| 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 | static bool metadata_spanwrap = false;
|
|---|
| 35 |
|
|---|
| 36 | // a few function prototypes
|
|---|
| 37 |
|
|---|
| 38 | static 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 |
|
|---|
| 43 | static bool parse_action (text_t::const_iterator &here, const text_t::const_iterator &end,
|
|---|
| 44 | format_t *formatlistptr, text_tset &metadata, bool &getParents);
|
|---|
| 45 |
|
|---|
| 46 | static text_t format_summary (const text_t& collection, recptproto* collectproto,
|
|---|
| 47 | ResultDocInfo_t &docinfo, displayclass &disp,
|
|---|
| 48 | text_tmap &options, ostream& logout);
|
|---|
| 49 | static text_t format_text (const text_t& collection, recptproto* collectproto,
|
|---|
| 50 | ResultDocInfo_t &docinfo, displayclass &disp,
|
|---|
| 51 | text_tmap &options, ostream& logout);
|
|---|
| 52 |
|
|---|
| 53 | static 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 |
|
|---|
| 59 | void 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 |
|
|---|
| 71 | void decision_t::clear() {
|
|---|
| 72 | command = dMeta;
|
|---|
| 73 | meta.clear();
|
|---|
| 74 | text.clear();
|
|---|
| 75 | }
|
|---|
| 76 |
|
|---|
| 77 | void 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 |
|
|---|
| 88 | void 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
|
|---|
| 109 | bool 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 |
|
|---|
| 127 | bool 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
|
|---|
| 138 | bool 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"
|
|---|
| 149 | bool 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 |
|
|---|
| 173 | text_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
|
|---|
| 186 | text_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??
|
|---|
| 217 | text_t iso639 (const text_t &langcode) {
|
|---|
| 218 | if (langcode.empty()) return "";
|
|---|
| 219 | return "_iso639:iso639"+langcode+"_";
|
|---|
| 220 | }
|
|---|
| 221 |
|
|---|
| 222 |
|
|---|
| 223 | text_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.
|
|---|
| 254 | text_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 |
|
|---|
| 333 | static 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 |
|
|---|
| 368 | static 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 |
|
|---|
| 416 | static 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 |
|
|---|
| 463 | static 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 |
|
|---|
| 545 | static 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 |
|
|---|
| 556 | static 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 == "metadata-spanwrap")
|
|---|
| 586 | formatlistptr->command = comMetadataSpanWrap;
|
|---|
| 587 |
|
|---|
| 588 | else if (meta == "/metadata-spanwrap")
|
|---|
| 589 | formatlistptr->command = comEndMetadataSpanWrap;
|
|---|
| 590 |
|
|---|
| 591 | else if (meta == "Summary")
|
|---|
| 592 | formatlistptr->command = comSummary;
|
|---|
| 593 |
|
|---|
| 594 | else if (meta == "DocImage")
|
|---|
| 595 | formatlistptr->command = comImage;
|
|---|
| 596 |
|
|---|
| 597 | else if (meta == "DocTOC")
|
|---|
| 598 | formatlistptr->command = comTOC;
|
|---|
| 599 |
|
|---|
| 600 | else if (meta == "DocumentButtonDetach")
|
|---|
| 601 | formatlistptr->command = comDocumentButtonDetach;
|
|---|
| 602 |
|
|---|
| 603 | else if (meta == "DocumentButtonHighlight")
|
|---|
| 604 | formatlistptr->command = comDocumentButtonHighlight;
|
|---|
| 605 |
|
|---|
| 606 | else if (meta == "DocumentButtonExpandContents")
|
|---|
| 607 | formatlistptr->command = comDocumentButtonExpandContents;
|
|---|
| 608 |
|
|---|
| 609 | else if (meta == "DocumentButtonExpandText")
|
|---|
| 610 | formatlistptr->command = comDocumentButtonExpandText;
|
|---|
| 611 |
|
|---|
| 612 | else if (meta == "DocOID")
|
|---|
| 613 | formatlistptr->command = comOID;
|
|---|
| 614 | else if (meta == "DocTopOID")
|
|---|
| 615 | formatlistptr->command = comTopOID;
|
|---|
| 616 | else if (meta == "DocRank")
|
|---|
| 617 | formatlistptr->command = comRank;
|
|---|
| 618 | else if (meta == "DocTermsFreqTotal")
|
|---|
| 619 | formatlistptr->command = comDocTermsFreqTotal;
|
|---|
| 620 | else if (meta.size() >= 10 && (substr(meta.begin(), meta.begin()+10) == "collection")) {
|
|---|
| 621 | formatlistptr->command = comCollection;
|
|---|
| 622 | parse_coll_meta(meta, formatlistptr->meta);
|
|---|
| 623 | }
|
|---|
| 624 | else {
|
|---|
| 625 | formatlistptr->command = comMeta;
|
|---|
| 626 | parse_meta (meta, formatlistptr->meta, metadata, getParents);
|
|---|
| 627 | }
|
|---|
| 628 | }
|
|---|
| 629 |
|
|---|
| 630 |
|
|---|
| 631 | static bool parse_string (const text_t &formatstring, format_t *formatlistptr,
|
|---|
| 632 | text_tset &metadata, bool &getParents) {
|
|---|
| 633 |
|
|---|
| 634 | text_t text;
|
|---|
| 635 | text_t::const_iterator here = formatstring.begin();
|
|---|
| 636 | text_t::const_iterator end = formatstring.end();
|
|---|
| 637 |
|
|---|
| 638 | while (here != end) {
|
|---|
| 639 |
|
|---|
| 640 | if (*here == '\\') {
|
|---|
| 641 | ++here;
|
|---|
| 642 | if (here != end) text.push_back (*here);
|
|---|
| 643 |
|
|---|
| 644 | } else if (*here == '{') {
|
|---|
| 645 | if (!text.empty()) {
|
|---|
| 646 | formatlistptr->command = comText;
|
|---|
| 647 | formatlistptr->text = text;
|
|---|
| 648 | formatlistptr->nextptr = new format_t();
|
|---|
| 649 | formatlistptr = formatlistptr->nextptr;
|
|---|
| 650 |
|
|---|
| 651 | text.clear();
|
|---|
| 652 | }
|
|---|
| 653 | if (parse_action (++here, end, formatlistptr, metadata, getParents)) {
|
|---|
| 654 |
|
|---|
| 655 | formatlistptr->nextptr = new format_t();
|
|---|
| 656 | formatlistptr = formatlistptr->nextptr;
|
|---|
| 657 | if (here == end) break;
|
|---|
| 658 | }
|
|---|
| 659 | } else if (*here == '[') {
|
|---|
| 660 | if (!text.empty()) {
|
|---|
