src/frontends/controllers/biblio.C

   1 /* This file is part of
   2  * ======================================================
   3  *
   4  *           LyX, The Document Processor
   5  *
   6  *           Copyright 2001 The LyX Team.
   7  *
   8  * ======================================================
   9  *
  10  * \file biblio.C
  11  * \author Angus Leeming <a.leeming@ic.ac.uk>
  12  */
  13
  14 #include <config.h>
  15
  16 #include <vector>
  17 #include <algorithm>
  18
  19 #ifdef __GNUG__
  20 #pragma implementation
  21 #endif
  22
  23 #include "LString.h"
  24 #include "biblio.h"
  25 #include "gettext.h" // for _()
  26 #include "helper_funcs.h"
  27 #include "support/lstrings.h"
  28 #include "support/LAssert.h"
  29 #include "support/LRegex.h"
  30
  31 using std::find;
  32 using std::min;
  33 using std::vector;
  34 using std::sort;
  35
  36 namespace biblio
  37 {
  38
  39 namespace {
  40
  41 using namespace biblio;
  42
  43 char const * const citeCommands[] = {
  44         "cite", "citet", "citep", "citealt", "citealp", "citeauthor",
  45         "citeyear", "citeyearpar" };
  46
  47 unsigned int const nCiteCommands =
  48         sizeof(citeCommands) / sizeof(char *);
  49
  50 CiteStyle const citeStyles[] = {
  51         CITE, CITET, CITEP, CITEALT, CITEALP,
  52         CITEAUTHOR, CITEYEAR, CITEYEARPAR };
  53
  54 unsigned int const nCiteStyles =
  55         sizeof(citeStyles) / sizeof(CiteStyle);
  56
  57 CiteStyle const citeStylesFull[] = {
  58         CITET, CITEP, CITEALT, CITEALP, CITEAUTHOR };
  59
  60 unsigned int const nCiteStylesFull =
  61         sizeof(citeStylesFull) / sizeof(CiteStyle);
  62
  63 CiteStyle const citeStylesUCase[] = {
  64         CITET, CITEP, CITEALT, CITEALP, CITEAUTHOR };
  65
  66 unsigned int const nCiteStylesUCase =
  67         sizeof(citeStylesUCase) / sizeof(CiteStyle);
  68
  69
  70 // The functions doing the dirty work for the search.
  71 vector<string>::const_iterator
  72 simpleSearch(InfoMap const & theMap,
  73              vector<string> const & keys,
  74              string const & expr,
  75              vector<string>::const_iterator start,
  76              Direction dir,
  77              bool caseSensitive)
  78 {
  79         string tmp = expr;
  80         if (!caseSensitive)
  81                 tmp = lowercase(tmp);
  82
  83         vector<string> searchwords = getVectorFromString(tmp, " ");
  84
  85         // Loop over all keys from start...
  86         for (vector<string>::const_iterator it = start;
  87              // End condition is direction-dependent.
  88              (dir == FORWARD) ? (it<keys.end()) : (it>=keys.begin());
  89              // increment is direction-dependent.
  90              (dir == FORWARD) ? (++it) : (--it)) {
  91
  92                 string data = (*it);
  93                 InfoMap::const_iterator info = theMap.find(*it);
  94                 if (info != theMap.end())
  95                         data += " " + info->second;
  96                 if (!caseSensitive)
  97                         data = lowercase(data);
  98
  99                 bool found = true;
 100
 101                 // Loop over all search words...
 102                 for (vector<string>::const_iterator sit = searchwords.begin();
 103                      sit != searchwords.end(); ++sit) {
 104                         if (data.find(*sit) == string::npos) {
 105                                 found = false;
 106                                 break;
 107                         }
 108                 }
 109
 110                 if (found) return it;
 111         }
 112
 113         return keys.end();
 114 }
 115
 116
 117 vector<string>::const_iterator
 118 regexSearch(InfoMap const & theMap,
 119             vector<string> const & keys,
 120             string const & expr,
 121             vector<string>::const_iterator start,
 122             Direction dir)
 123 {
 124         LRegex reg(expr);
 125
 126         for (vector<string>::const_iterator it = start;
 127              // End condition is direction-dependent.
 128              (dir == FORWARD) ? (it<keys.end()) : (it>=keys.begin());
 129              // increment is direction-dependent.
