source: gsdl/trunk/runtime-src/src/recpt/formattools.cpp@ 19046

Last change on this file since 19046 was 19046, checked in by davidb, 15 years ago

Qualifiers within metadata can now support a chain of parent: and child: terms, e.g. parent:child(All)
  • Property svn:executable set to *
  • Property svn:keywords set to Author Date Id Revision
File size: 48.8 KB
Line 
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 "recptprototools.h"
29#include "OIDtools.h"
30#include "summarise.h"
31
32#include <assert.h>
33
34// a few function prototypes
35
36static text_t format_string (const text_t& collection, recptproto* collectproto,
37 ResultDocInfo_t &docinfo, displayclass &disp,
38 format_t *formatlistptr, text_tmap &options,
39 ostream& logout);
40
41static bool parse_action (text_t::const_iterator &here, const text_t::const_iterator &end,
42 format_t *formatlistptr, text_tset &metadata, bool &getParents);
43
44static text_t format_summary (const text_t& collection, recptproto* collectproto,
45 ResultDocInfo_t &docinfo, displayclass &disp,
46 text_tmap &options, ostream& logout);
47static text_t format_text (const text_t& collection, recptproto* collectproto,
48 ResultDocInfo_t &docinfo, displayclass &disp,
49 text_tmap &options, ostream& logout);
50
51static text_t expand_metadata(const text_t &metavalue, const text_t& collection,
52 recptproto* collectproto, ResultDocInfo_t &docinfo,
53 displayclass &disp, text_tmap &options,
54 ostream &logout);
55
56
57void metadata_t::clear() {
58 metaname.clear();
59 metacommand = mNone;
60 mqualifier.parent = pNone;
61 mqualifier.sibling = sNone;
62 mqualifier.child = cNone;
63 pre_tree_traverse.clear();
64 parentoptions.clear();
65 siblingoptions.clear();
66 childoptions.clear();
67}
68
69void decision_t::clear() {
70 command = dMeta;
71 meta.clear();
72 text.clear();
73}
74
75void format_t::clear() {
76 command = comText;
77 decision.clear();
78 text.clear();
79 meta.clear();
80 nextptr = NULL;
81 ifptr = NULL;
82 elseptr = NULL;
83 orptr = NULL;
84}
85
86void formatinfo_t::clear() {
87 DocumentImages = false;
88 DocumentTitles = true;
89 DocumentHeading = "{Or}{[parent(Top):Title],[Title],untitled}<br>";
90 DocumentContents = true;
91 DocumentArrowsBottom = true;
92 DocumentArrowsTop = false;
93 DocumentSearchResultLinks = false;
94 DocumentButtons.erase (DocumentButtons.begin(), DocumentButtons.end());
95 // DocumentButtons.push_back ("Expand Text");
96 // DocumentButtons.push_back ("Expand Contents");
97 DocumentButtons.push_back ("Detach");
98 DocumentButtons.push_back ("Highlight");
99 RelatedDocuments = "";
100 DocumentText = "[Text]";
101 formatstrings.erase (formatstrings.begin(), formatstrings.end());
102 DocumentUseHTML = false;
103 AllowExtendedOptions = false;
104}
105
106// simply checks to see if formatstring begins with a <td> tag
107bool is_table_content (const text_t &formatstring) {
108 text_t::const_iterator here = formatstring.begin();
109 text_t::const_iterator end = formatstring.end();
110
111 while (here != end) {
112 if (*here != ' ') {
113 if ((*here == '<') && ((here+3) < end)) {
114 if ((*(here+1) == 't' || *(here+1) == 'T') &&
115 (*(here+2) == 'd' || *(here+2) == 'D') &&
116 (*(here+3) == '>' || *(here+3) == ' '))
117 return true;
118 } else return false;
119 }
120 ++here;
121 }
122 return false;
123}
124
125bool is_table_content (const format_t *formatlistptr) {
126
127 if (formatlistptr == NULL) return false;
128
129 if (formatlistptr->command == comText)
130 return is_table_content (formatlistptr->text);
131
132 return false;
133}
134
135// returns false if key isn't in formatstringmap
136bool get_formatstring (const text_t &key, const text_tmap &formatstringmap,
137 text_t &formatstring) {
138
139 formatstring.clear();
140 text_tmap::const_iterator it = formatstringmap.find(key);
141 if (it == formatstringmap.end()) return false;
142 formatstring = (*it).second;
143 return true;
144}
145
146// tries to find "key1key2" then "key1" then "key2"
147bool get_formatstring (const text_t &key1, const text_t &key2,
148 const text_tmap &formatstringmap,
149 text_t &formatstring) {
150
151 formatstring.clear();
152 text_tmap::const_iterator it = formatstringmap.find(key1 + key2);
153 if (it != formatstringmap.end()) {
154 formatstring = (*it).second;
155 return true;
156 }
157 it = formatstringmap.find(key1);
158 if (it != formatstringmap.end()) {
159 formatstring = (*it).second;
160 return true;
161 }
162 it = formatstringmap.find(key2);
163 if (it != formatstringmap.end()) {
164 formatstring = (*it).second;
165 return true;
166 }
167 return false;
168}
169
170
171text_t remove_namespace(const text_t &meta_name) {
172 text_t::const_iterator end = meta_name.end();
173 text_t::const_iterator it = findchar(meta_name.begin(), end, '.');
174 if (it != end) {
175 return substr(it+1, end);
176 }
177
178 return meta_name;
179
180}
181// returns a date of form _format:date_(year, month, day)
182// input is date of type yyyy-?mm-?dd
183// at least the year must be present in date
184text_t format_date (const text_t &date) {
185
186 if (date.size() < 4) return "";
187
188 text_t::const_iterator datebegin = date.begin();
189
190 text_t year = substr (datebegin, datebegin+4);
191 int chars_seen_so_far = 4;
192
193 if (date[chars_seen_so_far] == '-') ++chars_seen_so_far ;
194 if (date.size() < chars_seen_so_far+2) return "_format:date_("+year+")";
195
196 text_t month = substr (datebegin+chars_seen_so_far, datebegin+chars_seen_so_far+2);
197 int imonth = month.getint();
198 if (imonth <= 0 || imonth > 12) return "_format:date_("+year+")";
199
200 chars_seen_so_far += 2;
201 if (date[chars_seen_so_far] == '-') ++chars_seen_so_far ;
202
203 if (date.size() < chars_seen_so_far+2) return "_format:date_("+year+","+month+")";
204
205 text_t day = substr (datebegin+chars_seen_so_far, datebegin+chars_seen_so_far+2);
206 if (day[0] == '0') day = substr (day.begin()+1, day.end());
207 int iday = day.getint();
208 if (iday <= 0 || iday > 31) return "_format:date_("+year+","+month+")";
209
210 return "_format:date_("+year+","+month+","+day+")";
211}
212
213// converts an iso639 language code to its English equivalent
214// should we be checking that the macro exists??