| 661 | formatlistptr->command = comText;
|
|---|
| 662 | formatlistptr->text = text;
|
|---|
| 663 | formatlistptr->nextptr = new format_t();
|
|---|
| 664 | formatlistptr = formatlistptr->nextptr;
|
|---|
| 665 |
|
|---|
| 666 | text.clear();
|
|---|
| 667 | }
|
|---|
| 668 | text_t meta;
|
|---|
| 669 | ++here;
|
|---|
| 670 | while (*here != ']') {
|
|---|
| 671 | if (here == end) return false;
|
|---|
| 672 | meta.push_back (*here);
|
|---|
| 673 | ++here;
|
|---|
| 674 | }
|
|---|
| 675 | parse_meta (meta, formatlistptr, metadata, getParents);
|
|---|
| 676 | formatlistptr->nextptr = new format_t();
|
|---|
| 677 | formatlistptr = formatlistptr->nextptr;
|
|---|
| 678 |
|
|---|
| 679 | } else
|
|---|
| 680 | text.push_back (*here);
|
|---|
| 681 |
|
|---|
| 682 | if (here != end) ++here;
|
|---|
| 683 | }
|
|---|
| 684 | if (!text.empty()) {
|
|---|
| 685 | formatlistptr->command = comText;
|
|---|
| 686 | formatlistptr->text = text;
|
|---|
| 687 | formatlistptr->nextptr = new format_t();
|
|---|
| 688 | formatlistptr = formatlistptr->nextptr;
|
|---|
| 689 |
|
|---|
| 690 | }
|
|---|
| 691 | return true;
|
|---|
| 692 | }
|
|---|
| 693 |
|
|---|
| 694 |
|
|---|
| 695 | static bool parse_action (text_t::const_iterator &here, const text_t::const_iterator &end,
|
|---|
| 696 | format_t *formatlistptr, text_tset &metadata, bool &getParents) {
|
|---|
| 697 |
|
|---|
| 698 | text_t::const_iterator it = findchar (here, end, '}');
|
|---|
| 699 | if (it == end) return false;
|
|---|
| 700 |
|
|---|
| 701 | text_t com = substr (here, it);
|
|---|
| 702 | here = findchar (it, end, '{');
|
|---|
| 703 | if (here == end) return false;
|
|---|
| 704 | else ++here;
|
|---|
| 705 |
|
|---|
| 706 | if (com == "If" || com == "if" || com == "IF") formatlistptr->command = comIf;
|
|---|
| 707 | else if (com == "Or" || com == "or" || com == "OR") formatlistptr->command = comOr;
|
|---|
| 708 | else return false;
|
|---|
| 709 |
|
|---|
| 710 | int commacount = 0;
|
|---|
| 711 | text_t text;
|
|---|
| 712 | while (here != end) {
|
|---|
| 713 |
|
|---|
| 714 | if (*here == '\\') {
|
|---|
| 715 | ++here;
|
|---|
| 716 | if (here != end) text.push_back(*here);
|
|---|
| 717 |
|
|---|
| 718 | }
|
|---|
| 719 |
|
|---|
| 720 | else if (*here == ',' || *here == '}' || *here == '{') {
|
|---|
| 721 |
|
|---|
| 722 | if (formatlistptr->command == comOr) {
|
|---|
| 723 | // the {Or}{this, or this, or this, or this} statement
|
|---|
| 724 | format_t *or_ptr;
|
|---|
| 725 |
|
|---|
| 726 | // find the next unused orptr
|
|---|
| 727 | if (formatlistptr->orptr == NULL) {
|
|---|
| 728 | formatlistptr->orptr = new format_t();
|
|---|
| 729 | or_ptr = formatlistptr->orptr;
|
|---|
| 730 | } else {
|
|---|
| 731 | or_ptr = formatlistptr->orptr;
|
|---|
| 732 | while (or_ptr->nextptr != NULL)
|
|---|
| 733 | or_ptr = or_ptr->nextptr;
|
|---|
| 734 | or_ptr->nextptr = new format_t();
|
|---|
| 735 | or_ptr = or_ptr->nextptr;
|
|---|
| 736 | }
|
|---|
| 737 |
|
|---|
| 738 | if (!text.empty())
|
|---|
| 739 | {
|
|---|
| 740 | if (!parse_string(text, or_ptr, metadata, getParents)) { return false; }
|
|---|
| 741 | }
|
|---|
| 742 |
|
|---|
| 743 | if (*here == '{')
|
|---|
| 744 | {
|
|---|
| 745 | // Supports: {Or}{[Booktitle],[Title],{If}{[XXXX],aaa,bbb}}
|
|---|
| 746 | // but not : {Or}{[Booktitle],[Title]{If}{[XXXX],aaa,bbb}}
|
|---|
| 747 | // The latter can always be re-written:
|
|---|
| 748 | // {Or}{[Booktitle],{If}{[Title],[Title]{If}{[XXXX],aaa,bbb}}}
|
|---|
| 749 |
|
|---|
| 750 | if (!text.empty()) // already used up allocated format_t
|
|---|
| 751 | {
|
|---|
| 752 | // => allocate new one for detected action
|
|---|
| 753 | or_ptr->nextptr = new format_t();
|
|---|
| 754 | or_ptr = or_ptr->nextptr;
|
|---|
| 755 | }
|
|---|
| 756 | if (!parse_action(++here, end, or_ptr, metadata, getParents))
|
|---|
| 757 | {
|
|---|
| 758 | return false;
|
|---|
| 759 | }
|
|---|
| 760 | }
|
|---|
| 761 | else
|
|---|
| 762 | {
|
|---|
| 763 | if (*here == '}') break;
|
|---|
| 764 | }
|
|---|
| 765 | text.clear();
|
|---|
| 766 |
|
|---|
| 767 | }
|
|---|
| 768 |
|
|---|
| 769 | // Parse an {If}{decide,do,else} statement
|
|---|
| 770 | else {
|
|---|
| 771 |
|
|---|
| 772 | // Read the decision component.
|
|---|
| 773 | if (commacount == 0) {
|
|---|
| 774 | // Decsion can be a metadata element, or a piece of text.
|
|---|
| 775 | // Originally Stefan's code, updated 25/10/2000 by Gordon.
|
|---|
| 776 |
|
|---|
| 777 | text_t::const_iterator beginbracket = text.begin();
|
|---|
| 778 | text_t::const_iterator endbracket = (text.end() - 1);
|
|---|
| 779 |
|
|---|
| 780 | // Decision is based on a metadata element
|
|---|
| 781 | if ((*beginbracket == '[') && (*endbracket == ']')) {
|
|---|
| 782 | // Ignore the surrounding square brackets
|
|---|
| 783 | text_t meta = substr (beginbracket+1, endbracket);
|
|---|
| 784 | parse_meta (meta, formatlistptr->decision.meta, metadata, getParents);
|
|---|
| 785 | ++commacount;
|
|---|
| 786 | text.clear();
|
|---|
| 787 | }
|
|---|
| 788 |
|
|---|
| 789 | // Decision is a piece of text (probably a macro like _cgiargmode_).
|
|---|
| 790 | else {
|
|---|
| 791 |
|
|---|
| 792 | // hunt for any metadata in string, which might be uses in
|
|---|
| 793 | // to test a condition, e.g. [Format] eq 'PDF'
|
|---|
| 794 | format_t* dummyformat = new format_t();
|
|---|
| 795 | // update which metadata fields needed
|
|---|
| 796 | // (not interested in updatng formatlistptr)
|
|---|
| 797 | parse_string (text, dummyformat, metadata, getParents);
|
|---|
| 798 | delete dummyformat;
|
|---|
| 799 |
|
|---|
| 800 | formatlistptr->decision.command = dText;
|
|---|
| 801 | formatlistptr->decision.text = text;
|
|---|
| 802 | ++commacount;
|
|---|
| 803 | text.clear();
|
|---|
| 804 | }
|
|---|
| 805 | }
|
|---|
| 806 |
|
|---|
| 807 | // Read the "then" and "else" components of the {If} statement.
