]) +1; // fix up ingest error
|
---|
144 | var page = seqnum;
|
---|
145 |
|
---|
146 | if ((prev_id != null) && (id != prev_id)) {
|
---|
147 | // time to output previous item
|
---|
148 | var html_item = generate_item(line_num,prev_id,prev_pages);
|
---|
149 | $search_results.append(html_item);
|
---|
150 | line_num++;
|
---|
151 | prev_pages = [page];
|
---|
152 | }
|
---|
153 | else {
|
---|
154 | // accumulate pages
|
---|
155 | prev_pages.push(page)
|
---|
156 | }
|
---|
157 |
|
---|
158 | htids.push("htid:"+id);
|
---|
159 |
|
---|
160 | prev_id = id;
|
---|
161 | i++;
|
---|
162 | }
|
---|
163 |
|
---|
164 | var html_item = generate_item(line_num,prev_id,prev_pages);
|
---|
165 | $search_results.append(html_item);
|
---|
166 |
|
---|
167 |
|
---|
168 | document.location.href="#search-results-anchor";
|
---|
169 |
|
---|
170 | var next_prev = '<p style="width:100%;"><div id="search-prev" style="float: left;"><a>< Previous</a></div><div id="search-next" style="float: right;"><a>Next ></a></div></p>';
|
---|
171 |
|
---|
172 | $search_results.append(next_prev);
|
---|
173 | $('#search-prev').click(function(event) { show_new_results(-1*num_rows); });
|
---|
174 | $('#search-next').click(function(event) { show_new_results(num_rows); });
|
---|
175 |
|
---|
176 | var search_start = store_search_args.start;
|
---|
177 | if (search_start==0) {
|
---|
178 | $('#search-prev').hide();
|
---|
179 | }
|
---|
180 |
|
---|
181 | // Example URL for catalog metadata (multiple items)
|
---|
182 | // http://catalog.hathitrust.org/api/volumes/brief/json/id:552;lccn:70628581|isbn:0030110408
|
---|
183 |
|
---|
184 | var htids_str = htids.join("|",htids);
|
---|
185 | var cat_url = "http://catalog.hathitrust.org/api/volumes/brief/json/" + htids_str;
|
---|
186 | $.ajax({
|
---|
187 | url: cat_url,
|
---|
188 | dataType: 'jsonp',
|
---|
189 | jsonpCallback: "add_titles"
|
---|
190 | });
|
---|
191 | }
|
---|
192 |
|
---|
193 | var store_search_args = null;
|
---|
194 | var store_search_action = null;
|
---|
195 |
|
---|
196 | var group_by_vol_checked = 0;
|
---|
197 |
|
---|
198 |
|
---|
199 | function expand_field_lang_pos(q_text,langs_with_pos,langs_without_pos,search_all_checked)
|
---|
200 | {
|
---|
201 | var fields = [];
|
---|
202 | var universal_pos_tags = [ "VERB", "NOUN", "ADJ", "ADV", "ADP", "CONJ", "DET", "NUM", "PRT", "X" ];
|
---|
203 |
|
---|
204 | for (var li=0; li<langs_with_pos.length; li++) {
|
---|
205 | var lang = langs_with_pos[li];
|
---|
206 | var lang_enabled_id = lang + "-enabled";
|
---|
207 | var $lang_enabled_cb = $('#'+lang_enabled_id);
|
---|
208 | if ($lang_enabled_cb.is(':checked')) {
|
---|
209 | console.log("Extracting POS tags for: " + lang);
|
---|
210 |
|
---|
211 | for (var ti=0; ti<universal_pos_tags.length; ti++) {
|
---|
212 | var tag = universal_pos_tags[ti];
|
---|
213 | var lang_tag_id = lang+"-"+tag+"-htrctoken-cb";
|
---|
214 | var $lang_tag_cb = $('#'+lang_tag_id);
|
---|
215 | if (search_all_checked || ($lang_tag_cb.is(':checked'))) {
|
---|
216 | var lang_tag_field = lang+"_"+tag+"_htrctoken";
|
---|
217 | fields.push(lang_tag_field+":"+q_text);
|
---|
218 | }
|
---|
219 | }
|
---|
220 | }
|
---|
221 | }
|
---|
222 |
|
---|
