root/gsdl/trunk/runtime-src/src/recpt/formattools.cpp @ 19298

Revision 19298, 49.7 KB (checked in by davidb, 11 years ago)

optional support for metadata wrapped in span tags added. *OFF* by default.

  • Property svn:executable set to *
  • Property svn:keywords set to Author Date Id Revision
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
34bool spanwrap = false;
35
36// a few function prototypes
37
38static text_t format_string (const text_t& collection, recptproto* collectproto,
39                 ResultDocInfo_t &docinfo, displayclass &disp,
40                 format_t *formatlistptr, text_tmap &options,
41                 ostream& logout);
42
43static bool parse_action (text_t::const_iterator &here, const text_t::const_iterator &end,
44              format_t *formatlistptr, text_tset &metadata, bool &getParents);
45
46static text_t format_summary (const text_t& collection, recptproto* collectproto,
47                  ResultDocInfo_t &docinfo, displayclass &disp,
48                  text_tmap &options, ostream& logout);
49static text_t format_text (const text_t& collection, recptproto* collectproto,
50                  ResultDocInfo_t &docinfo, displayclass &disp,
51                  text_tmap &options, ostream& logout);
52
53static text_t expand_metadata(const text_t &metavalue, const text_t& collection,
54                  recptproto* collectproto, ResultDocInfo_t &docinfo,
55                  displayclass &disp, text_tmap &options,
56                  ostream &logout);
57
58
59void metadata_t::clear() {
60  metaname.clear();
61  metacommand = mNone;
62  mqualifier.parent  = pNone;
63  mqualifier.sibling = sNone;
64  mqualifier.child   = cNone;
65  pre_tree_traverse.clear();
66  parentoptions.clear();
67  siblingoptions.clear();
68  childoptions.clear();
69}
70
71void decision_t::clear() {
72  command = dMeta;
73  meta.clear();
74  text.clear();
75}
76
77void format_t::clear() {
78  command = comText;
79  decision.clear();
80  text.clear();
81  meta.clear();
82  nextptr = NULL;
83  ifptr = NULL;
84  elseptr = NULL;
85  orptr = NULL;
86}
87
88void formatinfo_t::clear() {
89  DocumentImages = false;
90  DocumentTitles = true;
91  DocumentHeading = "{Or}{[parent(Top):Title],[Title],untitled}<br>";
92  DocumentContents = true;
93  DocumentArrowsBottom = true;
94  DocumentArrowsTop = false;
95  DocumentSearchResultLinks = false;
96  DocumentButtons.erase (DocumentButtons.begin(), DocumentButtons.end());
97  //  DocumentButtons.push_back ("Expand Text");
98  //  DocumentButtons.push_back ("Expand Contents");
99  DocumentButtons.push_back ("Detach");
100  DocumentButtons.push_back ("Highlight");
101  RelatedDocuments = "";
102  DocumentText = "[Text]";
103  formatstrings.erase (formatstrings.begin(), formatstrings.end());
104  DocumentUseHTML = false;
105  AllowExtendedOptions = false;
106}
107
108// simply checks to see if formatstring begins with a <td> tag
109bool is_table_content (const text_t &formatstring) {
110  text_t::const_iterator here = formatstring.begin();
111  text_t::const_iterator end = formatstring.end();
112 
113  while (here != end) {
114    if (*here != ' ') {
115      if ((*here == '<') && ((here+3) < end)) {
116    if ((*(here+1) == 't' || *(here+1) == 'T') &&
117        (*(here+2) == 'd' || *(here+2) == 'D') &&
118        (*(here+3) == '>' || *(here+3) == ' '))
119      return true;
120      } else return false;
121    }
122    ++here;
123  }
124  return false;
125}
126
127bool is_table_content (const format_t *formatlistptr) {
128
129  if (formatlistptr == NULL) return false;
130 
131  if (formatlistptr->command == comText)
132    return is_table_content (formatlistptr->text);
133   
134  return false;
135}
136
137// returns false if key isn't in formatstringmap
138bool get_formatstring (const text_t &key, const text_tmap &formatstringmap,
139               text_t &formatstring) {
140
141  formatstring.clear();
142  text_tmap::const_iterator it = formatstringmap.find(key);
143  if (it == formatstringmap.end()) return false;
144  formatstring = (*it).second;
145  return true;
146}
147
148// tries to find "key1key2" then "key1" then "key2"
149bool get_formatstring (const text_t &key1, const text_t &key2, 
150               const text_tmap &formatstringmap,
151               text_t &formatstring) {
152
153  formatstring.clear();
154  text_tmap::const_iterator it = formatstringmap.find(key1 + key2);
155  if (it != formatstringmap.end()) {
156    formatstring = (*it).second;
157    return true;
158  }
159  it = formatstringmap.find(key1);
160  if (it != formatstringmap.end()) {
161    formatstring = (*it).second;
162    return true;
163  }
164  it = formatstringmap.find(key2);
165  if (it != formatstringmap.end()) {
166    formatstring = (*it).second;
167    return true;
168  }
169  return false;
170}
171
172
173text_t remove_namespace(const text_t &meta_name) {
174  text_t::const_iterator end = meta_name.end();
175  text_t::const_iterator it = findchar(meta_name.begin(), end, '.');
176  if (it != end) {
177    return substr(it+1, end);
178  }
179
180  return meta_name;
181
182}
183// returns a date of form _format:date_(year, month, day)
184// input is date of type yyyy-?mm-?dd
185// at least the year must be present in date
186text_t format_date (const text_t &date) {
187
188  if (date.size() < 4) return "";
189
190  text_t::const_iterator datebegin = date.begin();
191
192  text_t year = substr (datebegin, datebegin+4);
193  int chars_seen_so_far = 4;
194
195  if (date[chars_seen_so_far] == '-') ++chars_seen_so_far ;
196  if (date.size() < chars_seen_so_far+2) return "_format:date_("+year+")";
197 
198  text_t month = substr (datebegin+chars_seen_so_far, datebegin+chars_seen_so_far+2);
199  int imonth = month.getint();
200  if (imonth <= 0 || imonth > 12) return "_format:date_("+year+")";
201 
202  chars_seen_so_far += 2;
203  if (date[chars_seen_so_far] == '-') ++chars_seen_so_far ;
204 
205  if (date.size() < chars_seen_so_far+2) return "_format:date_("+year+","+month+")";
206
207  text_t day = substr (datebegin+chars_seen_so_far, datebegin+chars_seen_so_far+2);
208  if (day[0] == '0') day = substr (day.begin()+1, day.end());
209  int iday = day.getint();
210  if (iday <= 0 || iday > 31) return "_format:date_("+year+","+month+")";
211   
212  return "_format:date_("+year+","+month+","+day+")";
213
214
215// converts an iso639 language code to its English equivalent
216// should we be checking that the macro exists??