|
|---|
| 808 | else {
|
|---|
| 809 | format_t** nextlistptr = NULL;
|
|---|
| 810 | if (commacount == 1) {
|
|---|
| 811 | nextlistptr = &formatlistptr->ifptr;
|
|---|
| 812 | } else if (commacount == 2 ) {
|
|---|
| 813 | nextlistptr = &formatlistptr->elseptr;
|
|---|
| 814 | } else {
|
|---|
| 815 | return false;
|
|---|
| 816 | }
|
|---|
| 817 |
|
|---|
| 818 | if (!text.empty()) {
|
|---|
| 819 | if (*nextlistptr == NULL) {
|
|---|
| 820 | *nextlistptr = new format_t();
|
|---|
| 821 | } else {
|
|---|
| 822 |
|
|---|
| 823 | // skip to the end of any format_t statements already added
|
|---|
| 824 | while ((*nextlistptr)->nextptr != NULL)
|
|---|
| 825 | {
|
|---|
| 826 | nextlistptr = &(*nextlistptr)->nextptr;
|
|---|
| 827 | }
|
|---|
| 828 |
|
|---|
| 829 | (*nextlistptr)->nextptr = new format_t();
|
|---|
| 830 | nextlistptr = &(*nextlistptr)->nextptr;
|
|---|
| 831 | }
|
|---|
| 832 |
|
|---|
| 833 | if (!parse_string (text, *nextlistptr, metadata, getParents))
|
|---|
| 834 | {
|
|---|
| 835 | return false;
|
|---|
| 836 | }
|
|---|
| 837 | text.clear();
|
|---|
| 838 | }
|
|---|
| 839 |
|
|---|
| 840 | if (*here == '{')
|
|---|
| 841 | {
|
|---|
| 842 | if (*nextlistptr == NULL) {
|
|---|
| 843 | *nextlistptr = new format_t();
|
|---|
| 844 | } else {
|
|---|
| 845 | // skip to the end of any format_t statements already added
|
|---|
| 846 | while ((*nextlistptr)->nextptr != NULL)
|
|---|
| 847 | {
|
|---|
| 848 | nextlistptr = &(*nextlistptr)->nextptr;
|
|---|
| 849 | }
|
|---|
| 850 |
|
|---|
| 851 | (*nextlistptr)->nextptr = new format_t();
|
|---|
| 852 | nextlistptr = &(*nextlistptr)->nextptr;
|
|---|
| 853 | }
|
|---|
| 854 |
|
|---|
| 855 | if (!parse_action(++here, end, *nextlistptr, metadata, getParents))
|
|---|
| 856 | {
|
|---|
| 857 | return false;
|
|---|
| 858 | }
|
|---|
| 859 | }
|
|---|
| 860 | else
|
|---|
| 861 | {
|
|---|
| 862 | if (*here == '}') break;
|
|---|
| 863 | ++commacount;
|
|---|
| 864 | }
|
|---|
| 865 | }
|
|---|
| 866 | }
|
|---|
| 867 |
|
|---|
| 868 | } else text.push_back(*here);
|
|---|
| 869 |
|
|---|
| 870 | if (here != end) ++here;
|
|---|
| 871 | }
|
|---|
| 872 |
|
|---|
| 873 | return true;
|
|---|
| 874 | }
|
|---|
| 875 |
|
|---|
| 876 |
|
|---|
| 877 | bool parse_formatstring (const text_t &formatstring, format_t *formatlistptr,
|
|---|
| 878 | text_tset &metadata, bool &getParents) {
|
|---|
| 879 |
|
|---|
| 880 | formatlistptr->clear();
|
|---|
| 881 | getParents = false;
|
|---|
| 882 |
|
|---|
| 883 | return (parse_string (formatstring, formatlistptr, metadata, getParents));
|
|---|
| 884 | }
|
|---|
| 885 |
|
|---|
| 886 | // position -1 for all, -2 for the last, 0 for the first, or x for a particular piece
|
|---|
| 887 | // metainfo has all the values for the metadata, meta has the request metadata name and options, position tells which values to get
|
|---|
| 888 | static text_t get_formatted_meta_text(MetadataInfo_t &metainfo, const metadata_t &meta, int position, bool no_cgisafe = false)
|
|---|
| 889 | {
|
|---|
| 890 | text_t no_ns_metaname = remove_namespace(meta.metaname);
|
|---|
| 891 | text_t formatted_metatext;
|
|---|
| 892 | bool first = true;
|
|---|
| 893 |
|
|---|
| 894 | const int start_i=0;
|
|---|
| 895 | const int end_i = metainfo.values.size()-1;
|
|---|
| 896 |
|
|---|
| 897 | if (position == -1) { // all
|
|---|
| 898 | for (int i=start_i; i<=end_i; ++i) {
|
|---|
| 899 | if (!first) formatted_metatext += meta.siblingoptions;
|
|---|
| 900 |
|
|---|
| 901 | text_t fresh_metatext;
|
|---|
| 902 |
|
|---|
| 903 | if (meta.metacommand & mSpecial) {
|
|---|
| 904 | // special formatting
|
|---|
| 905 | if (no_ns_metaname == "Date") fresh_metatext = format_date (metainfo.values[i]);
|
|---|
| 906 | else if (no_ns_metaname == "Language") fresh_metatext = iso639(metainfo.values[i]);
|
|---|
| 907 | else fresh_metatext = "_format:"+meta.metaname+"_("+metainfo.values[i]+")";
|
|---|
| 908 | }
|
|---|
| 909 | else fresh_metatext = metainfo.values[i];
|
|---|
| 910 |
|
|---|
| 911 | if (metadata_spanwrap) {
|
|---|
| 912 | fresh_metatext = "<span class=\"editable-metadata\" metaname=\"" + meta.metaname + "\" metapos=\"" + i + "\">" + fresh_metatext + "</span>";
|
|---|
| 913 | }
|
|---|
| 914 | formatted_metatext += fresh_metatext;
|
|---|
| 915 |
|
|---|
| 916 | first = false;
|
|---|
| 917 |
|
|---|
| 918 | }
|
|---|
| 919 | } else {
|
|---|
| 920 | if (position == -2) { // end
|
|---|
| 921 | position = end_i;
|
|---|
| 922 | } else if (position < start_i || position > end_i) {
|
|---|
| 923 | return "";
|
|---|
| 924 | }
|
|---|
| 925 |
|
|---|
| 926 | text_t fresh_metatext;
|
|---|
| 927 | if (meta.metacommand & mSpecial) {
|
|---|
| 928 |
|
|---|
| 929 | // special formatting
|
|---|
| 930 | if (no_ns_metaname == "Date") fresh_metatext = format_date (metainfo.values[position]);
|
|---|
| 931 | else if (no_ns_metaname == "Language") fresh_metatext = iso639(metainfo.values[position]);
|
|---|
| 932 | else fresh_metatext = "_format:"+meta.metaname+"_("+metainfo.values[position]+")";
|
|---|
| 933 | }
|
|---|
| 934 | else fresh_metatext = metainfo.values[position];
|
|---|
| 935 |
|
|---|
| 936 | if (metadata_spanwrap) {
|
|---|
| 937 | fresh_metatext = "<span class=\"editable-metadata\" metaname=\"" + meta.metaname + "\" metapos=\"" + position + "\">" + fresh_metatext + "</span>";
|
|---|
| 938 | }
|
|---|
| 939 |
|
|---|
| 940 | formatted_metatext += fresh_metatext;
|
|---|
| 941 | }
|
|---|
| 942 |
|
|---|
| 943 | if (meta.metacommand & mCgiSafe && !no_cgisafe) return cgi_safe_unicode (formatted_metatext);
|
|---|
| 944 | else return formatted_metatext;
|
|---|
| 945 | }
|
|---|
| 946 |
|
|---|
| 947 | static text_t get_parent_meta (ResultDocInfo_t &docinfo, const metadata_t &meta, int siblings_values)
|
|---|
| 948 | {
|
|---|
| 949 |
|
|---|
| 950 | MetadataInfo_t *parent = docinfo.metadata[meta.metaname].parent;
|
|---|
| 951 | switch (meta.mqualifier.parent) {
|
|---|
| 952 | case pNone:
|
|---|
| 953 | return "Nothing!!";
|
|---|
| 954 | break;
|
|---|
| 955 |
|
|---|
| 956 | case pImmediate:
|
|---|
| 957 | if (parent != NULL) {
|
|---|
| 958 | return get_formatted_meta_text(*parent, meta, siblings_values);
|
|---|
| 959 | }
|
|---|
| 960 | break;
|
|---|
| 961 |
|
|---|
| 962 | case pTop:
|
|---|
| 963 | if (parent != NULL) {
|
|---|
| 964 | while (parent->parent != NULL) parent = parent->parent;
|
|---|
| 965 | return get_formatted_meta_text(*parent, meta, siblings_values);
|
|---|
| 966 | }
|
|---|
| 967 | break;
|
|---|
| 968 |
|
|---|
| 969 | case pAll:
|
|---|
| 970 | MetadataInfo_t *parent = docinfo.metadata[meta.metaname].parent;
|
|---|
| 971 | if (parent != NULL) {
|
|---|
| 972 | text_tarray tmparray;
|
|---|
| 973 | while (parent != NULL) {
|
|---|
| 974 | 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
