1 | // ==UserScript==
|
---|
2 | // @name HT-HTRC Mashup
|
---|
3 | // @author David Bainbridge
|
---|
4 | // @namespace org.hathitrust.researchcenter.mashup
|
---|
5 | // @description Hybrid interface between Hathitrust and HTRC
|
---|
6 | // @version 1.0
|
---|
7 | // @grant GM_xmlhttpRequest
|
---|
8 | // @require http://ajax.googleapis.com/ajax/libs/jquery/1.11.3/jquery.min.js
|
---|
9 | // @match https://www.hathitrust.org/*
|
---|
10 | // @match https://babel.hathitrust.org/cgi/ls*
|
---|
11 | // @match https://babel.hathitrust.org/cgi/mb*
|
---|
12 | // @match https://catalog.hathitrust.org/Search/Home*
|
---|
13 | // ==/UserScript==
|
---|
14 |
|
---|
15 |
|
---|
16 | var https_servlet_url="https://bedrock.resnet.cms.waikato.ac.nz/vol-checker/";
|
---|
17 |
|
---|
18 | function mashupInit() {
|
---|
19 |
|
---|
20 | var $fieldset = $('form fieldset').first();
|
---|
21 | var $search_button = $fieldset.find('button');
|
---|
22 | $search_button.css("width","47px");
|
---|
23 | $search_button.css("right","-55px");
|
---|
24 | $search_button.css("font-size","95%");
|
---|
25 | $search_button.find('span').first().hide();
|
---|
26 |
|
---|
27 | $fieldset.append('<button id="htrc-bw" class="button search" style="width:47px; right: -105px; background-image:url('+https_servlet_url + 'bookworm.png);" title="Bookworm"></button>');
|
---|
28 |
|
---|
29 | var $bw_button = $('#htrc-bw');
|
---|
30 | $bw_button.click(function(event) {
|
---|
31 | var input=$('#q1-input').val();
|
---|
32 | if (input === "") {
|
---|
33 | return;
|
---|
34 | }
|
---|
35 | event.preventDefault();
|
---|
36 |
|
---|
37 | var url_front = "https://bookworm.htrc.illinois.edu/#?%7B%22counttype%22%3A%22WordsPerMillion%22%2C%22search_limits%22%3A%5B";
|
---|
38 | var url_end = "%5D%7D";
|
---|
39 |
|
---|
40 | var word_front = "%7B%22word%22%3A%5B%22";
|
---|
41 | var word_end = "%22%5D%2C%22date_year%22%3A%7B%22%24gte%22%3A1750%2C%22%24lte%22%3A1923%7D%7D";
|
---|
42 |
|
---|
43 | var input_words = input.split(' ');
|
---|
44 | var word_len = input_words.length;
|
---|
45 |
|
---|
46 | var full_url = url_front;
|
---|
47 |
|
---|
48 | for (var i=0; i<word_len; i++) {
|
---|
49 | var word = input_words[i];
|
---|
50 | if (i>0) { full_url += "%2C"; } // append comma
|
---|
51 | full_url += word_front + word + word_end;
|
---|
52 | }
|
---|
53 |
|
---|
54 | full_url += url_end;
|
---|
55 |
|
---|
56 | // window.location.href = "https://bookworm.htrc.illinois.edu/#?%7B%22search_limits%22%3A%5B%7B%22word%22%3A%5B%22"+input+"%22%5D%2C%22date_year%22%3A%7B%22%24gte%22%3A1750%2C%22%24lte%22%3A1923%7D%7D%5D%7D";
|
---|
57 |
|
---|
58 | window.location.href = full_url;
|
---|
59 |
|
---|
60 | });
|
---|
61 |
|
---|
62 | var url = window.location.href;
|
---|
63 | var home_url_test = url.replace(/^https:\/\/www.hathitrust.org\//,"");
|
---|
64 | //console.log("*** home urltest = " + home_url_test);
|
---|
65 | if (home_url_test !== "") {
|
---|
66 | var $login_button = $('#login-button');
|
---|
67 | if ($login_button.length) {
|
---|
68 | $login_button.css('font-size','95%');
|
---|
69 | $login_button.css('right','-45px');
|
---|
70 | $login_button.css('width','80px');
|
---|
71 | }
|
---|
72 | }
|
---|
73 |
|
---|
74 |
|
---|
75 |
|
---|
76 | /*
|
---|
77 | each(function() {
|
---|
78 |
|
---|
79 | $('form button').each(function() {
|
---|
80 | console.log("*** this = " + this + "jquery this = " + $(this).val());
|
---|
81 | });
|
---|
82 | */
|
---|
83 |
|
---|
84 |
|
---|
85 | }
|
---|
86 |
|
---|
87 | function mashupAugmentResults()
|
---|
88 | {
|
---|
89 | //console.log("**** mashupAugmentResults() called");
|
---|
90 | var $results_a = $('#results_A');
|
---|
91 | var results_a_exists = $results_a.length;
|
---|
92 |
|
---|
93 | var result_class=".result";
|
---|
94 | var result_access_link_class=".result-access-link";
|
---|
95 |
|
---|
96 | if (!results_a_exists) {
|
---|
97 | // Not a main search result page
|
---|
98 | // => See if part of the collections page area
|
---|
99 | // => If it is, set things up so it will use that instead
|
---|
100 | $results_a = $('#form1');
|
---|
101 | results_a_exists = $results_a.length;
|
---|
102 | }
|
---|
103 |
|
---|
104 | if (!results_a_exists) {
|
---|
105 | // Perhaps a catalog page search?