217text_t iso639 (const text_t &langcode) {
218  if (langcode.empty()) return "";
219  return "_iso639:iso639"+langcode+"_";
220}
221
222
223text_t get_href (const text_t &link) {
224
225  text_t href;
226
227  text_t::const_iterator here = findchar(link.begin(), link.end(), '"');
228  text_t::const_iterator end = link.end();
229  if (here == end) return g_EmptyText;
230 
231  ++here;
232  while (here != end) {
233    if (*here == '"') break;
234    href.push_back(*here);
235    ++here;
236  }
237
238  return href;
239}
240
241//this function gets the information associated with the relation
242//metadata for the document associated with 'docinfo'. This relation
243//metadata consists of a line of pairs containing 'collection, document OID'
244//(this is the OID of the document related to the current document, and
245//the collection the related document belongs to). For each of these pairs
246//the title metadata is obtained and then an html link between the title
247//of the related doc and the document's position (the document will be
248//found in "<a href=\"_httpdocument_&c=collection&cl=search&d=OID">
249//(where collection is the related documents collection, and OID is the
250//related documents OID).  A list of these html links are made for as many
251//related documents as there are. This list is then returned. If there are
252//no related documents available for the current document then the string
253//'.. no related documents .. ' is returned.
254text_t get_related_docs(const text_t& collection, recptproto* collectproto,
255            ResultDocInfo_t &docinfo, ostream& logout){
256 
257  text_tset metadata;
258
259  //insert the metadata we wish to collect
260  metadata.insert("dc.Relation");
261  metadata.insert("Title"); 
262  metadata.insert("Subject"); //for emails, where title data doesn't apply
263 
264  FilterResponse_t response;
265  text_t relation = ""; //string for displaying relation metadata
266  text_t relationTitle = ""; //the related documents Title (or subject)
267  text_t relationOID = ""; //the related documents OID 
268
269  //get the information associated with the metadata for current doc
270  if (get_info (docinfo.OID, collection, "", metadata,
271        false, collectproto, response, logout)) {
272   
273    //if the relation metadata exists, store for displaying
274    if(!response.docInfo[0].metadata["dc.Relation"].values.empty()){
275      relationOID += response.docInfo[0].metadata["dc.Relation"].values[0];
276
277      //split relation data into pairs of collectionname,ID number
278      text_tarray relationpairs;
279      splitchar (relationOID.begin(), relationOID.end(), ' ', relationpairs);
280     
281      text_tarray::const_iterator currDoc = relationpairs.begin(); 
282      text_tarray::const_iterator lastDoc = relationpairs.end();
283
284      //iterate through the pairs to split and display
285      while(currDoc != lastDoc){
286   
287    //split pairs into collectionname and ID
288    text_tarray relationdata;
289    splitchar ((*currDoc).begin(), (*currDoc).end(), ',', relationdata);
290   
291    //get first element in the array (collection)
292    text_tarray::const_iterator doc_data = relationdata.begin();
293    text_t document_collection = *doc_data;
294    ++doc_data; //increment to get next item in array (oid)
295    text_t document_OID = *doc_data;
296   
297    //create html link to related document
298    relation += "<a href=\"_httpdocument_&c=" + document_collection;
299    relation += "&cl=search&d=" + document_OID;
300       
301    //get the information associated with the metadata for related doc
302    if (get_info (document_OID, document_collection, "", metadata,
303              false, collectproto, response, logout)) {
304     
305      //if title metadata doesn't exist, collect subject metadata
306      //if that doesn't exist, just call it 'related document'
307      if (!response.docInfo[0].metadata["Title"].values[0].empty())
308        relationTitle = response.docInfo[0].metadata["Title"].values[0];
309      else if (!response.docInfo[0].metadata["Subject"].values.empty())
310        relationTitle = response.docInfo[0].metadata["Subject"].values[0];
311      else relationTitle =  "RELATED DOCUMENT";
312     
313    }
314   
315    //link the related document's title to its page
316    relation += "\">" + relationTitle + "</a>";
317    relation += "  (" + document_collection + ")<br>";
318   
319    ++currDoc;
320      }
321    }
322   
323  }
324
325  if(relation.empty()) //no relation data for documnet
326    relation = ".. no related documents .. ";
327
328  return relation;
329}
330
331
332
333static void get_parent_options (text_t &instring, metadata_t &metaoption) {
334
335  assert (instring.size() > 7);
336  if (instring.size() <= 7) return;
337
338  text_t meta, com, op;
339  bool inbraces = false;
340  bool inquotes = false;
341  bool foundcolon = false;
342  text_t::const_iterator here = instring.begin()+6;
343  text_t::const_iterator end = instring.end();
344  while (here != end) {
345    if (foundcolon) meta.push_back (*here);
346    else if (*here == '(') inbraces = true;
347    else if (*here == ')') inbraces = false;
348    else if (*here == '\'' && !inquotes) inquotes = true;
349    else if (*here == '\'' && inquotes) inquotes = false;
350    else if (*here == ':' && !inbraces) foundcolon = true;
351    else if (inquotes) op.push_back (*here);
352    else com.push_back (*here);
353    ++here;
354  }
355
356  instring = meta;
357  if (com.empty())
358    metaoption.mqualifier.parent = pImmediate;
359  else if (com == "Top")
360    metaoption.mqualifier.parent = pTop;
361  else if (com == "All") {
362    metaoption.mqualifier.parent = pAll;
363    metaoption.parentoptions = op;
364  }
365}
366
367
368static void get_sibling_options (text_t &instring, metadata_t &metaoption) {
369
370  assert (instring.size() > 8);
371  if (instring.size() <= 8) return;
372  text_t meta, com, op;
373  bool inbraces = false;
374  bool inquotes = false;
375  bool foundcolon = false;
376  text_t::const_iterator here = instring.begin()+7;
377  text_t::const_iterator end = instring.end();
378  while (here != end) {
379    if (foundcolon) meta.push_back (*here);
380    else if (*here == '(') inbraces = true;
381    else if (*here == ')') inbraces = false;
382    else if (*here == '\'' && !inquotes) inquotes = true;
383    else if (*here == '\'' && inquotes) inquotes = false;
384    else if (*here == ':' && !inbraces) foundcolon = true;   
385    else if (inquotes) op.push_back (*here);
386    else com.push_back (*here);
387    ++here;
388  }
389
390  instring = meta;
391  metaoption.siblingoptions.clear();
392
393  if (com.empty()) {
394    metaoption.mqualifier.sibling = sAll;
395    metaoption.siblingoptions = " ";
396  }
397  else if (com == "first") {
398    metaoption.mqualifier.sibling = sNum;
399    metaoption.siblingoptions = "0";
400  }
401  else if (com == "last") {
402    metaoption.mqualifier.sibling = sNum;