215text_t iso639 (const text_t &langcode) {
216 if (langcode.empty()) return "";
217 return "_iso639:iso639"+langcode+"_";
218}
219
220
221text_t get_href (const text_t &link) {
222
223 text_t href;
224
225 text_t::const_iterator here = findchar(link.begin(), link.end(), '"');
226 text_t::const_iterator end = link.end();
227 if (here == end) return g_EmptyText;
228
229 ++here;
230 while (here != end) {
231 if (*here == '"') break;
232 href.push_back(*here);
233 ++here;
234 }
235
236 return href;
237}
238
239//this function gets the information associated with the relation
240//metadata for the document associated with 'docinfo'. This relation
241//metadata consists of a line of pairs containing 'collection, document OID'
242//(this is the OID of the document related to the current document, and
243//the collection the related document belongs to). For each of these pairs
244//the title metadata is obtained and then an html link between the title
245//of the related doc and the document's position (the document will be
246//found in "<a href=\"_httpdocument_&c=collection&cl=search&d=OID">
247//(where collection is the related documents collection, and OID is the
248//related documents OID). A list of these html links are made for as many
249//related documents as there are. This list is then returned. If there are
250//no related documents available for the current document then the string
251//'.. no related documents .. ' is returned.
252text_t get_related_docs(const text_t& collection, recptproto* collectproto,
253 ResultDocInfo_t &docinfo, ostream& logout){
254
255 text_tset metadata;
256
257 //insert the metadata we wish to collect
258 metadata.insert("dc.Relation");
259 metadata.insert("Title");
260 metadata.insert("Subject"); //for emails, where title data doesn't apply
261
262 FilterResponse_t response;
263 text_t relation = ""; //string for displaying relation metadata
264 text_t relationTitle = ""; //the related documents Title (or subject)
265 text_t relationOID = ""; //the related documents OID
266
267 //get the information associated with the metadata for current doc
268 if (get_info (docinfo.OID, collection, "", metadata,
269 false, collectproto, response, logout)) {
270
271 //if the relation metadata exists, store for displaying
272 if(!response.docInfo[0].metadata["dc.Relation"].values.empty()){
273 relationOID += response.docInfo[0].metadata["dc.Relation"].values[0];
274
275 //split relation data into pairs of collectionname,ID number
276 text_tarray relationpairs;
277 splitchar (relationOID.begin(), relationOID.end(), ' ', relationpairs);
278
279 text_tarray::const_iterator currDoc = relationpairs.begin();
280 text_tarray::const_iterator lastDoc = relationpairs.end();
281
282 //iterate through the pairs to split and display
283 while(currDoc != lastDoc){
284
285 //split pairs into collectionname and ID
286 text_tarray relationdata;
287 splitchar ((*currDoc).begin(), (*currDoc).end(), ',', relationdata);
288
289 //get first element in the array (collection)
290 text_tarray::const_iterator doc_data = relationdata.begin();
291 text_t document_collection = *doc_data;
292 ++doc_data; //increment to get next item in array (oid)
293 text_t document_OID = *doc_data;
294
295 //create html link to related document
296 relation += "<a href=\"_httpdocument_&c=" + document_collection;
297 relation += "&cl=search&d=" + document_OID;
298
299 //get the information associated with the metadata for related doc
300 if (get_info (document_OID, document_collection, "", metadata,
301 false, collectproto, response, logout)) {
302
303 //if title metadata doesn't exist, collect subject metadata
304 //if that doesn't exist, just call it 'related document'
305 if (!response.docInfo[0].metadata["Title"].values[0].empty())
306 relationTitle = response.docInfo[0].metadata["Title"].values[0];
307 else if (!response.docInfo[0].metadata["Subject"].values.empty())
308 relationTitle = response.docInfo[0].metadata["Subject"].values[0];
309 else relationTitle = "RELATED DOCUMENT";
310
311 }
312
313 //link the related document's title to its page
314 relation += "\">" + relationTitle + "</a>";
315 relation += " (" + document_collection + ")<br>";
316
317 ++currDoc;
318 }
319 }
320
321 }
322
323 if(relation.empty()) //no relation data for documnet
324 relation = ".. no related documents .. ";
325
326 return relation;
327}
328
329
330
331static void get_parent_options (text_t &instring, metadata_t &metaoption) {
332
333 assert (instring.size() > 7);
334 if (instring.size() <= 7) return;
335
336 text_t meta, com, op;
337 bool inbraces = false;
338 bool inquotes = false;
339 bool foundcolon = false;
340 text_t::const_iterator here = instring.begin()+6;
341 text_t::const_iterator end = instring.end();
342 while (here != end) {
343 if (foundcolon) meta.push_back (*here);
344 else if (*here == '(') inbraces = true;
345 else if (*here == ')') inbraces = false;
346 else if (*here == '\'' && !inquotes) inquotes = true;
347 else if (*here == '\'' && inquotes) inquotes = false;
348 else if (*here == ':' && !inbraces) foundcolon = true;
349 else if (inquotes) op.push_back (*here);
350 else com.push_back (*here);
351 ++here;
352 }
353
354 instring = meta;
355 if (com.empty())
356 metaoption.mqualifier.parent = pImmediate;
357 else if (com == "Top")
358 metaoption.mqualifier.parent = pTop;
359 else if (com == "All") {
360 metaoption.mqualifier.parent = pAll;
361 metaoption.parentoptions = op;
362 }
363}
364
365
366static void get_sibling_options (text_t &instring, metadata_t &metaoption) {
367
368 assert (instring.size() > 8);
369 if (instring.size() <= 8) return;
370 text_t meta, com, op;
371 bool inbraces = false;
372 bool inquotes = false;
373 bool foundcolon = false;
374 text_t::const_iterator here = instring.begin()+7;
375 text_t::const_iterator end = instring.end();
376 while (here != end) {
377 if (foundcolon) meta.push_back (*here);
378 else if (*here == '(') inbraces = true;
379 else if (*here == ')') inbraces = false;
380 else if (*here == '\'' && !inquotes) inquotes = true;
