source: gs3-extensions/html-to-expeditee/trunk/src/src/cgi-bin/html-to-expeditee.pl.in@ 24944

Last change on this file since 24944 was 24944, checked in by davidb, 12 years ago

More careful recursive traversal to classifier nodes; further developed of CSS to Expeditee attributes (background colour); and slightly less cludgy way to deal with full img URLS

File size: 9.3 KB
Line 
1#!@FULL_PERL_EXE@ -w
2
3# Need to specify the full path of Perl above, e.g. for Windows something like
4#!C:\\Perl32\\bin\\perl -w
5
6
7use strict;
8
9# Set this to 1 to work around IIS 6 craziness
10my $iis6_mode = 0;
11
12
13# IIS 6: for some reason, IIS runs this script with the working directory set to the Greenstone
14# directory rather than the cgi-bin directory, causing lots of stuff to fail
15if ($iis6_mode)
16{
17 # Change into cgi-bin directory
18 chdir("cgi-bin");
19}
20
21
22# We use require and an eval here (instead of "use") to catch any errors loading the module (for IIS)
23eval("require \"gsdlCGI.pm\"");
24if ($@)
25{
26 print STDOUT "Content-type:text/plain\n\n";
27 print STDOUT "ERROR: $@\n";
28 exit 0;
29}
30
31
32sub generate_html_form
33{
34 my ($isGSDL2,$site,$collect,$cl) = @_;
35
36 my $html_form = <<EOT;
37<html>
38 <head>
39 <title>HTML to Expeditee Frames</title>
40
41 <base href=".." />
42
43 <link type="text/css" href="ext/html-to-expeditee/jquery/css/le-frog/jquery-ui-1.8.16.custom.css" rel="stylesheet" />
44 <script type="text/javascript" src="ext/html-to-expeditee/jquery/js/jquery-1.6.2.min.js"></script>
45 <script type="text/javascript" src="ext/html-to-expeditee/jquery/js/jquery-ui-1.8.16.custom.min.js"></script>
46 <script type="text/javascript" src="ext/html-to-expeditee/js/gsajax-min.js"></script>
47 <script type="text/javascript" src="ext/html-to-expeditee/js/html-to-expeditee.js"></script>
48
49
50
51 </head>
52 <body>
53 <form class="ui-widget">
54 Convert the collection
55 <input type="text"
56 class="ui-corner-all" style="padding: 4px;"
57 name="collect" value="$collect" id="collect" />
58 to Expeditee frames by traversing the classifier
59 <input type="text"
60 class="ui-corner-all" style="padding: 4px;"
61 name="cl" value="$cl" id="cl" />
62
63 <input value="$site" id="site" type="hidden">
64 <input value="Go" id="go"
65 class="ui-button ui-widget ui-state-default ui-corner-all"
66 type="submit">
67
68 </form>
69 <script type="text/javascript">
70 var docOIDs = [];
71 var numDocOIDs;
72
73 \$(function() {
74 \$('#progress').progressbar();
75
76 \$('#go').button().click(function() {
77 var collect = document.getElementById("collect").value;
78 if (collect.match(/^\\s*\$/)) {
79 alert("No collection specified");
80 return false;
81 }
82
83 var cl = document.getElementById("cl").value;
84 if (cl.match(/^\\s*\$/)) {
85 alert("No classifier specified");
86 return false;
87 }
88 if (cl.match(/^\\d+\$/)) {
89 cl = "CL" + cl;
90 }
91
92 var gs2=$isGSDL2;
93
94 var url;
95 if (gs2) {
96 url = "library.cgi";
97 url += "?c="+collect +"&a=d&cl=" + cl;
98 }
99 else {
100 url = "dev";
101 url += "?c="+collect +"&a=b&rt=s&s=ClassifierBrowse&cl=" + cl;
102 url += "&excerptid=gs_content";
103 }
104
105
106 /* processing animation */
107
108 docOIDs = [];
109 var outstandingURLs = [];
110 var visitedURLs = {};
111
112 outstandingURLs.push(url);
113 visitedURLs[url] = 1;
114
115 while (outstandingURLs.length>0) {
116 url = outstandingURLs.shift();
