1 | /**********************************************************************
|
---|
2 | *
|
---|
3 | * formattools.cpp --
|
---|
4 | * Copyright (C) 1999 The New Zealand Digital Library Project
|
---|
5 | *
|
---|
6 | * A component of the Greenstone digital library software
|
---|
7 | * from the New Zealand Digital Library Project at the
|
---|
8 | * University of Waikato, New Zealand.
|
---|
9 | *
|
---|
10 | * This program is free software; you can redistribute it and/or modify
|
---|
11 | * it under the terms of the GNU General Public License as published by
|
---|
12 | * the Free Software Foundation; either version 2 of the License, or
|
---|
13 | * (at your option) any later version.
|
---|
14 | *
|
---|
15 | * This program is distributed in the hope that it will be useful,
|
---|
16 | * but WITHOUT ANY WARRANTY; without even the implied warranty of
|
---|
17 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
---|
18 | * GNU General Public License for more details.
|
---|
19 | *
|
---|
20 | * You should have received a copy of the GNU General Public License
|
---|
21 | * along with this program; if not, write to the Free Software
|
---|
22 | * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
|
---|
23 | *
|
---|
24 | *********************************************************************/
|
---|
25 |
|
---|
26 | #include "formattools.h"
|
---|
27 | #include "cgiutils.h"
|
---|
28 | #include "gsdltools.h"
|
---|
29 | #include "recptprototools.h"
|
---|
30 | #include "OIDtools.h"
|
---|
31 | #include "summarise.h"
|
---|
32 |
|
---|
33 | #include <assert.h>
|
---|
34 |
|
---|
35 | static bool metadata_wrap = false;
|
---|
36 | static text_t metadata_wrap_type = "";
|
---|
37 |
|
---|
38 | // a few function prototypes
|
---|
39 |
|
---|
40 | static text_t format_string (const text_t& collection, recptproto* collectproto,
|
---|
41 | ResultDocInfo_t &docinfo, displayclass &disp,
|
---|
42 | format_t *formatlistptr, text_tmap &options,
|
---|
43 | ostream& logout);
|
---|
44 |
|
---|
45 | static bool parse_action (text_t::const_iterator &here, const text_t::const_iterator &end,
|
---|
46 | format_t *formatlistptr, text_tset &metadata, bool &getParents);
|
---|
47 |
|
---|
48 | static text_t format_summary (const text_t& collection, recptproto* collectproto,
|
---|
49 | ResultDocInfo_t &docinfo, displayclass &disp,
|
---|
50 | text_tmap &options, ostream& logout);
|
---|
51 |
|
---|
52 | static text_t format_text (const text_t& collection, recptproto* collectproto,
|
---|
53 | ResultDocInfo_t &docinfo, displayclass &disp,
|
---|
54 | text_tmap &options, ostream& logout);
|
---|
55 |
|
---|
56 | static text_t expand_metadata(const text_t &metavalue, const text_t& collection,
|
---|
57 | recptproto* collectproto, ResultDocInfo_t &docinfo,
|
---|
58 | displayclass &disp, text_tmap &options,
|
---|
59 | ostream &logout);
|
---|
60 |
|
---|
61 | static text_t transform_to_GS3_format (format_t *formatlistptr, const text_t& nodeType);
|
---|
62 |
|
---|
63 | void metadata_t::clear() {
|
---|
64 | metaname.clear();
|
---|
65 | metacommand = mNone;
|
---|
66 | mqualifier.parent = pNone;
|
---|
67 | mqualifier.sibling = sNone;
|
---|
68 | mqualifier.child = cNone;
|
---|
69 | pre_tree_traverse.clear();
|
---|
70 | parentoptions.clear();
|
---|
71 | siblingoptions.clear();
|
---|
72 | childoptions.clear();
|
---|
73 | }
|
---|
74 |
|
---|
75 | void decision_t::clear() {
|
---|
76 | command = dMeta;
|
---|
77 | meta.clear();
|
---|
78 | text.clear();
|
---|
79 | }
|
---|
80 |
|
---|
81 | format_t::~format_t()
|
---|
82 | {
|
---|
83 | if (nextptr != NULL) delete nextptr;
|
---|
84 | if (ifptr != NULL) delete ifptr;
|
---|
85 | if (elseptr != NULL) delete elseptr;
|
---|
86 | if (orptr != NULL) delete orptr;
|
---|
87 | }
|
---|
88 |
|
---|
89 | void format_t::clear() {
|
---|
90 | command = comText;
|
---|
91 | decision.clear();
|
---|
92 | text.clear();
|
---|
93 | meta.clear();
|
---|
94 | nextptr = NULL;
|
---|
95 | ifptr = NULL;
|
---|
96 | elseptr = NULL;
|
---|
97 | orptr = NULL;
|
---|
98 | }
|
---|
99 |
|
---|
100 | void formatinfo_t::clear() {
|
---|
101 | DocumentImages = false;
|
---|
102 | DocumentTitles = true;
|
---|
103 | DocumentHeading = "{Or}{[parent(Top):Title],[Title],untitled}<br>";
|
---|
104 | DocumentContents = true;
|
---|
105 | DocumentArrowsBottom = true;
|
---|
106 | DocumentArrowsTop = false;
|
---|
107 | DocumentSearchResultLinks = false;
|
---|
108 | DocumentButtons.erase (DocumentButtons.begin(), DocumentButtons.end());
|
---|
109 | // DocumentButtons.push_back ("Expand Text");
|
---|
110 | // DocumentButtons.push_back ("Expand Contents");
|
---|
111 | DocumentButtons.push_back ("Detach");
|
---|
112 | DocumentButtons.push_back ("Highlight");
|
---|
113 | RelatedDocuments = "";
|
---|
114 | DocumentText = "[Text]";
|
---|
115 | formatstrings.erase (formatstrings.begin(), formatstrings.end());
|
---|
116 | DocumentUseHTML = false;
|
---|
117 | AllowExtendedOptions = false;
|
---|
118 | }
|
---|
119 |
|
---|
120 | // simply checks to see if formatstring begins with a <td> tag
|
---|
121 | bool is_table_content (const text_t &formatstring) {
|
---|
122 | text_t::const_iterator here = formatstring.begin();
|
---|
123 | text_t::const_iterator end = formatstring.end();
|
---|
124 |
|
---|
125 | while (here != end) {
|
---|
126 | if (*here != ' ') {
|
---|
127 | if ((*here == '<') && ((here+3) < end)) {
|
---|
128 | if ((*(here+1) == 't' || *(here+1) == 'T') &&
|
---|
129 | (*(here+2) == 'd' || *(here+2) == 'D') &&
|
---|
130 | (*(here+3) == '>' || *(here+3) == ' '))
|
---|
131 | //|| *(here+3) == '\t' || *(here+3) == '\n'))
|
---|
132 | return true;
|
---|
133 | } else return false;
|
---|
134 | }
|
---|
135 | ++here;
|
---|
136 | }
|
---|
137 | return false;
|
---|
138 | }
|
---|
139 |
|
---|
140 | bool is_table_content (const format_t *formatlistptr) {
|
---|
141 |
|
---|
142 | if (formatlistptr == NULL) return false;
|
---|
143 |
|
---|
144 | if (formatlistptr->command == comText)
|
---|
145 | return is_table_content (formatlistptr->text);
|
---|
146 |
|
---|
147 | return false;
|
---|
148 | }
|
---|
149 |
|
---|
150 | // returns false if key isn't in formatstringmap
|
---|
151 | bool get_formatstring (const text_t &key, const text_tmap &formatstringmap,
|
---|
152 | text_t &formatstring) {
|
---|
153 |
|
---|
154 | formatstring.clear();
|
---|
155 | text_tmap::const_iterator it = formatstringmap.find(key);
|
---|
156 | if (it == formatstringmap.end()) return false;
|
---|
157 | formatstring = (*it).second;
|
---|
158 | return true;
|
---|
159 | }
|
---|
160 |
|
---|
161 | // tries to find "key1key2" then "key1" then "key2"
|
---|
162 | bool get_formatstring (const text_t &key1, const text_t &key2,
|
---|
163 | const text_tmap &formatstringmap,
|
---|
164 | text_t &formatstring) {
|
---|
165 |
|
---|
166 | formatstring.clear();
|
---|
167 | text_tmap::const_iterator it = formatstringmap.find(key1 + key2);
|
---|
168 | if (it != formatstringmap.end()) {
|
---|
169 | formatstring = (*it).second;
|
---|
170 | return true;
|
---|
171 | }
|
---|
172 | it = formatstringmap.find(key1);
|
---|
173 | if (it != formatstringmap.end()) {
|
---|
174 | formatstring = (*it).second;
|
---|
175 | return true;
|
---|
176 | }
|
---|
177 | it = formatstringmap.find(key2);
|
---|
178 | if (it != formatstringmap.end()) {
|
---|
179 | formatstring = (*it).second;
|
---|
180 | return true;
|
---|
181 | }
|
---|
182 | return false;
|
---|
183 | }
|
---|
184 |
|
---|
185 |
|
---|
186 | text_t remove_namespace(const text_t &meta_name) {
|
---|
187 | text_t::const_iterator end = meta_name.end();
|
---|
188 | text_t::const_iterator it = findchar(meta_name.begin(), end, '.');
|
---|
189 | if (it != end) {
|
---|
190 | return substr(it+1, end);
|
---|
191 | }
|
---|
192 |
|
---|
193 | return meta_name;
|
---|
194 |
|
---|
195 | }
|
---|
196 | // returns a date of form _format:date_(year, month, day)
|
---|
197 | // input is date of type yyyy-?mm-?dd
|
---|
198 | // at least the year must be present in date
|
---|
199 | text_t format_date (const text_t &date) {
|
---|
200 |
|
---|
201 | if (date.size() < 4) return "";
|
---|
202 |
|
---|
203 | text_t::const_iterator datebegin = date.begin();
|
---|
204 |
|
---|
205 | text_t year = substr (datebegin, datebegin+4);
|
---|
206 | int chars_seen_so_far = 4;
|
---|
207 | if (chars_seen_so_far == date.size()) return "_format:date_("+year+")";
|
---|
208 |
|
---|
209 | if (date[chars_seen_so_far] == '-') ++chars_seen_so_far ;
|
---|
210 | if (date.size() < chars_seen_so_far+2) return "_format:date_("+year+")";
|
---|
211 |
|
---|
212 | text_t month = substr (datebegin+chars_seen_so_far, datebegin+chars_seen_so_far+2);
|
---|
213 | int imonth = month.getint();
|
---|
214 | if (imonth <= 0 || imonth > 12) return "_format:date_("+year+")";
|
---|
215 |
|
---|
216 | chars_seen_so_far += 2;
|
---|
217 | if (chars_seen_so_far == date.size()) return "_format:date_("+year+","+month+")";
|
---|
218 |
|
---|
219 | if (date[chars_seen_so_far] == '-') ++chars_seen_so_far ;
|
---|
220 | if (date.size() < chars_seen_so_far+2) return "_format:date_("+year+","+month+")";
|
---|
221 |
|
---|
222 | text_t day = substr (datebegin+chars_seen_so_far, datebegin+chars_seen_so_far+2);
|
---|
223 | if (day[0] == '0') day = substr (day.begin()+1, day.end());
|
---|
224 | int iday = day.getint();
|
---|
225 | if (iday <= 0 || iday > 31) return "_format:date_("+year+","+month+")";
|
---|
226 |
|
---|
227 | return "_format:date_("+year+","+month+","+day+")";
|
---|
228 | }
|
---|
229 |
|
---|
230 | // converts an iso639 language code to its English equivalent
|
---|
231 | // should we be checking that the macro exists??
|
---|
232 | text_t iso639 (const text_t &langcode) {
|
---|
233 | if (langcode.empty()) return "";
|
---|
234 | return "_iso639:iso639"+langcode+"_";
|
---|
235 | }
|
---|
236 |
|
---|
237 |
|
---|
238 | text_t get_href (const text_t &link) {
|
---|
239 |
|
---|
240 | text_t href;
|
---|
241 |
|
---|
242 | text_t::const_iterator here = findchar(link.begin(), link.end(), '"');
|
---|
243 | text_t::const_iterator end = link.end();
|
---|
244 | if (here == end) return g_EmptyText;
|
---|
245 |
|
---|
246 | ++here;
|
---|
247 | while (here != end) {
|
---|
248 | if (*here == '"') break;
|
---|
249 | href.push_back(*here);
|
---|
250 | ++here;
|
---|
251 | }
|
---|
252 |
|
---|
253 | return href;
|
---|
254 | }
|
---|
255 |
|
---|
256 | //this function gets the information associated with the relation
|
---|
257 | //metadata for the document associated with 'docinfo'. This relation
|
---|
258 | //metadata consists of a line of pairs containing 'collection, document OID'
|
---|
259 | //(this is the OID of the document related to the current document, and
|
---|
260 | //the collection the related document belongs to). For each of these pairs
|
---|
261 | //the title metadata is obtained and then an html link between the title
|
---|
262 | //of the related doc and the document's position (the document will be
|
---|
263 | //found in "<a href=\"_httpdocument_&c=collection&cl=search&d=OID">
|
---|
264 | //(where collection is the related documents collection, and OID is the
|
---|
265 | //related documents OID). A list of these html links are made for as many
|
---|
266 | //related documents as there are. This list is then returned. If there are
|
---|
267 | //no related documents available for the current document then the string
|
---|
268 | //'.. no related documents .. ' is returned.