381 else if (*here == '\'' && inquotes) inquotes = false;
382 else if (*here == ':' && !inbraces) foundcolon = true;
383 else if (inquotes) op.push_back (*here);
384 else com.push_back (*here);
385 ++here;
386 }
387
388 instring = meta;
389 metaoption.siblingoptions.clear();
390
391 if (com.empty()) {
392 metaoption.mqualifier.sibling = sAll;
393 metaoption.siblingoptions = " ";
394 }
395 else if (com == "first") {
396 metaoption.mqualifier.sibling = sNum;
397 metaoption.siblingoptions = "0";
398 }
399 else if (com == "last") {
400 metaoption.mqualifier.sibling = sNum;
401 metaoption.siblingoptions = "-2"; // == last
402 }
403 else if (com.getint()>0) {
404 metaoption.mqualifier.sibling = sNum;
405 int pos = com.getint()-1;
406 metaoption.siblingoptions +=pos;
407 }
408 else {
409 metaoption.mqualifier.sibling = sAll;
410 metaoption.siblingoptions = op;
411 }
412}
413
414static void get_child_options (text_t &instring, metadata_t &metaoption) {
415
416 assert (instring.size() > 6);
417 if (instring.size() <= 6) return;
418 text_t meta, com, op;
419 bool inbraces = false;
420 bool inquotes = false;
421 bool foundcolon = false;
422 text_t::const_iterator here = instring.begin()+5;
423 text_t::const_iterator end = instring.end();
424 while (here != end) {
425 if (foundcolon) meta.push_back (*here);
426 else if (*here == '(') inbraces = true;
427 else if (*here == ')') inbraces = false;
428 else if (*here == '\'' && !inquotes) inquotes = true;
429 else if (*here == '\'' && inquotes) inquotes = false;
430 else if (*here == ':' && !inbraces) foundcolon = true;
431 else if (inquotes) op.push_back (*here);
432 else com.push_back (*here);
433 ++here;
434 }
435
436 instring = meta;
437 if (com.empty()) {
438 metaoption.mqualifier.child = cAll;
439 metaoption.childoptions = " ";
440 }
441 else if (com == "first") {
442 metaoption.mqualifier.child = cNum;
443 metaoption.childoptions = ".fc";
444 }
445 else if (com == "last") {
446 metaoption.mqualifier.child = cNum;
447 metaoption.childoptions = ".lc";
448 }
449 else if (com.getint()>0) {
450 metaoption.mqualifier.child = cNum;
451 metaoption.childoptions = "."+com;
452 }
453 else {
454 metaoption.mqualifier.child = cAll;
455 metaoption.childoptions = op;
456 }
457}
458
459
460
461static void parse_meta (text_t &meta, metadata_t &metaoption,
462 text_tset &metadata, bool &getParents) {
463
464 if (meta.size() > 8 && (substr(meta.begin(), meta.begin()+8) == "cgisafe:")) {
465 metaoption.metacommand |= mCgiSafe;
466 meta = substr (meta.begin()+8, meta.end());
467 }
468 if (meta.size() > 7 && (substr(meta.begin(), meta.begin()+7) == "format:")) {
469 metaoption.metacommand |= mSpecial;
470 meta = substr (meta.begin()+7, meta.end());
471 }
472
473 bool had_parent_or_child = true;
474 bool prev_was_parent = false;
475 bool prev_was_child = false;
476
477 while (had_parent_or_child) {
478 if (meta.size() > 7
479 && (substr (meta.begin(), meta.begin()+6) == "parent")) {
480
481 // clear out sibling and child (cmd and options)
482 metaoption.metacommand &= ~(mChild|mSibling);
483 metaoption.childoptions.clear();
484 metaoption.siblingoptions.clear();
485
486 getParents = true;
487 metaoption.metacommand |= mParent;
488 get_parent_options (meta, metaoption);
489
490 if (prev_was_parent) {
491 metaoption.pre_tree_traverse += ".pr";
492 }
493 else if (prev_was_child) {
494 metaoption.pre_tree_traverse += ".fc";
495 }
496
497 prev_was_parent = true;
498 prev_was_child = false;
499 }
500 else if (meta.size() > 6
501 && (substr (meta.begin(), meta.begin()+5) == "child")) {
502
503 // clear out sibling and parent (cmd and options)
504 metaoption.metacommand &= ~(mParent|mSibling);
505 metaoption.parentoptions.clear();
506 metaoption.siblingoptions.clear();
507
508 metaoption.metacommand |= mChild;
509 get_child_options (meta, metaoption);
510 metadata.insert("contains");
511
512 if (prev_was_parent) {
513 metaoption.pre_tree_traverse += ".pr";
514 }
515 else if (prev_was_child) {
516 metaoption.pre_tree_traverse += ".fc";
517 }
518
519 prev_was_child = true;
520 prev_was_parent = false;
521 }
522 else {
523 prev_was_child = false;
524 prev_was_parent = false;
525 had_parent_or_child = false;
526 }
527 }
528
529 // parent/child can have sibling tacked on end also
530 if (meta.size() > 8 && (substr (meta.begin(), meta.begin()+7) == "sibling")) {
531 metaoption.metacommand |= mSibling;
532 get_sibling_options (meta, metaoption);
533 }
534
535 // check for ex. which may occur in format statements
536 if (meta.size()>3 && (substr(meta.begin(), meta.begin()+3) == "ex.")) {
537 meta = substr (meta.begin()+3, meta.end());
538 }
539 metadata.insert (meta);
540 metaoption.metaname = meta;
541}
542
543static void parse_coll_meta(text_t &meta, metadata_t &metaoption) {
544 if (meta == "collection") {
545 // no qualifiers
546 metaoption.metaname = g_EmptyText;
547 return;
548 }
549 meta = substr (meta.begin()+11, meta.end());
550 metaoption.metaname = meta;
551
552}
553
554static void parse_meta (text_t &meta, format_t *formatlistptr,
555 text_tset &metadata, bool &getParents) {
556
557 if (meta == "link")
558 formatlistptr->command = comLink;
559 else if (meta == "/link")
560 formatlistptr->command = comEndLink;
561
562 else if (meta == "href")
563 formatlistptr->command = comHref;
564
565 else if (meta == "num")
566 formatlistptr->command = comNum;
567
568 else if (meta == "icon")
569 formatlistptr->command = comIcon;
570
571 else if (meta == "Text")
572 formatlistptr->command = comDoc;
573
574 else if (meta == "RelatedDocuments")
575 formatlistptr->command = comRel;
576
577 else if (meta == "highlight")
578 formatlistptr->command = comHighlight;
579
580 else if (meta == "/highlight")
581 formatlistptr->command = comEndHighlight;
582
583 else if (meta == "Summary")
584 formatlistptr->command = comSummary;
585
586 else if (meta == "DocImage")
587 formatlistptr->command = comImage;
588
589 else if (meta == "DocTOC")
590 formatlistptr->command = comTOC;
591
592 else if (meta == "DocumentButtonDetach")