 130              (dir == FORWARD) ? (++it) : (--it)) {
 131
 132                 string data = (*it);
 133                 InfoMap::const_iterator info = theMap.find(*it);
 134                 if (info != theMap.end())
 135                         data += " " + info->second;
 136
 137                 if (reg.exec(data).size() > 0)
 138                         return it;
 139         }
 140
 141         return keys.end();
 142 }
 143
 144 string const familyName(string const & name)
 145 {
 146         // Very simple parser
 147         string fname = name;
 148
 149         string::size_type idx = fname.rfind(".");
 150         if (idx != string::npos)
 151                 fname = frontStrip(fname.substr(idx+1));
 152
 153         return fname;
 154 }
 155
 156
 157 string const getAbbreviatedAuthor(InfoMap const & map, string const & key)
 158 {
 159         lyx::Assert(!map.empty());
 160
 161         InfoMap::const_iterator it = map.find(key);
 162
 163         string author;
 164         if (it != map.end()) {
 165                 author = parseBibTeX(it->second, "author");
 166                 if (author.empty())
 167                         author = parseBibTeX(it->second, "editor");
 168
 169                 vector<string> authors = getVectorFromString(author, "and");
 170
 171                 if (!authors.empty()) {
 172                         author.erase();
 173
 174                         for (vector<string>::iterator it = authors.begin();
 175                              it != authors.end(); ++it) {
 176                                 *it = familyName(strip(*it));
 177                         }
 178
 179                         author = authors[0];
 180                         if (authors.size() == 2)
 181                                 author += _(" and ") + authors[1];
 182                         else if (authors.size() > 2)
 183                                 author += _(" et al.");
 184                 }
 185         }
 186
 187         if (author.empty())
 188                 author = _("Caesar et al.");
 189
 190         return author;
 191 }
 192
 193
 194 string const getYear(InfoMap const & map, string const & key)
 195 {
 196         lyx::Assert(!map.empty());
 197
 198         InfoMap::const_iterator it = map.find(key);
 199
 200         string year;
 201
 202         if (it != map.end())
 203                 year = parseBibTeX(it->second, "year");
 204
 205         if (year.empty())
 206                 year = "50BC";
 207
 208         return year;
 209 }
 210
 211 } // namespace anon
 212
 213
 214
 215
 216
 217
 218
 219 // A functor for use with std::sort, leading to case insensitive sorting
 220 struct compareNoCase: public std::binary_function<string, string, bool>
 221 {
 222         bool operator()(string const & s1, string const & s2) const {
 223                 return compare_no_case(s1, s2) < 0;
 224         }
 225 };
 226
 227 vector<string> const getKeys(InfoMap const & map)
 228 {
 229         vector<string> bibkeys;
 230
 231         for (InfoMap::const_iterator it = map.begin(); it != map.end(); ++it) {
 232                 bibkeys.push_back(it->first);
 233         }
 234
 235         sort(bibkeys.begin(), bibkeys.end(), compareNoCase());
 236         return bibkeys;
 237 }
 238
 239
 240 string const getInfo(InfoMap const & map, string const & key)
 241 {
 242         lyx::Assert(!map.empty());
 243
 244         InfoMap::const_iterator it = map.find(key);
 245         if (it == map.end()) return string();
 246
 247         // Search for all possible "required" keys
 248         string author = parseBibTeX(it->second, "author");
 249         if (author.empty())
 250                 author = parseBibTeX(it->second, "editor");
 251
 252         string year       = parseBibTeX(it->second, "year");
 253         string title      = parseBibTeX(it->second, "title");
 254         string booktitle  = parseBibTeX(it->second, "booktitle");
 255         string chapter    = parseBibTeX(it->second, "chapter");
 256         string pages      = parseBibTeX(it->second, "pages");
 257
 258         string media      = parseBibTeX(it->second, "journal");
 259         if (media.empty())
 260                 media = parseBibTeX(it->second, "publisher");
 261         if (media.empty())
 262                 media = parseBibTeX(it->second, "school");
 263         if (media.empty())
 264                 media = parseBibTeX(it->second, "institution");
 265
 266         ostringstream result;
 267         result << author;
 268         if (!year.empty())
 269                 result << ", " << year;
 270         if (!title.empty())
 271                 result << ", " << title;
 272         if (!booktitle.empty())
 273                 result << ", in " << booktitle;
 274         if (!chapter.empty())
 275                 result << ", Ch. " << chapter;
 276         if (!media.empty())
 277                 result << ", " << media;
 278         if (!pages.empty())
 279                 result << ", pp. " << pages;
 280
 281         if (result.str().empty()) // not a BibTeX record
 282                 result << it->second;
 283
 284         return result.str().c_str();
 285 }
 286
 287
 288 vector<string>::const_iterator
 289 searchKeys(InfoMap const & theMap,
 290            vector<string> const & keys,
 291            string const & expr,
 292            vector<string>::const_iterator start,
 293            Search type,
 294            Direction dir,
 295            bool caseSensitive)
 296 {
 297         // Preliminary checks
 298         if(start < keys.begin() || start >= keys.end())
 299                 return keys.end();
 300