|
---|
269 | text_t get_related_docs(const text_t& collection, recptproto* collectproto,
|
---|
270 | ResultDocInfo_t &docinfo, ostream& logout){
|
---|
271 |
|
---|
272 | text_tset metadata;
|
---|
273 |
|
---|
274 | //insert the metadata we wish to collect
|
---|
275 | metadata.insert("dc.Relation");
|
---|
276 | metadata.insert("Title");
|
---|
277 | metadata.insert("Subject"); //for emails, where title data doesn't apply
|
---|
278 |
|
---|
279 | FilterResponse_t response;
|
---|
280 | text_t relation = ""; //string for displaying relation metadata
|
---|
281 | text_t relationTitle = ""; //the related documents Title (or subject)
|
---|
282 | text_t relationOID = ""; //the related documents OID
|
---|
283 |
|
---|
284 | //get the information associated with the metadata for current doc
|
---|
285 | if (get_info (docinfo.OID, collection, "", metadata,
|
---|
286 | false, collectproto, response, logout)) {
|
---|
287 |
|
---|
288 | //if the relation metadata exists, store for displaying
|
---|
289 | if(!response.docInfo[0].metadata["dc.Relation"].values.empty()){
|
---|
290 | relationOID += response.docInfo[0].metadata["dc.Relation"].values[0];
|
---|
291 |
|
---|
292 | //split relation data into pairs of collectionname,ID number
|
---|
293 | text_tarray relationpairs;
|
---|
294 | splitchar (relationOID.begin(), relationOID.end(), ' ', relationpairs);
|
---|
295 |
|
---|
296 | text_tarray::const_iterator currDoc = relationpairs.begin();
|
---|
297 | text_tarray::const_iterator lastDoc = relationpairs.end();
|
---|
298 |
|
---|
299 | //iterate through the pairs to split and display
|
---|
300 | while(currDoc != lastDoc){
|
---|
301 |
|
---|
302 | //split pairs into collectionname and ID
|
---|
303 | text_tarray relationdata;
|
---|
304 | splitchar ((*currDoc).begin(), (*currDoc).end(), ',', relationdata);
|
---|
305 |
|
---|
306 | //get first element in the array (collection)
|
---|
307 | text_tarray::const_iterator doc_data = relationdata.begin();
|
---|
308 | text_t document_collection = *doc_data;
|
---|
309 | ++doc_data; //increment to get next item in array (oid)
|
---|
310 | text_t document_OID = *doc_data;
|
---|
311 |
|
---|
312 | //create html link to related document
|
---|
313 | relation += "<a href=\"_httpdocument_&c=" + document_collection;
|
---|
314 | relation += "&cl=search&d=" + document_OID;
|
---|
315 |
|
---|
316 | //get the information associated with the metadata for related doc
|
---|
317 | if (get_info (document_OID, document_collection, "", metadata,
|
---|
318 | false, collectproto, response, logout)) {
|
---|
319 |
|
---|
320 | //if title metadata doesn't exist, collect subject metadata
|
---|
321 | //if that doesn't exist, just call it 'related document'
|
---|
322 | if (!response.docInfo[0].metadata["Title"].values[0].empty())
|
---|
323 | relationTitle = response.docInfo[0].metadata["Title"].values[0];
|
---|
324 | else if (!response.docInfo[0].metadata["Subject"].values.empty())
|
---|
325 | relationTitle = response.docInfo[0].metadata["Subject"].values[0];
|
---|
326 | else relationTitle = "RELATED DOCUMENT";
|
---|
327 |
|
---|
328 | }
|
---|
329 |
|
---|
330 | //link the related document's title to its page
|
---|
331 | relation += "\">" + relationTitle + "</a>";
|
---|
332 | relation += " (" + document_collection + ")<br>";
|
---|
333 |
|
---|
334 | ++currDoc;
|
---|
335 | }
|
---|
336 | }
|
---|
337 |
|
---|
338 | }
|
---|
339 |
|
---|
340 | if(relation.empty()) //no relation data for documnet
|
---|
341 | relation = ".. no related documents .. ";
|
---|
342 |
|
---|
343 | return relation;
|
---|
344 | }
|
---|
345 |
|
---|
346 |
|
---|
347 |
|
---|
348 | static void get_parent_options (text_t &instring, metadata_t &metaoption) {
|
---|
349 |
|
---|
350 | assert (instring.size() > 7);
|
---|
351 | if (instring.size() <= 7) return;
|
---|
352 |
|
---|
353 | text_t meta, com, op;
|
---|
354 | bool inbraces = false;
|
---|
355 | bool inquotes = false;
|
---|
356 | bool foundcolon = false;
|
---|
357 | text_t::const_iterator here = instring.begin()+6;
|
---|
358 | text_t::const_iterator end = instring.end();
|
---|
359 | while (here != end) {
|
---|
360 | if (foundcolon) meta.push_back (*here);
|
---|
361 | else if (*here == '(') inbraces = true;
|
---|
362 | else if (*here == ')') inbraces = false;
|
---|
363 | else if (*here == '\'' && !inquotes) inquotes = true;
|
---|
364 | else if (*here == '\'' && inquotes) inquotes = false;
|
---|
365 | else if (*here == ':' && !inbraces) foundcolon = true;
|
---|
366 | else if (inquotes) op.push_back (*here);
|
---|
367 | else com.push_back (*here);
|
---|
368 | ++here;
|
---|
369 | }
|
---|
370 |
|
---|
371 | instring = meta;
|
---|
372 | if (com.empty())
|
---|
373 | metaoption.mqualifier.parent = pImmediate;
|
---|
374 | else if (com == "Top")
|
---|
375 | metaoption.mqualifier.parent = pTop;
|
---|
376 | else if (com == "All") {
|
---|
377 | metaoption.mqualifier.parent = pAll;
|
---|
378 | metaoption.parentoptions = op;
|
---|
379 | }
|
---|
380 | }
|
---|
381 |
|
---|
382 |
|
---|
383 | static void get_sibling_options (text_t &instring, metadata_t &metaoption) {
|
---|
384 |
|
---|
385 | assert (instring.size() > 8);
|
---|
386 | if (instring.size() <= 8) return;
|
---|
387 | text_t meta, com, op;
|
---|
388 | bool inbraces = false;
|
---|
389 | bool inquotes = false;
|
---|
390 | bool foundcolon = false;
|
---|
391 | text_t::const_iterator here = instring.begin()+7;
|
---|
392 | text_t::const_iterator end = instring.end();
|
---|
393 | while (here != end) {
|
---|
394 | if (foundcolon) meta.push_back (*here);
|
---|
395 | else if (*here == '(') inbraces = true;
|
---|
396 | else if (*here == ')') inbraces = false;
|
---|
397 | else if (*here == '\'' && !inquotes) inquotes = true;
|
---|
398 | else if (*here == '\'' && inquotes) inquotes = false;
|
---|
399 | else if (*here == ':' && !inbraces) foundcolon = true;
|
---|
400 | else if (inquotes) op.push_back (*here);
|
---|
401 | else com.push_back (*here);
|
---|
402 | ++here;
|
---|
403 | }
|
---|
404 |
|
---|
405 | instring = meta;
|
---|
406 | metaoption.siblingoptions.clear();
|
---|
407 |
|
---|
408 | if (com.empty()) {
|
---|
409 | metaoption.mqualifier.sibling = sAll;
|
---|
410 | metaoption.siblingoptions = " ";
|
---|
411 | }
|
---|
412 | else if (com == "first") {
|
---|
413 | metaoption.mqualifier.sibling = sNum;
|
---|
414 | metaoption.siblingoptions = "0";
|
---|
415 | }
|
---|
416 | else if (com == "last") {
|
---|
417 | metaoption.mqualifier.sibling = sNum;
|
---|
418 | metaoption.siblingoptions = "-2"; // == last
|
---|
419 | }
|
---|
420 | else if (com.getint()>0) {
|
---|
421 | metaoption.mqualifier.sibling = sNum;
|
---|
422 | int pos = com.getint()-1;
|
---|
423 | metaoption.siblingoptions +=pos;
|
---|
424 | }
|
---|
425 | else {
|
---|
426 | metaoption.mqualifier.sibling = sAll;
|
---|
427 | metaoption.siblingoptions = op;
|
---|
428 | }
|
---|
429 | }
|
---|
430 |
|
---|
431 | static void get_child_options (text_t &instring, metadata_t &metaoption) {
|
---|
432 |
|
---|
433 | assert (instring.size() > 6);
|
---|
434 | if (instring.size() <= 6) return;
|
---|
435 | text_t meta, com, op;
|
---|
436 | bool inbraces = false;
|
---|
437 | bool inquotes = false;
|
---|
438 | bool foundcolon = false;
|
---|
439 | text_t::const_iterator here = instring.begin()+5;
|
---|
440 | text_t::const_iterator end = instring.end();
|
---|
441 | while (here != end) {
|
---|
442 | if (foundcolon) meta.push_back (*here);
|
---|
443 | else if (*here == '(') inbraces = true;
|
---|
444 | else if (*here == ')') inbraces = false;
|
---|
445 | else if (*here == '\'' && !inquotes) inquotes = true;
|
---|
446 | else if (*here == '\'' && inquotes) inquotes = false;
|
---|
447 | else if (*here == ':' && !inbraces) foundcolon = true;
|
---|
448 | else if (inquotes) op.push_back (*here);
|
---|
449 | else com.push_back (*here);
|
---|
450 | ++here;
|
---|
451 | }
|
---|
452 |
|
---|
453 | instring = meta;
|
---|
454 | if (com.empty()) {
|
---|
455 | metaoption.mqualifier.child = cAll;
|
---|
456 | metaoption.childoptions = " ";
|
---|
457 | }
|
---|
458 | else if (com == "first") {
|
---|
459 | metaoption.mqualifier.child = cNum;
|
---|
460 | metaoption.childoptions = ".fc";
|
---|
461 | }
|
---|
462 | else if (com == "last") {
|
---|
463 | metaoption.mqualifier.child = cNum;
|
---|
464 | metaoption.childoptions = ".lc";
|
---|
465 | }
|
---|
466 | else if (com.getint()>0) {
|
---|
467 | metaoption.mqualifier.child = cNum;
|
---|
468 | metaoption.childoptions = "."+com;
|
---|
469 | }
|
---|
470 | else {
|
---|
471 | metaoption.mqualifier.child = cAll;
|
---|
472 | metaoption.childoptions = op;
|
---|
473 | }
|
---|
474 | }
|
---|
475 |
|
---|
476 |
|
---|
477 | static void get_truncate_options (text_t &instring, metadata_t &metaoption)
|
---|
478 | {
|
---|
479 | assert (instring.size() > ((text_t) "truncate").size());
|
---|
480 | if (instring.size() <= ((text_t) "truncate").size()) return;
|
---|
481 | text_t meta, com;
|
---|
482 | bool inbraces = false;
|
---|
483 | bool foundcolon = false;
|
---|
484 | text_t::const_iterator here = instring.begin() + ((text_t) "truncate").size();
|
---|
485 | text_t::const_iterator end = instring.end();
|
---|
486 | while (here != end) {
|
---|
487 | if (foundcolon) meta.push_back (*here);
|
---|
488 | else if (*here == '(') inbraces = true;
|
---|
489 | else if (*here == ')') inbraces = false;
|
---|
490 | else if (*here == ':' && !inbraces) foundcolon = true;
|
---|
491 | else com.push_back (*here);
|
---|
492 | ++here;
|
---|
493 | }
|
---|
494 |
|
---|
495 | instring = meta;
|
---|
496 |
|
---|
497 | if (!com.empty())
|
---|
498 | {
|
---|
499 | metaoption.siblingoptions = com;
|
---|
500 | }
|
---|
501 | else
|
---|
502 | {
|
---|
503 | // Default is 100 characters if not specified
|
---|
504 | metaoption.siblingoptions = "100";
|
---|
505 | }
|
---|
506 | }
|
---|
507 |
|
---|
508 |
|
---|
509 |
|
---|
510 | static void parse_meta (text_t &meta, metadata_t &metaoption,
|
---|
511 | text_tset &metadata, bool &getParents) {
|
---|
512 |
|
---|
513 | // Look for the various format statement modifiers
|
---|
514 | // This needs to be done in a loop otherwise not all combinations will be accepted, but actually the order
|
---|
515 | // is irrelevant because this is not stored in metaoption.metacommand anyway
|
---|
516 | bool keep_trying = true;
|
---|
517 | while (keep_trying)
|
---|
518 | {
|
---|
519 | keep_trying = false;
|
---|
520 |
|
---|
521 | if (meta.size() > ((text_t) "cgisafe:").size() && starts_with(meta, "cgisafe:"))
|
---|
522 | {
|
---|
523 | metaoption.metacommand |= mCgiSafe;
|
---|
524 | meta = substr(meta.begin() + ((text_t) "cgisafe:").size(), meta.end());
|
---|
525 | keep_trying = true;
|
---|
526 | }
|
---|
527 | if (meta.size() > ((text_t) "format:").size() && starts_with(meta, "format:"))
|
---|
528 | {
|
---|
529 | metaoption.metacommand |= mSpecial;
|
---|
530 | meta = substr(meta.begin() + ((text_t) "format:").size(), meta.end());
|
---|
531 | keep_trying = true;
|
---|
532 | }
|
---|
533 |
|
---|
534 | // New "truncate" special formatting option
|
---|
535 | if (meta.size() > ((text_t) "truncate").size() && starts_with(meta, "truncate")) // No colons due to truncate(X)
|
---|
536 | {
|
---|
537 | metaoption.metacommand |= mTruncate;
|
---|
538 | get_truncate_options (meta, metaoption);
|
---|
539 | keep_trying = true;
|
---|
540 | }
|
---|
541 | // New "htmlsafe" special formatting option
|
---|
542 | if (meta.size() > ((text_t) "htmlsafe:").size() && starts_with(meta, "htmlsafe:"))
|
---|
543 | {
|
---|
544 | metaoption.metacommand |= mHTMLSafe;
|
---|
545 | meta = substr(meta.begin() + ((text_t) "htmlsafe:").size(), meta.end());
|
---|
546 | keep_trying = true;
|
---|
547 | }
|
---|
548 | // New "xmlsafe" special formatting option
|
---|
549 | if (meta.size() > ((text_t) "xmlsafe:").size() && starts_with(meta, "xmlsafe:"))
|
---|
550 | {
|
---|
551 | metaoption.metacommand |= mXMLSafe;
|
---|
552 | meta = substr(meta.begin() + ((text_t) "xmlsafe:").size(), meta.end());
|
---|
553 | keep_trying = true;
|
---|
554 | }
|
---|
555 | // New "dmsafe" special formatting option
|
---|
556 | if (meta.size() > ((text_t) "dmsafe:").size() && starts_with(meta, "dmsafe:"))
|
---|
557 | {
|
---|
558 | metaoption.metacommand |= mDMSafe;
|
---|
559 | meta = substr(meta.begin() + ((text_t) "dmsafe:").size(), meta.end());
|
---|
560 | keep_trying = true;
|
---|
561 | }
|
---|
562 | }
|
---|
563 |
|
---|
564 | bool had_parent_or_child = true;
|
---|
565 | bool prev_was_parent = false;
|
---|
566 | bool prev_was_child = false;
|
---|
567 |
|
---|
568 | while (had_parent_or_child) {
|
---|
569 | if (meta.size() > 7
|
---|
570 | && (substr (meta.begin(), meta.begin()+6) == "parent")) {
|
---|
571 |
|
---|
572 | // clear out sibling and child (cmd and options)
|
---|
573 | metaoption.metacommand &= ~(mChild|mSibling);
|
---|
574 | metaoption.childoptions.clear();
|
---|
575 | metaoption.siblingoptions.clear();
|
---|
576 |
|
---|
577 | getParents = true;
|
---|
578 | metaoption.metacommand |= mParent;
|
---|
579 | get_parent_options (meta, metaoption);