223 | for (var li=0; li<langs_without_pos.length; li++) {
|
---|
224 | var lang = langs_without_pos[li];
|
---|
225 | var lang_enabled_id = lang + "-enabled";
|
---|
226 | var $lang_enabled_cb = $('#'+lang_enabled_id);
|
---|
227 |
|
---|
228 | if (search_all_checked || ($lang_enabled_cb.is(':checked'))) {
|
---|
229 | console.log("Adding in non-POS field for: " + lang);
|
---|
230 | var lang_tag_field = lang+"_htrctoken";
|
---|
231 | fields.push(lang_tag_field+":"+q_text);
|
---|
232 | }
|
---|
233 | }
|
---|
234 |
|
---|
235 | var fields_str = fields.join(" OR ");
|
---|
236 |
|
---|
237 | return fields_str;
|
---|
238 | }
|
---|
239 |
|
---|
240 | function expand_query_field_and_boolean(query,langs_with_pos,langs_without_pos,search_all_checked)
|
---|
241 | {
|
---|
242 | // boolean terms
|
---|
243 | // => pos and lang field
|
---|
244 |
|
---|
245 | var query_terms = query.split(/\s+/);
|
---|
246 | var query_terms_len = query_terms.length;
|
---|
247 |
|
---|
248 | var bool_query_term = [];
|
---|
249 |
|
---|
250 | var i=0;
|
---|
251 | var prev_bool = "";
|
---|
252 |
|
---|
253 | for (var i=0; i<query_terms_len; i++) {
|
---|
254 | var term = query_terms[i];
|
---|
255 | if (term.match(/^(and|or)$/i)) {
|
---|
256 | prev_bool = term.toUpperCase();
|
---|
257 | }
|
---|
258 | else {
|
---|
259 | if (i>0) {
|
---|
260 | if (prev_bool == "") {
|
---|
261 | prev_bool = "AND";
|
---|
262 | }
|
---|
263 | }
|
---|
264 |
|
---|
265 | var expanded_term = expand_field_lang_pos(term,langs_with_pos,langs_without_pos,search_all_checked)
|
---|
266 |
|
---|
267 | term = "(" + expanded_term + ")";
|
---|
268 |
|
---|
269 | if (prev_bool != "") {
|
---|
270 | bool_query_term.push(prev_bool);
|
---|
271 | prev_bool = "";
|
---|
272 | }
|
---|
273 | bool_query_term.push(term);
|
---|
274 | }
|
---|
275 | }
|
---|
276 |
|
---|
277 | var bool_query = bool_query_term.join(" ");
|
---|
278 |
|
---|
279 | return bool_query;
|
---|
280 | }
|
---|
281 |
|
---|
282 |
|
---|
283 | function submit_action(event)
|
---|
284 | {
|
---|
285 | event.preventDefault();
|
---|
286 |
|
---|
287 | $('.search-in-progress').css("cursor","wait");
|
---|
288 |
|
---|
289 | store_search_action = $('#search-form').attr("action");
|
---|
290 |
|
---|
291 | var arg_indent = $('#indent').attr('value');
|
---|
292 | var arg_wt = $('#wt').attr('value');
|
---|
293 | var arg_start = $('#start').attr('value');
|
---|
294 | var arg_rows = $('#rows').attr('value');
|
---|
295 | var q_text = $('#q').val();
|
---|
296 |
|
---|
297 | group_by_vol_checked = $('#group-results-by-id:checked').length;
|
---|
298 | var search_all_checked = $('#search-all-langs:checked').length;
|
---|
299 |
|
---|
300 | if (q_text === "") {
|
---|
301 | $('.search-in-progress').css("cursor","auto");
|
---|
302 | alert("No query term entered");
|
---|
303 | return;
|
---|
304 | }
|
---|
305 |
|
---|
306 | arg_q = expand_query_field_and_boolean(q_text,langs_with_pos,langs_without_pos,search_all_checked);
|
---|
307 |
|
---|
308 | if (arg_q == "") {
|
---|
309 | $('.search-in-progress').css("cursor","auto");
|
---|
310 | alert("No languages selected");
|
---|
311 | return;
|
---|
312 | }
|
---|
313 |
|
---|
314 |
|
---|
315 | // Example search on one of the htrc-full-ef fields is:
|
---|
316 | // q=en_NOUN_htrctoken:farming
|
---|
317 |
|
---|
318 | store_search_args = { q: arg_q, indent: arg_indent, wt: arg_wt, start: arg_start, rows: arg_rows };
|
---|
319 |
|
---|
320 | if (group_by_vol_checked) {
|
---|
321 | store_search_args.sort="id asc";
|
---|
322 | }
|
---|
323 |
|
---|
324 |
|
---|
325 | $.ajax({
|
---|
326 | type: 'GET',
|
---|
327 | url: store_search_action,
|
---|
328 | data: store_search_args,
|
---|
329 | dataType: 'json',
|
---|
330 | success: show_results,
|
---|
331 | error: ajax_error
|
---|
332 | });
|
---|
333 |
|
---|
334 | }
|
---|
335 |
|
---|
336 | function generate_pos_langs()
|
---|
337 | {
|
---|
338 | var pos_checkbox = [
|
---|
339 | { pos:"VERB", label:"Verbs", tooltip:"Verbs (all tenses and modes)" },
|
---|
340 | { pos:"NOUN", label:"Nouns", tooltip:"Nouns (common and proper)" },
|
---|
341 | { pos:"ADJ", label:"Adjectives", tooltip: null },
|
---|
342 | { pos:"ADV", label:"Adverbs", tooltip: null },
|
---|
343 | { pos:"ADP", label:"Adpositions", tooltip:"Adpositions (prepositions and postpositions)" },
|
---|
344 | { pos:"CONJ", label:"Conjunctions",tooltip: null },
|
---|
345 | { pos:"DET", label:"Determiners", tooltip: null },
|
---|
346 | { pos:"NUM", label:"Numbers", tooltip:"Cardinal numbers" },
|
---|
347 | { pos:"PRT", label:"Particles", tooltip:"Particles or other function words" },
|
---|
348 | { pos:"X", label:"Other", tooltip:"Other words, such as foreign words, typos, abbreviations"}
|
---|
349 | ];
|
---|
350 |
|
---|
351 | var $pos_fieldsets = $('#pos-fieldsets');
|
---|
352 |
|
---|
353 | for (var li=0; li<langs_with_pos.length; li++) {
|
---|
354 |
|
---|
355 | var l = langs_with_pos[li];
|
---|
356 | var lang_full = isoLangs[l].name;
|
---|
357 | var lang_native_full = isoLangs[l].nativeName;
|
---|
358 | var opt_title = (lang_full !== lang_native_full) ? 'title="'+lang_native_full+'"' : "";
|
---|
359 |
|
---|
360 | var opt_enabled = (l == "en") ? 'checked="checked"' : "";
|
---|
361 |
|
---|
362 | var legend = "";
|
---|
363 | legend += ' <legend style="margin-bottom: 5px; padding-top: 15px;">\n';
|
---|
364 | legend += ' <input type="checkbox" name="'+l+'-enabled" id="'+l+'-enabled" ' + opt_enabled +'/>\n';
|
---|
365 | legend += ' <span ' + opt_title + '>'+lang_full+':</span>\n';
|
---|
366 | legend += ' </legend>\n';
|
---|
367 |
|
---|
368 |
|
---|
369 | var check_box_list = [];
|
---|
370 |
|
---|
371 | for (var pi=0; pi<pos_checkbox.length; pi++) {
|
---|
372 | var pos_info = pos_checkbox[pi];
|
---|
373 | var pos = pos_info.pos;
|
---|
374 | var label = pos_info.label;
|
---|
375 | var tooltip = pos_info.tooltip;
|
---|
376 | var opt_tooltip = (tooltip != null) ? 'title="'+tooltip+'"' : "";
|
---|
377 |
|
---|
378 | var check_box = "";
|
---|
379 | check_box += ' <input type="checkbox" name="'+l+'-'+pos+'-htrctoken-cb" id="'+l+'-'+pos+'-htrctoken-cb" checked="checked" />\n';
|
---|
380 | check_box += ' <label for="'+l+'-'+pos+'-htrctoken-cb" '+opt_tooltip+'>'+label+'</label>\n';
|
---|
381 |
|
---|
382 | check_box_list.push(check_box);
|
---|
383 | }
|
---|
384 |
|
---|
385 | var fieldset = "";
|
---|
386 | var opt_showhide_class = (li>0) ? 'class="show-hide-lang"' : "";
|
---|