593 formatlistptr->command = comDocumentButtonDetach;
594
595 else if (meta == "DocumentButtonHighlight")
596 formatlistptr->command = comDocumentButtonHighlight;
597
598 else if (meta == "DocumentButtonExpandContents")
599 formatlistptr->command = comDocumentButtonExpandContents;
600
601 else if (meta == "DocumentButtonExpandText")
602 formatlistptr->command = comDocumentButtonExpandText;
603
604 else if (meta == "DocOID")
605 formatlistptr->command = comOID;
606 else if (meta == "DocTopOID")
607 formatlistptr->command = comTopOID;
608 else if (meta == "DocRank")
609 formatlistptr->command = comRank;
610 else if (meta == "DocTermsFreqTotal")
611 formatlistptr->command = comDocTermsFreqTotal;
612 else if (meta.size() >= 10 && (substr(meta.begin(), meta.begin()+10) == "collection")) {
613 formatlistptr->command = comCollection;
614 parse_coll_meta(meta, formatlistptr->meta);
615 }
616 else {
617 formatlistptr->command = comMeta;
618 parse_meta (meta, formatlistptr->meta, metadata, getParents);
619 }
620}
621
622
623static bool parse_string (const text_t &formatstring, format_t *formatlistptr,
624 text_tset &metadata, bool &getParents) {
625
626 text_t text;
627 text_t::const_iterator here = formatstring.begin();
628 text_t::const_iterator end = formatstring.end();
629
630 while (here != end) {
631
632 if (*here == '\\') {
633 ++here;
634 if (here != end) text.push_back (*here);
635
636 } else if (*here == '{') {
637 if (!text.empty()) {
638 formatlistptr->command = comText;
639 formatlistptr->text = text;
640 formatlistptr->nextptr = new format_t();
641 formatlistptr = formatlistptr->nextptr;
642
643 text.clear();
644 }
645 if (parse_action (++here, end, formatlistptr, metadata, getParents)) {
646
647 formatlistptr->nextptr = new format_t();
648 formatlistptr = formatlistptr->nextptr;
649 if (here == end) break;
650 }
651 } else if (*here == '[') {
652 if (!text.empty()) {
653 formatlistptr->command = comText;
654 formatlistptr->text = text;
655 formatlistptr->nextptr = new format_t();
656 formatlistptr = formatlistptr->nextptr;
657
658 text.clear();
659 }
660 text_t meta;
661 ++here;
662 while (*here != ']') {
663 if (here == end) return false;
664 meta.push_back (*here);
665 ++here;
666 }
667 parse_meta (meta, formatlistptr, metadata, getParents);
668 formatlistptr->nextptr = new format_t();
669 formatlistptr = formatlistptr->nextptr;
670
671 } else
672 text.push_back (*here);
673
674 if (here != end) ++here;
675 }
676 if (!text.empty()) {
677 formatlistptr->command = comText;
678 formatlistptr->text = text;
679 formatlistptr->nextptr = new format_t();
680 formatlistptr = formatlistptr->nextptr;
681
682 }
683 return true;
684}
685
686
687static bool parse_action (text_t::const_iterator &here, const text_t::const_iterator &end,
688 format_t *formatlistptr, text_tset &metadata, bool &getParents) {
689
690 text_t::const_iterator it = findchar (here, end, '}');
691 if (it == end) return false;
692
693 text_t com = substr (here, it);
694 here = findchar (it, end, '{');
695 if (here == end) return false;
696 else ++here;
697
698 if (com == "If" || com == "if" || com == "IF") formatlistptr->command = comIf;
699 else if (com == "Or" || com == "or" || com == "OR") formatlistptr->command = comOr;
700 else return false;
701
702 int commacount = 0;
703 text_t text;
704 while (here != end) {
705
706 if (*here == '\\') {
707 ++here;
708 if (here != end) text.push_back(*here);
709
710 }
711
712 else if (*here == ',' || *here == '}' || *here == '{') {
713
714 if (formatlistptr->command == comOr) {
715 // the {Or}{this, or this, or this, or this} statement
716 format_t *or_ptr;
717
718 // find the next unused orptr
719 if (formatlistptr->orptr == NULL) {
720 formatlistptr->orptr = new format_t();
721 or_ptr = formatlistptr->orptr;
722 } else {
723 or_ptr = formatlistptr->orptr;
724 while (or_ptr->nextptr != NULL)
725 or_ptr = or_ptr->nextptr;
726 or_ptr->nextptr = new format_t();
727 or_ptr = or_ptr->nextptr;
728 }
729
730 if (!text.empty())
731 {
732 if (!parse_string(text, or_ptr, metadata, getParents)) { return false; }
733 }
734
735 if (*here == '{')
736 {
737 // Supports: {Or}{[Booktitle],[Title],{If}{[XXXX],aaa,bbb}}
738 // but not : {Or}{[Booktitle],[Title]{If}{[XXXX],aaa,bbb}}
739 // The latter can always be re-written:
740 // {Or}{[Booktitle],{If}{[Title],[Title]{If}{[XXXX],aaa,bbb}}}
741
742 if (!text.empty()) // already used up allocated format_t
743 {
744 // => allocate new one for detected action
745 or_ptr->nextptr = new format_t();
746 or_ptr = or_ptr->nextptr;
747 }
748 if (!parse_action(++here, end, or_ptr, metadata, getParents))
749 {
750 return false;
751 }
752 }
753 else
754 {
755 if (*here == '}') break;
756 }
757 text.clear();
758
759 }
760
761 // Parse an {If}{decide,do,else} statement
762 else {
763
764 // Read the decision component.
765 if (commacount == 0) {
766 // Decsion can be a metadata element, or a piece of text.
767 // Originally Stefan's code, updated 25/10/2000 by Gordon.
768
769 text_t::const_iterator beginbracket = text.begin();
770 text_t::const_iterator endbracket = (text.end() - 1);
771
772 // Decision is based on a metadata element
773 if ((*beginbracket == '[') && (*endbracket == ']')) {
774 // Ignore the surrounding square brackets
775 text_t meta = substr (beginbracket+1, endbracket);
776 parse_meta (meta, formatlistptr->decision.meta, metadata, getParents);
777 ++commacount;
778 text.clear();
779 }
780
781 // Decision is a piece of text (probably a macro like _cgiargmode_).
782 else {
783
784 // hunt for any metadata in string, which might be uses in
785 // to test a condition, e.g. [Format] eq 'PDF'
786 format_t* dummyformat = new format_t();
787 // update which metadata fields needed
788 // (not interested in updatng formatlistptr)
789 parse_string (text, dummyformat, metadata, getParents);
790 delete dummyformat;
791
792 formatlistptr->decision.command = dText;
793 formatlistptr->decision.text = text;
794 ++commacount;
795 text.clear();
796 }
797 }
798
799 // Read the "then" and "else" components of the {If} statement.