117
118 var clHtml = urlGetSync(url);
119
120 var aElems;
121 var workingTrav = document.getElementById("workingTraverse");
122 workingTrav.innerHTML = clHtml;
123 aElems = workingTrav.getElementsByTagName("a");
124
125
126 /* any links with a=(b|d) ... cl=??? => outstandingURLS */
127 /* any links with a=d ... d=??? => docOIDS */
128
129 var actionRE = new RegExp("(\\\\?|&)a=(?:d|b)(&|\$)");
130 var clRE = new RegExp("(\\\\?|&)cl=" + cl + "(\\\\.\\\\d+)+(&|\$)");
131 var docRE = new RegExp("(?:\\\\?|&)d=(.*?)(?:&|\$)");
132
133 for (var i=0; i<aElems.length; i++) {
134 var aElem = aElems[i];
135 var href=aElem.href;
136 if (href && href.match(actionRE)) {
137 if (href.match(clRE)) {
138 if (!visitedURLs[href]) {
139 // console.log("found a new CL line: " + href);
140 outstandingURLs.push(href);
141 visitedURLs[href] = 1;
142 }
143 }
144 else if (href.match(docRE)) {
145
146 var docMatch = docRE.exec(href);
147 var docOID = docMatch[1];
148
149 // console.log("found a new doc line: " + docOID);
150
151 docOIDs.push(docOID);
152 }
153 }
154 }
155 }
156
157 numDocOIDs = docOIDs.length;
158
159 var iframe = document.getElementById("iframe");
160
161 var docOID = docOIDs.shift();
162 //console.log("doc oid = " + docOID);
163
164 var url;
165 if (gs2) {
166 url = "library.cgi";
167 url += "?c="+collect +"&a=d&d=" + docOID;
168 }
169 else {
170 url = "dev";
171 url += "?c="+collect +"&a=d&d=" + docOID;
172 url += "&ed=1&dt=hierarchy";
173 // url += "&excerptid=gs_content";
174 }
175
176 /* load iframe with document drawn from Greenstone collection */
177 iframe.src = url;
178
179 var progressbar = document.getElementById("progressbar");
180 progressbar.style.display = "block";
181
182
183 return false; });
184 });
185
186 function pageLoaded()
187 {
188 var iframe = document.getElementById("iframe");
189
190 if (iframe.src) {
191 if (iframe.style.display != "block") {
192 iframe.height = "90%";
193 iframe.style.display = "block";
194 }
195
196 var iframeDoc;
197 if ( iframe.contentDocument )
198 { /* FF */
199 iframeDoc = iframe.contentDocument;
200 }
201 else if ( iframe.contentWindow )
202 { /* IE */
203 iframeDoc = iframe.contentWindow.document;
204 }
205
206 var frameID = (numDocOIDs - docOIDs.length);
207 var progressPercent = frameID/numDocOIDs * 100;
208
209 \$(function() {
210 \$('#progressbar').progressbar({ value: progressPercent })
211 });
212
213 var gsContent = iframeDoc.getElementById("gs_content");
214
215 /*
216 var gsContentPos = getElementPosition(gsContent);
217 var pxl = gsContentPos.xl;
218 var pxr = gsContentPos.xr;
219 */
220
221 var expFrameTree = htmlToExpeditee(gsContent);
222
223 var expFrame = JSON.stringify(expFrameTree);
224 //alert(expFrame);
225
226 var collect = document.getElementById("collect").value;
227 var site = document.getElementById("site").value;
228
229 var url = "cgi-bin/html-to-expeditee.pl";
230 var params = "c=" + collect;
231 if (site.match(/\\w/)) {
232 params += "&site=" + site;
233 }
234 params += "&a=generate-frame&fn=" + frameID;
235 params += "&json=" + escape(expFrame);
236
237 var clHtml = urlPostSync(url,params);
238
239 if (!clHtml.match(/html-to-expeditee saved frame/)) {
240 alert("Error processing url: " + url);
241 }
242
243 if (docOIDs.length>0) {
244 var docOID = docOIDs.shift();
245
246 // console.log("doc oid = " + docOID);
247
248 // is the following line used ????