 301         string search_expr = frontStrip(strip(expr));
 302         if (search_expr.empty())
 303                 return keys.end();
 304
 305         if (type == SIMPLE)
 306                 return simpleSearch(theMap, keys, search_expr, start, dir,
 307                                     caseSensitive);
 308
 309         return regexSearch(theMap, keys, search_expr, start, dir);
 310 }
 311
 312
 313 string const parseBibTeX(string data, string const & findkey)
 314 {
 315         string keyvalue;
 316
 317         for (string::iterator it=data.begin(); it<data.end(); ++it) {
 318                 if ((*it) == '\n' || (*it) == '\t')
 319                         (*it)= ' ';
 320         }
 321
 322         data = frontStrip(data);
 323         while (!data.empty() && data[0] != '=' &&
 324                (data.find(' ') != string::npos || data.find('=') != string::npos)) {
 325
 326                 string::size_type keypos = min(data.find(' '), data.find('='));
 327                 string key = lowercase(data.substr(0, keypos));
 328
 329                 data = data.substr(keypos, data.length()-1);
 330                 data = frontStrip(strip(data));
 331                 if (data.length() > 1 && data[0]=='=') {
 332                         data = frontStrip(data.substr(1, data.length()-1));
 333                         if (!data.empty()) {
 334                                 keypos = 1;
 335                                 string value;
 336                                 char enclosing;
 337
 338                                 if (data[0]=='{') {
 339                                         enclosing = '}';
 340                                 } else if (data[0]=='"') {
 341                                         enclosing = '"';
 342                                 } else {
 343                                         keypos=0;
 344                                         enclosing=' ';
 345                                 }
 346
 347                                 if (keypos &&
 348                                     data.find(enclosing)!=string::npos &&
 349                                     data.length()>1) {
 350                                         string tmp = data.substr(keypos,
 351                                                                  data.length()-1);
 352                                         while (tmp.find('{') != string::npos &&
 353                                                tmp.find('}') != string::npos &&
 354                                                tmp.find('{') < tmp.find('}') &&
 355                                                tmp.find('{') < tmp.find(enclosing)) {
 356
 357                                                 keypos += tmp.find('{')+1;
 358                                                 tmp = data.substr(keypos,
 359                                                                   data.length()-1);
 360                                                 keypos += tmp.find('}')+1;
 361                                                 tmp = data.substr(keypos,
 362                                                                   data.length()-1);
 363                                         }
 364
 365                                         if (tmp.find(enclosing)==string::npos)
 366                                                 return keyvalue;
 367                                         else {
 368                                                 keypos += tmp.find(enclosing);
 369                                                 tmp = data.substr(keypos,
 370                                                                   data.length()-1);
 371                                         }
 372
 373                                         value = data.substr(1, keypos-1);
 374
 375                                         if (keypos+1<data.length()-1)
 376                                                 data = frontStrip(data.substr(keypos+1, data.length()-1));
 377                                         else
 378                                                 data = "";
 379
 380                                 } else if (!keypos &&
 381                                            (data.find(' ') ||
 382                                             data.find(','))) {
 383                                         keypos = data.length()-1;
 384                                         if (data.find(' ') != string::npos)
 385                                                 keypos = data.find(' ');
 386                                         if (data.find(',') != string::npos &&
 387                                             keypos > data.find(','))
 388                                                 keypos = data.find(',');
 389
 390                                         value = data.substr(0, keypos);
 391
 392                                         if (keypos+1<data.length()-1)
 393                                                 data = frontStrip(data.substr(keypos+1, data.length()-1));
 394                                         else
 395                                                 data = "";
 396                                 }
 397                                 else
 398                                         return keyvalue;
 399
 400                                 if (findkey == key) {
 401                                         keyvalue = value;
 402                                         return keyvalue;
 403                                 }
 404
 405                                 data = frontStrip(frontStrip(data,','));
 406                         }
 407                 }