|
---|
580 |
|
---|
581 | if (prev_was_parent) {
|
---|
582 | metaoption.pre_tree_traverse += ".pr";
|
---|
583 | }
|
---|
584 | else if (prev_was_child) {
|
---|
585 | metaoption.pre_tree_traverse += ".fc";
|
---|
586 | }
|
---|
587 |
|
---|
588 | prev_was_parent = true;
|
---|
589 | prev_was_child = false;
|
---|
590 | }
|
---|
591 | else if (meta.size() > 6
|
---|
592 | && (substr (meta.begin(), meta.begin()+5) == "child")) {
|
---|
593 |
|
---|
594 | // clear out sibling and parent (cmd and options)
|
---|
595 | metaoption.metacommand &= ~(mParent|mSibling);
|
---|
596 | metaoption.parentoptions.clear();
|
---|
597 | metaoption.siblingoptions.clear();
|
---|
598 |
|
---|
599 | metaoption.metacommand |= mChild;
|
---|
600 | get_child_options (meta, metaoption);
|
---|
601 | metadata.insert("contains");
|
---|
602 |
|
---|
603 | if (prev_was_parent) {
|
---|
604 | metaoption.pre_tree_traverse += ".pr";
|
---|
605 | }
|
---|
606 | else if (prev_was_child) {
|
---|
607 | metaoption.pre_tree_traverse += ".fc";
|
---|
608 | }
|
---|
609 |
|
---|
610 | prev_was_child = true;
|
---|
611 | prev_was_parent = false;
|
---|
612 | }
|
---|
613 | else {
|
---|
614 | prev_was_child = false;
|
---|
615 | prev_was_parent = false;
|
---|
616 | had_parent_or_child = false;
|
---|
617 | }
|
---|
618 | }
|
---|
619 |
|
---|
620 | // parent/child can have sibling tacked on end also
|
---|
621 | if (meta.size() > 8 && (substr (meta.begin(), meta.begin()+7) == "sibling")) {
|
---|
622 | metaoption.metacommand |= mSibling;
|
---|
623 | get_sibling_options (meta, metaoption);
|
---|
624 | }
|
---|
625 |
|
---|
626 | // check for ex. which may occur in format statements
|
---|
627 | // remove "ex." prefix, but only if there are no other metadata set qualifiers
|
---|
628 | // in the metaname, since we want to retain prefixes like "ex.dc." as-is
|
---|
629 | text_t::iterator period = findchar(meta.begin(), meta.end(), '.');
|
---|
630 | text_t::iterator lastperiod = findlastchar(meta.begin(), meta.end(), '.');
|
---|
631 |
|
---|
632 | if (meta.size()>3 && (substr(meta.begin(), meta.begin()+3) == "ex.") && period == lastperiod) {
|
---|
633 | meta = substr (meta.begin()+3, meta.end());
|
---|
634 | }
|
---|
635 | metadata.insert (meta);
|
---|
636 | metaoption.metaname = meta;
|
---|
637 | }
|
---|
638 |
|
---|
639 | static void parse_coll_meta(text_t &meta, metadata_t &metaoption) {
|
---|
640 | if (meta == "collection") {
|
---|
641 | // no qualifiers
|
---|
642 | metaoption.metaname = g_EmptyText;
|
---|
643 | return;
|
---|
644 | }
|
---|
645 | meta = substr (meta.begin()+11, meta.end());
|
---|
646 | metaoption.metaname = meta;
|
---|
647 |
|
---|
648 | }
|
---|
649 |
|
---|
650 | static void parse_meta (text_t &meta, format_t *formatlistptr,
|
---|
651 | text_tset &metadata, bool &getParents) {
|
---|
652 |
|
---|
653 | // check for ex. which may occur in format statements
|
---|
654 | // remove "ex." prefix, but only if there are no other metadata set qualifiers
|
---|
655 | // in the metaname, since we want to retain prefixes like "ex.dc." as-is
|
---|
656 | text_t::iterator period = findchar(meta.begin(), meta.end(), '.');
|
---|
657 | text_t::iterator lastperiod = findlastchar(meta.begin(), meta.end(), '.');
|
---|
658 |
|
---|
659 | if (meta.size()>3 && (substr(meta.begin(), meta.begin()+3) == "ex.") && period == lastperiod) {
|
---|
660 | meta = substr (meta.begin()+3, meta.end());
|
---|
661 | }
|
---|
662 | if (meta == "link")
|
---|
663 | formatlistptr->command = comLink;
|
---|
664 | else if (meta == "/link")
|
---|
665 | formatlistptr->command = comEndLink;
|
---|
666 |
|
---|
667 | // the metaname "srclink_file" is deprecated, use "srclinkFile"
|
---|
668 | else if (meta == "srclink") {
|
---|
669 | formatlistptr->command = comAssocLink;
|
---|
670 | formatlistptr->meta.metaname = "srclinkFile";
|
---|
671 | metadata.insert("srclinkFile");
|
---|
672 | }
|
---|
673 | else if (meta == "srchref") {
|
---|
674 | formatlistptr->command = comAssocLink;
|
---|
675 | formatlistptr->text = "href";
|
---|
676 | formatlistptr->meta.metaname = "srclinkFile";
|
---|
677 | metadata.insert("srclinkFile");
|
---|
678 | }
|
---|
679 | else if (meta == "/srclink") {
|
---|
680 | formatlistptr->command = comEndAssocLink;
|
---|
681 | formatlistptr->meta.metaname = "srclinkFile";
|
---|
682 | }
|
---|
683 | // and weblink etc
|
---|
684 | else if (meta == "href")
|
---|
685 | formatlistptr->command = comHref;
|
---|
686 |
|
---|
687 | else if (meta == "num")
|
---|
688 | formatlistptr->command = comNum;
|
---|
689 |
|
---|
690 | else if (meta == "icon")
|
---|
691 | formatlistptr->command = comIcon;
|
---|
692 |
|
---|
693 | else if (meta == "Text")
|
---|
694 | formatlistptr->command = comDoc;
|
---|
695 |
|
---|
696 | else if (meta == "RelatedDocuments")
|
---|
697 | formatlistptr->command = comRel;
|
---|
698 |
|
---|
699 | else if (meta == "highlight")
|
---|
700 | formatlistptr->command = comHighlight;
|
---|
701 |
|
---|
702 | else if (meta == "/highlight")
|
---|
703 | formatlistptr->command = comEndHighlight;
|
---|
704 |
|
---|
705 | else if (meta == "metadata-spanwrap")
|
---|
706 | formatlistptr->command = comMetadataSpanWrap;
|
---|
707 |
|
---|
708 | else if (meta == "/metadata-spanwrap")
|
---|
709 | formatlistptr->command = comEndMetadataSpanWrap;
|
---|
710 |
|
---|
711 | else if (meta == "metadata-divwrap")
|
---|
712 | formatlistptr->command = comMetadataDivWrap;
|
---|
713 |
|
---|
714 | else if (meta == "/metadata-divwrap")
|
---|
715 | formatlistptr->command = comEndMetadataDivWrap;
|
---|
716 |
|
---|
717 | else if (meta == "Summary")
|
---|
718 | formatlistptr->command = comSummary;
|
---|
719 |
|
---|
720 | else if (meta == "DocImage")
|
---|
721 | formatlistptr->command = comImage;
|
---|
722 |
|
---|
723 | else if (meta == "DocTOC")
|
---|
724 | formatlistptr->command = comTOC;
|
---|
725 |
|
---|
726 | else if (meta == "DocumentButtonDetach")
|
---|
727 | formatlistptr->command = comDocumentButtonDetach;
|
---|
728 |
|
---|
729 | else if (meta == "DocumentButtonHighlight")
|
---|
730 | formatlistptr->command = comDocumentButtonHighlight;
|
---|
731 |
|
---|
732 | else if (meta == "DocumentButtonExpandContents")
|
---|
733 | formatlistptr->command = comDocumentButtonExpandContents;
|
---|
734 |
|
---|
735 | else if (meta == "DocumentButtonExpandText")
|
---|
736 | formatlistptr->command = comDocumentButtonExpandText;
|
---|
737 |
|
---|
738 | else if (meta == "DocOID")
|
---|
739 | formatlistptr->command = comOID;
|
---|
740 | else if (meta == "DocTopOID")
|
---|
741 | formatlistptr->command = comTopOID;
|
---|
742 | else if (meta == "DocRank")
|
---|
743 | formatlistptr->command = comRank;
|
---|
744 | else if (meta == "DocTermsFreqTotal")
|
---|
745 | formatlistptr->command = comDocTermsFreqTotal;
|
---|
746 | else if (meta.size() >= 10 && (substr(meta.begin(), meta.begin()+10) == "collection")) {
|
---|
747 | formatlistptr->command = comCollection;
|
---|
748 | parse_coll_meta(meta, formatlistptr->meta);
|
---|
749 | }
|
---|
750 | else {
|
---|
751 | formatlistptr->command = comMeta;
|
---|
752 | parse_meta (meta, formatlistptr->meta, metadata, getParents);
|
---|
753 | }
|
---|
754 | }
|
---|
755 |
|
---|
756 |
|
---|
757 | static bool parse_string (const text_t &formatstring, format_t *formatlistptr,
|
---|
758 | text_tset &metadata, bool &getParents) {
|
---|
759 |
|
---|
760 | text_t text;
|
---|
761 | text_t::const_iterator here = formatstring.begin();
|
---|
762 | text_t::const_iterator end = formatstring.end();
|
---|
763 |
|
---|
764 | while (here != end) {
|
---|
765 |
|
---|
766 | if (*here == '\\') {
|
---|
767 | ++here;
|
---|
768 | if (here != end) text.push_back (*here);
|
---|
769 |
|
---|
770 | } else if (*here == '{') {
|
---|
771 | if (!text.empty()) {
|
---|
772 | formatlistptr->command = comText;
|
---|
773 | formatlistptr->text = text;
|
---|
774 | formatlistptr->nextptr = new format_t();
|
---|
775 | formatlistptr = formatlistptr->nextptr;
|
---|
776 |
|
---|
777 | text.clear();
|
---|
778 | }
|
---|
779 | if (parse_action (++here, end, formatlistptr, metadata, getParents)) {
|
---|
780 |
|
---|
781 | formatlistptr->nextptr = new format_t();
|
---|
782 | formatlistptr = formatlistptr->nextptr;
|
---|
783 | if (here == end) break;
|
---|
784 | }
|
---|
785 | } else if (*here == '[') {
|
---|
786 | if (!text.empty()) {
|
---|
787 | formatlistptr->command = comText;
|
---|
788 | formatlistptr->text = text;
|
---|
789 | formatlistptr->nextptr = new format_t();
|
---|
790 | formatlistptr = formatlistptr->nextptr;
|
---|
791 |
|
---|
792 | text.clear();
|
---|
793 | }
|
---|
794 | text_t meta;
|
---|
795 | ++here;
|
---|
796 | while (*here != ']') {
|
---|
797 | if (here == end) return false;
|
---|
798 | meta.push_back (*here);
|
---|
799 | ++here;
|
---|
800 | }
|
---|
801 | parse_meta (meta, formatlistptr, metadata, getParents);
|
---|
802 | formatlistptr->nextptr = new format_t();
|
---|
803 | formatlistptr = formatlistptr->nextptr;
|
---|
804 |
|
---|
805 | } else
|
---|
806 | text.push_back (*here);
|
---|
807 |
|
---|
808 | if (here != end) ++here;
|
---|
809 | }
|
---|
810 | if (!text.empty()) {
|
---|
811 | formatlistptr->command = comText;
|
---|
812 | formatlistptr->text = text;
|
---|
813 | formatlistptr->nextptr = new format_t();
|
---|
814 | formatlistptr = formatlistptr->nextptr;
|
---|
815 |
|
---|
816 | }
|
---|
817 | return true;
|
---|
818 | }
|
---|
819 |
|
---|
820 |
|
---|
821 | static bool parse_action (text_t::const_iterator &here, const text_t::const_iterator &end,
|
---|
822 | format_t *formatlistptr, text_tset &metadata, bool &getParents) {
|
---|
823 |
|
---|
824 | text_t::const_iterator it = findchar (here, end, '}');
|
---|
825 | if (it == end) return false;
|
---|
826 |
|
---|
827 | text_t com = substr (here, it);
|
---|
828 | here = findchar (it, end, '{');
|
---|
829 | if (here == end) return false;
|
---|
830 | else ++here;
|
---|
831 |
|
---|
832 | if (com == "If" || com == "if" || com == "IF") formatlistptr->command = comIf;
|
---|
833 | else if (com == "Or" || com == "or" || com == "OR") formatlistptr->command = comOr;
|
---|
834 | else return false;
|
---|
835 |
|
---|
836 | int commacount = 0;
|
---|
837 | text_t text;
|
---|
838 | while (here != end) {
|
---|
839 |
|
---|
840 | if (*here == '\\') {
|
---|
841 | ++here;
|
---|
842 | if (here != end) text.push_back(*here);
|
---|
843 |
|
---|
844 | }
|
---|
845 |
|
---|
846 | else if (*here == ',' || *here == '}' || *here == '{') {
|
---|
847 |
|
---|
848 | if (formatlistptr->command == comOr) {
|
---|
849 | // the {Or}{this, or this, or this, or this} statement
|
---|
850 | format_t *or_ptr;
|
---|
851 |
|
---|
852 | // find the next unused orptr
|
---|
853 | if (formatlistptr->orptr == NULL) {
|
---|
854 | formatlistptr->orptr = new format_t();
|
---|
855 | or_ptr = formatlistptr->orptr;
|
---|
856 | } else {
|
---|
857 | or_ptr = formatlistptr->orptr;
|
---|
858 | while (or_ptr->nextptr != NULL)
|
---|
859 | or_ptr = or_ptr->nextptr;
|
---|
860 | or_ptr->nextptr = new format_t();
|
---|
861 | or_ptr = or_ptr->nextptr;
|
---|
862 | }
|
---|
863 |
|
---|
864 | if (!text.empty())
|
---|
865 | {
|
---|
866 | if (!parse_string(text, or_ptr, metadata, getParents)) { return false; }
|
---|
867 | }
|
---|
868 |
|
---|
869 | if (*here == '{')
|
---|
870 | {
|
---|
871 | // Supports: {Or}{[Booktitle],[Title],{If}{[XXXX],aaa,bbb}}
|
---|
872 | // but not : {Or}{[Booktitle],[Title]{If}{[XXXX],aaa,bbb}}
|
---|
873 | // The latter can always be re-written:
|
---|
874 | // {Or}{[Booktitle],{If}{[Title],[Title]{If}{[XXXX],aaa,bbb}}}
|
---|
875 |
|
---|
876 | if (!text.empty()) // already used up allocated format_t
|
---|
877 | {
|
---|
878 | // => allocate new one for detected action
|
---|
879 | or_ptr->nextptr = new format_t();
|
---|
880 | or_ptr = or_ptr->nextptr;
|
---|
881 | }
|
---|
882 | if (!parse_action(++here, end, or_ptr, metadata, getParents))
|
---|
883 | {
|
---|
884 | return false;
|
---|
885 | }
|
---|
886 | }
|
---|
887 | else
|
---|
888 | {
|
---|
889 | if (*here == '}') break;
|
---|
890 | }
|
---|
891 | text.clear();
|
---|
892 |
|
---|
893 | }
|
---|
894 |
|
---|
895 | // Parse an {If}{decide,do,else} statement
|
---|
896 | else {
|
---|
897 |
|
---|
898 | // Read the decision component.
|
---|
899 | if (commacount == 0) {
|
---|
900 | // Decsion can be a metadata element, or a piece of text.
|
---|
901 | // Originally Stefan's code, updated 25/10/2000 by Gordon.
|
---|
902 |
|
---|
903 | text_t::const_iterator beginbracket = text.begin();
|
---|
904 | text_t::const_iterator endbracket = (text.end() - 1);
|
---|
905 |
|
---|
906 | // Decision is based on a metadata element
|
---|
907 | if ((*beginbracket == '[') && (*endbracket == ']')) {
|
---|
908 | // Ignore the surrounding square brackets
|
---|
909 | text_t meta = substr (beginbracket+1, endbracket);
|
---|
910 | parse_meta (meta, formatlistptr->decision.meta, metadata, getParents);
|
---|
911 | ++commacount;
|
---|
912 | text.clear();
|
---|
913 | }
|
---|
914 |
|
---|
915 | // Decision is a piece of text (probably a macro like _cgiargmode_).