|
|---|
| 975 | parent = parent->parent;
|
|---|
| 976 | }
|
|---|
| 977 | // now join them up - use teh parent separator
|
|---|
| 978 | bool first = true;
|
|---|
| 979 | text_t tmp;
|
|---|
| 980 | text_tarray::reverse_iterator here = tmparray.rbegin();
|
|---|
| 981 | text_tarray::reverse_iterator end = tmparray.rend();
|
|---|
| 982 | while (here != end) {
|
|---|
| 983 | if (!first) tmp += meta.parentoptions;
|
|---|
| 984 | tmp += *here;
|
|---|
| 985 | first = false;
|
|---|
| 986 | ++here;
|
|---|
| 987 | }
|
|---|
| 988 | if (meta.metacommand & mCgiSafe) return cgi_safe_unicode (tmp);
|
|---|
| 989 | else return tmp;
|
|---|
| 990 | }
|
|---|
| 991 | }
|
|---|
| 992 | return "";
|
|---|
| 993 |
|
|---|
| 994 | }
|
|---|
| 995 |
|
|---|
| 996 | static text_t get_child_meta (const text_t& collection,
|
|---|
| 997 | recptproto* collectproto,
|
|---|
| 998 | ResultDocInfo_t &docinfo, displayclass &disp,
|
|---|
| 999 | const metadata_t &meta, text_tmap &options,
|
|---|
| 1000 | ostream& logout, int siblings_values)
|
|---|
| 1001 | {
|
|---|
| 1002 | if (docinfo.metadata["contains"].values[0].size()==0) return ""; // no children
|
|---|
| 1003 |
|
|---|
| 1004 | const text_t& pre_tree_trav = meta.pre_tree_traverse;
|
|---|
| 1005 | const text_t& child_metaname = meta.metaname;
|
|---|
| 1006 | const text_t& child_field = meta.childoptions;
|
|---|
| 1007 | text_tset child_metadata;
|
|---|
| 1008 | child_metadata.insert(child_metaname);
|
|---|
| 1009 |
|
|---|
| 1010 | FilterResponse_t child_response;
|
|---|
| 1011 | if (meta.mqualifier.child == cNum) {
|
|---|
| 1012 | // just one child
|
|---|
| 1013 | //get the information associated with the metadata for child doc
|
|---|
| 1014 | if (!get_info (docinfo.OID+pre_tree_trav+child_field, collection, "",
|
|---|
| 1015 | child_metadata, false, collectproto, child_response,
|
|---|
| 1016 | logout)) return ""; // invalid child number
|
|---|
| 1017 |
|
|---|
| 1018 | if (child_response.docInfo.empty()) return false; // no info for the child
|
|---|
| 1019 |
|
|---|
| 1020 | ResultDocInfo_t& child_docinfo = child_response.docInfo[0];
|
|---|
| 1021 | MetadataInfo_t& metaname_rec = child_docinfo.metadata[child_metaname];
|
|---|
| 1022 |
|
|---|
| 1023 | text_t child_metavalue = get_formatted_meta_text(metaname_rec,meta,siblings_values);
|
|---|
| 1024 | return expand_metadata(child_metavalue,collection,collectproto,
|
|---|
| 1025 | child_docinfo,disp,options,logout);
|
|---|
| 1026 | }
|
|---|
| 1027 |
|
|---|
| 1028 |
|
|---|
| 1029 | if (meta.mqualifier.child != cAll) return false; // invalid qualifier
|
|---|
| 1030 |
|
|---|
| 1031 |
|
|---|
| 1032 | if (!pre_tree_trav.empty()) {
|
|---|
| 1033 | // need to get relevant "contains" metadata for new (e.g. pre tree trav) node
|
|---|
| 1034 | FilterResponse_t trav_response;
|
|---|
| 1035 |
|
|---|
| 1036 | text_tset trav_metadata;
|
|---|
| 1037 | trav_metadata.insert("contains");
|
|---|
| 1038 |
|
|---|
| 1039 | if (!get_info (docinfo.OID+pre_tree_trav, collection, "",
|
|---|
| 1040 | trav_metadata, false, collectproto, trav_response,
|
|---|
| 1041 | logout)) return ""; // invalid pre_tree_trav
|
|---|
| 1042 |
|
|---|
| 1043 | if (trav_response.docInfo.empty()) return false; // no info for the pre_tree_trav OID
|
|---|
| 1044 |
|
|---|
| 1045 | ResultDocInfo_t& trav_docinfo = trav_response.docInfo[0];
|
|---|
| 1046 | // use this for rest of routine
|
|---|
| 1047 | docinfo = trav_docinfo;
|
|---|
| 1048 | }
|
|---|
| 1049 |
|
|---|
| 1050 | // we need to get all children
|
|---|
| 1051 | text_t result = "";
|
|---|
| 1052 | text_tarray children;
|
|---|
| 1053 | text_t contains = docinfo.metadata["contains"].values[0];
|
|---|
| 1054 | splitchar (contains.begin(), contains.end(), ';', children);
|
|---|
| 1055 | text_tarray::const_iterator here = children.begin();
|
|---|
| 1056 | text_tarray::const_iterator end = children.end();
|
|---|
| 1057 | bool first = true;
|
|---|
| 1058 | while (here !=end) {
|
|---|
| 1059 | text_t oid = *here;
|
|---|
| 1060 | here++;
|
|---|
| 1061 | if (*(oid.begin()) == '"') translate_parent (oid, docinfo.OID);
|
|---|
| 1062 |
|
|---|
| 1063 | //get the information associated with the metadata for child doc
|
|---|
| 1064 | if (!get_info (oid, collection, "", child_metadata,
|
|---|
| 1065 | false, collectproto, child_response, logout) ||
|
|---|
| 1066 | child_response.docInfo.empty()) {
|
|---|
| 1067 | first = false;
|
|---|
| 1068 | continue;
|
|---|
| 1069 | }
|
|---|
| 1070 |
|
|---|
| 1071 |
|
|---|
| 1072 | ResultDocInfo_t& child_docinfo = child_response.docInfo[0];
|
|---|
| 1073 | MetadataInfo_t& metaname_rec = child_docinfo.metadata[child_metaname];
|
|---|
| 1074 |
|
|---|
| 1075 | text_t child_metavalue = get_formatted_meta_text(metaname_rec,meta,siblings_values);
|
|---|
| 1076 | if (!first) result += child_field;
|
|---|
| 1077 | first = false;
|
|---|
| 1078 | // need to do this here cos otherwise we are in the wrong document
|
|---|
| 1079 | result += expand_metadata(child_metavalue,collection,collectproto,
|
|---|
| 1080 | child_docinfo,disp,options,logout);
|
|---|
| 1081 | }
|
|---|
| 1082 | return result;
|
|---|
| 1083 |
|
|---|
| 1084 | }
|
|---|
| 1085 |
|
|---|
| 1086 | static text_t get_meta (const text_t& collection, recptproto* collectproto,
|
|---|
| 1087 | ResultDocInfo_t &docinfo, displayclass &disp,
|
|---|
| 1088 | const metadata_t &meta, text_tmap &options,
|
|---|
| 1089 | ostream& logout) {
|
|---|
| 1090 |
|
|---|
| 1091 | // make sure we have the requested metadata
|
|---|
| 1092 | MetadataInfo_tmap::iterator it = docinfo.metadata.find (meta.metaname);
|
|---|
| 1093 | if (it == docinfo.metadata.end()) return "";
|
|---|
| 1094 |
|
|---|
| 1095 | int siblings_values = 0; // default is no siblings, just the first metadata available
|
|---|
| 1096 | if (meta.metacommand & mSibling) {
|
|---|
| 1097 | if (meta.mqualifier.sibling == sAll) {
|
|---|
| 1098 | siblings_values = -1; //all
|
|---|
| 1099 | } else if (meta.mqualifier.sibling == sNum) {
|
|---|
| 1100 | siblings_values = meta.siblingoptions.getint();
|
|---|
| 1101 | }
|
|---|
| 1102 | }
|
|---|
| 1103 | if (meta.metacommand & mParent) {
|
|---|
| 1104 | return get_parent_meta(docinfo,meta,siblings_values);
|
|---|
| 1105 | }
|
|---|
| 1106 |
|
|---|
| 1107 | else if (meta.metacommand & mChild) {
|
|---|
| 1108 | return get_child_meta(collection,collectproto,docinfo,disp,meta,
|
|---|
| 1109 | options,logout, siblings_values);
|
|---|
| 1110 | }
|
|---|
| 1111 | else if (meta.metacommand & mSibling) { // only siblings
|
|---|
| 1112 | MetadataInfo_t& metaname_rec = docinfo.metadata[meta.metaname];
|
|---|
| 1113 | return get_formatted_meta_text(docinfo.metadata[meta.metaname],meta, siblings_values);
|
|---|
| 1114 | }
|
|---|
| 1115 | else {
|
|---|