|
---|
106 | $results_a = $('form[name="addForm"]');
|
---|
107 | results_a_exists = $results_a.length;
|
---|
108 | result_class = ".resultitem";
|
---|
109 | result_access_link_class = ".AccessLink";
|
---|
110 | }
|
---|
111 |
|
---|
112 | if (results_a_exists) {
|
---|
113 | //console.log("*** results_a = " + $results_a.html());
|
---|
114 |
|
---|
115 | var ids = [];
|
---|
116 | //var file_safe_ids = [];
|
---|
117 |
|
---|
118 | //$results_a.find('.result').each(function() {
|
---|
119 | // var $result_access_link = $(this).find('.result-access-link');
|
---|
120 |
|
---|
121 | $results_a.find(result_class).each(function() {
|
---|
122 | var $result_access_link = $(this).find(result_access_link_class);
|
---|
123 |
|
---|
124 | //console.log("*** result access link = " + $result_access_link.html());
|
---|
125 |
|
---|
126 | var $id_link = $result_access_link.find('ul>li>a').first();
|
---|
127 | //console.log("*** id_link = " + $id_link.html());
|
---|
128 | var data_clicklog = $id_link.attr("data_clicklog");
|
---|
129 | if (!data_clicklog) {
|
---|
130 | // fake one from the second <a> href (if present)
|
---|
131 | $id_link2 = $result_access_link.find('ul>li>a:eq(1)');
|
---|
132 | if ($id_link2.length>0) {
|
---|
133 | //console.log("*** id_link2 = " + $id_link2.html());
|
---|
134 | var id2_href = $id_link2.attr("href");
|
---|
135 | var id2 = id2_href.replace(/^https:\/\/hdl.handle.net\/\d+\//,"");
|
---|
136 | data_clicklog = "{\"id\":\"" + id2 + "\"}";
|
---|
137 | }
|
---|
138 | else {
|
---|
139 | data_clicklog="";
|
---|
140 | }
|
---|
141 | }
|
---|
142 | //console.log("*** data clicklog= " + data_clicklog);
|
---|
143 |
|
---|
144 |
|
---|
145 | var data_json_str = data_clicklog.replace(/^[a-z]+\|/,"");
|
---|
146 | //console.log("*** data json= " + data_json_str);
|
---|
147 |
|
---|
148 | var id = null;
|
---|
149 | if (data_json_str !== "") {
|
---|
150 | var data_json = JSON.parse(data_json_str);
|
---|
151 | id = data_json.id;
|
---|
152 | //console.log("*** (catalog extracted) id= '" + id + "'");
|
---|
153 | }
|
---|
154 | else {
|
---|
155 | var $second_id_link = $result_access_link.find('ul>li>a').eq(1);
|
---|
156 | //console.log("*** 2nd link len = " + $second_id_link);
|
---|
157 | var id_href = $second_id_link.attr("href");
|
---|
158 | //console.log("*** id href = " + id_href);
|
---|
159 | if (id_href) {
|
---|
160 | //console.log("*** id href = " + id_href);
|
---|
161 | var encoded_id = id_href.replace(/^.*id=/,"");
|
---|
162 | id = decodeURIComponent(encoded_id);
|
---|
163 | }
|
---|
164 | }
|
---|
165 | if (id) {
|
---|
166 | ids.push(id);
|
---|
167 |
|
---|
168 | //var file_safe_id = id.replace(/:/g,"+").replace(/\//g,"=");
|
---|
169 | //file_safe_ids.push(file_safe_id);
|
---|
170 |
|
---|
171 | $result_access_link.attr("id","htrc-mashup-" + id);
|
---|
172 | //$result_access_link.attr("id","htrc-mashup-" + file_safe_id);
|
---|
173 | }
|
---|
174 | });
|
---|
175 |
|
---|
176 | var ids_str = ids.join(",");
|
---|
177 | //var file_safe_ids_str = file_safe_ids.join(",");
|
---|
178 |
|
---|
179 | //console.log("*** ids= " + JSON.stringify(ids));
|
---|
180 | //console.log("*** ids_str = " + ids_str);
|
---|
181 |
|
---|
182 | console.log("**** Away to request: " + https_servlet_url + "VolumeCheck?ids=" + encodeURIComponent(ids_str));
|
---|
183 |
|
---|
184 | var check_ids_url = https_servlet_url + "VolumeCheck";
|
---|
185 | //var check_data = { "ids": encodeURIComponent(ids_str) };
|
---|
186 | var check_data = { "ids": ids_str };
|
---|
187 |
|
---|
188 | $.post(check_ids_url,check_data)
|
---|
189 | .done(function(data,textStatus,response) {
|
---|
190 | console.log("Adding in HTRC cross-checks");
|
---|
191 | var ids_exist = data;
|
---|
192 | for (var k in ids_exist) {
|
---|
193 | //console.log("*** k = '" + k + "'");
|
---|
194 |