403    metaoption.siblingoptions = "-2"; // == last
404  }
405  else if (com.getint()>0) {
406    metaoption.mqualifier.sibling = sNum;
407    int pos = com.getint()-1;
408    metaoption.siblingoptions +=pos;
409  }
410  else {
411    metaoption.mqualifier.sibling = sAll;
412    metaoption.siblingoptions = op;
413  }
414}
415
416static void get_child_options (text_t &instring, metadata_t &metaoption) {
417
418  assert (instring.size() > 6);
419  if (instring.size() <= 6) return;
420  text_t meta, com, op;
421  bool inbraces = false;
422  bool inquotes = false;
423  bool foundcolon = false;
424  text_t::const_iterator here = instring.begin()+5;
425  text_t::const_iterator end = instring.end();
426  while (here != end) {
427    if (foundcolon) meta.push_back (*here);
428    else if (*here == '(') inbraces = true;
429    else if (*here == ')') inbraces = false;
430    else if (*here == '\'' && !inquotes) inquotes = true;
431    else if (*here == '\'' && inquotes) inquotes = false;
432    else if (*here == ':' && !inbraces) foundcolon = true;
433    else if (inquotes) op.push_back (*here);
434    else com.push_back (*here);
435    ++here;
436  }
437
438  instring = meta;
439  if (com.empty()) {
440    metaoption.mqualifier.child = cAll;
441    metaoption.childoptions = " ";
442  }
443  else if (com == "first") {
444    metaoption.mqualifier.child = cNum;
445    metaoption.childoptions = ".fc";
446  }
447  else if (com == "last") {
448    metaoption.mqualifier.child = cNum;
449    metaoption.childoptions = ".lc";
450  }
451  else if (com.getint()>0) {
452    metaoption.mqualifier.child = cNum;
453    metaoption.childoptions = "."+com;
454  }
455  else {
456    metaoption.mqualifier.child = cAll;
457    metaoption.childoptions = op;
458  }
459}
460
461
462
463static void parse_meta (text_t &meta, metadata_t &metaoption,
464            text_tset &metadata, bool &getParents) {
465
466  if (meta.size() > 8 && (substr(meta.begin(), meta.begin()+8) == "cgisafe:")) {
467    metaoption.metacommand |= mCgiSafe;
468    meta = substr (meta.begin()+8, meta.end());
469  }
470  if (meta.size() > 7 && (substr(meta.begin(), meta.begin()+7) == "format:")) {   
471    metaoption.metacommand |= mSpecial;
472    meta = substr (meta.begin()+7, meta.end());
473  }
474
475  bool had_parent_or_child = true;
476  bool prev_was_parent = false;
477  bool prev_was_child  = false;
478
479  while (had_parent_or_child) {
480    if (meta.size() > 7
481    && (substr (meta.begin(), meta.begin()+6) == "parent")) {
482
483      // clear out sibling and child (cmd and options)
484      metaoption.metacommand &= ~(mChild|mSibling);
485      metaoption.childoptions.clear();
486      metaoption.siblingoptions.clear();
487
488      getParents = true;
489      metaoption.metacommand |= mParent;
490      get_parent_options (meta, metaoption);
491
492      if (prev_was_parent) {
493    metaoption.pre_tree_traverse += ".pr";
494      }
495      else if (prev_was_child) {
496    metaoption.pre_tree_traverse += ".fc";
497      }
498
499      prev_was_parent = true;
500      prev_was_child  = false;
501    }
502    else if (meta.size() > 6
503         && (substr (meta.begin(), meta.begin()+5) == "child")) {
504
505      // clear out sibling and parent (cmd and options)
506      metaoption.metacommand &= ~(mParent|mSibling);
507      metaoption.parentoptions.clear();
508      metaoption.siblingoptions.clear();
509
510      metaoption.metacommand |= mChild;
511      get_child_options (meta, metaoption);
512      metadata.insert("contains");
513
514      if (prev_was_parent) {
515    metaoption.pre_tree_traverse += ".pr";
516      }
517      else if (prev_was_child) {
518    metaoption.pre_tree_traverse += ".fc";
519      }
520
521      prev_was_child  = true;
522      prev_was_parent = false;
523    }
524    else {
525      prev_was_child  = false;
526      prev_was_parent = false;
527      had_parent_or_child = false;
528    }
529  }
530
531  // parent/child can have sibling tacked on end also
532  if (meta.size() > 8 && (substr (meta.begin(), meta.begin()+7) == "sibling")) {
533    metaoption.metacommand |= mSibling;
534    get_sibling_options (meta, metaoption);
535  }
536 
537  // check for ex. which may occur in format statements
538  if (meta.size()>3 && (substr(meta.begin(), meta.begin()+3) == "ex.")) {
539    meta = substr (meta.begin()+3, meta.end());
540  }
541  metadata.insert (meta);
542  metaoption.metaname = meta;
543}
544
545static void parse_coll_meta(text_t &meta, metadata_t &metaoption) {
546  if (meta == "collection") {
547    // no qualifiers
548    metaoption.metaname = g_EmptyText;
549    return;
550  }
551  meta = substr (meta.begin()+11, meta.end());
552  metaoption.metaname = meta;
553 
554}
555
556static void parse_meta (text_t &meta, format_t *formatlistptr,
557            text_tset &metadata, bool &getParents) {
558 
559  if (meta == "link")
560    formatlistptr->command = comLink;
561  else if (meta == "/link")
562    formatlistptr->command = comEndLink;
563
564  else if (meta == "href")
565    formatlistptr->command = comHref;
566
567  else if (meta == "num")
568    formatlistptr->command = comNum;
569
570  else if (meta == "icon")
571    formatlistptr->command = comIcon;
572
573  else if (meta == "Text")
574    formatlistptr->command = comDoc;
575 
576  else if (meta == "RelatedDocuments")
577   formatlistptr->command = comRel;
578
579  else if (meta == "highlight")
580    formatlistptr->command = comHighlight;
581
582  else if (meta == "/highlight")
583    formatlistptr->command = comEndHighlight;
584
585  else if (meta == "Summary")
586    formatlistptr->command = comSummary;
587
588  else if (meta == "DocImage")
589    formatlistptr->command = comImage;
590
591  else if (meta == "DocTOC")
592    formatlistptr->command = comTOC;
593
594  else if (meta == "DocumentButtonDetach")
595     formatlistptr->command = comDocumentButtonDetach;
596 
597  else if (meta == "DocumentButtonHighlight")
598     formatlistptr->command = comDocumentButtonHighlight;
599 
600  else if (meta == "DocumentButtonExpandContents")
601    formatlistptr->command = comDocumentButtonExpandContents;
602
603  else if (meta == "DocumentButtonExpandText")
604     formatlistptr->command = comDocumentButtonExpandText;
605
606  else if (meta == "DocOID")
607     formatlistptr->command = comOID;
608  else if (meta == "DocTopOID")
609    formatlistptr->command = comTopOID;
610  else if (meta == "DocRank")
611    formatlistptr->command = comRank;
612  else if (meta == "DocTermsFreqTotal")
613    formatlistptr->command = comDocTermsFreqTotal;
614  else if (meta.size() >= 10 && (substr(meta.begin(), meta.begin()+10) == "collection")) {
615    formatlistptr->command = comCollection;
616    parse_coll_meta(meta, formatlistptr->meta);
617  }
618  else {
619    formatlistptr->command = comMeta;
620    parse_meta (meta, formatlistptr->meta, metadata, getParents);
621  }
622}
623
624
625static bool parse_string (const text_t &formatstring, format_t *formatlistptr,
626              text_tset &metadata, bool &getParents) {