|
---|
916 | else {
|
---|
917 |
|
---|
918 | // hunt for any metadata in string, which might be uses in
|
---|
919 | // to test a condition, e.g. [Format] eq 'PDF'
|
---|
920 | format_t* dummyformat = new format_t();
|
---|
921 | // update which metadata fields needed
|
---|
922 | // (not interested in updatng formatlistptr)
|
---|
923 | parse_string (text, dummyformat, metadata, getParents);
|
---|
924 | delete dummyformat;
|
---|
925 |
|
---|
926 | formatlistptr->decision.command = dText;
|
---|
927 | formatlistptr->decision.text = text;
|
---|
928 | ++commacount;
|
---|
929 | text.clear();
|
---|
930 | }
|
---|
931 | }
|
---|
932 |
|
---|
933 | // Read the "then" and "else" components of the {If} statement.
|
---|
934 | else {
|
---|
935 | format_t** nextlistptr = NULL;
|
---|
936 | if (commacount == 1) {
|
---|
937 | nextlistptr = &formatlistptr->ifptr;
|
---|
938 | } else if (commacount == 2 ) {
|
---|
939 | nextlistptr = &formatlistptr->elseptr;
|
---|
940 | } else {
|
---|
941 | return false;
|
---|
942 | }
|
---|
943 |
|
---|
944 | if (!text.empty()) {
|
---|
945 | if (*nextlistptr == NULL) {
|
---|
946 | *nextlistptr = new format_t();
|
---|
947 | } else {
|
---|
948 |
|
---|
949 | // skip to the end of any format_t statements already added
|
---|
950 | while ((*nextlistptr)->nextptr != NULL)
|
---|
951 | {
|
---|
952 | nextlistptr = &(*nextlistptr)->nextptr;
|
---|
953 | }
|
---|
954 |
|
---|
955 | (*nextlistptr)->nextptr = new format_t();
|
---|
956 | nextlistptr = &(*nextlistptr)->nextptr;
|
---|
957 | }
|
---|
958 |
|
---|
959 | if (!parse_string (text, *nextlistptr, metadata, getParents))
|
---|
960 | {
|
---|
961 | return false;
|
---|
962 | }
|
---|
963 | text.clear();
|
---|
964 | }
|
---|
965 |
|
---|
966 | if (*here == '{')
|
---|
967 | {
|
---|
968 | if (*nextlistptr == NULL) {
|
---|
969 | *nextlistptr = new format_t();
|
---|
970 | } else {
|
---|
971 | // skip to the end of any format_t statements already added
|
---|
972 | while ((*nextlistptr)->nextptr != NULL)
|
---|
973 | {
|
---|
974 | nextlistptr = &(*nextlistptr)->nextptr;
|
---|
975 | }
|
---|
976 |
|
---|
977 | (*nextlistptr)->nextptr = new format_t();
|
---|
978 | nextlistptr = &(*nextlistptr)->nextptr;
|
---|
979 | }
|
---|
980 |
|
---|
981 | if (!parse_action(++here, end, *nextlistptr, metadata, getParents))
|
---|
982 | {
|
---|
983 | return false;
|
---|
984 | }
|
---|
985 | }
|
---|
986 | else
|
---|
987 | {
|
---|
988 | if (*here == '}') break;
|
---|
989 | ++commacount;
|
---|
990 | }
|
---|
991 | }
|
---|
992 | }
|
---|
993 |
|
---|
994 | } else text.push_back(*here);
|
---|
995 |
|
---|
996 | if (here != end) ++here;
|
---|
997 | }
|
---|
998 |
|
---|
999 | return true;
|
---|
1000 | }
|
---|
1001 |
|
---|
1002 |
|
---|
1003 | static text_t wrap_metatext(const text_t& metatext, const text_t& OID,
|
---|
1004 | const text_t metaname, int metapos=-1)
|
---|
1005 | {
|
---|
1006 |
|
---|
1007 | text_t tag_type = metadata_wrap_type;
|
---|
1008 | text_t editable_type = (metaname == "Text") ? "text" : "metadata";
|
---|
1009 |
|
---|
1010 | text_t wrapped_metatext = "<" + tag_type + " ";
|
---|
1011 | wrapped_metatext += "class=\"editable-" + editable_type + "\" ";
|
---|
1012 |
|
---|
1013 | wrapped_metatext += "docoid=\"" + OID + "\" ";
|
---|
1014 | wrapped_metatext += "metaname=\"" + metaname + "\"";
|
---|
1015 |
|
---|
1016 | if (metapos>=0) {
|
---|
1017 | text_t metapos_str = metapos;
|
---|
1018 | wrapped_metatext += " metapos=\"" + metapos_str + "\"";
|
---|
1019 | }
|
---|
1020 |
|
---|
1021 | wrapped_metatext += ">" + metatext + "</" + tag_type + ">";
|
---|
1022 |
|
---|
1023 | return wrapped_metatext;
|
---|
1024 | }
|
---|
1025 |
|
---|
1026 |
|
---|
1027 |
|
---|
1028 | bool parse_formatstring (const text_t &formatstring, format_t *formatlistptr,
|
---|
1029 | text_tset &metadata, bool &getParents) {
|
---|
1030 |
|
---|
1031 | formatlistptr->clear();
|
---|
1032 | getParents = false;
|
---|
1033 |
|
---|
1034 | return (parse_string (formatstring, formatlistptr, metadata, getParents));
|
---|
1035 | }
|
---|
1036 |
|
---|
1037 | // position -1 for all, -2 for the last, 0 for the first, or x for a particular piece
|
---|
1038 | // metainfo has all the values for the metadata, meta has the request metadata name and options, position tells which values to get
|
---|
1039 |
|
---|
1040 | static text_t get_formatted_meta_text(const text_t& OID, MetadataInfo_t &metainfo, const metadata_t &meta, int position, bool no_cgisafe = false)
|
---|
1041 | {
|
---|
1042 | text_t no_ns_metaname = remove_namespace(meta.metaname);
|
---|
1043 | text_t formatted_metatext;
|
---|
1044 | bool first = true;
|
---|
1045 |
|
---|
1046 | const int start_i=0;
|
---|
1047 | const int end_i = metainfo.values.size()-1;
|
---|
1048 |
|
---|
1049 | if (position == -1) { // all
|
---|
1050 | for (int i=start_i; i<=end_i; ++i) {
|
---|
1051 | if (!first) formatted_metatext += meta.siblingoptions;
|
---|
1052 |
|
---|
1053 | text_t fresh_metatext;
|
---|
1054 |
|
---|
1055 | if (meta.metacommand & mSpecial) {
|
---|
1056 | // special formatting
|
---|
1057 | if (no_ns_metaname == "Date") fresh_metatext = format_date (metainfo.values[i]);
|
---|
1058 | else if (no_ns_metaname == "Language") fresh_metatext = iso639(metainfo.values[i]);
|
---|
1059 | else fresh_metatext = "_format:"+meta.metaname+"_("+metainfo.values[i]+")";
|
---|
1060 | }
|
---|
1061 | else fresh_metatext = metainfo.values[i];
|
---|
1062 |
|
---|
1063 | // New "truncate" special formatting option
|
---|
1064 | if (meta.metacommand & mTruncate)
|
---|
1065 | {
|
---|
1066 | int truncate_length = meta.siblingoptions.getint();
|
---|
1067 | text_t truncated_value = fresh_metatext;
|
---|
1068 | if (truncated_value.size() > truncate_length)
|
---|
1069 | {
|
---|
1070 | truncated_value = substr(truncated_value.begin(), truncated_value.begin() + truncate_length) + "... _texttruncated_";
|
---|
1071 | }
|
---|
1072 | fresh_metatext = truncated_value;
|
---|
1073 | }
|
---|
1074 | // New "xmlsafe" special formatting option
|
---|
1075 | if (meta.metacommand & mXMLSafe)
|
---|
1076 | {
|
---|
1077 | // Make it XML-safe
|
---|
1078 | text_t text_xml_safe = "";
|
---|
1079 | text_t::const_iterator text_iterator = fresh_metatext.begin();
|
---|
1080 | while (text_iterator != fresh_metatext.end())
|
---|
1081 | {
|
---|
1082 | if (*text_iterator == '&') text_xml_safe += "&";
|
---|
1083 | else if (*text_iterator == '<') text_xml_safe += "<";
|
---|
1084 | else if (*text_iterator == '>') text_xml_safe += ">";
|
---|
1085 | else text_xml_safe.push_back(*text_iterator);
|
---|
1086 | text_iterator++;
|
---|
1087 | }
|
---|
1088 | fresh_metatext = text_xml_safe;
|
---|
1089 | }
|
---|
1090 | // New "htmlsafe" special formatting option
|
---|
1091 | if (meta.metacommand & mHTMLSafe)
|
---|
1092 | {
|
---|
1093 | // Make it HTML-safe
|
---|
1094 | text_t text_html_safe = "";
|
---|
1095 | text_t::const_iterator text_iterator = fresh_metatext.begin();
|
---|
1096 | while (text_iterator != fresh_metatext.end())
|
---|
1097 | {
|
---|
1098 | if (*text_iterator == '&') text_html_safe += "&";
|
---|
1099 | else if (*text_iterator == '<') text_html_safe += "<";
|
---|
1100 | else if (*text_iterator == '>') text_html_safe += ">";
|
---|
1101 | else if (*text_iterator == '"') text_html_safe += """;
|
---|
1102 | else text_html_safe.push_back(*text_iterator);
|
---|
1103 | text_iterator++;
|
---|
1104 | }
|
---|
1105 | fresh_metatext = text_html_safe;
|
---|
1106 | }
|
---|
1107 | // New "dmsafe" special formatting option (always apply to "srclinkFile" metadata)
|
---|
1108 | // (The metaname "srclink_file" is deprecated, use "srclinkFile")
|
---|
1109 | if (meta.metacommand & mDMSafe || meta.metaname == "srclinkFile")
|
---|
1110 | {
|
---|
1111 | // Make it macro-safe
|
---|
1112 | text_t text_dm_safe = dm_safe(fresh_metatext);
|
---|
1113 | fresh_metatext = text_dm_safe;
|
---|
1114 | }
|
---|
1115 |
|
---|
1116 | if (metadata_wrap) {
|
---|
1117 | fresh_metatext = wrap_metatext(fresh_metatext,OID,meta.metaname,i);
|
---|
1118 | }
|
---|
1119 | formatted_metatext += fresh_metatext;
|
---|
1120 |
|
---|
1121 | first = false;
|
---|
1122 |
|
---|
1123 | }
|
---|
1124 | } else {
|
---|
1125 | if (position == -2) { // end
|
---|
1126 | position = end_i;
|
---|
1127 | } else if (position < start_i || position > end_i) {
|
---|
1128 | return "";
|
---|
1129 | }
|
---|
1130 |
|
---|
1131 | text_t fresh_metatext;
|
---|
1132 | if (meta.metacommand & mSpecial) {
|
---|
1133 |
|
---|
1134 | // special formatting
|
---|
1135 | if (no_ns_metaname == "Date") fresh_metatext = format_date (metainfo.values[position]);
|
---|
1136 | else if (no_ns_metaname == "Language") fresh_metatext = iso639(metainfo.values[position]);
|
---|
1137 | else fresh_metatext = "_format:"+meta.metaname+"_("+metainfo.values[position]+")";
|
---|
1138 | }
|
---|
1139 | else fresh_metatext = metainfo.values[position];
|
---|
1140 |
|
---|
1141 | // New "truncate" special formatting option
|
---|
1142 | if (meta.metacommand & mTruncate)
|
---|
1143 | {
|
---|
1144 | int truncate_length = meta.siblingoptions.getint();
|
---|
1145 | text_t truncated_value = fresh_metatext;
|
---|
1146 | if (truncated_value.size() > truncate_length)
|
---|
1147 | {
|
---|
1148 | truncated_value = substr(truncated_value.begin(), truncated_value.begin() + truncate_length) + "... _texttruncated_";
|
---|
1149 | }
|
---|
1150 | fresh_metatext = truncated_value;
|
---|
1151 | }
|
---|
1152 | // New "xmlsafe" special formatting option
|
---|
1153 | if (meta.metacommand & mXMLSafe)
|
---|
1154 | {
|
---|
1155 | // Make it XML-safe
|
---|
1156 | text_t text_xml_safe = "";
|
---|
1157 | text_t::const_iterator text_iterator = fresh_metatext.begin();
|
---|
1158 | while (text_iterator != fresh_metatext.end())
|
---|
1159 | {
|
---|
1160 | if (*text_iterator == '&') text_xml_safe += "&";
|
---|
1161 | else if (*text_iterator == '<') text_xml_safe += "<";
|
---|
1162 | else if (*text_iterator == '>') text_xml_safe += ">";
|
---|
1163 | else text_xml_safe.push_back(*text_iterator);
|
---|
1164 | text_iterator++;
|
---|
1165 | }
|
---|
1166 | fresh_metatext = text_xml_safe;
|
---|
1167 | }
|
---|
1168 | // New "htmlsafe" special formatting option
|
---|
1169 | if (meta.metacommand & mHTMLSafe)
|
---|
1170 | {
|
---|
1171 | // Make it HTML-safe
|
---|
1172 | text_t text_html_safe = "";
|
---|
1173 | text_t::const_iterator text_iterator = fresh_metatext.begin();
|
---|
1174 | while (text_iterator != fresh_metatext.end())
|
---|
1175 | {
|
---|
1176 | if (*text_iterator == '&') text_html_safe += "&";
|
---|
1177 | else if (*text_iterator == '<') text_html_safe += "<";
|
---|
1178 | else if (*text_iterator == '>') text_html_safe += ">";
|
---|
1179 | else if (*text_iterator == '"') text_html_safe += """;
|
---|
1180 | else if (*text_iterator == '\'') text_html_safe += "'";
|
---|
1181 | else if (*text_iterator == ',') text_html_safe += ",";
|
---|
1182 | else text_html_safe.push_back(*text_iterator);
|
---|
1183 | text_iterator++;
|
---|
1184 | }
|
---|
1185 | fresh_metatext = text_html_safe;
|
---|
1186 | }
|
---|
1187 | // New "dmsafe" special formatting option (always apply to "srclinkFile" metadata)
|
---|
1188 | // (The metaname "srclink_file" is deprecated, use "srclinkFile")
|
---|
1189 | if (meta.metacommand & mDMSafe || meta.metaname == "srclinkFile")
|
---|
1190 | {
|
---|
1191 | // Make it macro-safe
|
---|
1192 | text_t text_dm_safe = dm_safe(fresh_metatext);
|
---|
1193 | fresh_metatext = text_dm_safe;
|
---|
1194 | }
|
---|
1195 |
|
---|
1196 | if (metadata_wrap) {
|
---|
1197 | fresh_metatext = wrap_metatext(fresh_metatext,OID,meta.metaname,position);
|
---|
1198 | }
|
---|
1199 |
|
---|
1200 | formatted_metatext += fresh_metatext;
|
---|
1201 | }
|
---|
1202 |
|
---|
1203 | if (meta.metacommand & mCgiSafe && !no_cgisafe) return cgi_safe_unicode (formatted_metatext);
|
---|
1204 | else return formatted_metatext;
|
---|
1205 | }
|
---|
1206 |
|
---|
1207 | static text_t get_parent_meta (ResultDocInfo_t &docinfo, const metadata_t &meta, int siblings_values)
|
---|
1208 | {
|
---|
1209 |
|
---|
1210 | MetadataInfo_t *parent = docinfo.metadata[meta.metaname].parent;
|
---|
1211 |
|
---|
1212 | switch (meta.mqualifier.parent) {
|
---|
1213 | case pNone:
|
---|
1214 | return "Nothing!!";
|
---|
1215 | break;
|
---|
1216 |
|
---|
1217 | case pImmediate:
|
---|
1218 | if (parent != NULL) {
|
---|
1219 | text_t parent_oid = get_parent(docinfo.OID);
|
---|
1220 | return get_formatted_meta_text(parent_oid,*parent, meta, siblings_values);
|
---|
1221 | }
|
---|
1222 | break;
|
---|
1223 |
|
---|
1224 | case pTop:
|
---|
1225 | if (parent != NULL) {
|
---|
1226 | text_t parent_oid = get_parent(docinfo.OID);
|
---|
1227 |
|
---|
1228 | while (parent->parent != NULL) {
|
---|
1229 | parent = parent->parent;
|
---|
1230 | parent_oid = get_parent(parent_oid);
|
---|
1231 | }
|
---|
1232 | return get_formatted_meta_text(parent_oid,*parent, meta, siblings_values);
|
---|
1233 | }
|
---|
1234 | break;
|
---|
1235 |
|
---|
1236 | case pAll:
|
---|
1237 | MetadataInfo_t *parent = docinfo.metadata[meta.metaname].parent;
|
---|
1238 | if (parent != NULL) {
|
---|