387 |
|
---|
388 | if (li==1) {
|
---|
389 | fieldset += '<button id="show-hide-lang">Show other languages ...</button>';
|
---|
390 | }
|
---|
391 |
|
---|
392 | fieldset += '<fieldset ' + opt_showhide_class + '>\n';
|
---|
393 | fieldset += legend;
|
---|
394 | fieldset += ' <div id="'+l+'-pos-choice">\n';
|
---|
395 |
|
---|
396 | var check_box_join = check_box_list.join(' ');
|
---|
397 | fieldset += check_box_join;
|
---|
398 |
|
---|
399 | fieldset += ' </div>\n';
|
---|
400 | fieldset += '</fieldset>\n';
|
---|
401 |
|
---|
402 | $pos_fieldsets.append(fieldset);
|
---|
403 | $('#'+l+'-enabled').click(lang_pos_toggle);
|
---|
404 |
|
---|
405 | if (l == "en") {
|
---|
406 | $('#en-pos-choice *').prop('disabled',false);
|
---|
407 | }
|
---|
408 | else {
|
---|
409 | $('#'+l+'-pos-choice *').prop('disabled',true);
|
---|
410 | }
|
---|
411 | }
|
---|
412 |
|
---|
413 | show_hide_lang()
|
---|
414 | }
|
---|
415 |
|
---|
416 | function show_hide_lang()
|
---|
417 | {
|
---|
418 | $( "#show-hide-lang" ).click(function(event) {
|
---|
419 | event.preventDefault();
|
---|
420 | if($('.show-hide-lang:visible').length) {
|
---|
421 | $('.show-hide-lang').hide("slide", {direction: "up" } , 1000);
|
---|
422 | $('#show-hide-lang').html("Show other languages ...");
|
---|
423 | }
|
---|
424 | else {
|
---|
425 | $('.show-hide-lang').show("slide", {direction: "up" } , 1000);
|
---|
426 | $('#show-hide-lang').html("Hide other languages ...");
|
---|
427 | }
|
---|
428 | });
|
---|
429 | }
|
---|
430 |
|
---|
431 | function generate_other_langs()
|
---|
432 | {
|
---|
433 | // setup other languages
|
---|
434 | // for each 'langs_without_pos' generate HTML of the form:
|
---|
435 | // <input type="checkbox" name="fr-enabled" id="fr-enabled" />French
|
---|
436 | var $other_langs = $('#other-langs');
|
---|
437 |
|
---|
438 | for (var i=0; i<langs_without_pos.length; i++) {
|
---|
439 | var lang = langs_without_pos[i];
|
---|
440 | var labeled_checkbox = '<nobr>';
|
---|
441 |
|
---|
442 | labeled_checkbox += '<input type="checkbox" name="'+lang+'-enabled" id="'+lang+'-enabled" />';
|
---|
443 | /*
|
---|
444 | if (lang === "zh-cn") {
|
---|
445 | console.log("Mapping zh-cn => zh");
|
---|
446 | lang = "zh";
|
---|
447 | }
|
---|
448 | if (lang === "zh-tw") {
|
---|
449 | console.log("Mapping zh-tw => zh");
|
---|
450 | lang = "zh";
|
---|
451 | }
|
---|
452 | */
|
---|
453 | var lang_full = isoLangs[lang].name;
|
---|
454 | var lang_native_full = isoLangs[lang].nativeName;
|
---|
455 | var opt_title = (lang_full !== lang_native_full) ? 'title="'+lang_native_full+'"' : "";
|
---|
456 |
|
---|
457 | labeled_checkbox += '<label for="'+lang+'-enabled" style="padding-left: 5px; padding-right: 10px;" ' + opt_title + '>'+lang_full+'</label>';
|
---|
458 |
|
---|
459 | labeled_checkbox += '</nobr> ';
|
---|
460 |
|
---|
461 | $other_langs.append(labeled_checkbox);
|
---|
462 |
|
---|
463 | }
|
---|
464 | }
|
---|
465 |
|
---|
466 | $(function() {
|
---|
467 | generate_pos_langs();
|
---|
468 |
|
---|
469 | generate_other_langs();
|
---|
470 |
|
---|
471 | if ($('#search-submit').length>0) {
|
---|
472 | $('#search-submit').click(submit_action);
|
---|
473 | }
|
---|
474 |
|
---|
475 | });
|
---|