627
628  text_t text;
629  text_t::const_iterator here = formatstring.begin();
630  text_t::const_iterator end = formatstring.end();
631
632  while (here != end) {
633
634    if (*here == '\\') {
635      ++here;
636      if (here != end) text.push_back (*here);
637
638    } else if (*here == '{') {
639      if (!text.empty()) {
640    formatlistptr->command = comText;
641    formatlistptr->text = text;
642    formatlistptr->nextptr = new format_t();
643    formatlistptr = formatlistptr->nextptr;
644   
645    text.clear();
646      }
647      if (parse_action (++here, end, formatlistptr, metadata, getParents)) {
648
649    formatlistptr->nextptr = new format_t();
650    formatlistptr = formatlistptr->nextptr;
651    if (here == end) break;
652      }
653    } else if (*here == '[') {
654      if (!text.empty()) {
655    formatlistptr->command = comText;
656    formatlistptr->text = text;
657    formatlistptr->nextptr = new format_t();
658    formatlistptr = formatlistptr->nextptr;
659
660    text.clear();
661      }
662      text_t meta;
663      ++here;
664      while (*here != ']') {
665    if (here == end) return false;
666    meta.push_back (*here);
667    ++here;
668      }
669      parse_meta (meta, formatlistptr, metadata, getParents);
670      formatlistptr->nextptr = new format_t();
671      formatlistptr = formatlistptr->nextptr;
672
673    } else
674      text.push_back (*here);
675
676    if (here != end) ++here;
677  }
678  if (!text.empty()) {
679    formatlistptr->command = comText;
680    formatlistptr->text = text;
681    formatlistptr->nextptr = new format_t();
682    formatlistptr = formatlistptr->nextptr;
683
684  }
685  return true;
686}
687
688
689static bool parse_action (text_t::const_iterator &here, const text_t::const_iterator &end,
690              format_t *formatlistptr, text_tset &metadata, bool &getParents) {
691
692  text_t::const_iterator it = findchar (here, end, '}');
693  if (it == end) return false;
694
695  text_t com = substr (here, it);
696  here = findchar (it, end, '{');
697  if (here == end) return false;
698  else ++here;
699
700  if (com == "If" || com == "if" || com == "IF") formatlistptr->command = comIf;
701  else if (com == "Or" || com == "or" || com == "OR") formatlistptr->command = comOr;
702  else return false;
703
704  int commacount = 0;
705  text_t text;
706  while (here != end) {
707
708    if (*here == '\\') {
709      ++here;
710      if (here != end) text.push_back(*here);
711     
712    }
713 
714    else if (*here == ',' || *here == '}' || *here == '{') {
715
716      if (formatlistptr->command == comOr) {
717    // the {Or}{this, or this, or this, or this} statement
718    format_t *or_ptr;
719   
720    // find the next unused orptr
721    if (formatlistptr->orptr == NULL) {
722      formatlistptr->orptr = new format_t();
723      or_ptr = formatlistptr->orptr;
724    } else {
725      or_ptr = formatlistptr->orptr;
726      while (or_ptr->nextptr != NULL)
727        or_ptr = or_ptr->nextptr;
728      or_ptr->nextptr = new format_t();
729      or_ptr = or_ptr->nextptr;
730    }
731
732    if (!text.empty())
733      {
734        if (!parse_string(text, or_ptr, metadata, getParents)) { return false; }
735      }
736
737    if (*here == '{')
738      {
739        // Supports: {Or}{[Booktitle],[Title],{If}{[XXXX],aaa,bbb}}
740        // but not : {Or}{[Booktitle],[Title]{If}{[XXXX],aaa,bbb}}
741        // The latter can always be re-written:
742        // {Or}{[Booktitle],{If}{[Title],[Title]{If}{[XXXX],aaa,bbb}}}
743       
744        if (!text.empty()) // already used up allocated format_t
745          {
746        // => allocate new one for detected action
747        or_ptr->nextptr = new format_t();
748        or_ptr = or_ptr->nextptr;
749          }
750        if (!parse_action(++here, end, or_ptr, metadata, getParents))
751          {
752        return false;
753          }
754      }
755    else
756      {
757        if (*here == '}') break;
758      }
759    text.clear();
760
761      }
762
763      // Parse an {If}{decide,do,else} statement
764      else {
765   
766    // Read the decision component. 
767    if (commacount == 0) {
768      // Decsion can be a metadata element, or a piece of text.
769      // Originally Stefan's code, updated 25/10/2000 by Gordon.
770
771      text_t::const_iterator beginbracket = text.begin();
772      text_t::const_iterator endbracket = (text.end() - 1);
773
774      // Decision is based on a metadata element
775      if ((*beginbracket == '[') && (*endbracket == ']')) {
776        // Ignore the surrounding square brackets
777        text_t meta = substr (beginbracket+1, endbracket);
778        parse_meta (meta, formatlistptr->decision.meta, metadata, getParents);
779        ++commacount;
780        text.clear();
781      }
782
783      // Decision is a piece of text (probably a macro like _cgiargmode_).
784      else {
785
786        // hunt for any metadata in string, which might be uses in
787        // to test a condition, e.g. [Format] eq 'PDF'
788        format_t* dummyformat = new format_t();
789        // update which metadata fields needed
790        // (not interested in updatng formatlistptr)
791        parse_string (text, dummyformat, metadata, getParents);
792        delete dummyformat;
793
794        formatlistptr->decision.command = dText;
795        formatlistptr->decision.text = text;
796        ++commacount;
797        text.clear();
798      }
799    }
800
801    // Read the "then" and "else" components of the {If} statement.
802    else {
803      format_t** nextlistptr = NULL;
804      if (commacount == 1) {
805        nextlistptr = &formatlistptr->ifptr;
806      } else if (commacount == 2 ) {
807        nextlistptr = &formatlistptr->elseptr;
808      } else {
809        return false;
810      }
811
812      if (!text.empty()) {
813        if (*nextlistptr == NULL) {
814          *nextlistptr = new format_t();
815        } else {
816
817          // skip to the end of any format_t statements already added
818          while ((*nextlistptr)->nextptr != NULL)
819          {
820        nextlistptr = &(*nextlistptr)->nextptr;
821          }
822
823          (*nextlistptr)->nextptr = new format_t();
824          nextlistptr = &(*nextlistptr)->nextptr;
825        }
826
827        if (!parse_string (text, *nextlistptr, metadata, getParents))
828          {
829        return false;
830          }
831        text.clear();
832      }
833     
834      if (*here == '{')
835        {
836          if (*nextlistptr == NULL) {
837        *nextlistptr = new format_t();
838          } else {
839        // skip to the end of any format_t statements already added
840        while ((*nextlistptr)->nextptr != NULL)
841          {
842            nextlistptr = &(*nextlistptr)->nextptr;
843          }
844
845        (*nextlistptr)->nextptr = new format_t();
846        nextlistptr = &(*nextlistptr)->nextptr;
847          }
848
849          if (!parse_action(++here, end, *nextlistptr, metadata, getParents))