| 1116 |
|
|---|
| 1117 | // straightforward metadata request (nothing fancy)
|
|---|
| 1118 |
|
|---|
| 1119 | text_t classifier_metaname = docinfo.classifier_metadata_type;
|
|---|
| 1120 | int metaname_index
|
|---|
| 1121 | = (classifier_metaname == meta.metaname) ? docinfo.classifier_metadata_offset : 0;
|
|---|
| 1122 | return get_formatted_meta_text(docinfo.metadata[meta.metaname], meta, metaname_index);
|
|---|
| 1123 | }
|
|---|
| 1124 |
|
|---|
| 1125 | return "";
|
|---|
| 1126 | }
|
|---|
| 1127 |
|
|---|
| 1128 | static text_t get_or (const text_t& collection, recptproto* collectproto,
|
|---|
| 1129 | ResultDocInfo_t &docinfo, displayclass &disp,
|
|---|
| 1130 | format_t *orptr, text_tmap &options,
|
|---|
| 1131 | ostream& logout) {
|
|---|
| 1132 |
|
|---|
| 1133 | while (orptr != NULL) {
|
|---|
| 1134 |
|
|---|
| 1135 | if (metadata_spanwrap) {
|
|---|
| 1136 | // need to be a bit more careful about this
|
|---|
| 1137 | // => test for it *without* spanwrap, and if defined, then
|
|---|
| 1138 | // got back and generate it again, this time with spanwrap on
|
|---|
| 1139 |
|
|---|
| 1140 | metadata_spanwrap = false;
|
|---|
| 1141 | text_t test_tmp = format_string (collection,collectproto,docinfo, disp, orptr,
|
|---|
| 1142 | options, logout);
|
|---|
| 1143 | metadata_spanwrap = true;
|
|---|
| 1144 | if (!test_tmp.empty()) {
|
|---|
| 1145 |
|
|---|
| 1146 | return format_string (collection,collectproto,docinfo, disp, orptr,
|
|---|
| 1147 | options, logout);
|
|---|
| 1148 | }
|
|---|
| 1149 | }
|
|---|
| 1150 | else {
|
|---|
| 1151 | text_t tmp = format_string (collection,collectproto,docinfo, disp, orptr,
|
|---|
| 1152 | options, logout);
|
|---|
| 1153 | if (!tmp.empty()) return tmp;
|
|---|
| 1154 | }
|
|---|
| 1155 |
|
|---|
| 1156 | orptr = orptr->nextptr;
|
|---|
| 1157 | }
|
|---|
| 1158 | return "";
|
|---|
| 1159 | }
|
|---|
| 1160 |
|
|---|
| 1161 | static bool char_is_whitespace(const char c)
|
|---|
| 1162 | {
|
|---|
| 1163 | return ((c == ' ') || (c == '\t') || (c == '\n') || (c == '\r'));
|
|---|
| 1164 |
|
|---|
| 1165 | }
|
|---|
| 1166 |
|
|---|
| 1167 | static int scan_over_whitespace(const text_t& outstring, const int start_pos)
|
|---|
| 1168 | {
|
|---|
| 1169 | int pos = start_pos;
|
|---|
| 1170 | while (pos<outstring.size()) {
|
|---|
| 1171 | if (!char_is_whitespace(outstring[pos])) {
|
|---|
| 1172 | break;
|
|---|
| 1173 | }
|
|---|
| 1174 | ++pos;
|
|---|
| 1175 | }
|
|---|
| 1176 |
|
|---|
| 1177 | return pos;
|
|---|
| 1178 | }
|
|---|
| 1179 |
|
|---|
| 1180 | static int rscan_over_whitespace(const text_t& outstring, const int start_pos)
|
|---|
| 1181 | {
|
|---|
| 1182 | int pos = start_pos;
|
|---|
| 1183 | while (pos>=0) {
|
|---|
| 1184 | if (!char_is_whitespace(outstring[pos])) {
|
|---|
| 1185 | break;
|
|---|
| 1186 | }
|
|---|
| 1187 | --pos;
|
|---|
| 1188 | }
|
|---|
| 1189 |
|
|---|
| 1190 | return pos;
|
|---|
| 1191 | }
|
|---|
| 1192 |
|
|---|
| 1193 | static int rscan_for_whitespace(const text_t& outstring, const int start_pos)
|
|---|
| 1194 | {
|
|---|
| 1195 | int pos = start_pos;
|
|---|
| 1196 | while (pos>=0) {
|
|---|
| 1197 | if (char_is_whitespace(outstring[pos])) {
|
|---|
| 1198 | break;
|
|---|
| 1199 | }
|
|---|
| 1200 | --pos;
|
|---|
| 1201 | }
|
|---|
| 1202 |
|
|---|
| 1203 | return pos;
|
|---|
| 1204 | }
|
|---|
| 1205 |
|
|---|
| 1206 |
|
|---|
| 1207 | static int rscan_for(const text_t& outstring, const int start_pos,
|
|---|
| 1208 | const char find_c)
|
|---|
| 1209 | {
|
|---|
| 1210 | int pos = start_pos;
|
|---|
| 1211 | while (pos>=0) {
|
|---|
| 1212 | char c = outstring[pos];
|
|---|
| 1213 | if (outstring[pos] == find_c) {
|
|---|
| 1214 | break;
|
|---|
| 1215 | }
|
|---|
| 1216 | --pos;
|
|---|
| 1217 | }
|
|---|
| 1218 |
|
|---|
| 1219 | return pos;
|
|---|
| 1220 | }
|
|---|
| 1221 |
|
|---|
| 1222 | text_t extract_substr(const text_t& outstring, const int start_pos,
|
|---|
| 1223 | const int end_pos)
|
|---|
| 1224 | {
|
|---|
| 1225 | text_t extracted_str;
|
|---|
| 1226 | extracted_str.clear();
|
|---|
| 1227 |
|
|---|
| 1228 | for (int pos=start_pos; pos<=end_pos; ++pos) {
|
|---|
| 1229 | extracted_str.push_back(outstring[pos]);
|
|---|
| 1230 | }
|
|---|
| 1231 |
|
|---|
| 1232 | return extracted_str;
|
|---|
| 1233 | }
|
|---|
| 1234 |
|
|---|
| 1235 |
|
|---|
| 1236 | static text_t expand_potential_metadata(const text_t& collection,
|
|---|
| 1237 | recptproto* collectproto,
|
|---|
| 1238 | ResultDocInfo_t &docinfo,
|
|---|
| 1239 | displayclass &disp,
|
|---|
| 1240 | const text_t& intext,
|
|---|
| 1241 | text_tmap &options,
|
|---|
| 1242 | ostream& logout)
|
|---|
| 1243 | {
|
|---|
| 1244 | text_t outtext;
|
|---|
| 1245 |
|
|---|
| 1246 | // decide if dealing with metadata or text
|
|---|
| 1247 |
|
|---|
| 1248 | text_t::const_iterator beginbracket = intext.begin();
|
|---|
| 1249 | text_t::const_iterator endbracket = (intext.end() - 1);
|
|---|
| 1250 |
|
|---|
| 1251 | // Decision is based on a metadata element
|
|---|
| 1252 | if ((*beginbracket == '[') && (*endbracket == ']')) {
|
|---|
| 1253 | // Ignore the surrounding square brackets
|
|---|
| 1254 | text_t meta_text = substr (beginbracket+1, endbracket);
|
|---|
| 1255 |
|
|---|
| 1256 | if (meta_text == "Text") {
|
|---|
| 1257 | outtext = format_text(collection, collectproto, docinfo, disp, options, logout);
|
|---|
| 1258 | }
|
|---|
| 1259 | else {
|
|---|
| 1260 |
|
|---|
| 1261 | text_tset metadata;
|
|---|
| 1262 | bool getParents =false;
|
|---|
| 1263 | metadata_t meta;
|
|---|
| 1264 |
|
|---|
| 1265 | parse_meta (meta_text, meta, metadata, getParents);
|
|---|
| 1266 | outtext
|
|---|
| 1267 | = get_meta (collection,collectproto,docinfo,disp,meta,options,logout);
|
|---|
| 1268 | }
|
|---|
| 1269 |
|
|---|
| 1270 | }
|
|---|
| 1271 | else {
|
|---|
| 1272 | outtext = intext;
|
|---|
| 1273 | }
|
|---|
| 1274 |
|
|---|
| 1275 | return outtext;
|
|---|
| 1276 | }
|
|---|
| 1277 |
|
|---|
| 1278 |
|
|---|
| 1279 |
|
|---|
| 1280 |
|
|---|
| 1281 | static bool uses_expression(const text_t& collection, recptproto* collectproto,
|
|---|
| 1282 | ResultDocInfo_t &docinfo,
|
|---|
| 1283 | displayclass &disp,
|
|---|
| 1284 | const text_t& outstring, text_t& lhs_expr,
|
|---|
| 1285 | text_t& op_expr, text_t& rhs_expr,
|
|---|
| 1286 | text_tmap &options,
|
|---|
| 1287 | ostream& logout)
|
|---|
| 1288 | {
|
|---|
| 1289 | // Note: the string may not be of the form: str1 op str2, however
|
|---|
| 1290 | // to deterine this we have to process it on the assumption it is,
|
|---|
| 1291 | // and if at any point an 'erroneous' value is encountered, return
|
|---|
| 1292 | // false and let something else have a go at evaluating it
|
|---|
| 1293 |
|
|---|
| 1294 | // Starting at the end of the string and working backwards ..