800 else {
801 format_t** nextlistptr = NULL;
802 if (commacount == 1) {
803 nextlistptr = &formatlistptr->ifptr;
804 } else if (commacount == 2 ) {
805 nextlistptr = &formatlistptr->elseptr;
806 } else {
807 return false;
808 }
809
810 if (!text.empty()) {
811 if (*nextlistptr == NULL) {
812 *nextlistptr = new format_t();
813 } else {
814
815 // skip to the end of any format_t statements already added
816 while ((*nextlistptr)->nextptr != NULL)
817 {
818 nextlistptr = &(*nextlistptr)->nextptr;
819 }
820
821 (*nextlistptr)->nextptr = new format_t();
822 nextlistptr = &(*nextlistptr)->nextptr;
823 }
824
825 if (!parse_string (text, *nextlistptr, metadata, getParents))
826 {
827 return false;
828 }
829 text.clear();
830 }
831
832 if (*here == '{')
833 {
834 if (*nextlistptr == NULL) {
835 *nextlistptr = new format_t();
836 } else {
837 // skip to the end of any format_t statements already added
838 while ((*nextlistptr)->nextptr != NULL)
839 {
840 nextlistptr = &(*nextlistptr)->nextptr;
841 }
842
843 (*nextlistptr)->nextptr = new format_t();
844 nextlistptr = &(*nextlistptr)->nextptr;
845 }
846
847 if (!parse_action(++here, end, *nextlistptr, metadata, getParents))
848 {
849 return false;
850 }
851 }
852 else
853 {
854 if (*here == '}') break;
855 ++commacount;
856 }
857 }
858 }
859
860 } else text.push_back(*here);
861
862 if (here != end) ++here;
863 }
864
865 return true;
866}
867
868
869bool parse_formatstring (const text_t &formatstring, format_t *formatlistptr,
870 text_tset &metadata, bool &getParents) {
871
872 formatlistptr->clear();
873 getParents = false;
874
875 return (parse_string (formatstring, formatlistptr, metadata, getParents));
876}
877
878// position -1 for all, -2 for the last, 0 for the first, or x for a particular piece
879// metainfo has all the values for the metadata, meta has the request metadata name and options, position tells which values to get
880static text_t get_formatted_meta_text(MetadataInfo_t &metainfo, const metadata_t &meta, int position, bool no_cgisafe = false)
881{
882 text_t no_ns_metaname = remove_namespace(meta.metaname);
883 text_t tmp;
884 bool first = true;
885
886 const int start_i=0;
887 const int end_i = metainfo.values.size()-1;
888
889 if (position == -1) { // all
890 for (int i=start_i; i<=end_i; ++i) {
891 if (!first) tmp += meta.siblingoptions;
892 if (meta.metacommand & mSpecial) {
893 // special formatting
894 if (no_ns_metaname == "Date") tmp += format_date (metainfo.values[i]);
895 else if (no_ns_metaname == "Language") tmp += iso639(metainfo.values[i]);
896 else tmp += "_format:"+meta.metaname+"_("+metainfo.values[i]+")";
897 }
898 else tmp += metainfo.values[i];
899 first = false;
900
901 }
902 } else {
903 if (position == -2) { // end
904 position = end_i;
905 } else if (position < start_i || position > end_i) {
906 return "";
907 }
908 if (meta.metacommand & mSpecial) {
909 // special formatting
910 if (no_ns_metaname == "Date") tmp += format_date (metainfo.values[position]);
911 else if (no_ns_metaname == "Language") tmp += iso639(metainfo.values[position]);
912 else tmp += "_format:"+meta.metaname+"_("+metainfo.values[position]+")";
913 }
914 else tmp += metainfo.values[position];
915 }
916 if (meta.metacommand & mCgiSafe && !no_cgisafe) return cgi_safe_unicode (tmp);
917 else return tmp;
918}
919
920static text_t get_parent_meta (ResultDocInfo_t &docinfo, const metadata_t &meta, int siblings_values)
921{
922
923 MetadataInfo_t *parent = docinfo.metadata[meta.metaname].parent;
924 switch (meta.mqualifier.parent) {
925 case pNone:
926 return "Nothing!!";
927 break;
928
929 case pImmediate:
930 if (parent != NULL) {
931 return get_formatted_meta_text(*parent, meta, siblings_values);
932 }
933 break;
934
935 case pTop:
936 if (parent != NULL) {
937 while (parent->parent != NULL) parent = parent->parent;
938 return get_formatted_meta_text(*parent, meta, siblings_values);
939 }
940 break;
941
942 case pAll:
943 MetadataInfo_t *parent = docinfo.metadata[meta.metaname].parent;
944 if (parent != NULL) {
945 text_tarray tmparray;
946 while (parent != NULL) {
947 tmparray.push_back (get_formatted_meta_text(*parent, meta, siblings_values, true)); // set no_cgisafe to true, as we'll do it once we have all the metadata
948 parent = parent->parent;
949 }
950 // now join them up - use teh parent separator
951 bool first = true;
952 text_t tmp;
953 text_tarray::reverse_iterator here = tmparray.rbegin();
954 text_tarray::reverse_iterator end = tmparray.rend();
955 while (here != end) {
956 if (!first) tmp += meta.parentoptions;
957 tmp += *here;
958 first = false;
959 ++here;
960 }
961 if (meta.metacommand & mCgiSafe) return cgi_safe_unicode (tmp);
962 else return tmp;
963 }
964 }
965 return "";
966
967}
968
969static text_t get_child_meta (const text_t& collection,
970 recptproto* collectproto,
971 ResultDocInfo_t &docinfo, displayclass &disp,
972 const metadata_t &meta, text_tmap &options,
973 ostream& logout, int siblings_values)
974{
975 if (docinfo.metadata["contains"].values[0].size()==0) return ""; // no children
976
977 const text_t& pre_tree_trav = meta.pre_tree_traverse;
978 const text_t& child_metaname = meta.metaname;
979 const text_t& child_field = meta.childoptions;
980 text_tset child_metadata;
981 child_metadata.insert(child_metaname);
982
983 FilterResponse_t child_response;
984 if (meta.mqualifier.child == cNum) {
985 // just one child
986 //get the information associated with the metadata for child doc
987 if (!get_info (docinfo.OID+pre_tree_trav+child_field, collection, "",
988 child_metadata, false, collectproto, child_response,
989 logout)) return ""; // invalid child number
990
991 if (child_response.docInfo.empty()) return false; // no info for the child
992
993 ResultDocInfo_t& child_docinfo = child_response.docInfo[0];
994 MetadataInfo_t& metaname_rec = child_docinfo.metadata[child_metaname];
995
996 text_t child_metavalue = get_formatted_meta_text(metaname_rec,meta,siblings_values);
997 return expand_metadata(child_metavalue,collection,collectproto,
998 child_docinfo,disp,options,logout);
999 }
1000
1001
1002 if (meta.mqualifier.child != cAll) return false; // invalid qualifier
1003
1004
1005 if (!pre_tree_trav.empty()) {
1006 // need to get relevant "contains" metadata for new (e.g. pre tree trav) node
1007 FilterResponse_t trav_response;
1008
1009 text_tset trav_metadata;
1010 trav_metadata.insert("contains");
1011
1012 if (!get_info (docinfo.OID+pre_tree_trav, collection, "",
1013 trav_metadata, false, collectproto, trav_response,
1014 logout)) return ""; // invalid pre_tree_trav
1015
1016 if (trav_response.docInfo.empty()) return false; // no info for the pre_tree_trav OID
1017
1018 ResultDocInfo_t& trav_docinfo = trav_response.docInfo[0];