850        {
851          return false;
852        }
853        }
854      else
855        {
856          if (*here == '}') break;
857          ++commacount;
858        }
859    }
860      }
861     
862    } else text.push_back(*here);
863   
864    if (here != end) ++here;
865  }
866
867  return true;
868}
869
870
871bool parse_formatstring (const text_t &formatstring, format_t *formatlistptr,
872             text_tset &metadata, bool &getParents) {
873
874  formatlistptr->clear();
875  getParents = false;
876
877  return (parse_string (formatstring, formatlistptr, metadata, getParents));
878}
879
880// position -1 for all, -2 for the last, 0 for the first, or x for a particular piece
881// metainfo has all the values for the metadata, meta has the request metadata name and options, position tells which values to get
882static text_t get_formatted_meta_text(MetadataInfo_t &metainfo, const metadata_t &meta, int position, bool no_cgisafe = false)
883{
884  text_t no_ns_metaname = remove_namespace(meta.metaname);
885  text_t formatted_metatext;
886  bool first = true;
887 
888  const int start_i=0;
889  const int end_i = metainfo.values.size()-1;
890 
891  if (position == -1) { // all
892    for (int i=start_i; i<=end_i; ++i) {
893      if (!first) formatted_metatext += meta.siblingoptions;
894     
895      text_t fresh_metatext;
896
897      if (meta.metacommand & mSpecial) {
898    // special formatting
899    if (no_ns_metaname == "Date") fresh_metatext = format_date (metainfo.values[i]);
900    else if (no_ns_metaname == "Language") fresh_metatext = iso639(metainfo.values[i]);
901    else fresh_metatext = "_format:"+meta.metaname+"_("+metainfo.values[i]+")";
902      }
903      else fresh_metatext = metainfo.values[i];
904
905      if (spanwrap) {
906    fresh_metatext = "<span metaname=\"" + meta.metaname + "\" metapos=\"" + i + "\">" + fresh_metatext + "</span>";
907      }
908      formatted_metatext += fresh_metatext;
909
910      first = false;
911     
912    }
913  } else {
914    if (position == -2) { // end
915      position = end_i;
916    } else if (position < start_i || position > end_i) {
917      return "";
918    }
919
920    text_t fresh_metatext;
921    if (meta.metacommand & mSpecial) {
922
923      // special formatting
924      if (no_ns_metaname == "Date") fresh_metatext = format_date (metainfo.values[position]);
925      else if (no_ns_metaname == "Language") fresh_metatext = iso639(metainfo.values[position]);
926      else fresh_metatext = "_format:"+meta.metaname+"_("+metainfo.values[position]+")";
927    }
928    else fresh_metatext = metainfo.values[position];
929
930    if (spanwrap) {
931      fresh_metatext = "<span metaname=\"" + meta.metaname + "\" metapos=\"" + position + "\">" + fresh_metatext + "</span>";
932    }
933
934    formatted_metatext += fresh_metatext;
935  }
936
937  if (meta.metacommand & mCgiSafe && !no_cgisafe) return cgi_safe_unicode (formatted_metatext);
938  else return formatted_metatext;
939}
940
941static text_t get_parent_meta (ResultDocInfo_t &docinfo, const metadata_t &meta, int siblings_values)
942{
943 
944  MetadataInfo_t *parent = docinfo.metadata[meta.metaname].parent;
945  switch (meta.mqualifier.parent) {
946  case pNone:
947    return "Nothing!!";
948    break;
949
950  case pImmediate:
951    if (parent != NULL) {
952      return get_formatted_meta_text(*parent, meta, siblings_values);
953    }
954    break;
955
956  case pTop:
957    if (parent != NULL) {
958      while (parent->parent != NULL) parent = parent->parent;
959      return get_formatted_meta_text(*parent, meta, siblings_values);
960    }
961    break;
962
963  case pAll:
964    MetadataInfo_t *parent = docinfo.metadata[meta.metaname].parent;
965    if (parent != NULL) {
966      text_tarray tmparray;
967      while (parent != NULL) {
968    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
969    parent = parent->parent;
970      }
971      // now join them up - use teh parent separator
972      bool first = true;
973      text_t tmp;
974      text_tarray::reverse_iterator here = tmparray.rbegin();
975      text_tarray::reverse_iterator end = tmparray.rend();
976      while (here != end) {
977    if (!first) tmp += meta.parentoptions;
978    tmp += *here;
979    first = false;
980    ++here;
981      }
982      if (meta.metacommand & mCgiSafe) return cgi_safe_unicode (tmp);
983      else return tmp;
984    }
985  }
986  return "";
987
988}
989
990static text_t get_child_meta (const text_t& collection,
991                  recptproto* collectproto,
992                  ResultDocInfo_t &docinfo, displayclass &disp,
993                  const metadata_t &meta, text_tmap &options,
994                  ostream& logout, int siblings_values)
995{
996  if (docinfo.metadata["contains"].values[0].size()==0) return ""; // no children
997 
998  const text_t& pre_tree_trav = meta.pre_tree_traverse;
999  const text_t& child_metaname = meta.metaname;
1000  const text_t& child_field = meta.childoptions;
1001  text_tset child_metadata;
1002  child_metadata.insert(child_metaname);
1003
1004  FilterResponse_t child_response;
1005  if (meta.mqualifier.child == cNum) {
1006    // just one child
1007    //get the information associated with the metadata for child doc
1008    if (!get_info (docinfo.OID+pre_tree_trav+child_field, collection, "",
1009           child_metadata, false, collectproto, child_response,
1010           logout)) return ""; // invalid child number
1011
1012      if (child_response.docInfo.empty()) return false; // no info for the child
1013 
1014      ResultDocInfo_t& child_docinfo = child_response.docInfo[0];
1015      MetadataInfo_t& metaname_rec = child_docinfo.metadata[child_metaname];
1016 
1017      text_t child_metavalue = get_formatted_meta_text(metaname_rec,meta,siblings_values);
1018      return expand_metadata(child_metavalue,collection,collectproto,
1019             child_docinfo,disp,options,logout);
1020  }
1021 
1022   
1023  if (meta.mqualifier.child != cAll) return false; // invalid qualifier
1024
1025
1026  if (!pre_tree_trav.empty()) {
1027    // need to get relevant "contains" metadata for new (e.g. pre tree trav) node
1028    FilterResponse_t trav_response;
1029
1030    text_tset trav_metadata;
1031    trav_metadata.insert("contains");
1032
1033    if (!get_info (docinfo.OID+pre_tree_trav, collection, "",
1034           trav_metadata, false, collectproto, trav_response,
1035           logout)) return ""; // invalid pre_tree_trav
1036
1037    if (trav_response.docInfo.empty()) return false; // no info for the pre_tree_trav OID
1038 
1039    ResultDocInfo_t& trav_docinfo = trav_response.docInfo[0];
1040    // use this for rest of routine
1041    docinfo = trav_docinfo;
1042  }
1043 
1044  // we need to get all children
1045  text_t result = "";
1046  text_tarray children;
1047  text_t contains = docinfo.metadata["contains"].values[0];
1048  splitchar (contains.begin(), contains.end(), ';', children);
1049  text_tarray::const_iterator here = children.begin();