1239 | text_t parent_oid = get_parent(docinfo.OID);
|
---|
1240 |
|
---|
1241 | text_tarray tmparray;
|
---|
1242 | while (parent != NULL) {
|
---|
1243 | tmparray.push_back (get_formatted_meta_text(parent_oid,*parent, meta, siblings_values, true)); // set no_cgisafe to true, as we'll do it once we have all the metadata
|
---|
1244 | parent = parent->parent;
|
---|
1245 | parent_oid = get_parent(parent_oid);
|
---|
1246 |
|
---|
1247 | }
|
---|
1248 | // now join them up - use teh parent separator
|
---|
1249 | bool first = true;
|
---|
1250 | text_t tmp;
|
---|
1251 | text_tarray::reverse_iterator here = tmparray.rbegin();
|
---|
1252 | text_tarray::reverse_iterator end = tmparray.rend();
|
---|
1253 | while (here != end) {
|
---|
1254 | if (!first) tmp += meta.parentoptions;
|
---|
1255 | tmp += *here;
|
---|
1256 | first = false;
|
---|
1257 | ++here;
|
---|
1258 | }
|
---|
1259 | if (meta.metacommand & mCgiSafe) return cgi_safe_unicode (tmp);
|
---|
1260 | else return tmp;
|
---|
1261 | }
|
---|
1262 | }
|
---|
1263 | return "";
|
---|
1264 |
|
---|
1265 | }
|
---|
1266 |
|
---|
1267 | static text_t get_child_meta (const text_t& collection,
|
---|
1268 | recptproto* collectproto,
|
---|
1269 | ResultDocInfo_t &docinfo, displayclass &disp,
|
---|
1270 | const metadata_t &meta, text_tmap &options,
|
---|
1271 | ostream& logout, int siblings_values)
|
---|
1272 | {
|
---|
1273 | if (docinfo.metadata["contains"].values[0].size()==0) return ""; // no children
|
---|
1274 |
|
---|
1275 | const text_t& pre_tree_trav = meta.pre_tree_traverse;
|
---|
1276 | const text_t& child_metaname = meta.metaname;
|
---|
1277 | const text_t& child_field = meta.childoptions;
|
---|
1278 | text_tset child_metadata;
|
---|
1279 | child_metadata.insert(child_metaname);
|
---|
1280 |
|
---|
1281 | FilterResponse_t child_response;
|
---|
1282 | if (meta.mqualifier.child == cNum) {
|
---|
1283 | // just one child
|
---|
1284 | //get the information associated with the metadata for child doc
|
---|
1285 | if (!get_info (docinfo.OID+pre_tree_trav+child_field, collection, "",
|
---|
1286 | child_metadata, false, collectproto, child_response,
|
---|
1287 | logout)) return ""; // invalid child number
|
---|
1288 |
|
---|
1289 | if (child_response.docInfo.empty()) return false; // no info for the child
|
---|
1290 |
|
---|
1291 | ResultDocInfo_t& child_docinfo = child_response.docInfo[0];
|
---|
1292 | MetadataInfo_t& metaname_rec = child_docinfo.metadata[child_metaname];
|
---|
1293 |
|
---|
1294 | text_t child_metavalue
|
---|
1295 | = get_formatted_meta_text(child_docinfo.OID,metaname_rec,meta,siblings_values);
|
---|
1296 | return expand_metadata(child_metavalue,collection,collectproto,
|
---|
1297 | child_docinfo,disp,options,logout);
|
---|
1298 | }
|
---|
1299 |
|
---|
1300 |
|
---|
1301 | if (meta.mqualifier.child != cAll) return false; // invalid qualifier
|
---|
1302 |
|
---|
1303 |
|
---|
1304 | if (!pre_tree_trav.empty()) {
|
---|
1305 | // need to get relevant "contains" metadata for new (e.g. pre tree trav) node
|
---|
1306 | FilterResponse_t trav_response;
|
---|
1307 |
|
---|
1308 | text_tset trav_metadata;
|
---|
1309 | trav_metadata.insert("contains");
|
---|
1310 |
|
---|
1311 | if (!get_info (docinfo.OID+pre_tree_trav, collection, "",
|
---|
1312 | trav_metadata, false, collectproto, trav_response,
|
---|
1313 | logout)) return ""; // invalid pre_tree_trav
|
---|
1314 |
|
---|
1315 | if (trav_response.docInfo.empty()) return false; // no info for the pre_tree_trav OID
|
---|
1316 |
|
---|
1317 | ResultDocInfo_t& trav_docinfo = trav_response.docInfo[0];
|
---|
1318 |
|
---|
1319 | // use this for rest of routine
|
---|
1320 | docinfo = trav_docinfo;
|
---|
1321 | }
|
---|
1322 |
|
---|
1323 | // we need to get all children
|
---|
1324 | text_t result = "";
|
---|
1325 | text_tarray children;
|
---|
1326 | text_t contains = docinfo.metadata["contains"].values[0];
|
---|
1327 | splitchar (contains.begin(), contains.end(), ';', children);
|
---|
1328 | text_tarray::const_iterator here = children.begin();
|
---|
1329 | text_tarray::const_iterator end = children.end();
|
---|
1330 | bool first = true;
|
---|
1331 | while (here !=end) {
|
---|
1332 | text_t oid = *here;
|
---|
1333 | here++;
|
---|
1334 | if (*(oid.begin()) == '"') translate_parent (oid, docinfo.OID);
|
---|
1335 |
|
---|
1336 | //get the information associated with the metadata for child doc
|
---|
1337 | if (!get_info (oid, collection, "", child_metadata,
|
---|
1338 | false, collectproto, child_response, logout) ||
|
---|
1339 | child_response.docInfo.empty()) {
|
---|
1340 | first = false;
|
---|
1341 | continue;
|
---|
1342 | }
|
---|
1343 |
|
---|
1344 |
|
---|
1345 | ResultDocInfo_t& child_docinfo = child_response.docInfo[0];
|
---|
1346 | MetadataInfo_t& metaname_rec = child_docinfo.metadata[child_metaname];
|
---|
1347 |
|
---|
1348 | text_t child_metavalue
|
---|
1349 | = get_formatted_meta_text(child_docinfo.OID,metaname_rec,meta,siblings_values);
|
---|
1350 |
|
---|
1351 |
|
---|
1352 | if (!first) result += child_field;
|
---|
1353 | first = false;
|
---|
1354 | // need to do this here cos otherwise we are in the wrong document
|
---|
1355 | text_t em = expand_metadata(child_metavalue,collection,collectproto,
|
---|
1356 | child_docinfo,disp,options,logout);
|
---|
1357 |
|
---|
1358 | result += em;
|
---|
1359 | }
|
---|
1360 | return result;
|
---|
1361 |
|
---|
1362 | }
|
---|
1363 |
|
---|
1364 | static text_t get_meta (const text_t& collection, recptproto* collectproto,
|
---|
1365 | ResultDocInfo_t &docinfo, displayclass &disp,
|
---|
1366 | const metadata_t &meta, text_tmap &options,
|
---|
1367 | ostream& logout) {
|
---|
1368 |
|
---|
1369 | // make sure we have the requested metadata
|
---|
1370 | MetadataInfo_tmap::iterator it = docinfo.metadata.find (meta.metaname);
|
---|
1371 | if (it == docinfo.metadata.end()) return "";
|
---|
1372 |
|
---|
1373 | int siblings_values = 0; // default is no siblings, just the first metadata available
|
---|
1374 | if (meta.metacommand & mSibling) {
|
---|
1375 | if (meta.mqualifier.sibling == sAll) {
|
---|
1376 | siblings_values = -1; //all
|
---|
1377 | } else if (meta.mqualifier.sibling == sNum) {
|
---|
1378 | siblings_values = meta.siblingoptions.getint();
|
---|
1379 | }
|
---|
1380 | }
|
---|
1381 | if (meta.metacommand & mParent) {
|
---|
1382 | return get_parent_meta(docinfo,meta,siblings_values);
|
---|
1383 | }
|
---|
1384 |
|
---|
1385 | else if (meta.metacommand & mChild) {
|
---|
1386 | return get_child_meta(collection,collectproto,docinfo,disp,meta,
|
---|
1387 | options,logout, siblings_values);
|
---|
1388 | }
|
---|
1389 | else if (meta.metacommand & mSibling) { // only siblings
|
---|
1390 | MetadataInfo_t& metaname_rec = docinfo.metadata[meta.metaname];
|
---|
1391 | return get_formatted_meta_text(docinfo.OID,docinfo.metadata[meta.metaname],meta, siblings_values);
|
---|
1392 | }
|
---|
1393 | else {
|
---|
1394 |
|
---|
1395 | // straightforward metadata request (nothing fancy)
|
---|
1396 |
|
---|
1397 | text_t classifier_metaname = docinfo.classifier_metadata_type;
|
---|
1398 | int metaname_index
|
---|
1399 | = (classifier_metaname == meta.metaname) ? docinfo.classifier_metadata_offset : 0;
|
---|
1400 | return get_formatted_meta_text(docinfo.OID,docinfo.metadata[meta.metaname], meta, metaname_index);
|
---|
1401 | }
|
---|
1402 |
|
---|
1403 | return "";
|
---|
1404 | }
|
---|
1405 |
|
---|
1406 | static text_t get_or (const text_t& collection, recptproto* collectproto,
|
---|
1407 | ResultDocInfo_t &docinfo, displayclass &disp,
|
---|
1408 | format_t *orptr, text_tmap &options,
|
---|
1409 | ostream& logout) {
|
---|
1410 |
|
---|
1411 | while (orptr != NULL) {
|
---|
1412 |
|
---|
1413 | if (metadata_wrap) {
|
---|
1414 | // need to be a bit more careful about this
|
---|
1415 | // => test for it *without* spanwrap or divwrap, and if defined, then
|
---|
1416 | // got back and generate it again, this time with spanwrap/divwrap on
|
---|
1417 |
|
---|
1418 | metadata_wrap = false;
|
---|
1419 | text_t test_tmp = format_string (collection,collectproto,docinfo, disp, orptr,
|
---|
1420 | options, logout);
|
---|
1421 | metadata_wrap = true;
|
---|
1422 | if (!test_tmp.empty()) {
|
---|
1423 |
|
---|
1424 | return format_string (collection,collectproto,docinfo, disp, orptr,
|
---|
1425 | options, logout);
|
---|
1426 | }
|
---|
1427 | }
|
---|
1428 | else {
|
---|
1429 | text_t tmp = format_string (collection,collectproto,docinfo, disp, orptr,
|
---|
1430 | options, logout);
|
---|
1431 | if (!tmp.empty()) return tmp;
|
---|
1432 | }
|
---|
1433 |
|
---|
1434 | orptr = orptr->nextptr;
|
---|
1435 | }
|
---|
1436 | return "";
|
---|
1437 | }
|
---|
1438 |
|
---|
1439 | static bool char_is_whitespace(const char c)
|
---|
1440 | {
|
---|
1441 | return ((c == ' ') || (c == '\t') || (c == '\n') || (c == '\r'));
|
---|
1442 |
|
---|
1443 | }
|
---|
1444 |
|
---|
1445 | static int scan_over_whitespace(const text_t& outstring, const int start_pos)
|
---|
1446 | {
|
---|
1447 | int pos = start_pos;
|
---|
1448 | while (pos<outstring.size()) {
|
---|
1449 | if (!char_is_whitespace(outstring[pos])) {
|
---|
1450 | break;
|
---|
1451 | }
|
---|
1452 | ++pos;
|
---|
1453 | }
|
---|
1454 |
|
---|
1455 | return pos;
|
---|
1456 | }
|
---|
1457 |
|
---|
1458 | static int rscan_over_whitespace(const text_t& outstring, const int start_pos)
|
---|
1459 | {
|
---|
1460 | int pos = start_pos;
|
---|
1461 | while (pos>=0) {
|
---|
1462 | if (!char_is_whitespace(outstring[pos])) {
|
---|
1463 | break;
|
---|
1464 | }
|
---|
1465 | --pos;
|
---|
1466 | }
|
---|
1467 |
|
---|
1468 | return pos;
|
---|
1469 | }
|
---|
1470 |
|
---|
1471 | static int rscan_for_whitespace(const text_t& outstring, const int start_pos)
|
---|
1472 | {
|
---|
1473 | int pos = start_pos;
|
---|
1474 | while (pos>=0) {
|
---|
1475 | if (char_is_whitespace(outstring[pos])) {
|
---|
1476 | break;
|
---|
1477 | }
|
---|
1478 | --pos;
|
---|
1479 | }
|
---|
1480 |
|
---|
1481 | return pos;
|
---|
1482 | }
|
---|
1483 |
|
---|
1484 |
|
---|
1485 | static int rscan_for(const text_t& outstring, const int start_pos,
|
---|
1486 | const char find_c)
|
---|
1487 | {
|
---|
1488 | int pos = start_pos;
|
---|
1489 | while (pos>=0) {
|
---|
1490 | char c = outstring[pos];
|
---|
1491 | if (outstring[pos] == find_c) {
|
---|
1492 | break;
|
---|
1493 | }
|
---|
1494 | --pos;
|
---|
1495 | }
|
---|
1496 |
|
---|
1497 | return pos;
|
---|
1498 | }
|
---|
1499 |
|
---|
1500 | text_t extract_substr(const text_t& outstring, const int start_pos,
|
---|
1501 | const int end_pos)
|
---|
1502 | {
|
---|
1503 | text_t extracted_str;
|
---|
1504 | extracted_str.clear();
|
---|
1505 |
|
---|
1506 | for (int pos=start_pos; pos<=end_pos; ++pos) {
|
---|
1507 | extracted_str.push_back(outstring[pos]);
|
---|
1508 | }
|
---|
1509 |
|
---|
1510 | return extracted_str;
|
---|
1511 | }
|
---|
1512 |
|
---|
1513 |
|
---|
1514 | static text_t expand_potential_metadata(const text_t& collection,
|
---|
1515 | recptproto* collectproto,
|
---|
1516 | ResultDocInfo_t &docinfo,
|
---|
1517 | displayclass &disp,
|
---|
1518 | const text_t& intext,
|
---|
1519 | text_tmap &options,
|
---|
1520 | ostream& logout)
|
---|
1521 | {
|
---|
1522 | text_t outtext;
|
---|
1523 |
|
---|
1524 | // decide if dealing with metadata or text
|
---|
1525 |
|
---|
1526 | text_t::const_iterator beginbracket = intext.begin();
|
---|
1527 | text_t::const_iterator endbracket = (intext.end() - 1);
|
---|
1528 |
|
---|
1529 | // Decision is based on a metadata element
|
---|
1530 | if ((*beginbracket == '[') && (*endbracket == ']')) {
|
---|
1531 | // Ignore the surrounding square brackets
|
---|
1532 | text_t meta_text = substr (beginbracket+1, endbracket);
|
---|
1533 |
|
---|
1534 | if (meta_text == "Text") {
|
---|
1535 | outtext = format_text(collection, collectproto, docinfo, disp, options, logout);
|
---|
1536 | }
|
---|
1537 | else {
|
---|
1538 |
|
---|
1539 | text_tset metadata;
|
---|
1540 | bool getParents =false;
|
---|
1541 | metadata_t meta;
|
---|
1542 |
|
---|
1543 | parse_meta (meta_text, meta, metadata, getParents);
|
---|
1544 | outtext
|
---|
1545 | = get_meta (collection,collectproto,docinfo,disp,meta,options,logout);
|
---|
1546 | }
|
---|
1547 |
|
---|
1548 | }
|
---|
1549 | else {
|
---|
1550 | outtext = intext;
|
---|
1551 | }
|
---|
1552 |
|
---|
1553 | return outtext;
|
---|
1554 | }
|
---|
1555 |
|
---|
1556 |
|
---|
1557 | static bool uses_expression(const text_t& collection, recptproto* collectproto,
|
---|
1558 | ResultDocInfo_t &docinfo,
|
---|
1559 | displayclass &disp,
|
---|
1560 | const text_t& outstring, text_t& lhs_expr,
|
---|
1561 | text_t& op_expr, text_t& rhs_expr,
|
---|
1562 | text_tmap &options,
|
---|
1563 | ostream& logout)
|
---|
1564 | {
|
---|
1565 | // Note: the string may not be of the form: str1 op str2, however
|
---|
1566 | // to deterine this we have to process it on the assumption it is,
|
---|
1567 | // and if at any point an 'erroneous' value is encountered, return
|
---|
1568 | // false and let something else have a go at evaluating it
|
---|
1569 |
|
---|
1570 | // Starting at the end of the string and working backwards ..