1019 // use this for rest of routine
1020 docinfo = trav_docinfo;
1021 }
1022
1023 // we need to get all children
1024 text_t result = "";
1025 text_tarray children;
1026 text_t contains = docinfo.metadata["contains"].values[0];
1027 splitchar (contains.begin(), contains.end(), ';', children);
1028 text_tarray::const_iterator here = children.begin();
1029 text_tarray::const_iterator end = children.end();
1030 bool first = true;
1031 while (here !=end) {
1032 text_t oid = *here;
1033 here++;
1034 if (*(oid.begin()) == '"') translate_parent (oid, docinfo.OID);
1035
1036 //get the information associated with the metadata for child doc
1037 if (!get_info (oid, collection, "", child_metadata,
1038 false, collectproto, child_response, logout) ||
1039 child_response.docInfo.empty()) {
1040 first = false;
1041 continue;
1042 }
1043
1044
1045 ResultDocInfo_t& child_docinfo = child_response.docInfo[0];
1046 MetadataInfo_t& metaname_rec = child_docinfo.metadata[child_metaname];
1047
1048 text_t child_metavalue = get_formatted_meta_text(metaname_rec,meta,siblings_values);
1049 if (!first) result += child_field;
1050 first = false;
1051 // need to do this here cos otherwise we are in the wrong document
1052 result += expand_metadata(child_metavalue,collection,collectproto,
1053 child_docinfo,disp,options,logout);
1054 }
1055 return result;
1056
1057}
1058
1059static text_t get_meta (const text_t& collection, recptproto* collectproto,
1060 ResultDocInfo_t &docinfo, displayclass &disp,
1061 const metadata_t &meta, text_tmap &options,
1062 ostream& logout) {
1063
1064 // make sure we have the requested metadata
1065 MetadataInfo_tmap::iterator it = docinfo.metadata.find (meta.metaname);
1066 if (it == docinfo.metadata.end()) return "";
1067
1068 int siblings_values = 0; // default is no siblings, just the first metadata available
1069 if (meta.metacommand & mSibling) {
1070 if (meta.mqualifier.sibling == sAll) {
1071 siblings_values = -1; //all
1072 } else if (meta.mqualifier.sibling == sNum) {
1073 siblings_values = meta.siblingoptions.getint();
1074 }
1075 }
1076 if (meta.metacommand & mParent) {
1077 return get_parent_meta(docinfo,meta,siblings_values);
1078 }
1079
1080 else if (meta.metacommand & mChild) {
1081 return get_child_meta(collection,collectproto,docinfo,disp,meta,
1082 options,logout, siblings_values);
1083 }
1084 else if (meta.metacommand & mSibling) { // only siblings
1085 MetadataInfo_t& metaname_rec = docinfo.metadata[meta.metaname];
1086 return get_formatted_meta_text(docinfo.metadata[meta.metaname],meta, siblings_values);
1087 }
1088 else {
1089
1090 // straightforward metadata request (nothing fancy)
1091
1092 text_t classifier_metaname = docinfo.classifier_metadata_type;
1093 int metaname_index
1094 = (classifier_metaname == meta.metaname) ? docinfo.classifier_metadata_offset : 0;
1095 return get_formatted_meta_text(docinfo.metadata[meta.metaname], meta, metaname_index);
1096 }
1097
1098 return "";
1099}
1100
1101static text_t get_or (const text_t& collection, recptproto* collectproto,
1102 ResultDocInfo_t &docinfo, displayclass &disp,
1103 format_t *orptr, text_tmap &options,
1104 ostream& logout) {
1105
1106 text_t tmp;
1107 while (orptr != NULL) {
1108
1109 tmp = format_string (collection,collectproto,docinfo, disp, orptr,
1110 options, logout);
1111 if (!tmp.empty()) return tmp;
1112
1113 orptr = orptr->nextptr;
1114 }
1115 return "";
1116}
1117
1118static bool char_is_whitespace(const char c)
1119{
1120 return ((c == ' ') || (c == '\t') || (c == '\n') || (c == '\r'));
1121
1122}
1123
1124static int scan_over_whitespace(const text_t& outstring, const int start_pos)
1125{
1126 int pos = start_pos;
1127 while (pos<outstring.size()) {
1128 if (!char_is_whitespace(outstring[pos])) {
1129 break;
1130 }
1131 ++pos;
1132 }
1133
1134 return pos;
1135}
1136
1137static int rscan_over_whitespace(const text_t& outstring, const int start_pos)
1138{
1139 int pos = start_pos;
1140 while (pos>=0) {
1141 if (!char_is_whitespace(outstring[pos])) {
1142 break;
1143 }
1144 --pos;
1145 }
1146
1147 return pos;
1148}
1149
1150static int rscan_for_whitespace(const text_t& outstring, const int start_pos)
1151{
1152 int pos = start_pos;
1153 while (pos>=0) {
1154 if (char_is_whitespace(outstring[pos])) {
1155 break;
1156 }
1157 --pos;
1158 }
1159
1160 return pos;
1161}
1162
1163
1164static int rscan_for(const text_t& outstring, const int start_pos,
1165 const char find_c)
1166{
1167 int pos = start_pos;
1168 while (pos>=0) {
1169 char c = outstring[pos];
1170 if (outstring[pos] == find_c) {
1171 break;
1172 }
1173 --pos;
1174 }
1175
1176 return pos;
1177}
1178
1179text_t extract_substr(const text_t& outstring, const int start_pos,
1180 const int end_pos)
1181{
1182 text_t extracted_str;
1183 extracted_str.clear();
1184
1185 for (int pos=start_pos; pos<=end_pos; ++pos) {
1186 extracted_str.push_back(outstring[pos]);
1187 }
1188
1189 return extracted_str;
1190}
1191
1192
1193static text_t expand_potential_metadata(const text_t& collection,
1194 recptproto* collectproto,
1195 ResultDocInfo_t &docinfo,
1196 displayclass &disp,
1197 const text_t& intext,
1198 text_tmap &options,
1199 ostream& logout)
1200{
1201 text_t outtext;
1202
1203 // decide if dealing with metadata or text
1204
1205 text_t::const_iterator beginbracket = intext.begin();
1206 text_t::const_iterator endbracket = (intext.end() - 1);
1207
1208 // Decision is based on a metadata element
1209 if ((*beginbracket == '[') && (*endbracket == ']')) {
1210 // Ignore the surrounding square brackets
1211 text_t meta_text = substr (beginbracket+1, endbracket);
1212
1213 if (meta_text == "Text") {
1214 outtext = format_text(collection, collectproto, docinfo, disp, options, logout);
1215 } else {
1216
1217 text_tset metadata;
1218 bool getParents =false;
1219 metadata_t meta;
1220
1221 parse_meta (meta_text, meta, metadata, getParents);
1222 outtext
1223 = get_meta (collection,collectproto,docinfo,disp,meta,options,logout);
1224 }
1225
1226 }
1227 else {
1228 outtext = intext;
1229 }
1230
1231 return outtext;
1232}
1233
1234
1235
1236
1237static bool uses_expression(const text_t& collection, recptproto* collectproto,
1238 ResultDocInfo_t &docinfo,
1239 displayclass &disp,
1240 const text_t& outstring, text_t& lhs_expr,
1241 text_t& op_expr, text_t& rhs_expr,
1242 text_tmap &options,
1243 ostream& logout)
1244{
1245 // Note: the string may not be of the form: str1 op str2, however
1246 // to deterine this we have to process it on the assumption it is,
1247 // and if at any point an 'erroneous' value is encountered, return
1248 // false and let something else have a go at evaluating it
1249
1250 // Starting at the end of the string and working backwards ..