1050  text_tarray::const_iterator end = children.end();
1051  bool first = true;
1052  while (here !=end) {
1053    text_t oid = *here;
1054    here++;
1055    if (*(oid.begin()) == '"') translate_parent (oid, docinfo.OID);
1056   
1057    //get the information associated with the metadata for child doc
1058    if (!get_info (oid, collection, "", child_metadata,
1059           false, collectproto, child_response, logout) ||
1060    child_response.docInfo.empty()) {
1061      first = false;
1062      continue;
1063    }
1064   
1065   
1066    ResultDocInfo_t& child_docinfo = child_response.docInfo[0];
1067    MetadataInfo_t& metaname_rec = child_docinfo.metadata[child_metaname];
1068   
1069    text_t child_metavalue = get_formatted_meta_text(metaname_rec,meta,siblings_values);
1070    if (!first) result += child_field;
1071    first = false;
1072    // need to do this here cos otherwise we are in the wrong document
1073    result +=  expand_metadata(child_metavalue,collection,collectproto,
1074                   child_docinfo,disp,options,logout);
1075  }
1076  return result;
1077   
1078}
1079
1080static text_t get_meta (const text_t& collection, recptproto* collectproto,
1081            ResultDocInfo_t &docinfo, displayclass &disp,
1082            const metadata_t &meta, text_tmap &options,
1083            ostream& logout) {
1084 
1085  // make sure we have the requested metadata
1086  MetadataInfo_tmap::iterator it = docinfo.metadata.find (meta.metaname);
1087  if (it == docinfo.metadata.end()) return "";
1088
1089  int siblings_values = 0; // default is no siblings, just the first metadata available
1090  if (meta.metacommand & mSibling) {
1091    if (meta.mqualifier.sibling == sAll) {
1092      siblings_values = -1; //all
1093    } else if (meta.mqualifier.sibling == sNum) {
1094      siblings_values = meta.siblingoptions.getint();
1095    }
1096  }
1097  if (meta.metacommand & mParent) {
1098    return get_parent_meta(docinfo,meta,siblings_values);
1099  }
1100
1101  else if (meta.metacommand & mChild) {
1102    return get_child_meta(collection,collectproto,docinfo,disp,meta,
1103                options,logout, siblings_values);
1104  }
1105  else if (meta.metacommand & mSibling) { // only siblings
1106    MetadataInfo_t& metaname_rec = docinfo.metadata[meta.metaname];
1107    return get_formatted_meta_text(docinfo.metadata[meta.metaname],meta, siblings_values);
1108  }
1109  else {
1110   
1111    // straightforward metadata request (nothing fancy)
1112
1113    text_t classifier_metaname = docinfo.classifier_metadata_type;
1114    int metaname_index
1115      = (classifier_metaname == meta.metaname) ? docinfo.classifier_metadata_offset : 0;
1116    return get_formatted_meta_text(docinfo.metadata[meta.metaname], meta, metaname_index);
1117  }
1118 
1119  return "";
1120}
1121
1122static text_t get_or (const text_t& collection, recptproto* collectproto,
1123              ResultDocInfo_t &docinfo, displayclass &disp,
1124              format_t *orptr, text_tmap &options,
1125              ostream& logout) {
1126
1127  text_t tmp;
1128  while (orptr != NULL) {
1129
1130    tmp = format_string (collection,collectproto,docinfo, disp, orptr,
1131             options, logout);
1132    if (!tmp.empty()) return tmp;
1133
1134    orptr = orptr->nextptr;
1135  }
1136  return "";
1137}
1138
1139static bool char_is_whitespace(const char c)
1140{
1141  return ((c == ' ') || (c == '\t') || (c == '\n') || (c == '\r'));
1142
1143}
1144
1145static int scan_over_whitespace(const text_t& outstring, const int start_pos)
1146{
1147  int pos = start_pos;
1148  while (pos<outstring.size()) {
1149    if (!char_is_whitespace(outstring[pos])) {
1150      break;
1151    }
1152    ++pos;
1153  }
1154
1155  return pos;
1156}
1157
1158static int rscan_over_whitespace(const text_t& outstring, const int start_pos)
1159{
1160  int pos = start_pos;
1161  while (pos>=0) {
1162    if (!char_is_whitespace(outstring[pos])) {
1163      break;
1164    }
1165    --pos;
1166  }
1167
1168  return pos;
1169}
1170
1171static int rscan_for_whitespace(const text_t& outstring, const int start_pos)
1172{
1173  int pos = start_pos;
1174  while (pos>=0) {
1175    if (char_is_whitespace(outstring[pos])) {
1176      break;
1177    }
1178    --pos;
1179  }
1180
1181  return pos;
1182}
1183
1184
1185static int rscan_for(const text_t& outstring, const int start_pos,
1186             const char find_c)
1187{
1188  int pos = start_pos;
1189  while (pos>=0) {
1190    char c = outstring[pos];
1191    if (outstring[pos] == find_c) {
1192      break;
1193    }
1194    --pos;
1195  }
1196
1197  return pos;
1198}
1199
1200text_t extract_substr(const text_t& outstring, const int start_pos,
1201              const int end_pos)
1202{
1203  text_t extracted_str;
1204  extracted_str.clear();
1205
1206  for (int pos=start_pos; pos<=end_pos; ++pos) {
1207    extracted_str.push_back(outstring[pos]);
1208  }
1209
1210  return extracted_str;
1211}
1212
1213
1214static text_t expand_potential_metadata(const text_t& collection,
1215                    recptproto* collectproto,
1216                    ResultDocInfo_t &docinfo,
1217                    displayclass &disp,
1218                    const text_t& intext,
1219                    text_tmap &options,
1220                    ostream& logout)
1221{
1222  text_t outtext;
1223
1224  // decide if dealing with metadata or text
1225
1226  text_t::const_iterator beginbracket = intext.begin();
1227  text_t::const_iterator endbracket = (intext.end() - 1);
1228
1229  // Decision is based on a metadata element
1230  if ((*beginbracket == '[') && (*endbracket == ']')) {
1231    // Ignore the surrounding square brackets
1232    text_t meta_text = substr (beginbracket+1, endbracket);
1233
1234    if (meta_text == "Text") {
1235      outtext = format_text(collection, collectproto, docinfo, disp, options, logout);
1236    } else {
1237
1238      text_tset metadata;
1239      bool getParents =false;
1240      metadata_t meta;
1241     
1242      parse_meta (meta_text, meta, metadata, getParents);   
1243      outtext
1244    = get_meta (collection,collectproto,docinfo,disp,meta,options,logout);
1245    }
1246
1247  }
1248  else {
1249    outtext = intext;
1250  }
1251
1252  return outtext;
1253}
1254
1255
1256
1257
1258static bool uses_expression(const text_t& collection, recptproto* collectproto,
1259                ResultDocInfo_t &docinfo,
1260                displayclass &disp,
1261                const text_t& outstring, text_t& lhs_expr,
1262                text_t& op_expr, text_t& rhs_expr,
1263                text_tmap &options,
1264                ostream& logout)
1265{
1266  // Note: the string may not be of the form: str1 op str2, however
1267  // to deterine this we have to process it on the assumption it is,
1268  // and if at any point an 'erroneous' value is encountered, return
1269  // false and let something else have a go at evaluating it
1270
1271  // Starting at the end of the string and working backwards ..