|
---|
195 | //var id_str = '#htrc-mashup-'+k;
|
---|
196 | //console.log("*** id str = " + id_str);
|
---|
197 | // var $id_div = $(id_str);
|
---|
198 |
|
---|
199 | //var $id_div = $('#'+'htrc-mashup-'+k);
|
---|
200 | var id_div = document.getElementById('htrc-mashup-'+k);
|
---|
201 | //console.log("*** id_div = " + id_div);
|
---|
202 | var $id_div = $(id_div);
|
---|
203 |
|
---|
204 | if (ids_exist[k]) {
|
---|
205 | var encoded_id=encodeURIComponent(k);
|
---|
206 | var ef_url = https_servlet_url + "VolumeCheck?download-id=" + encoded_id;
|
---|
207 | var atag = "<a href=\""+ ef_url +"\"><span class=\"icomoon icomoon-download\"></span>Download Extracted Features</a>";
|
---|
208 |
|
---|
209 | $id_div.find("ul").append("<li title=\""+k+"\" style=\"color: #924a0b;\">"+atag+"</li>"); // â
|
---|
210 | }
|
---|
211 | else {
|
---|
212 | // â, â
|
---|
213 | $id_div.find("ul").append("<li title=\""+k+"\" style=\"color: red;\">HTRC unfriendly â</li>");
|
---|
214 | }
|
---|
215 |
|
---|
216 | //console.log("*** id div len = " + $id_div.length);
|
---|
217 |
|
---|
218 | //$.append("<li>HTRC friendly</li>");
|
---|
219 | }
|
---|
220 | })
|
---|
221 | .fail(function(esponse,textStatus,errorThrown) {
|
---|
222 | alert( "error:" + errorThrown);
|
---|
223 | });
|
---|
224 |
|
---|
225 |
|
---|
226 | //console.log("*** GM httpRequest made ");
|
---|
227 | }
|
---|
228 |
|
---|
229 | // Look for collection set download button
|
---|
230 |
|
---|
231 |
|
---|
232 | var $col_download_button = $('form button[data-tracking-action="MB Download Metadata');
|
---|
233 | if ($col_download_button.length >0) {
|
---|
234 | var $download_form=$col_download_button.parent();
|
---|
235 | $download_form.append("<button id=\"col-to-workset-download\" class=\"btn btn-mini\" style=\"margin-top: 12px;\"><i class=\"icomoon icomoon-download\"></i> Convert to HTRC Workset</button>");
|
---|
236 | var $workset_download_button = $('#col-to-workset-download');
|
---|
237 | $workset_download_button.click(function(event) {
|
---|
238 | event.preventDefault();
|
---|
239 |
|
---|
240 | var action = $download_form.attr("action");
|
---|
241 | //var action_url = action;
|
---|
242 |
|
---|
243 | var action_url = https_servlet_url + "VolumeCheck";
|
---|
244 |
|
---|
245 | // Extract hidden elems from, e.g.
|
---|
246 | // <input type="hidden" name="c" value="464226859" />
|
---|
247 | // <input type="hidden" name="a" value="download" />
|
---|
248 | // <input type="hidden" name="format" value="text" />
|
---|
249 | var $hidden_inputs = $download_form.find("input[type=\"hidden\"]");
|
---|
250 | for (var i=0; i<$hidden_inputs.length; i++) {
|
---|
251 | var hidden_input = $hidden_inputs[i];
|
---|
252 |
|
---|
253 | if (i===0) {
|
---|
254 | action_url += "?";
|
---|
255 | }
|
---|
256 | else {
|
---|
257 | action_url += "&";
|
---|
258 | }
|
---|
259 |
|
---|
260 | var $hidden_input = $(hidden_input);
|
---|
261 | var hi_name = $hidden_input.attr("name");
|
---|
262 | var hi_val = $hidden_input.attr("value");
|
---|
263 | if (hi_name === "c") {
|
---|
264 | hi_name = "convert-col";
|
---|
265 | }
|
---|
266 | action_url += hi_name + "=" + hi_val;
|
---|
267 |
|
---|
268 | }
|
---|
269 |
|
---|
270 | // <title>Collections: Ancestry and Genealogy | HathiTrust Digital Library</title>
|
---|
271 | var all_col_title = $('title').text();
|
---|
272 | var col_title = all_col_title.replace(/^Collections: /,"").replace(/ \|.*?$/,"");
|
---|
273 | action_url += "&col-title=" + encodeURIComponent(col_title);
|
---|
274 |
|
---|
275 | console.log("workset download url: " + action_url);
|
---|
276 | window.location.href = action_url;
|
---|
277 | });
|
---|
278 |
|
---|
279 | }
|
---|
280 |
|
---|
281 |
|
---|
282 | }
|
---|
283 |
|
---|
284 | mashupInit();
|
---|
285 |
|
---|
286 | mashupAugmentResults();
|
---|
287 |
|
---|