1251
1252 const int outstring_len = outstring.size();
1253
1254 // skip over white space
1255 int rhs_end = rscan_over_whitespace(outstring,outstring_len-1);
1256
1257 if (rhs_end<=0) {
1258 // no meaningful text or (rhs_end==0) no room for operator
1259 return false;
1260 }
1261
1262 // check for ' or " and then scan over token
1263 const char potential_quote = outstring[rhs_end];
1264 int rhs_start=rhs_end;
1265 bool quoted = false;
1266
1267 if ((potential_quote == '\'') || (potential_quote == '\"')) {
1268 --rhs_end;
1269 rhs_start = rscan_for(outstring,rhs_end-1,potential_quote) +1;
1270 quoted = true;
1271 }
1272 else {
1273 rhs_start = rscan_for_whitespace(outstring,rhs_end-1) +1;
1274 }
1275
1276 if ((rhs_end-rhs_start)<0) {
1277 // no meaningful rhs expression
1278 return false;
1279 }
1280
1281 // form rhs_expr
1282 rhs_expr = extract_substr(outstring,rhs_start,rhs_end);
1283
1284 // skip over white space
1285 const int to_whitespace = (quoted) ? 2 : 1;
1286
1287 int op_end = rscan_over_whitespace(outstring,rhs_start-to_whitespace);
1288 int op_start = rscan_for_whitespace(outstring,op_end-1)+1;
1289
1290
1291 if (op_end-op_start<0) {
1292 // no meaningful expression operator
1293 return false;
1294 }
1295
1296 op_expr = extract_substr(outstring,op_start,op_end);
1297
1298
1299 // check for operator
1300 if ((op_expr != "eq") && (op_expr != "ne") && (op_expr != "gt") &&
1301 (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")) {
1302
1303 // not a valid operator
1304 return false;
1305 }
1306
1307 int lhs_end = rscan_over_whitespace(outstring,op_start-1);
1308 if (lhs_end<0) {
1309 // no meaningful lhs expression
1310 return false;
1311 }
1312
1313 int lhs_start = scan_over_whitespace(outstring,0);
1314
1315 // form lhs_expr from remainder of string
1316 lhs_expr = extract_substr(outstring,lhs_start,lhs_end);
1317
1318 // Now we know we have a valid expression, look up any
1319 // metadata terms
1320
1321 rhs_expr = expand_potential_metadata(collection,collectproto,docinfo,
1322 disp,rhs_expr,options,logout);
1323 lhs_expr = expand_potential_metadata(collection,collectproto,docinfo,
1324 disp,lhs_expr,options,logout);
1325
1326 return true;
1327}
1328
1329static bool eval_expression_true(const text_t& lhs_expr,const text_t& op_expr,
1330 const text_t& rhs_expr, ostream& logout)
1331{
1332 if (op_expr == "eq") return (lhs_expr == rhs_expr);
1333 else if (op_expr == "ne" ) return (lhs_expr != rhs_expr);
1334 else if (op_expr == "gt") return (lhs_expr > rhs_expr);
1335 else if (op_expr == "ge") return (lhs_expr >= rhs_expr);
1336 else if (op_expr == "lt") return (lhs_expr < rhs_expr);
1337 else if (op_expr == "le") return (lhs_expr <= rhs_expr);
1338 else if (op_expr == "==") return (lhs_expr.getint() == rhs_expr.getint());
1339 else if (op_expr == "!=") return (lhs_expr.getint() != rhs_expr.getint());
1340 else if (op_expr == ">") return (lhs_expr.getint() > rhs_expr.getint());
1341 else if (op_expr == ">=") return (lhs_expr.getint() >= rhs_expr.getint());
1342 else if (op_expr == "<") return (lhs_expr.getint() < rhs_expr.getint());
1343 else if (op_expr == "<=") return (lhs_expr.getint() <= rhs_expr.getint());
1344 else if (op_expr == "sw") return (starts_with(lhs_expr,rhs_expr));
1345 else if (op_expr == "ew") return (ends_with(lhs_expr,rhs_expr));
1346 else {
1347 logout << "Error: '" << op_expr << "' is not a recognised operator." << endl;
1348 }
1349
1350 return false;
1351}
1352
1353
1354static text_t get_if (const text_t& collection, recptproto* collectproto,
1355 ResultDocInfo_t &docinfo, displayclass &disp,
1356 const decision_t &decision,
1357 format_t *ifptr, format_t *elseptr,
1358 text_tmap &options, ostream& logout)
1359{
1360 // If the decision component is a metadata element, then evaluate it
1361 // to see whether we output the "then" or the "else" clause
1362 if (decision.command == dMeta) {
1363 if (get_meta (collection,collectproto,docinfo,disp,decision.meta,options,
1364 logout) != "") {
1365 if (ifptr != NULL)
1366 return get_formatted_string (collection,collectproto,docinfo, disp, ifptr,
1367 options, logout);
1368 }
1369 else {
1370 if (elseptr != NULL)
1371 return get_formatted_string (collection,collectproto,docinfo, disp, elseptr,
1372 options, logout);
1373 }
1374 }
1375
1376 // If the decision component is text, then evaluate it (it is probably a
1377 // macro like _cgiargmode_) to decide what to output.
1378 else if (decision.command == dText) {
1379
1380 text_t outstring;
1381 disp.expandstring (decision.text, outstring);
1382
1383 // Check for if expression in form: str1 op str2
1384 // (such as [x] eq "y")
1385 text_t lhs_expr, op_expr, rhs_expr;
1386 if (uses_expression(collection,collectproto,docinfo, disp, outstring,lhs_expr,op_expr,rhs_expr, options,logout)) {
1387 if (eval_expression_true(lhs_expr,op_expr,rhs_expr,logout)) {
1388 if (ifptr != NULL) {
1389 return get_formatted_string (collection, collectproto, docinfo, disp, ifptr,
1390 options, logout);
1391 }
1392 else {
1393 return "";
1394 }
1395 } else {
1396 if (elseptr != NULL) {
1397 return get_formatted_string (collection, collectproto, docinfo, disp, elseptr,
1398 options, logout);
1399 }
1400 else {
1401 return "";
1402 }
1403 }
1404 }
1405
1406
1407 // This is a tad tricky. When we expand a string like _cgiargmode_, that is
1408 // a cgi argument macro that has not been set, it evaluates to itself.
1409 // Therefore, were have to say that a piece of text evalautes true if
1410 // it is non-empty and if it is a cgi argument evaulating to itself.