|
|---|
| 1295 |
|
|---|
| 1296 | const int outstring_len = outstring.size();
|
|---|
| 1297 |
|
|---|
| 1298 | // skip over white space
|
|---|
| 1299 | int rhs_end = rscan_over_whitespace(outstring,outstring_len-1);
|
|---|
| 1300 |
|
|---|
| 1301 | if (rhs_end<=0) {
|
|---|
| 1302 | // no meaningful text or (rhs_end==0) no room for operator
|
|---|
| 1303 | return false;
|
|---|
| 1304 | }
|
|---|
| 1305 |
|
|---|
| 1306 | // check for ' or " and then scan over token
|
|---|
| 1307 | const char potential_quote = outstring[rhs_end];
|
|---|
| 1308 | int rhs_start=rhs_end;
|
|---|
| 1309 | bool quoted = false;
|
|---|
| 1310 |
|
|---|
| 1311 | if ((potential_quote == '\'') || (potential_quote == '\"')) {
|
|---|
| 1312 | --rhs_end;
|
|---|
| 1313 | rhs_start = rscan_for(outstring,rhs_end-1,potential_quote) +1;
|
|---|
| 1314 | quoted = true;
|
|---|
| 1315 | }
|
|---|
| 1316 | else {
|
|---|
| 1317 | rhs_start = rscan_for_whitespace(outstring,rhs_end-1) +1;
|
|---|
| 1318 | }
|
|---|
| 1319 |
|
|---|
| 1320 | if ((rhs_end-rhs_start)<0) {
|
|---|
| 1321 | // no meaningful rhs expression
|
|---|
| 1322 | return false;
|
|---|
| 1323 | }
|
|---|
| 1324 |
|
|---|
| 1325 | // form rhs_expr
|
|---|
| 1326 | rhs_expr = extract_substr(outstring,rhs_start,rhs_end);
|
|---|
| 1327 |
|
|---|
| 1328 | // skip over white space
|
|---|
| 1329 | const int to_whitespace = (quoted) ? 2 : 1;
|
|---|
| 1330 |
|
|---|
| 1331 | int op_end = rscan_over_whitespace(outstring,rhs_start-to_whitespace);
|
|---|
| 1332 | int op_start = rscan_for_whitespace(outstring,op_end-1)+1;
|
|---|
| 1333 |
|
|---|
| 1334 | if ((op_end<0) && (op_start<0)) {
|
|---|
| 1335 | // no meaningful expression operator
|
|---|
| 1336 | return false;
|
|---|
| 1337 | }
|
|---|
| 1338 |
|
|---|
| 1339 | if (op_end-op_start<0) {
|
|---|
| 1340 | // no meaningful expression operator
|
|---|
| 1341 | return false;
|
|---|
| 1342 | }
|
|---|
| 1343 |
|
|---|
| 1344 | op_expr = extract_substr(outstring,op_start,op_end);
|
|---|
| 1345 |
|
|---|
| 1346 |
|
|---|
| 1347 | // check for operator
|
|---|
| 1348 | if ((op_expr != "eq") && (op_expr != "ne") && (op_expr != "gt") &&
|
|---|
| 1349 | (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")) {
|
|---|
| 1350 |
|
|---|
| 1351 | // not a valid operator
|
|---|
| 1352 | return false;
|
|---|
| 1353 | }
|
|---|
| 1354 |
|
|---|
| 1355 | int lhs_end = rscan_over_whitespace(outstring,op_start-1);
|
|---|
| 1356 | if (lhs_end<0) {
|
|---|
| 1357 | // no meaningful lhs expression
|
|---|
| 1358 | return false;
|
|---|
| 1359 | }
|
|---|
| 1360 |
|
|---|
| 1361 | int lhs_start = scan_over_whitespace(outstring,0);
|
|---|
| 1362 |
|
|---|
| 1363 | // form lhs_expr from remainder of string
|
|---|
| 1364 | lhs_expr = extract_substr(outstring,lhs_start,lhs_end);
|
|---|
| 1365 |
|
|---|
| 1366 | // Now we know we have a valid expression, look up any
|
|---|
| 1367 | // metadata terms
|
|---|
| 1368 |
|
|---|
| 1369 | rhs_expr = expand_potential_metadata(collection,collectproto,docinfo,
|
|---|
| 1370 | disp,rhs_expr,options,logout);
|
|---|
| 1371 | lhs_expr = expand_potential_metadata(collection,collectproto,docinfo,
|
|---|
| 1372 | disp,lhs_expr,options,logout);
|
|---|
| 1373 |
|
|---|
| 1374 | return true;
|
|---|
| 1375 | }
|
|---|
| 1376 |
|
|---|
| 1377 | static bool eval_expression_true(const text_t& lhs_expr,const text_t& op_expr,
|
|---|
| 1378 | const text_t& rhs_expr, ostream& logout)
|
|---|
| 1379 | {
|
|---|
| 1380 | if (op_expr == "eq") return (lhs_expr == rhs_expr);
|
|---|
| 1381 | else if (op_expr == "ne" ) return (lhs_expr != rhs_expr);
|
|---|
| 1382 | else if (op_expr == "gt") return (lhs_expr > rhs_expr);
|
|---|
| 1383 | else if (op_expr == "ge") return (lhs_expr >= rhs_expr);
|
|---|
| 1384 | else if (op_expr == "lt") return (lhs_expr < rhs_expr);
|
|---|
| 1385 | else if (op_expr == "le") return (lhs_expr <= rhs_expr);
|
|---|
| 1386 | else if (op_expr == "==") return (lhs_expr.getint() == rhs_expr.getint());
|
|---|
| 1387 | else if (op_expr == "!=") return (lhs_expr.getint() != rhs_expr.getint());
|
|---|
| 1388 | else if (op_expr == ">") return (lhs_expr.getint() > rhs_expr.getint());
|
|---|
| 1389 | else if (op_expr == ">=") return (lhs_expr.getint() >= rhs_expr.getint());
|
|---|
| 1390 | else if (op_expr == "<") return (lhs_expr.getint() < rhs_expr.getint());
|
|---|
| 1391 | else if (op_expr == "<=") return (lhs_expr.getint() <= rhs_expr.getint());
|
|---|
| 1392 | else if (op_expr == "sw") return (starts_with(lhs_expr,rhs_expr));
|
|---|
| 1393 | else if (op_expr == "ew") return (ends_with(lhs_expr,rhs_expr));
|
|---|
| 1394 | else {
|
|---|
| 1395 | logout << "Error: '" << op_expr << "' is not a recognised operator." << endl;
|
|---|
| 1396 | }
|
|---|
| 1397 |
|
|---|
| 1398 | return false;
|
|---|
| 1399 | }
|
|---|
| 1400 |
|
|---|
| 1401 |
|
|---|
| 1402 | static text_t get_if (const text_t& collection, recptproto* collectproto,
|
|---|
| 1403 | ResultDocInfo_t &docinfo, displayclass &disp,
|
|---|
| 1404 | const decision_t &decision,
|
|---|
| 1405 | format_t *ifptr, format_t *elseptr,
|
|---|
| 1406 | text_tmap &options, ostream& logout)
|
|---|
| 1407 | {
|
|---|
| 1408 | // If the decision component is a metadata element, then evaluate it
|
|---|
| 1409 | // to see whether we output the "then" or the "else" clause
|
|---|
| 1410 | if (decision.command == dMeta) {
|
|---|
| 1411 |
|
|---|
| 1412 | bool store_metadata_spanwrap = metadata_spanwrap;
|
|---|
| 1413 | metadata_spanwrap = 0;
|
|---|
| 1414 |
|
|---|
| 1415 | // temporarily suspend metadata-spanwrap (if on) so can test if metadata item really exits or not
|
|---|
| 1416 | bool metadata_exists
|
|---|
| 1417 | = (get_meta (collection,collectproto,docinfo,disp,decision.meta,options,
|
|---|
| 1418 | logout) != "");
|
|---|
| 1419 |
|
|---|
| 1420 | metadata_spanwrap = store_metadata_spanwrap;
|
|---|
| 1421 |
|
|---|
| 1422 | if (metadata_exists) {
|
|---|
| 1423 | if (ifptr != NULL)
|
|---|
| 1424 | return get_formatted_string (collection,collectproto,docinfo, disp, ifptr,
|
|---|
| 1425 | options, logout);
|
|---|
| 1426 | }
|
|---|
| 1427 | else {
|
|---|
| 1428 | if (elseptr != NULL)
|
|---|
| 1429 | return get_formatted_string (collection,collectproto,docinfo, disp, elseptr,
|
|---|
| 1430 | options, logout);
|
|---|
| 1431 | }
|
|---|
| 1432 | }
|
|---|
| 1433 |
|
|---|
| 1434 | // If the decision component is text, then evaluate it (it is probably a
|
|---|
| 1435 | // macro like _cgiargmode_) to decide what to output.