1272
1273  const int outstring_len = outstring.size();
1274
1275  // skip over white space
1276  int rhs_end = rscan_over_whitespace(outstring,outstring_len-1);
1277
1278  if (rhs_end<=0) {
1279    // no meaningful text or (rhs_end==0) no room for operator
1280    return false;
1281  }
1282
1283  // check for ' or " and then scan over token
1284  const char potential_quote = outstring[rhs_end];
1285  int rhs_start=rhs_end;
1286  bool quoted = false;
1287
1288  if ((potential_quote == '\'') || (potential_quote == '\"')) {
1289    --rhs_end;
1290    rhs_start = rscan_for(outstring,rhs_end-1,potential_quote) +1;
1291    quoted = true;
1292  }
1293  else {
1294    rhs_start = rscan_for_whitespace(outstring,rhs_end-1) +1;
1295  }
1296
1297  if ((rhs_end-rhs_start)<0) {
1298    // no meaningful rhs expression
1299    return false;
1300  }
1301
1302  // form rhs_expr
1303  rhs_expr = extract_substr(outstring,rhs_start,rhs_end);
1304
1305  // skip over white space
1306  const int to_whitespace = (quoted) ? 2 : 1;
1307
1308  int op_end = rscan_over_whitespace(outstring,rhs_start-to_whitespace);
1309  int op_start = rscan_for_whitespace(outstring,op_end-1)+1;
1310
1311  if ((op_end<0) && (op_start<0)) {
1312    // no meaningful expression operator
1313    return false;
1314  }
1315
1316  if (op_end-op_start<0) {
1317    // no meaningful expression operator
1318    return false;
1319  }
1320
1321  op_expr = extract_substr(outstring,op_start,op_end);
1322
1323
1324  // check for operator
1325  if ((op_expr != "eq") && (op_expr != "ne") && (op_expr != "gt") &&
1326     (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")) {
1327
1328    // not a valid operator
1329    return false;
1330  }
1331
1332  int lhs_end = rscan_over_whitespace(outstring,op_start-1);
1333  if (lhs_end<0) {
1334    // no meaningful lhs expression
1335    return false;
1336  }
1337
1338  int lhs_start = scan_over_whitespace(outstring,0);
1339
1340  // form lhs_expr from remainder of string
1341  lhs_expr = extract_substr(outstring,lhs_start,lhs_end);
1342
1343  // Now we know we have a valid expression, look up any
1344  // metadata terms
1345
1346  rhs_expr = expand_potential_metadata(collection,collectproto,docinfo,
1347                       disp,rhs_expr,options,logout);
1348  lhs_expr = expand_potential_metadata(collection,collectproto,docinfo,
1349                       disp,lhs_expr,options,logout);
1350
1351  return true;
1352}
1353
1354static bool eval_expression_true(const text_t& lhs_expr,const text_t& op_expr,
1355                 const text_t& rhs_expr, ostream& logout)
1356{
1357  if (op_expr == "eq") return (lhs_expr == rhs_expr);
1358  else if (op_expr == "ne" ) return (lhs_expr != rhs_expr);
1359  else if (op_expr == "gt") return (lhs_expr > rhs_expr);
1360  else if (op_expr == "ge") return (lhs_expr >= rhs_expr);
1361  else if (op_expr == "lt") return (lhs_expr < rhs_expr);
1362  else if (op_expr == "le") return (lhs_expr <= rhs_expr);
1363  else if (op_expr == "==") return (lhs_expr.getint() == rhs_expr.getint());
1364  else if (op_expr == "!=") return (lhs_expr.getint() != rhs_expr.getint());
1365  else if (op_expr == ">") return (lhs_expr.getint() > rhs_expr.getint());
1366  else if (op_expr == ">=") return (lhs_expr.getint() >= rhs_expr.getint());
1367  else if (op_expr == "<") return (lhs_expr.getint() < rhs_expr.getint());
1368  else if (op_expr == "<=") return (lhs_expr.getint() <= rhs_expr.getint());
1369  else if (op_expr == "sw") return (starts_with(lhs_expr,rhs_expr));
1370  else if (op_expr == "ew") return (ends_with(lhs_expr,rhs_expr));
1371  else {
1372    logout << "Error: '" << op_expr << "' is not a recognised operator." << endl;
1373  }
1374 
1375  return false;
1376}
1377
1378
1379static text_t get_if (const text_t& collection, recptproto* collectproto,
1380              ResultDocInfo_t &docinfo, displayclass &disp,
1381              const decision_t &decision,
1382              format_t *ifptr, format_t *elseptr,
1383              text_tmap &options, ostream& logout)
1384{
1385  // If the decision component is a metadata element, then evaluate it
1386  // to see whether we output the "then" or the "else" clause
1387  if (decision.command == dMeta) {
1388
1389    bool store_spanwrap = spanwrap;
1390    spanwrap = 0;
1391
1392    // temporarily suspend spanwrap (if on) so can test if metadata item really exits or not
1393    bool metadata_exists
1394      = (get_meta (collection,collectproto,docinfo,disp,decision.meta,options,
1395           logout) != "");
1396
1397    spanwrap = store_spanwrap;
1398
1399    if (metadata_exists) {
1400      if (ifptr != NULL)
1401    return get_formatted_string (collection,collectproto,docinfo, disp, ifptr,
1402                     options, logout);
1403    }
1404    else {
1405      if (elseptr != NULL)
1406    return get_formatted_string (collection,collectproto,docinfo, disp, elseptr,
1407                     options, logout);
1408    }
1409  }
1410
1411  // If the decision component is text, then evaluate it (it is probably a
1412  // macro like _cgiargmode_) to decide what to output.
1413  else if (decision.command == dText) {
1414
1415    text_t outstring;
1416    disp.expandstring (decision.text, outstring);
1417
1418    // Check for if expression in form: str1 op str2
1419    // (such as [x] eq "y")
1420    text_t lhs_expr, op_expr, rhs_expr;
1421    if (uses_expression(collection,collectproto,docinfo, disp, outstring,lhs_expr,op_expr,rhs_expr, options,logout)) {
1422      if (eval_expression_true(lhs_expr,op_expr,rhs_expr,logout)) {
1423    if (ifptr != NULL) {
1424      return get_formatted_string (collection, collectproto, docinfo, disp, ifptr,
1425                       options, logout);
1426    }
1427    else {
1428      return "";
1429    }
1430      } else {
1431    if (elseptr != NULL) {
1432      return get_formatted_string (collection, collectproto, docinfo, disp, elseptr,
1433                       options, logout);
1434    }
1435    else {
1436      return "";
1437    }
1438      }
1439    }
1440
1441
1442    // This is a tad tricky.  When we expand a string like _cgiargmode_, that is
1443    // a cgi argument macro that has not been set, it evaluates to itself.
1444    // Therefore, were have to say that a piece of text evalautes true if
1445    // it is non-empty and if it is a cgi argument evaulating to itself.