|
---|
1571 |
|
---|
1572 | const int outstring_len = outstring.size();
|
---|
1573 |
|
---|
1574 | // skip over white space
|
---|
1575 | int rhs_end = rscan_over_whitespace(outstring,outstring_len-1);
|
---|
1576 |
|
---|
1577 | if (rhs_end<=0) {
|
---|
1578 | // no meaningful text or (rhs_end==0) no room for operator
|
---|
1579 | return false;
|
---|
1580 | }
|
---|
1581 |
|
---|
1582 | // check for ' or " and then scan over token
|
---|
1583 | const char potential_quote = outstring[rhs_end];
|
---|
1584 | int rhs_start=rhs_end;
|
---|
1585 | bool quoted = false;
|
---|
1586 |
|
---|
1587 | if ((potential_quote == '\'') || (potential_quote == '\"')) {
|
---|
1588 | --rhs_end;
|
---|
1589 | rhs_start = rscan_for(outstring,rhs_end-1,potential_quote) +1;
|
---|
1590 | quoted = true;
|
---|
1591 | }
|
---|
1592 | else {
|
---|
1593 | rhs_start = rscan_for_whitespace(outstring,rhs_end-1) +1;
|
---|
1594 | }
|
---|
1595 |
|
---|
1596 | if ((rhs_end-rhs_start)<0) {
|
---|
1597 | // no meaningful rhs expression
|
---|
1598 | return false;
|
---|
1599 | }
|
---|
1600 |
|
---|
1601 | // form rhs_expr
|
---|
1602 | rhs_expr = extract_substr(outstring,rhs_start,rhs_end);
|
---|
1603 |
|
---|
1604 | // skip over white space
|
---|
1605 | const int to_whitespace = (quoted) ? 2 : 1;
|
---|
1606 |
|
---|
1607 | int op_end = rscan_over_whitespace(outstring,rhs_start-to_whitespace);
|
---|
1608 | int op_start = rscan_for_whitespace(outstring,op_end-1)+1;
|
---|
1609 |
|
---|
1610 | if ((op_end<0) && (op_start<0)) {
|
---|
1611 | // no meaningful expression operator
|
---|
1612 | return false;
|
---|
1613 | }
|
---|
1614 |
|
---|
1615 | if (op_end-op_start<0) {
|
---|
1616 | // no meaningful expression operator
|
---|
1617 | return false;
|
---|
1618 | }
|
---|
1619 |
|
---|
1620 | op_expr = extract_substr(outstring,op_start,op_end);
|
---|
1621 |
|
---|
1622 |
|
---|
1623 | // check for operator
|
---|
1624 | if ((op_expr != "eq") && (op_expr != "ne") && (op_expr != "gt") &&
|
---|
1625 | (op_expr != "ge") && (op_expr != "lt") && (op_expr != "le") && (op_expr != "==") && (op_expr != "!=") && (op_expr != ">") && (op_expr != ">=") && (op_expr != "<") && (op_expr != "<=") && (op_expr != "sw") && (op_expr != "ew")) {
|
---|
1626 |
|
---|
1627 | // not a valid operator
|
---|
1628 | return false;
|
---|
1629 | }
|
---|
1630 |
|
---|
1631 | int lhs_end = rscan_over_whitespace(outstring,op_start-1);
|
---|
1632 | if (lhs_end<0) {
|
---|
1633 | // no meaningful lhs expression
|
---|
1634 | return false;
|
---|
1635 | }
|
---|
1636 |
|
---|
1637 | int lhs_start = scan_over_whitespace(outstring,0);
|
---|
1638 |
|
---|
1639 | // form lhs_expr from remainder of string
|
---|
1640 | lhs_expr = extract_substr(outstring,lhs_start,lhs_end);
|
---|
1641 |
|
---|
1642 | // Now we know we have a valid expression, look up any
|
---|
1643 | // metadata terms
|
---|
1644 |
|
---|
1645 | rhs_expr = expand_potential_metadata(collection,collectproto,docinfo,
|
---|
1646 | disp,rhs_expr,options,logout);
|
---|
1647 | lhs_expr = expand_potential_metadata(collection,collectproto,docinfo,
|
---|
1648 | disp,lhs_expr,options,logout);
|
---|
1649 |
|
---|
1650 | return true;
|
---|
1651 | }
|
---|
1652 |
|
---|
1653 | static bool eval_expression_true(const text_t& lhs_expr,const text_t& op_expr,
|
---|
1654 | const text_t& rhs_expr, ostream& logout)
|
---|
1655 | {
|
---|
1656 | if (op_expr == "eq") return (lhs_expr == rhs_expr);
|
---|
1657 | else if (op_expr == "ne" ) return (lhs_expr != rhs_expr);
|
---|
1658 | else if (op_expr == "gt") return (lhs_expr > rhs_expr);
|
---|
1659 | else if (op_expr == "ge") return (lhs_expr >= rhs_expr);
|
---|
1660 | else if (op_expr == "lt") return (lhs_expr < rhs_expr);
|
---|
1661 | else if (op_expr == "le") return (lhs_expr <= rhs_expr);
|
---|
1662 | else if (op_expr == "==") return (lhs_expr.getint() == rhs_expr.getint());
|
---|
1663 | else if (op_expr == "!=") return (lhs_expr.getint() != rhs_expr.getint());
|
---|
1664 | else if (op_expr == ">") return (lhs_expr.getint() > rhs_expr.getint());
|
---|
1665 | else if (op_expr == ">=") return (lhs_expr.getint() >= rhs_expr.getint());
|
---|
1666 | else if (op_expr == "<") return (lhs_expr.getint() < rhs_expr.getint());
|
---|
1667 | else if (op_expr == "<=") return (lhs_expr.getint() <= rhs_expr.getint());
|
---|
1668 | else if (op_expr == "sw") return (starts_with(lhs_expr,rhs_expr));
|
---|
1669 | else if (op_expr == "ew") return (ends_with(lhs_expr,rhs_expr));
|
---|
1670 | else {
|
---|
1671 | logout << "Error: '" << op_expr << "' is not a recognised operator." << endl;
|
---|
1672 | }
|
---|
1673 |
|
---|
1674 | return false;
|
---|
1675 | }
|
---|
1676 |
|
---|
1677 |
|
---|
1678 | static text_t get_if (const text_t& collection, recptproto* collectproto,
|
---|
1679 | ResultDocInfo_t &docinfo, displayclass &disp,
|
---|
1680 | const decision_t &decision,
|
---|
1681 | format_t *ifptr, format_t *elseptr,
|
---|
1682 | text_tmap &options, ostream& logout)
|
---|
1683 | {
|
---|
1684 | // If the decision component is a metadata element, then evaluate it
|
---|
1685 | // to see whether we output the "then" or the "else" clause
|
---|
1686 | if (decision.command == dMeta) {
|
---|
1687 |
|
---|
1688 | bool store_metadata_wrap = metadata_wrap;
|
---|
1689 | metadata_wrap = 0;
|
---|
1690 |
|
---|
1691 | // temporarily suspend metadata-XXXwrap (if on) so can test if metadata item really exits or not
|
---|
1692 | bool metadata_exists
|
---|
1693 | = (get_meta (collection,collectproto,docinfo,disp,decision.meta,options,
|
---|
1694 | logout) != "");
|
---|
1695 |
|
---|
1696 | metadata_wrap = store_metadata_wrap;
|
---|
1697 |
|
---|
1698 | if (metadata_exists) {
|
---|
1699 | if (ifptr != NULL)
|
---|
1700 | return get_formatted_string (collection,collectproto,docinfo, disp, ifptr,
|
---|
1701 | options, logout);
|
---|
1702 | }
|
---|
1703 | else {
|
---|
1704 | if (elseptr != NULL)
|
---|
1705 | return get_formatted_string (collection,collectproto,docinfo, disp, elseptr,
|
---|
1706 | options, logout);
|
---|
1707 | }
|
---|
1708 | }
|
---|
1709 |
|
---|
1710 | // If the decision component is text, then evaluate it (it is probably a
|
---|
1711 | // macro like _cgiargmode_) to decide what to output.
|
---|
1712 | else if (decision.command == dText) {
|
---|
1713 |
|
---|
1714 | text_t outstring;
|
---|
1715 | disp.expandstring (decision.text, outstring);
|
---|
1716 |
|
---|
1717 | // Check for if expression in form: str1 op str2
|
---|
1718 | // (such as [x] eq "y")
|
---|
1719 | text_t lhs_expr, op_expr, rhs_expr;
|
---|
1720 | if (uses_expression(collection,collectproto,docinfo, disp, outstring,lhs_expr,op_expr,rhs_expr, options,logout)) {
|
---|
1721 | if (eval_expression_true(lhs_expr,op_expr,rhs_expr,logout)) {
|
---|
1722 | if (ifptr != NULL) {
|
---|
1723 | return get_formatted_string (collection, collectproto, docinfo, disp, ifptr,
|
---|
1724 | options, logout);
|
---|
1725 | }
|
---|
1726 | else {
|
---|
1727 | return "";
|
---|
1728 | }
|
---|
1729 | } else {
|
---|
1730 | if (elseptr != NULL) {
|
---|
1731 | return get_formatted_string (collection, collectproto, docinfo, disp, elseptr,
|
---|
1732 | options, logout);
|
---|
1733 | }
|
---|
1734 | else {
|
---|
1735 | return "";
|
---|
1736 | }
|
---|
1737 | }
|
---|
1738 | }
|
---|
1739 |
|
---|
1740 |
|
---|
1741 | // This is a tad tricky. When we expand a string like _cgiargmode_, that is
|
---|
1742 | // a cgi argument macro that has not been set, it evaluates to itself.
|
---|
1743 | // Therefore, we have to say that a piece of text evaluates true if
|
---|
1744 | // it is non-empty and if it is a cgi argument evaulating to itself.