|
|---|
| 1436 | else if (decision.command == dText) {
|
|---|
| 1437 |
|
|---|
| 1438 | text_t outstring;
|
|---|
| 1439 | disp.expandstring (decision.text, outstring);
|
|---|
| 1440 |
|
|---|
| 1441 | // Check for if expression in form: str1 op str2
|
|---|
| 1442 | // (such as [x] eq "y")
|
|---|
| 1443 | text_t lhs_expr, op_expr, rhs_expr;
|
|---|
| 1444 | if (uses_expression(collection,collectproto,docinfo, disp, outstring,lhs_expr,op_expr,rhs_expr, options,logout)) {
|
|---|
| 1445 | if (eval_expression_true(lhs_expr,op_expr,rhs_expr,logout)) {
|
|---|
| 1446 | if (ifptr != NULL) {
|
|---|
| 1447 | return get_formatted_string (collection, collectproto, docinfo, disp, ifptr,
|
|---|
| 1448 | options, logout);
|
|---|
| 1449 | }
|
|---|
| 1450 | else {
|
|---|
| 1451 | return "";
|
|---|
| 1452 | }
|
|---|
| 1453 | } else {
|
|---|
| 1454 | if (elseptr != NULL) {
|
|---|
| 1455 | return get_formatted_string (collection, collectproto, docinfo, disp, elseptr,
|
|---|
| 1456 | options, logout);
|
|---|
| 1457 | }
|
|---|
| 1458 | else {
|
|---|
| 1459 | return "";
|
|---|
| 1460 | }
|
|---|
| 1461 | }
|
|---|
| 1462 | }
|
|---|
| 1463 |
|
|---|
| 1464 |
|
|---|
| 1465 | // This is a tad tricky. When we expand a string like _cgiargmode_, that is
|
|---|
| 1466 | // a cgi argument macro that has not been set, it evaluates to itself.
|
|---|
| 1467 | // Therefore, were have to say that a piece of text evalautes true if
|
|---|
| 1468 | // it is non-empty and if it is a cgi argument evaulating to itself.
|
|---|
| 1469 |
|
|---|
| 1470 | if ((outstring != "") && !((outstring == decision.text) && (outstring[0] == '_'))) {
|
|---|
| 1471 | if (ifptr != NULL)
|
|---|
| 1472 | return get_formatted_string (collection, collectproto, docinfo, disp, ifptr,
|
|---|
| 1473 | options, logout);
|
|---|
| 1474 | } else {
|
|---|
| 1475 | if (elseptr != NULL)
|
|---|
| 1476 | return get_formatted_string (collection, collectproto, docinfo, disp, elseptr,
|
|---|
| 1477 | options, logout);
|
|---|
| 1478 | }
|
|---|
| 1479 | }
|
|---|
| 1480 |
|
|---|
| 1481 | return "";
|
|---|
| 1482 | }
|
|---|
| 1483 |
|
|---|
| 1484 | bool includes_metadata(const text_t& text)
|
|---|
| 1485 | {
|
|---|
| 1486 | text_t::const_iterator here = text.begin();
|
|---|
| 1487 | text_t::const_iterator end = text.end();
|
|---|
| 1488 | while (here != end) {
|
|---|
| 1489 | if (*here == '[') return true;
|
|---|
| 1490 | ++here;
|
|---|
| 1491 | }
|
|---|
| 1492 |
|
|---|
| 1493 | return false;
|
|---|
| 1494 | }
|
|---|
| 1495 |
|
|---|
| 1496 | static text_t expand_metadata(const text_t &metavalue, const text_t& collection,
|
|---|
| 1497 | recptproto* collectproto,
|
|---|
| 1498 | ResultDocInfo_t &docinfo,
|
|---|
| 1499 | displayclass &disp, text_tmap &options,
|
|---|
| 1500 | ostream &logout) {
|
|---|
| 1501 |
|
|---|
| 1502 | if (includes_metadata(metavalue)) {
|
|---|
| 1503 |
|
|---|
| 1504 | // text has embedded metadata in it => expand it
|
|---|
| 1505 | FilterRequest_t request;
|
|---|
| 1506 | FilterResponse_t response;
|
|---|
| 1507 |
|
|---|
| 1508 | request.getParents = false;
|
|---|
| 1509 |
|
|---|
| 1510 | format_t *expanded_formatlistptr = new format_t();
|
|---|
| 1511 | parse_formatstring (metavalue, expanded_formatlistptr,
|
|---|
| 1512 | request.fields, request.getParents);
|
|---|
| 1513 |
|
|---|
| 1514 | // retrieve metadata
|
|---|
| 1515 | get_info(docinfo.OID, collection, "", request.fields, request.getParents,
|
|---|
| 1516 | collectproto, response, logout);
|
|---|
| 1517 |
|
|---|
| 1518 | if (!response.docInfo.empty()) {
|
|---|
| 1519 |
|
|---|
| 1520 | text_t expanded_metavalue
|
|---|
| 1521 | = get_formatted_string(collection, collectproto,
|
|---|
| 1522 | response.docInfo[0], disp, expanded_formatlistptr,
|
|---|
| 1523 | options, logout);
|
|---|
| 1524 |
|
|---|
| 1525 | return expanded_metavalue;
|
|---|
| 1526 | }
|
|---|
| 1527 | else {
|
|---|
| 1528 | return metavalue;
|
|---|
| 1529 | }
|
|---|
| 1530 | }
|
|---|
| 1531 | else {
|
|---|
| 1532 |
|
|---|
| 1533 | return metavalue;
|
|---|
| 1534 | }
|
|---|
| 1535 | }
|
|---|
| 1536 |
|
|---|
| 1537 | text_t get_collection_meta(const text_t& collection, recptproto* collectproto,
|
|---|
| 1538 | displayclass &disp,
|
|---|
| 1539 | text_t meta_name, ostream& logout) {
|
|---|
| 1540 |
|
|---|
| 1541 | ColInfoResponse_t collectinfo;
|
|---|
| 1542 | comerror_t err;
|
|---|
| 1543 | collectproto->get_collectinfo (collection, collectinfo,err,logout);
|
|---|
| 1544 | text_t meta_value = "";
|
|---|
| 1545 | text_t lang;
|
|---|
| 1546 | disp.expandstring("_cgiargl_",lang);
|
|---|
| 1547 | if (lang.empty()) {
|
|---|
| 1548 | lang = "en";
|
|---|
| 1549 | }
|
|---|
| 1550 |
|
|---|
| 1551 | if (err == noError) {
|
|---|
| 1552 | meta_value = collectinfo.get_collectionmeta(meta_name, lang);
|
|---|
| 1553 | }
|
|---|
| 1554 | return meta_value;
|
|---|
| 1555 |
|
|---|
| 1556 |
|
|---|
| 1557 | }
|
|---|
| 1558 | text_t format_string (const text_t& collection, recptproto* collectproto,
|
|---|
| 1559 | ResultDocInfo_t &docinfo, displayclass &disp,
|
|---|
| 1560 | format_t *formatlistptr, text_tmap &options,
|
|---|
| 1561 | ostream& logout) {
|
|---|
| 1562 |
|
|---|
| 1563 | if (formatlistptr == NULL) return "";
|
|---|
| 1564 |
|
|---|
| 1565 | switch (formatlistptr->command) {
|
|---|
| 1566 | case comOID:
|
|---|
| 1567 | return docinfo.OID;
|
|---|
| 1568 | case comTopOID:
|
|---|
| 1569 | {
|
|---|
| 1570 | text_t top_id;
|
|---|
| 1571 | get_top(docinfo.OID, top_id);
|
|---|
| 1572 | return top_id;
|
|---|
| 1573 | }
|
|---|
| 1574 | case comRank:
|
|---|
| 1575 | return text_t(docinfo.ranking);
|
|---|
| 1576 | case comText:
|
|---|
| 1577 | return formatlistptr->text;
|
|---|
| 1578 | case comLink:
|
|---|
| 1579 | return options["link"];
|
|---|
| 1580 | case comEndLink:
|
|---|
| 1581 | if (options["link"].empty()) return "";
|
|---|
| 1582 | else return "</a>";
|
|---|
| 1583 | case comHref:
|
|---|
| 1584 | return get_href(options["link"]);
|
|---|
| 1585 | case comIcon:
|
|---|
| 1586 | return options["icon"];
|
|---|
| 1587 | case comNum:
|
|---|
| 1588 | return docinfo.result_num;
|
|---|
| 1589 | case comRel: //if [RelatedDocuments] appears in format string, collect relation data
|
|---|
| 1590 | return get_related_docs(collection, collectproto, docinfo, logout);
|
|---|
| 1591 |
|
|---|
| 1592 | case comSummary:
|
|---|