1446
1447    if ((outstring != "") && !((outstring == decision.text) && (outstring[0] == '_'))) {
1448      if (ifptr != NULL)
1449    return get_formatted_string (collection, collectproto, docinfo, disp, ifptr,
1450                     options, logout);
1451    } else {
1452      if (elseptr != NULL)
1453    return get_formatted_string (collection, collectproto, docinfo, disp, elseptr,
1454                     options, logout);
1455    }
1456  }
1457 
1458  return "";
1459}
1460
1461bool includes_metadata(const text_t& text)
1462{
1463  text_t::const_iterator here = text.begin();
1464  text_t::const_iterator end = text.end();
1465  while (here != end) {
1466    if (*here == '[') return true;
1467    ++here;
1468  }
1469
1470  return false;
1471}
1472
1473static text_t expand_metadata(const text_t &metavalue, const text_t& collection,
1474                  recptproto* collectproto,
1475                  ResultDocInfo_t &docinfo,
1476                  displayclass &disp, text_tmap &options,
1477                  ostream &logout) {
1478     
1479  if (includes_metadata(metavalue)) {
1480   
1481    // text has embedded metadata in it => expand it
1482    FilterRequest_t request;
1483    FilterResponse_t response;
1484   
1485    request.getParents = false;
1486   
1487    format_t *expanded_formatlistptr = new format_t();
1488    parse_formatstring (metavalue, expanded_formatlistptr,
1489            request.fields, request.getParents);
1490   
1491    // retrieve metadata
1492    get_info(docinfo.OID, collection, "", request.fields, request.getParents,
1493         collectproto, response, logout);
1494   
1495    if (!response.docInfo.empty()) {
1496     
1497      text_t expanded_metavalue
1498    = get_formatted_string(collection, collectproto,
1499                   response.docInfo[0], disp, expanded_formatlistptr,
1500                   options, logout);
1501     
1502      return expanded_metavalue;
1503    }
1504    else {
1505      return metavalue;
1506    }
1507  }
1508  else {
1509   
1510    return metavalue;
1511  }
1512}
1513
1514text_t get_collection_meta(const text_t& collection, recptproto* collectproto,
1515               displayclass &disp,
1516               text_t meta_name, ostream& logout) {
1517 
1518  ColInfoResponse_t collectinfo;
1519  comerror_t err;
1520  collectproto->get_collectinfo (collection, collectinfo,err,logout);
1521  text_t meta_value = "";
1522  text_t lang;
1523  disp.expandstring("_cgiargl_",lang);
1524  if (lang.empty()) {
1525    lang = "en";
1526  }
1527
1528  if (err == noError) {
1529    meta_value = collectinfo.get_collectionmeta(meta_name, lang);
1530  }
1531  return meta_value;
1532 
1533
1534}
1535text_t format_string (const text_t& collection, recptproto* collectproto,
1536              ResultDocInfo_t &docinfo, displayclass &disp,
1537              format_t *formatlistptr, text_tmap &options,
1538              ostream& logout) {
1539
1540  if (formatlistptr == NULL) return "";
1541
1542  switch (formatlistptr->command) {
1543     case comOID:
1544    return docinfo.OID;
1545  case comTopOID:
1546    {
1547      text_t top_id;
1548      get_top(docinfo.OID, top_id);
1549      return top_id;
1550    }
1551  case comRank:
1552    return text_t(docinfo.ranking);
1553     case comText:
1554    return formatlistptr->text;
1555     case comLink:
1556    return options["link"];
1557     case comEndLink:
1558    if (options["link"].empty()) return "";
1559    else return "</a>";
1560     case comHref:
1561    return get_href(options["link"]);
1562     case comIcon:
1563    return options["icon"];
1564     case comNum:
1565    return docinfo.result_num;
1566     case comRel: //if [RelatedDocuments] appears in format string, collect relation data
1567    return get_related_docs(collection, collectproto, docinfo, logout);
1568     case comSummary:
1569    return format_summary(collection, collectproto, docinfo, disp, options, logout);
1570     case comMeta:
1571    {
1572       const text_t& metavalue =  get_meta (collection,collectproto, docinfo, disp,formatlistptr->meta,options,logout);
1573       return expand_metadata(metavalue, collection, collectproto, docinfo, disp, options, logout);
1574    }
1575     case comDoc:
1576    return format_text(collection, collectproto, docinfo, disp, options, logout);
1577    //return options["text"];
1578     case comImage:
1579    return expand_metadata(options["DocImage"], collection, collectproto, docinfo, disp, options, logout);
1580     case comTOC:
1581    return options["DocTOC"];
1582     case comDocumentButtonDetach:
1583    return options["DocumentButtonDetach"];
1584     case comDocumentButtonHighlight:
1585    return options["DocumentButtonHighlight"];
1586     case comDocumentButtonExpandContents:
1587    return options["DocumentButtonExpandContents"];
1588     case comDocumentButtonExpandText:
1589    return options["DocumentButtonExpandText"];
1590     case comHighlight:
1591    if (options["highlight"] == "1") return "<b>";
1592    break;
1593     case comEndHighlight:
1594    if (options["highlight"] == "1") return "</b>";
1595    break;
1596     case comIf:
1597    return get_if (collection, collectproto, docinfo, disp,
1598               formatlistptr->decision, formatlistptr->ifptr,
1599               formatlistptr->elseptr, options, logout);
1600     case comOr:
1601    return get_or (collection,collectproto, docinfo, disp, formatlistptr->orptr,
1602               options, logout);
1603     case comDocTermsFreqTotal:
1604       return docinfo.num_terms_matched;
1605     case comCollection:
1606       if (formatlistptr->meta.metaname == g_EmptyText) {
1607     return collection;
1608       }
1609       return get_collection_meta(collection, collectproto, disp, formatlistptr->meta.metaname, logout);
1610   
1611  }
1612  return "";
1613}
1614
1615text_t get_formatted_string (const text_t& collection, recptproto* collectproto,
1616                 ResultDocInfo_t &docinfo, displayclass &disp,
1617                 format_t *formatlistptr, text_tmap &options,
1618                 ostream& logout) {
1619
1620   text_t ft;
1621   while (formatlistptr != NULL)
1622      {
1623     ft += format_string (collection, collectproto, docinfo, disp, formatlistptr,
1624                  options, logout);
1625     formatlistptr = formatlistptr->nextptr;
1626      }
1627   
1628   return ft;
1629}
1630
1631
1632// 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?
1633text_t format_text (const text_t& collection, recptproto* collectproto,
1634            ResultDocInfo_t &docinfo, displayclass &disp,
1635            text_tmap &options, ostream& logout) {
1636  if(!options["text"].empty()) {
1637    return options["text"];
1638  }
1639  // else get document text here
1640  DocumentRequest_t docrequest;
1641  DocumentResponse_t docresponse;
1642  comerror_t err;
1643  docrequest.OID = docinfo.OID;
1644  collectproto->get_document (collection, docrequest, docresponse, err, logout);
1645  return docresponse.doc;
1646
1647}
1648 
1649/* FUNCTION NAME: format_summary
1650 * DESC: this is invoked when a [Summary] special metadata is processed.
1651 * RETURNS: a query-biased summary for the document */
1652
1653text_t format_summary (const text_t& collection, recptproto* collectproto,
1654               ResultDocInfo_t &docinfo, displayclass &disp,
1655               text_tmap &options, ostream& logout) {
1656
1657  // GRB: added code here to ensure that the cstr (and other collections)
1658  //      uses the document metadata item Summary, rather than compressing
1659  //      the text of the document, processed via the methods in
1660  //      summarise.cpp
1661  if (docinfo.metadata.count("Summary") > 0 &&
1662      docinfo.metadata["Summary"].values.size() > 0) {
1663    return docinfo.metadata["Summary"].values[0];
1664  }
1665
1666  text_t textToSummarise, query;
1667  if(options["text"].empty()) { // get document text
1668     DocumentRequest_t docrequest;
1669     DocumentResponse_t docresponse;
1670     comerror_t err;
1671     docrequest.OID = docinfo.OID;
1672     collectproto->get_document (collection, docrequest, docresponse, err, logout);
1673     textToSummarise = docresponse.doc;
1674  } else // in practice, this would not happen, because text is only
1675         // loaded with the [Text] command
1676     textToSummarise = options["text"];
1677  disp.expandstring("_cgiargq_",query);
1678  return summarise(textToSummarise,query,80);
1679}
Note: See TracBrowser for help on using the browser.