1411
1412 if ((outstring != "") && !((outstring == decision.text) && (outstring[0] == '_'))) {
1413 if (ifptr != NULL)
1414 return get_formatted_string (collection, collectproto, docinfo, disp, ifptr,
1415 options, logout);
1416 } else {
1417 if (elseptr != NULL)
1418 return get_formatted_string (collection, collectproto, docinfo, disp, elseptr,
1419 options, logout);
1420 }
1421 }
1422
1423 return "";
1424}
1425
1426bool includes_metadata(const text_t& text)
1427{
1428 text_t::const_iterator here = text.begin();
1429 text_t::const_iterator end = text.end();
1430 while (here != end) {
1431 if (*here == '[') return true;
1432 ++here;
1433 }
1434
1435 return false;
1436}
1437
1438static text_t expand_metadata(const text_t &metavalue, const text_t& collection,
1439 recptproto* collectproto,
1440 ResultDocInfo_t &docinfo,
1441 displayclass &disp, text_tmap &options,
1442 ostream &logout) {
1443
1444 if (includes_metadata(metavalue)) {
1445
1446 // text has embedded metadata in it => expand it
1447 FilterRequest_t request;
1448 FilterResponse_t response;
1449
1450 request.getParents = false;
1451
1452 format_t *expanded_formatlistptr = new format_t();
1453 parse_formatstring (metavalue, expanded_formatlistptr,
1454 request.fields, request.getParents);
1455
1456 // retrieve metadata
1457 get_info(docinfo.OID, collection, "", request.fields, request.getParents,
1458 collectproto, response, logout);
1459
1460 if (!response.docInfo.empty()) {
1461
1462 text_t expanded_metavalue
1463 = get_formatted_string(collection, collectproto,
1464 response.docInfo[0], disp, expanded_formatlistptr,
1465 options, logout);
1466
1467 return expanded_metavalue;
1468 }
1469 else {
1470 return metavalue;
1471 }
1472 }
1473 else {
1474
1475 return metavalue;
1476 }
1477}
1478
1479text_t get_collection_meta(const text_t& collection, recptproto* collectproto,
1480 displayclass &disp,
1481 text_t meta_name, ostream& logout) {
1482
1483 ColInfoResponse_t collectinfo;
1484 comerror_t err;
1485 collectproto->get_collectinfo (collection, collectinfo,err,logout);
1486 text_t meta_value = "";
1487 text_t lang;
1488 disp.expandstring("_cgiargl_",lang);
1489 if (lang.empty()) {
1490 lang = "en";
1491 }
1492
1493 if (err == noError) {
1494 meta_value = collectinfo.get_collectionmeta(meta_name, lang);
1495 }
1496 return meta_value;
1497
1498
1499}
1500text_t format_string (const text_t& collection, recptproto* collectproto,
1501 ResultDocInfo_t &docinfo, displayclass &disp,
1502 format_t *formatlistptr, text_tmap &options,
1503 ostream& logout) {
1504
1505 if (formatlistptr == NULL) return "";
1506
1507 switch (formatlistptr->command) {
1508 case comOID:
1509 return docinfo.OID;
1510 case comTopOID:
1511 {
1512 text_t top_id;
1513 get_top(docinfo.OID, top_id);
1514 return top_id;
1515 }
1516 case comRank:
1517 return text_t(docinfo.ranking);
1518 case comText:
1519 return formatlistptr->text;
1520 case comLink:
1521 return options["link"];
1522 case comEndLink:
1523 if (options["link"].empty()) return "";
1524 else return "</a>";
1525 case comHref:
1526 return get_href(options["link"]);
1527 case comIcon:
1528 return options["icon"];
1529 case comNum:
1530 return docinfo.result_num;
1531 case comRel: //if [RelatedDocuments] appears in format string, collect relation data
1532 return get_related_docs(collection, collectproto, docinfo, logout);
1533 case comSummary:
1534 return format_summary(collection, collectproto, docinfo, disp, options, logout);
1535 case comMeta:
1536 {
1537 const text_t& metavalue = get_meta (collection,collectproto, docinfo, disp,formatlistptr->meta,options,logout);
1538 return expand_metadata(metavalue, collection, collectproto, docinfo, disp, options, logout);
1539 }
1540 case comDoc:
1541 return format_text(collection, collectproto, docinfo, disp, options, logout);
1542 //return options["text"];
1543 case comImage:
1544 return expand_metadata(options["DocImage"], collection, collectproto, docinfo, disp, options, logout);
1545 case comTOC:
1546 return options["DocTOC"];
1547 case comDocumentButtonDetach:
1548 return options["DocumentButtonDetach"];
1549 case comDocumentButtonHighlight:
1550 return options["DocumentButtonHighlight"];
1551 case comDocumentButtonExpandContents:
1552 return options["DocumentButtonExpandContents"];
1553 case comDocumentButtonExpandText:
1554 return options["DocumentButtonExpandText"];
1555 case comHighlight:
1556 if (options["highlight"] == "1") return "<b>";
1557 break;
1558 case comEndHighlight:
1559 if (options["highlight"] == "1") return "</b>";
1560 break;
1561 case comIf:
1562 return get_if (collection, collectproto, docinfo, disp,
1563 formatlistptr->decision, formatlistptr->ifptr,
1564 formatlistptr->elseptr, options, logout);
1565 case comOr:
1566 return get_or (collection,collectproto, docinfo, disp, formatlistptr->orptr,
1567 options, logout);
1568 case comDocTermsFreqTotal:
1569 return docinfo.num_terms_matched;
1570 case comCollection:
1571 if (formatlistptr->meta.metaname == g_EmptyText) {
1572 return collection;
1573 }
1574 return get_collection_meta(collection, collectproto, disp, formatlistptr->meta.metaname, logout);
1575
1576 }
1577 return "";
1578}
1579
1580text_t get_formatted_string (const text_t& collection, recptproto* collectproto,
1581 ResultDocInfo_t &docinfo, displayclass &disp,
1582 format_t *formatlistptr, text_tmap &options,
1583 ostream& logout) {
1584
1585 text_t ft;
1586 while (formatlistptr != NULL)
1587 {
1588 ft += format_string (collection, collectproto, docinfo, disp, formatlistptr,
1589 options, logout);
1590 formatlistptr = formatlistptr->nextptr;
1591 }
1592
1593 return ft;
1594}
1595
1596
1597// we have only preloaded the text in DocumentAction. But you may want to get the text in query. so copy what we have done with format_summary and get the text here. probably is quite expensive?
1598text_t format_text (const text_t& collection, recptproto* collectproto,
1599 ResultDocInfo_t &docinfo, displayclass &disp,
1600 text_tmap &options, ostream& logout) {
1601 if(!options["text"].empty()) {
1602 return options["text"];
1603 }
1604 // else get document text here
1605 DocumentRequest_t docrequest;
1606 DocumentResponse_t docresponse;
1607 comerror_t err;
1608 docrequest.OID = docinfo.OID;
1609 collectproto->get_document (collection, docrequest, docresponse, err, logout);
1610 return docresponse.doc;
1611
1612}
1613
1614/* FUNCTION NAME: format_summary
1615 * DESC: this is invoked when a [Summary] special metadata is processed.
1616 * RETURNS: a query-biased summary for the document */
1617
1618text_t format_summary (const text_t& collection, recptproto* collectproto,
1619 ResultDocInfo_t &docinfo, displayclass &disp,
1620 text_tmap &options, ostream& logout) {
1621
1622 // GRB: added code here to ensure that the cstr (and other collections)
1623 // uses the document metadata item Summary, rather than compressing
1624 // the text of the document, processed via the methods in
1625 // summarise.cpp
1626 if (docinfo.metadata.count("Summary") > 0 &&
1627 docinfo.metadata["Summary"].values.size() > 0) {
1628 return docinfo.metadata["Summary"].values[0];
1629 }
1630
1631 text_t textToSummarise, query;
1632 if(options["text"].empty()) { // get document text
1633 DocumentRequest_t docrequest;
1634 DocumentResponse_t docresponse;
1635 comerror_t err;
1636 docrequest.OID = docinfo.OID;
1637 collectproto->get_document (collection, docrequest, docresponse, err, logout);
1638 textToSummarise = docresponse.doc;
1639 } else // in practice, this would not happen, because text is only
1640 // loaded with the [Text] command
1641 textToSummarise = options["text"];
1642 disp.expandstring("_cgiargq_",query);
1643 return summarise(textToSummarise,query,80);
1644}
Note: See TracBrowser for help on using the repository browser.