| 1593 | {
|
|---|
| 1594 | text_t summary = format_summary(collection, collectproto, docinfo, disp, options, logout);
|
|---|
| 1595 | if (metadata_spanwrap) {
|
|---|
| 1596 | summary = "<span class=\"editable-metadata\" metaname=\"Summary\">"+summary+"</span>";
|
|---|
| 1597 | }
|
|---|
| 1598 | return summary;
|
|---|
| 1599 | }
|
|---|
| 1600 |
|
|---|
| 1601 | case comMeta:
|
|---|
| 1602 | {
|
|---|
| 1603 | const text_t& metavalue = get_meta (collection,collectproto, docinfo, disp,formatlistptr->meta,options,logout);
|
|---|
| 1604 | return expand_metadata(metavalue, collection, collectproto, docinfo, disp, options, logout);
|
|---|
| 1605 | }
|
|---|
| 1606 |
|
|---|
| 1607 | case comDoc:
|
|---|
| 1608 | return format_text(collection, collectproto, docinfo, disp, options, logout);
|
|---|
| 1609 |
|
|---|
| 1610 | case comImage:
|
|---|
| 1611 | return expand_metadata(options["DocImage"], collection, collectproto, docinfo, disp, options, logout);
|
|---|
| 1612 | case comTOC:
|
|---|
| 1613 | return options["DocTOC"];
|
|---|
| 1614 | case comDocumentButtonDetach:
|
|---|
| 1615 | return options["DocumentButtonDetach"];
|
|---|
| 1616 | case comDocumentButtonHighlight:
|
|---|
| 1617 | return options["DocumentButtonHighlight"];
|
|---|
| 1618 | case comDocumentButtonExpandContents:
|
|---|
| 1619 | return options["DocumentButtonExpandContents"];
|
|---|
| 1620 | case comDocumentButtonExpandText:
|
|---|
| 1621 | return options["DocumentButtonExpandText"];
|
|---|
| 1622 | case comHighlight:
|
|---|
| 1623 | if (options["highlight"] == "1") return "<b>";
|
|---|
| 1624 | break;
|
|---|
| 1625 | case comEndHighlight:
|
|---|
| 1626 | if (options["highlight"] == "1") return "</b>";
|
|---|
| 1627 | break;
|
|---|
| 1628 | case comMetadataSpanWrap:
|
|---|
| 1629 | metadata_spanwrap=true; return "";
|
|---|
| 1630 | break;
|
|---|
| 1631 | case comEndMetadataSpanWrap:
|
|---|
| 1632 | metadata_spanwrap=false; return "";
|
|---|
| 1633 | break;
|
|---|
| 1634 | case comIf:
|
|---|
| 1635 | return get_if (collection, collectproto, docinfo, disp,
|
|---|
| 1636 | formatlistptr->decision, formatlistptr->ifptr,
|
|---|
| 1637 | formatlistptr->elseptr, options, logout);
|
|---|
| 1638 | case comOr:
|
|---|
| 1639 | return get_or (collection,collectproto, docinfo, disp, formatlistptr->orptr,
|
|---|
| 1640 | options, logout);
|
|---|
| 1641 | case comDocTermsFreqTotal:
|
|---|
| 1642 | return docinfo.num_terms_matched;
|
|---|
| 1643 | case comCollection:
|
|---|
| 1644 | if (formatlistptr->meta.metaname == g_EmptyText) {
|
|---|
| 1645 | return collection;
|
|---|
| 1646 | }
|
|---|
| 1647 | return get_collection_meta(collection, collectproto, disp, formatlistptr->meta.metaname, logout);
|
|---|
| 1648 |
|
|---|
| 1649 | }
|
|---|
| 1650 | return "";
|
|---|
| 1651 | }
|
|---|
| 1652 |
|
|---|
| 1653 | text_t get_formatted_string (const text_t& collection, recptproto* collectproto,
|
|---|
| 1654 | ResultDocInfo_t &docinfo, displayclass &disp,
|
|---|
| 1655 | format_t *formatlistptr, text_tmap &options,
|
|---|
| 1656 | ostream& logout) {
|
|---|
| 1657 |
|
|---|
| 1658 | text_t ft;
|
|---|
| 1659 | while (formatlistptr != NULL)
|
|---|
| 1660 | {
|
|---|
| 1661 | ft += format_string (collection, collectproto, docinfo, disp, formatlistptr,
|
|---|
| 1662 | options, logout);
|
|---|
| 1663 | formatlistptr = formatlistptr->nextptr;
|
|---|
| 1664 | }
|
|---|
| 1665 |
|
|---|
| 1666 | return ft;
|
|---|
| 1667 | }
|
|---|
| 1668 |
|
|---|
| 1669 |
|
|---|
| 1670 | // 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?
|
|---|
| 1671 | text_t format_text (const text_t& collection, recptproto* collectproto,
|
|---|
| 1672 | ResultDocInfo_t &docinfo, displayclass &disp,
|
|---|
| 1673 | text_tmap &options, ostream& logout)
|
|---|
| 1674 | {
|
|---|
| 1675 | text_t text;
|
|---|
| 1676 |
|
|---|
| 1677 | if(!options["text"].empty()) {
|
|---|
| 1678 | text = options["text"];
|
|---|
| 1679 | }
|
|---|
| 1680 | else {
|
|---|
| 1681 | // get document text here
|
|---|
| 1682 | DocumentRequest_t docrequest;
|
|---|
| 1683 | DocumentResponse_t docresponse;
|
|---|
| 1684 | comerror_t err;
|
|---|
| 1685 | docrequest.OID = docinfo.OID;
|
|---|
| 1686 | collectproto->get_document (collection, docrequest, docresponse, err, logout);
|
|---|
| 1687 | text = docresponse.doc;
|
|---|
| 1688 | }
|
|---|
| 1689 |
|
|---|
| 1690 | if (metadata_spanwrap) {
|
|---|
| 1691 | text = "<div class=\"editable-text\" metaname=\"Text\">"+text+"</div>";
|
|---|
| 1692 | }
|
|---|
| 1693 |
|
|---|
| 1694 | return text;
|
|---|
| 1695 | }
|
|---|
| 1696 |
|
|---|
| 1697 | /* FUNCTION NAME: format_summary
|
|---|
| 1698 | * DESC: this is invoked when a [Summary] special metadata is processed.
|
|---|
| 1699 | * RETURNS: a query-biased summary for the document */
|
|---|
| 1700 |
|
|---|
| 1701 | text_t format_summary (const text_t& collection, recptproto* collectproto,
|
|---|
| 1702 | ResultDocInfo_t &docinfo, displayclass &disp,
|
|---|
| 1703 | text_tmap &options, ostream& logout) {
|
|---|
| 1704 |
|
|---|
| 1705 | // GRB: added code here to ensure that the cstr (and other collections)
|
|---|
| 1706 | // uses the document metadata item Summary, rather than compressing
|
|---|
| 1707 | // the text of the document, processed via the methods in
|
|---|
| 1708 | // summarise.cpp
|
|---|
| 1709 | if (docinfo.metadata.count("Summary") > 0 &&
|
|---|
| 1710 | docinfo.metadata["Summary"].values.size() > 0) {
|
|---|
| 1711 | return docinfo.metadata["Summary"].values[0];
|
|---|
| 1712 | }
|
|---|
| 1713 |
|
|---|
| 1714 | text_t textToSummarise, query;
|
|---|
| 1715 | if(options["text"].empty()) { // get document text
|
|---|
| 1716 | DocumentRequest_t docrequest;
|
|---|
| 1717 | DocumentResponse_t docresponse;
|
|---|
| 1718 | comerror_t err;
|
|---|
| 1719 | docrequest.OID = docinfo.OID;
|
|---|
| 1720 | collectproto->get_document (collection, docrequest, docresponse, err, logout);
|
|---|
| 1721 | textToSummarise = docresponse.doc;
|
|---|
| 1722 | }
|
|---|
| 1723 | else {
|
|---|
| 1724 | // in practice, this would not happen, because text is only
|
|---|
| 1725 | // loaded with the [Text] command
|
|---|
| 1726 | textToSummarise = options["text"];
|
|---|
| 1727 | }
|
|---|
| 1728 |
|
|---|
| 1729 | disp.expandstring("_cgiargq_",query);
|
|---|
| 1730 | return summarise(textToSummarise,query,80);
|
|---|
| 1731 | }
|
|---|