249 var cl = document.getElementById("cl").value;
250
251 var gs2 = $isGSDL2;
252 var url;
253
254 if (gs2) {
255 url = "library.cgi";
256 url += "?c="+collect +"&a=d&d=" + docOID;
257 }
258 else {
259 url = "dev";
260 url += "?c="+collect +"&a=d&d=" + docOID;
261 url += "&ed=1&dt=hierarchy";
262 // url += "&excerptid=gs_content";
263 }
264
265 iframe.src = url;
266 }
267 else {
268 var progressbar = document.getElementById("progressbar");
269 progressbar.style.display = "none";
270
271 iframe.style.display = "none";
272 delete iframe.src;
273 }
274 }
275
276 }
277
278 </script>
279
280 <div id="progressbar" width="100%"
281 style="display: none; margin: 10px; height: 10px;"></div>
282 <div id="workingTraverse" style="display: none"></div>
283
284 <hr style="margin: 10px;">
285
286 <iframe width="100%" id="iframe" style="display: none;"
287 onload="pageLoaded()"></iframe>
288
289 </body>
290</html>
291EOT
292
293 print "Content-type:text/html\n\n";
294 print $html_form;
295}
296
297sub main
298{
299
300# $ENV{'QUERY_STRING'} = "a=...";set-import-metadata&c=espresso-music&d=HASH012d6f72cde5dc48162f4a1d.1&metaname=annotation&metapos=0&metavalue=adfadfad";
301# $ENV{'REQUEST_METHOD'} = "GET";
302
303 my $gsdl_cgi = new gsdlCGI();
304
305 # Load the Greenstone modules that we need to use
306 $gsdl_cgi->setup_gsdl();
307
308 my $gsdlhome = $ENV{'GSDLHOME'};
309 $gsdl_cgi->checked_chdir($gsdlhome);
310
311 require cgiactions::HtmlToExpediteeAction;
312
313 # Useful debug statement for seeing what packages have been included
314#### printf("%-45s%-s\n",$_,$INC{$_}) foreach (sort keys %INC);
315
316 $gsdl_cgi->parse_cgi_args();
317
318 # We don't want the gsdlCGI module to return errors and warnings in XML
319 $gsdl_cgi->{'xml'} = 0;
320
321 my $fn = $gsdl_cgi->clean_param("fn"); # frame number
322
323 if (defined $fn) {
324
325 my $action = new HtmlToExpediteeAction($gsdl_cgi,$iis6_mode);
326
327 $action->do_action();
328 }
329 else {
330 # generate form, pre-filled out with any useful values such
331 # as the collection and classifier value
332
333 my $collect = $gsdl_cgi->clean_param("c");
334 my $cl = $gsdl_cgi->clean_param("cl");
335
336 # Establish collect_dir using defining 'site' along the way if GS3
337
338 my $site = undef;
339 my $isGSDL2 = undef;
340
341 if ($gsdl_cgi->greenstone_version() == 2) {
342 $isGSDL2 = 1;
343 }
344 else {
345 $isGSDL2 = 0;
346
347 # GS3 (and possible future versions) make use of 'site'
348 $site = $gsdl_cgi->clean_param("site");
349 if (!defined $site) {
350 $gsdl_cgi->generate_error("No site specified.");
351 }
352 }
353
354 generate_html_form($isGSDL2,$site,$collect,$cl);
355 }
356}
357
358
359
360&main();
Note: See TracBrowser for help on using the repository browser.