|
---|
1745 |
|
---|
1746 | if ((outstring != "") && !((outstring == decision.text) && (outstring[0] == '_'))) {
|
---|
1747 | if (ifptr != NULL)
|
---|
1748 | return get_formatted_string (collection, collectproto, docinfo, disp, ifptr,
|
---|
1749 | options, logout);
|
---|
1750 | } else {
|
---|
1751 | if (elseptr != NULL)
|
---|
1752 | return get_formatted_string (collection, collectproto, docinfo, disp, elseptr,
|
---|
1753 | options, logout);
|
---|
1754 | }
|
---|
1755 | }
|
---|
1756 |
|
---|
1757 | return "";
|
---|
1758 | }
|
---|
1759 |
|
---|
1760 | bool includes_metadata(const text_t& text)
|
---|
1761 | {
|
---|
1762 | text_t::const_iterator here = text.begin();
|
---|
1763 | text_t::const_iterator end = text.end();
|
---|
1764 |
|
---|
1765 | char startbracket = '[';
|
---|
1766 | char endbracket = ']';
|
---|
1767 |
|
---|
1768 | char bracket = startbracket;
|
---|
1769 | while (here != end) {
|
---|
1770 | if (*here == bracket) {
|
---|
1771 | if(bracket == startbracket) {
|
---|
1772 | // seen a [, next look for a ] to confirm it's metadata
|
---|
1773 | bracket = endbracket;
|
---|
1774 | } else if(bracket == endbracket) {
|
---|
1775 | // found [ ... ] in text, so we think it includes metadata
|
---|
1776 | return true;
|
---|
1777 | }
|
---|
1778 | }
|
---|
1779 | ++here;
|
---|
1780 | }
|
---|
1781 |
|
---|
1782 | return false;
|
---|
1783 | }
|
---|
1784 |
|
---|
1785 | static text_t expand_metadata(const text_t &metavalue, const text_t& collection,
|
---|
1786 | recptproto* collectproto,
|
---|
1787 | ResultDocInfo_t &docinfo,
|
---|
1788 | displayclass &disp, text_tmap &options,
|
---|
1789 | ostream &logout) {
|
---|
1790 |
|
---|
1791 | if (includes_metadata(metavalue)) {
|
---|
1792 |
|
---|
1793 | // text has embedded metadata in it => expand it
|
---|
1794 | FilterRequest_t request;
|
---|
1795 | FilterResponse_t response;
|
---|
1796 |
|
---|
1797 | request.getParents = false;
|
---|
1798 |
|
---|
1799 | format_t *expanded_formatlistptr = new format_t();
|
---|
1800 | parse_formatstring (metavalue, expanded_formatlistptr,
|
---|
1801 | request.fields, request.getParents);
|
---|
1802 |
|
---|
1803 | // retrieve metadata
|
---|
1804 | get_info(docinfo.OID, collection, "", request.fields, request.getParents,
|
---|
1805 | collectproto, response, logout);
|
---|
1806 |
|
---|
1807 | if (!response.docInfo.empty()) {
|
---|
1808 |
|
---|
1809 | text_t expanded_metavalue
|
---|
1810 | = get_formatted_string(collection, collectproto,
|
---|
1811 | response.docInfo[0], disp, expanded_formatlistptr,
|
---|
1812 | options, logout);
|
---|
1813 |
|
---|
1814 | return expanded_metavalue;
|
---|
1815 | }
|
---|
1816 | else {
|
---|
1817 | return metavalue;
|
---|
1818 | }
|
---|
1819 | }
|
---|
1820 | else {
|
---|
1821 |
|
---|
1822 | return metavalue;
|
---|
1823 | }
|
---|
1824 | }
|
---|
1825 |
|
---|
1826 | text_t get_collection_meta(const text_t& collection, recptproto* collectproto,
|
---|
1827 | displayclass &disp,
|
---|
1828 | text_t meta_name, ostream& logout) {
|
---|
1829 |
|
---|
1830 | ColInfoResponse_t collectinfo;
|
---|
1831 | comerror_t err;
|
---|
1832 | collectproto->get_collectinfo (collection, collectinfo,err,logout);
|
---|
1833 | text_t meta_value = "";
|
---|
1834 | text_t lang;
|
---|
1835 | disp.expandstring("_cgiargl_",lang);
|
---|
1836 | if (lang.empty()) {
|
---|
1837 | lang = "en";
|
---|
1838 | }
|
---|
1839 |
|
---|
1840 | if (err == noError) {
|
---|
1841 | meta_value = collectinfo.get_collectionmeta(meta_name, lang);
|
---|
1842 | }
|
---|
1843 | return meta_value;
|
---|
1844 |
|
---|
1845 |
|
---|
1846 | }
|
---|
1847 | text_t format_string (const text_t& collection, recptproto* collectproto,
|
---|
1848 | ResultDocInfo_t &docinfo, displayclass &disp,
|
---|
1849 | format_t *formatlistptr, text_tmap &options,
|
---|
1850 | ostream& logout) {
|
---|
1851 |
|
---|
1852 | if (formatlistptr == NULL) return "";
|
---|
1853 |
|
---|
1854 | switch (formatlistptr->command) {
|
---|
1855 | case comOID:
|
---|
1856 | return docinfo.OID;
|
---|
1857 | case comTopOID:
|
---|
1858 | {
|
---|
1859 | text_t top_id;
|
---|
1860 | get_top(docinfo.OID, top_id);
|
---|
1861 | return top_id;
|
---|
1862 | }
|
---|
1863 | case comRank:
|
---|
1864 | return text_t(docinfo.ranking);
|
---|
1865 | case comText:
|
---|
1866 | return formatlistptr->text;
|
---|
1867 | case comLink:
|
---|
1868 | return options["link"];
|
---|
1869 | case comEndLink:
|
---|
1870 | {
|
---|
1871 | if (options["link"].empty()) return "";
|
---|
1872 | else return "</a>";
|
---|
1873 | }
|
---|
1874 | case comHref:
|
---|
1875 | return get_href(options["link"]);
|
---|
1876 | case comIcon:
|
---|
1877 | return options["icon"];
|
---|
1878 | case comNum:
|
---|
1879 | return docinfo.result_num;
|
---|
1880 | case comRel: //if [RelatedDocuments] appears in format string, collect relation data
|
---|
1881 | return get_related_docs(collection, collectproto, docinfo, logout);
|
---|
1882 |
|
---|
1883 | case comSummary:
|
---|
1884 | return format_summary(collection, collectproto, docinfo, disp, options, logout);
|
---|
1885 | case comAssocLink:
|
---|
1886 | {
|
---|
1887 | text_t link_filename = get_meta(collection, collectproto, docinfo, disp, formatlistptr->meta, options, logout);
|
---|
1888 | if (!link_filename.empty()) {
|
---|
1889 | text_t href= expand_metadata(options["assocfilepath"], collection, collectproto, docinfo, disp, options, logout) + link_filename;
|
---|
1890 | if (formatlistptr->text == "href") {
|
---|
1891 | return href;
|
---|
1892 | }
|
---|
1893 | return "<a href=\""+ href + "\">";
|
---|
1894 | }
|
---|
1895 | return "";
|
---|
1896 | }
|
---|
1897 | case comEndAssocLink:
|
---|
1898 | {
|
---|
1899 | text_t link_filename = get_meta(collection, collectproto, docinfo, disp, formatlistptr->meta, options, logout);
|
---|
1900 | if (!link_filename.empty()) {
|
---|
1901 | return "</a>";
|
---|
1902 | }
|
---|
1903 | return "";
|
---|
1904 | }
|
---|
1905 | case comMeta:
|
---|
1906 | {
|
---|
1907 | const text_t& metavalue = get_meta (collection,collectproto, docinfo, disp,formatlistptr->meta,options,logout);
|
---|
1908 | return expand_metadata(metavalue, collection, collectproto, docinfo, disp, options, logout);
|
---|
1909 | }
|
---|
1910 |
|
---|
1911 | case comDoc:
|
---|
1912 | return format_text(collection, collectproto, docinfo, disp, options, logout);
|
---|
1913 |
|
---|
1914 | case comImage:
|
---|
1915 | return expand_metadata(options["DocImage"], collection, collectproto, docinfo, disp, options, logout);
|
---|
1916 | case comTOC:
|
---|
1917 | return options["DocTOC"];
|
---|
1918 | case comDocumentButtonDetach:
|
---|
1919 | return options["DocumentButtonDetach"];
|
---|
1920 | case comDocumentButtonHighlight:
|
---|
1921 | return options["DocumentButtonHighlight"];
|
---|
1922 | case comDocumentButtonExpandContents:
|
---|
1923 | return options["DocumentButtonExpandContents"];
|
---|
1924 | case comDocumentButtonExpandText:
|
---|
1925 | return options["DocumentButtonExpandText"];
|
---|
1926 | case comHighlight:
|
---|
1927 | if (options["highlight"] == "1") return "<b>";
|
---|
1928 | break;
|
---|
1929 | case comEndHighlight:
|
---|
1930 | if (options["highlight"] == "1") return "</b>";
|
---|
1931 | break;
|
---|
1932 | case comMetadataSpanWrap:
|
---|
1933 | metadata_wrap=true; metadata_wrap_type="span"; return "";
|
---|
1934 | break;
|
---|
1935 | case comEndMetadataSpanWrap:
|
---|
1936 | metadata_wrap=false; metadata_wrap_type=""; return "";
|
---|
1937 | break;
|
---|
1938 | case comMetadataDivWrap:
|
---|
1939 | metadata_wrap=true; metadata_wrap_type="div"; return "";
|
---|
1940 | break;
|
---|
1941 | case comEndMetadataDivWrap:
|
---|
1942 | metadata_wrap=false; metadata_wrap_type=""; return "";
|
---|
1943 | break;
|
---|
1944 | case comIf:
|
---|
1945 | return get_if (collection, collectproto, docinfo, disp,
|
---|
1946 | formatlistptr->decision, formatlistptr->ifptr,
|
---|
1947 | formatlistptr->elseptr, options, logout);
|
---|
1948 | case comOr:
|
---|
1949 | return get_or (collection,collectproto, docinfo, disp, formatlistptr->orptr,
|
---|
1950 | options, logout);
|
---|
1951 | case comDocTermsFreqTotal:
|
---|
1952 | return docinfo.num_terms_matched;
|
---|
1953 | case comCollection:
|
---|
1954 | if (formatlistptr->meta.metaname == g_EmptyText) {
|
---|
1955 | return collection;
|
---|
1956 | }
|
---|
1957 | return get_collection_meta(collection, collectproto, disp, formatlistptr->meta.metaname, logout);
|
---|
1958 |
|
---|
1959 | }
|
---|
1960 | return "";
|
---|
1961 | }
|
---|
1962 |
|
---|
1963 | text_t get_formatted_string (const text_t& collection, recptproto* collectproto,
|
---|
1964 | ResultDocInfo_t &docinfo, displayclass &disp,
|
---|
1965 | format_t *formatlistptr, text_tmap &options,
|
---|
1966 | ostream& logout) {
|
---|
1967 |
|
---|
1968 | text_t ft;
|
---|
1969 | while (formatlistptr != NULL)
|
---|
1970 | {
|
---|
1971 | ft += format_string (collection, collectproto, docinfo, disp, formatlistptr,
|
---|
1972 | options, logout);
|
---|
1973 | formatlistptr = formatlistptr->nextptr;
|
---|
1974 | }
|
---|
1975 |
|
---|
1976 | return ft;
|
---|
1977 | }
|
---|
1978 |
|
---|
1979 |
|
---|
1980 | // we have only preloaded the text in DocumentAction. But you may want
|
---|
1981 | // to get the text in query, so copy what we have done with
|
---|
1982 | // format_summary and get the text here. Probably is quite expensive?
|
---|
1983 | text_t format_text (const text_t& collection, recptproto* collectproto,
|
---|
1984 | ResultDocInfo_t &docinfo, displayclass &disp,
|
---|
1985 | text_tmap &options, ostream& logout)
|
---|
1986 | {
|
---|
1987 | text_t text;
|
---|
1988 |
|
---|
1989 | if (!options["text"].empty()) {
|
---|
1990 | text = options["text"];
|
---|
1991 | }
|
---|
1992 | else {
|
---|
1993 | // get document text here
|
---|
1994 | DocumentRequest_t docrequest;
|
---|
1995 | DocumentResponse_t docresponse;
|
---|
1996 | comerror_t err;
|
---|
1997 | docrequest.OID = docinfo.OID;
|
---|
1998 | collectproto->get_document (collection, docrequest, docresponse, err, logout);
|
---|
1999 | text = docresponse.doc;
|
---|
2000 | }
|
---|
2001 |
|
---|
2002 | if (metadata_wrap) {
|
---|
2003 | text = wrap_metatext(text,docinfo.OID,"Text");
|
---|
2004 | }
|
---|
2005 |
|
---|
2006 | return text;
|
---|
2007 | }
|
---|
2008 |
|
---|
2009 | /* FUNCTION NAME: format_summary
|
---|
2010 | * DESC: this is invoked when a [Summary] special metadata is processed.
|
---|
2011 | * RETURNS: a query-biased summary for the document */
|
---|
2012 |
|
---|
2013 | text_t format_summary (const text_t& collection, recptproto* collectproto,
|
---|
2014 | ResultDocInfo_t &docinfo, displayclass &disp,
|
---|
2015 | text_tmap &options, ostream& logout) {
|
---|
2016 |
|
---|
2017 | // GRB: added code here to ensure that the cstr (and other collections)
|
---|
2018 | // uses the document metadata item Summary, rather than compressing
|
---|
2019 | // the text of the document, processed via the methods in
|
---|
2020 | // summarise.cpp
|
---|
2021 |
|
---|
2022 | text_t summary;
|
---|
2023 |
|
---|
2024 | if (docinfo.metadata.count("Summary") > 0 &&
|
---|
2025 | docinfo.metadata["Summary"].values.size() > 0) {
|
---|
2026 | summary = docinfo.metadata["Summary"].values[0];
|
---|
2027 | }
|
---|
2028 | else {
|
---|
2029 |
|
---|
2030 | text_t textToSummarise, query;
|
---|
2031 |
|
---|
2032 | if(options["text"].empty()) { // get document text
|
---|
2033 | DocumentRequest_t docrequest;
|
---|
2034 | DocumentResponse_t docresponse;
|
---|
2035 | comerror_t err;
|
---|
2036 | docrequest.OID = docinfo.OID;
|
---|
2037 | collectproto->get_document (collection, docrequest, docresponse, err, logout);
|
---|
2038 | textToSummarise = docresponse.doc;
|
---|
2039 | }
|
---|
2040 | else {
|
---|
2041 | // in practice, this would not happen, because text is only
|
---|
2042 | // loaded with the [Text] command
|
---|
2043 | textToSummarise = options["text"];
|
---|
2044 | }
|
---|
2045 |
|
---|
2046 | disp.expandstring("_cgiargq_",query);
|
---|
2047 | summary = summarise(textToSummarise,query,80);
|
---|
2048 | //summary = substr(textToSummarise.begin(),textToSummarise.begin()+80);
|
---|
2049 | }
|
---|
2050 |
|
---|
2051 | summary.replace("'","'");
|
---|
2052 | summary.replace("\n","
");
|
---|
2053 |
|
---|
2054 | if (metadata_wrap) {
|
---|
2055 | summary = wrap_metatext(summary,docinfo.OID,"Summary");
|
---|
2056 | }
|
---|
2057 |
|
---|
2058 | return summary;
|
---|
2059 | }
|
---|
2060 |
|
---|
2061 | //-------------- GS3 related functions --------------
|
---|
2062 | // copy of the other uses_expression function, but without using the extra GS2-runtime-specific parameters
|
---|
2063 | static bool uses_expression(const text_t& outstring, text_t& lhs_expr,
|
---|
2064 | text_t& op_expr, text_t& rhs_expr)
|
---|
2065 | {
|
---|
2066 | // Note: the string may not be of the form: str1 op str2, however
|
---|
2067 | // to deterine this we have to process it on the assumption it is,
|
---|
2068 | // and if at any point an 'erroneous' value is encountered, return
|
---|
2069 | // false and let something else have a go at evaluating it
|
---|
2070 |
|
---|
2071 | // Starting at the end of the string and working backwards ..