 408                 else return keyvalue;
 409         }
 410         return keyvalue;
 411 }
 412
 413
 414 CitationStyle const getCitationStyle(string const & command)
 415 {
 416         if (command.empty()) return CitationStyle();
 417
 418         CitationStyle cs;
 419         string cmd = command;
 420
 421         if (cmd[0] == 'C') {
 422                 cs.forceUCase = true;
 423                 cmd[0] = 'c';
 424         }
 425
 426         size_t n = cmd.size()-1;
 427         if (cmd[n] == '*') {
 428                 cs.full = true;
 429                 cmd = cmd.substr(0,n);
 430         }
 431
 432         char const * const * const last = citeCommands + nCiteCommands;
 433         char const * const * const ptr = std::find(citeCommands, last, cmd);
 434
 435         if (ptr != last) {
 436                 size_t idx = ptr - citeCommands;
 437                 cs.style = citeStyles[idx];
 438         }
 439
 440         return cs;
 441 }
 442
 443
 444 string const getCiteCommand(CiteStyle command, bool full, bool forceUCase)
 445 {
 446         string cite = citeCommands[command];
 447         if (full) {
 448                 CiteStyle const * last = citeStylesFull + nCiteStylesFull;
 449                 if (std::find(citeStylesFull, last, command) != last)
 450                         cite += "*";
 451         }
 452
 453         if (forceUCase) {
 454                 CiteStyle const * last = citeStylesUCase + nCiteStylesUCase;
 455                 if (std::find(citeStylesUCase, last, command) != last)
 456                         cite[0] = 'C';
 457         }
 458
 459         return cite;
 460 }
 461
 462
 463 vector<CiteStyle> const getCiteStyles(bool usingNatbib)
 464 {
 465         unsigned int nStyles = 1;
 466         unsigned int start = 0;
 467         if (usingNatbib) {
 468                 nStyles = nCiteStyles - 1;
 469                 start = 1;
 470         }
 471
 472         vector<CiteStyle> styles(nStyles);
 473
 474         vector<CiteStyle>::size_type i = 0;
 475         int j = start;
 476         for (; i != styles.size(); ++i, ++j) {
 477                 styles[i] = citeStyles[j];
 478         }
 479
 480         return styles;
 481 }
 482
 483
 484 vector<string> const
 485 getNumericalStrings(string const & key,
 486                     InfoMap const & map, vector<CiteStyle> const & styles)
 487 {
 488         if (map.empty()) {
 489                 vector<string> vec(1);
 490                 vec[0] = _("No database");
 491                 return vec;
 492         }
 493
 494         vector<string> vec(styles.size());
 495
 496         string const author = getAbbreviatedAuthor(map, key);
 497         string const year   = getYear(map, key);
 498
 499         for (vector<string>::size_type i = 0; i != vec.size(); ++i) {
 500                 string str;
 501
 502                 switch (styles[i]) {
 503                 case CITE:
 504                 case CITEP:
 505                         str = "[#ID]";
 506                         break;
 507
 508                 case CITET:
 509                         str = author + " [#ID]";
 510                         break;
 511
 512                 case CITEALT:
 513                         str = author + " #ID";
 514                         break;
 515
 516                 case CITEALP:
 517                         str = "#ID";
 518                         break;
 519
 520                 case CITEAUTHOR:
 521                         str = author;
 522                         break;
 523
 524                 case CITEYEAR:
 525                         str = year;
 526                         break;
 527
 528                 case CITEYEARPAR:
 529                         str = "(" + year + ")";
 530                         break;
 531                 }
 532
 533                 vec[i] = str;
 534         }
 535
 536         return vec;
 537 }
 538
 539
 540 vector<string> const
 541 getAuthorYearStrings(string const & key,
 542                     InfoMap const & map, vector<CiteStyle> const & styles)
 543 {
 544         if (map.empty()) {
 545                 vector<string> vec(1);
 546                 vec[0] = _("No database");
 547                 return vec;
 548         }
 549
 550         vector<string> vec(styles.size());
 551
 552         string const author = getAbbreviatedAuthor(map, key);
 553         string const year   = getYear(map, key);
 554
 555         for (vector<string>::size_type i = 0; i != vec.size(); ++i) {
 556                 string str;
 557
 558                 switch (styles[i]) {
 559                 case CITET:
 560                         str = author + " (" + year + ")";
 561                         break;
 562
 563                 case CITE:
 564                 case CITEP:
 565                         str = "(" + author + ", " + year + ")";
 566                         break;
 567
 568                 case CITEALT:
 569                         str = author + " " + year ;
 570                         break;
 571
 572                 case CITEALP:
 573                         str = author + ", " + year ;
 574                         break;
 575
 576                 case CITEAUTHOR:
 577                         str = author;
 578                         break;
 579
 580                 case CITEYEAR:
 581                         str = year;
 582                         break;
 583
 584                 case CITEYEARPAR:
 585                         str = "(" + year + ")";
 586                         break;
 587                 }
 588
 589                 vec[i] = str;
 590         }
 591
 592         return vec;
 593 }
 594
 595 } // namespace biblio