|
---|
2072 |
|
---|
2073 | const int outstring_len = outstring.size();
|
---|
2074 |
|
---|
2075 | // skip over white space
|
---|
2076 | int rhs_end = rscan_over_whitespace(outstring,outstring_len-1);
|
---|
2077 |
|
---|
2078 | if (rhs_end<=0) {
|
---|
2079 | // no meaningful text or (rhs_end==0) no room for operator
|
---|
2080 | return false;
|
---|
2081 | }
|
---|
2082 |
|
---|
2083 | // check for ' or " and then scan over token
|
---|
2084 | const char potential_quote = outstring[rhs_end];
|
---|
2085 | int rhs_start=rhs_end;
|
---|
2086 | bool quoted = false;
|
---|
2087 |
|
---|
2088 | if ((potential_quote == '\'') || (potential_quote == '\"')) {
|
---|
2089 | --rhs_end;
|
---|
2090 | rhs_start = rscan_for(outstring,rhs_end-1,potential_quote) +1;
|
---|
2091 | quoted = true;
|
---|
2092 | }
|
---|
2093 | else {
|
---|
2094 | rhs_start = rscan_for_whitespace(outstring,rhs_end-1) +1;
|
---|
2095 | }
|
---|
2096 |
|
---|
2097 | if ((rhs_end-rhs_start)<0) {
|
---|
2098 | // no meaningful rhs expression
|
---|
2099 | return false;
|
---|
2100 | }
|
---|
2101 |
|
---|
2102 | // form rhs_expr
|
---|
2103 | rhs_expr = extract_substr(outstring,rhs_start,rhs_end);
|
---|
2104 |
|
---|
2105 | // skip over white space
|
---|
2106 | const int to_whitespace = (quoted) ? 2 : 1;
|
---|
2107 |
|
---|
2108 | int op_end = rscan_over_whitespace(outstring,rhs_start-to_whitespace);
|
---|
2109 | int op_start = rscan_for_whitespace(outstring,op_end-1)+1;
|
---|
2110 |
|
---|
2111 | if ((op_end<0) && (op_start<0)) {
|
---|
2112 | // no meaningful expression operator
|
---|
2113 | return false;
|
---|
2114 | }
|
---|
2115 |
|
---|
2116 | if (op_end-op_start<0) {
|
---|
2117 | // no meaningful expression operator
|
---|
2118 | return false;
|
---|
2119 | }
|
---|
2120 |
|
---|
2121 | op_expr = extract_substr(outstring,op_start,op_end);
|
---|
2122 |
|
---|
2123 |
|
---|
2124 | // check for operator
|
---|
2125 | if ((op_expr != "eq") && (op_expr != "ne") && (op_expr != "gt") &&
|
---|
2126 | (op_expr != "ge") && (op_expr != "lt") && (op_expr != "le") && (op_expr != "==") && (op_expr != "!=") && (op_expr != ">") && (op_expr != ">=") && (op_expr != "<") && (op_expr != "<=") && (op_expr != "sw") && (op_expr != "ew")) {
|
---|
2127 |
|
---|
2128 | // not a valid operator
|
---|
2129 | return false;
|
---|
2130 | }
|
---|
2131 |
|
---|
2132 | int lhs_end = rscan_over_whitespace(outstring,op_start-1);
|
---|
2133 | if (lhs_end<0) {
|
---|
2134 | // no meaningful lhs expression
|
---|
2135 | return false;
|
---|
2136 | }
|
---|
2137 |
|
---|
2138 | int lhs_start = scan_over_whitespace(outstring,0);
|
---|
2139 |
|
---|
2140 | // form lhs_expr from remainder of string
|
---|
2141 | lhs_expr = extract_substr(outstring,lhs_start,lhs_end);
|
---|
2142 |
|
---|
2143 | return true;
|
---|
2144 | }
|
---|
2145 |
|
---|
2146 | // [ex.Title] -> ex.Title
|
---|
2147 | static text_t remove_bracket_bookends(const text_t &str) {
|
---|
2148 |
|
---|
2149 | if(str[0] == '[' && str[str.size()-1] == ']') {
|
---|
2150 | return substr (str.begin()+1, str.end()-1);
|
---|
2151 | } else {
|
---|
2152 | return str;
|
---|
2153 | }
|
---|
2154 | }
|
---|
2155 |
|
---|
2156 | static text_t get_gs3_if (const decision_t &decision, format_t *ifptr, format_t *elseptr, const text_t& nodeType)
|
---|
2157 | {
|
---|
2158 | text_t ifstmt ="<gsf:switch>";
|
---|
2159 |
|
---|
2160 |
|
---|
2161 | if (decision.command == dMeta) {
|
---|
2162 | ifstmt += "<gsf:metadata name=\"";
|
---|
2163 | ifstmt += remove_bracket_bookends(decision.meta.metaname);
|
---|
2164 | ifstmt += "\"/>";
|
---|
2165 | ifstmt += "<gsf:when test=\"exists\">";
|
---|
2166 | }
|
---|
2167 |
|
---|
2168 | else { //if(decision.command == dText)
|
---|
2169 |
|
---|
2170 | text_t outstring = decision.text;
|
---|
2171 |
|
---|
2172 | // Check for if expression in form: str1 op str2
|
---|
2173 | // (such as [x] eq "y")
|
---|
2174 | text_t lhs_expr, op_expr, rhs_expr;
|
---|
2175 | if (uses_expression(outstring,lhs_expr,op_expr,rhs_expr)) {
|
---|
2176 |
|
---|
2177 | text_t if_operator = op_expr;
|
---|
2178 | if (op_expr == "eq" || op_expr == "==") {
|
---|
2179 | if_operator = "equals";
|
---|
2180 | } else if (op_expr == "ne" || op_expr == "!=") {
|
---|
2181 | if_operator = "notEquals";
|
---|
2182 | } else if (op_expr == "gt" || op_expr == ">") {
|
---|
2183 | if_operator = "greaterThan";
|
---|
2184 | } else if (op_expr == "lt" || op_expr == "<") {
|
---|
2185 | if_operator = "lessThan";
|
---|
2186 | } else if (op_expr == "ge" || op_expr == ">=") {
|
---|
2187 | if_operator = "greaterThanOrEquals";
|
---|
2188 | } else if (op_expr == "le" || op_expr == "<=") {
|
---|
2189 | if_operator = "lessThanOrEquals";
|
---|
2190 | } else if (op_expr == "sw") {
|
---|
2191 | if_operator = "startsWith";
|
---|
2192 | } else if (op_expr == "ew") {
|
---|
2193 | if_operator = "endsWith";
|
---|
2194 | }
|
---|
2195 |
|
---|
2196 | ifstmt += "<gsf:metadata name=\"";
|
---|
2197 | ifstmt += remove_bracket_bookends(lhs_expr);
|
---|
2198 | ifstmt += "\"/>";
|
---|
2199 |
|
---|
2200 | ifstmt += "<gsf:when test=\"";
|
---|
2201 | ifstmt += if_operator; // the test operator
|
---|
2202 | ifstmt += "\" test-value=\"";
|
---|
2203 | ifstmt += remove_bracket_bookends(rhs_expr); // the test-value
|
---|
2204 | ifstmt += "\">";
|
---|
2205 | }
|
---|
2206 | else {
|
---|
2207 | ifstmt += "<gsf:metadata name=\"";
|
---|
2208 | ifstmt += remove_bracket_bookends(decision.text);
|
---|
2209 | ifstmt += "\"/>";
|
---|
2210 | ifstmt += "<gsf:when test=\"exists\">";
|
---|
2211 | }
|
---|
2212 | }
|
---|
2213 |
|
---|
2214 | // if portion
|
---|
2215 | text_t if_body = "";
|
---|
2216 | while(ifptr != NULL) { // body of if can contain a list of items to be transformed into GS3 format stmts
|
---|
2217 | if_body += transform_to_GS3_format (ifptr, nodeType);
|
---|
2218 | ifptr = ifptr->nextptr;
|
---|
2219 | }
|
---|
2220 | ifstmt += if_body;
|
---|
2221 | ifstmt += "</gsf:when>";
|
---|
2222 |
|
---|
2223 | // else portion
|
---|
2224 | if(elseptr != NULL) {
|
---|
2225 |
|
---|
2226 | ifstmt += "<gsf:otherwise>";
|
---|
2227 | text_t else_body = ""; // body of else can contain a list of items to be transformed into GS3 format stmts
|
---|
2228 | while(elseptr != NULL) {
|
---|
2229 | else_body += transform_to_GS3_format (elseptr, nodeType);
|
---|
2230 | elseptr = elseptr->nextptr;
|
---|
2231 | }
|
---|
2232 | ifstmt += else_body;
|
---|
2233 | ifstmt += "</gsf:otherwise>";
|
---|
2234 | }
|
---|
2235 |
|
---|
2236 | ifstmt += "</gsf:switch>";
|
---|
2237 | return ifstmt;
|
---|
2238 | }
|
---|
2239 |
|
---|
2240 |
|
---|
2241 | static text_t get_gs3_or (format_t *orptr, const text_t& nodeType) {
|
---|
2242 | text_t result = "<gsf:choose-metadata>";
|
---|
2243 |
|
---|
2244 | while(orptr != NULL) {
|
---|
2245 | text_t or_body = transform_to_GS3_format (orptr, nodeType);
|
---|
2246 | if (!or_body.empty()) {
|
---|
2247 | result += or_body;
|
---|
2248 | }
|
---|
2249 |
|
---|
2250 | orptr = orptr->nextptr;
|
---|
2251 | }
|
---|
2252 | result += "</gsf:choose-metadata>";
|
---|
2253 | return result;
|
---|
2254 | }
|
---|
2255 |
|
---|
2256 | // what about all the <td>? Does that get stored in formatlistptr, such as under the ->text field?
|
---|
2257 | text_t get_GS3_formatstring (format_t *formatlistptr, const text_t& nodeType) {
|
---|
2258 | text_t result;
|
---|
2259 |
|
---|
2260 | while (formatlistptr != NULL) {
|
---|
2261 | result += transform_to_GS3_format(formatlistptr, nodeType);
|
---|
2262 | formatlistptr = formatlistptr->nextptr;
|
---|
2263 | }
|
---|
2264 |
|
---|
2265 | return result;
|
---|
2266 | }
|
---|
2267 |
|
---|
2268 | text_t transform_to_GS3_format (format_t *formatlistptr, const text_t& nodeType) {
|
---|
2269 | if (formatlistptr == NULL) return "";
|
---|
2270 |
|
---|
2271 | switch (formatlistptr->command) {
|
---|
2272 | case comOID:
|
---|
2273 | return "<gsf:OID/>";
|
---|
2274 | case comTopOID:
|
---|
2275 | return "<gsf:metadata name='OID' select='root' />"; // for now try this
|
---|
2276 | case comRank:
|
---|
2277 | return "<gsf:rank/>";
|
---|
2278 | case comText:
|
---|
2279 | return formatlistptr->text; // [text]? or any string that is not a command or reserved
|
---|
2280 | case comLink:
|
---|
2281 | if(nodeType == "classifier") {
|
---|
2282 | return "<gsf:link type='classifier'>";
|
---|
2283 | } else { // if nodeType is document or not set
|
---|
2284 | return "<gsf:link type='document'>";
|
---|
2285 | }
|
---|
2286 | case comEndLink:
|
---|
2287 | return "</gsf:link>";
|
---|
2288 | case comHref:
|
---|
2289 | return "<gsf:lib name=\"href\"/>"; // ??? in gslib xsl, output comment marking not sure what this maps to
|
---|
2290 | case comIcon:
|
---|
2291 | if(nodeType == "classifier") {
|
---|
2292 | return "<gsf:icon type='classifier'/>";
|
---|
2293 | } else { // if nodeType is document or not set
|
---|
2294 | return "<gsf:icon type='document'/>";
|
---|
2295 | }
|
---|
2296 | case comNum:
|
---|
2297 | return "<gsf:lib name=\"num\"/>"; // ??? in gslib xsl, output comment marking not sure what this maps to
|
---|
2298 | case comRel: //if [RelatedDocuments] appears in format string, collect relation data
|
---|
2299 | return "<gsf:lib name=\"RelatedDocuments\"/>"; // output comment marking this as deprecated or to be implemented for GS3 in gslib xslt
|
---|
2300 | case comSummary:
|
---|
2301 | return "<gsf:lib name=\"Summary\"/>"; // in gslib xslt output comment marking this as to be implemented for GS3
|
---|
2302 | // need to invent this for GS3 based on what GS2 does
|
---|
2303 | case comAssocLink:
|
---|
2304 | return "<gsf:link type='source'>";
|
---|
2305 | case comEndAssocLink:
|
---|
2306 | return "</gsf:link>";
|
---|
2307 | case comMeta:
|
---|
2308 | return "<gsf:metadata name=\"" + formatlistptr->meta.metaname + "\" />";//?
|
---|
2309 | case comDoc:
|
---|
2310 | return "<gsf:text/>";
|
---|
2311 | case comImage: // the cover img seems to be handled by some magic code in GS3
|
---|
2312 | return "<gsf:lib name=\"image\"/>"; // in gslib xslt output a comment marking this as not working in the same way in GS3 as in GS2. TOC currently not relocatable
|
---|
2313 | case comTOC:
|
---|
2314 | return "<gsf:lib name=\"TOC\"/>"; // in gslib xslt output a comment marking this as not working in the same way in GS3 as in GS2. TOC currently not relocatable
|
---|
2315 | // need to think about whether an equivalent actually exists
|
---|
2316 | // return "<gsf:option name=\"TOC\" value=\"true\"/>"; // this is wrong
|
---|
2317 | case comDocumentButtonDetach:
|
---|
2318 | return "<gsf:lib name=\"DocumentButtonDetach\"/>"; // output comment marking this as deprecated in gslib xslt
|
---|
2319 | case comDocumentButtonHighlight:
|
---|
2320 | return "<gsf:lib name=\"DocumentButtonHighlight\"/>"; // output comment marking this as deprecated in gslib xslt
|
---|
2321 | case comDocumentButtonExpandContents:
|
---|
2322 | return "<gsf:lib name=\"DocumentButtonExpandContents\"/>"; // output comment marking this as deprecated in gslib xslt
|
---|
2323 | case comDocumentButtonExpandText:
|
---|
2324 | return "<gsf:lib name=\"DocumentButtonExpandText\"/>"; // output comment marking this as deprecated in gslib xslt
|
---|
2325 | case comHighlight:
|
---|
2326 | return "<span class=\"highlight\">";
|
---|
2327 | break;
|
---|
2328 | case comEndHighlight:
|
---|
2329 | return "</span>";
|
---|
2330 | break;
|
---|
2331 | case comMetadataSpanWrap:
|
---|
2332 | metadata_wrap=true; metadata_wrap_type="span"; return "";
|
---|
2333 | break;
|
---|
2334 | case comEndMetadataSpanWrap:
|
---|
2335 | metadata_wrap=false; metadata_wrap_type=""; return "";
|
---|
2336 | break;
|
---|
2337 | case comMetadataDivWrap:
|
---|
2338 | metadata_wrap=true; metadata_wrap_type="div"; return "";
|
---|
2339 | break;
|
---|
2340 | case comEndMetadataDivWrap:
|
---|
2341 | metadata_wrap=false; metadata_wrap_type=""; return "";
|
---|
2342 | break;
|
---|
2343 | case comIf:
|
---|
2344 | if(formatlistptr->decision.meta.metaname == "numleafdocs") {
|
---|
2345 | if(nodeType == "classifier") {
|
---|
2346 | text_t if_body = "";
|
---|
2347 | while(formatlistptr->ifptr != NULL) { // body of if can contain a list of items to be transformed into GS3 format stmts
|
---|
2348 | if_body += transform_to_GS3_format (formatlistptr->ifptr, nodeType);
|
---|
2349 | formatlistptr->ifptr = formatlistptr->ifptr->nextptr;
|
---|
2350 | }
|
---|
2351 | return if_body;
|
---|
2352 | } else if(nodeType == "document") {
|
---|
2353 | text_t else_body = ""; // body of else can contain a list of items to be transformed into GS3 format stmts
|
---|
2354 | while(formatlistptr->elseptr != NULL) {
|
---|
2355 | else_body += transform_to_GS3_format (formatlistptr->elseptr, nodeType);
|
---|
2356 | formatlistptr->elseptr = formatlistptr->elseptr->nextptr;
|
---|
2357 | }
|
---|
2358 | return else_body;
|
---|
2359 | }
|
---|
2360 | }
|
---|
2361 |
|
---|
2362 | // if nodeType not specified as classifier or document, or if the If test is not for numleafdocs' existence
|
---|
2363 | return get_gs3_if (formatlistptr->decision, formatlistptr->ifptr, formatlistptr->elseptr, nodeType);
|
---|
2364 | case comOr:
|
---|
2365 | return get_gs3_or (formatlistptr->orptr, nodeType);
|
---|
2366 | //return "<gsf:choose-metadata>"+get_gs3_or (formatlistptr->orptr, nodeType)+"</gsf:choose-metadata>";
|
---|
2367 | case comDocTermsFreqTotal:
|
---|
2368 | return "<gsf:lib name=\"DocTermsFreqTotal\"/>";
|
---|
2369 | case comCollection: // trying to get all the metadata for a collection. How is this done in GS3???
|
---|
2370 | return "<gsf:lib name=\"collection\"/>";
|
---|
2371 | }
|
---|
2372 | return "";
|
---|
2373 | }
|
---|