src/BiblioInfo.cpp

   1 /**
   2  * \file BiblioInfo.cpp
   3  * This file is part of LyX, the document processor.
   4  * Licence details can be found in the file COPYING.
   5  *
   6  * \author Angus Leeming
   7  * \author Herbert Voß
   8  * \author Richard Heck
   9  *
  10  * Full author contact details are available in file CREDITS.
  11  */
  12
  13 #include <config.h>
  14
  15 #include "BiblioInfo.h"
  16 #include "Buffer.h"
  17 #include "BufferParams.h"
  18 #include "buffer_funcs.h"
  19 #include "Encoding.h"
  20 #include "InsetIterator.h"
  21 #include "Paragraph.h"
  22
  23 #include "insets/Inset.h"
  24 #include "insets/InsetBibitem.h"
  25 #include "insets/InsetBibtex.h"
  26 #include "insets/InsetInclude.h"
  27
  28 #include "support/docstream.h"
  29 #include "support/gettext.h"
  30 #include "support/lassert.h"
  31 #include "support/lstrings.h"
  32 #include "support/textutils.h"
  33
  34 #include "boost/regex.hpp"
  35
  36 using namespace std;
  37 using namespace lyx::support;
  38
  39
  40 namespace lyx {
  41
  42 //////////////////////////////////////////////////////////////////////
  43 //
  44 // BibTeXInfo
  45 //
  46 //////////////////////////////////////////////////////////////////////
  47
  48 BibTeXInfo::BibTeXInfo(docstring const & key, docstring const & type)
  49         : is_bibtex_(true), bib_key_(key), entry_type_(type), info_()
  50 {}
  51
  52
  53 bool BibTeXInfo::hasField(docstring const & field) const
  54 {
  55         return count(field) == 1;
  56 }
  57
  58
  59 docstring const & BibTeXInfo::operator[](docstring const & field) const
  60 {
  61         BibTeXInfo::const_iterator it = find(field);
  62         if (it != end())
  63                 return it->second;
  64         static docstring const empty_value = docstring();
  65         return empty_value;
  66 }
  67
  68
  69 docstring const & BibTeXInfo::operator[](string const & field) const
  70 {
  71         return operator[](from_ascii(field));
  72 }
  73
  74
  75 docstring BibTeXInfo::getValueForKey(string const & key,
  76                 BibTeXInfo const * const xref) const
  77 {
  78         docstring const ret = operator[](key);
  79         if (!ret.empty() || !xref)
  80                 return ret;
  81         return (*xref)[key];
  82 }
  83
  84
  85 docstring familyName(docstring const & name)
  86 {
  87         if (name.empty())
  88                 return docstring();
  89
  90         // first we look for a comma, and take the last name to be everything
  91         // preceding the right-most one, so that we also get the "jr" part.
  92         docstring::size_type idx = name.rfind(',');
  93         if (idx != docstring::npos)
  94                 return ltrim(name.substr(0, idx));
  95
  96         // OK, so now we want to look for the last name. We're going to
  97         // include the "von" part. This isn't perfect.
  98         // Split on spaces, to get various tokens.
  99         vector<docstring> pieces = getVectorFromString(name, from_ascii(" "));
 100         // If we only get two, assume the last one is the last name
 101         if (pieces.size() <= 2)
 102                 return pieces.back();
 103
 104         // Now we look for the first token that begins with a lower case letter.
 105         vector<docstring>::const_iterator it = pieces.begin();
 106         vector<docstring>::const_iterator en = pieces.end();
 107         for (; it != en; ++it) {
 108                 if ((*it).size() == 0)
 109                         continue;
 110                 char_type const c = (*it)[0];
 111                 if (isLower(c))
 112                         break;
 113         }
 114
 115         if (it == en) // we never found a "von"
 116                 return pieces.back();
 117
 118         // reconstruct what we need to return
 119         docstring retval;
 120         bool first = true;
 121         for (; it != en; ++it) {
 122                 if (!first)
 123                         retval += " ";
 124                 else
 125                         first = false;
 126                 retval += *it;
 127         }
 128         return retval;
 129 }
 130
 131 docstring const BibTeXInfo::getAbbreviatedAuthor() const
 132 {
 133         if (!is_bibtex_) {
 134                 docstring const opt = trim(operator[]("label"));
 135                 if (opt.empty())
 136                         return docstring();
 137
 138                 docstring authors;
 139                 split(opt, authors, '(');
 140                 return authors;
 141         }
 142
 143         docstring author = operator[]("author");
 144         if (author.empty()) {
 145                 author = operator[]("editor");
 146                 if (author.empty())
 147                         return bib_key_;
 148         }
 149
 150         // OK, we've got some names. Let's format them.
 151         // Try to split the author list on " and "
 152         vector<docstring> const authors =
 153                 getVectorFromString(author, from_ascii(" and "));
 154
 155         if (authors.size() == 2)
 156                 return bformat(_("%1$s and %2$s"),
 157                         familyName(authors[0]), familyName(authors[1]));
 158
 159         if (authors.size() > 2)
 160                 return bformat(_("%1$s et al."), familyName(authors[0]));
 161
 162         return familyName(authors[0]);
 163 }
 164
 165
 166 docstring const BibTeXInfo::getYear() const
 167 {
 168         if (!is_bibtex_) {
 169                 docstring const opt = trim(operator[]("label"));
 170                 if (opt.empty())
 171                         return docstring();
 172
 173                 docstring authors;
 174                 docstring const tmp = split(opt, authors, '(');
 175                 docstring year;
 176                 split(tmp, year, ')');
 177                 return year;
 178         }
 179
 180         docstring year = operator[]("year");
 181         if (year.empty())
 182                 year = _("No year");
 183         return year;
 184 }
 185
 186
 187 docstring const BibTeXInfo::getXRef() const
 188 {
 189         if (!is_bibtex_)
 190                 return docstring();
 191         return operator[]("crossref");
 192 }
 193
 194
 195 namespace {
 196
 197         docstring convertLaTeXCommands(docstring const & str)
 198         {
 199                 docstring val = str;
 200                 docstring ret;
 201
 202                 bool scanning_cmd = false;
 203                 bool scanning_math = false;
 204                 bool escaped = false; // used to catch \$, etc.
 205                 while (val.size()) {
 206                         char_type const ch = val[0];
 207
 208                         // if we're scanning math, we output everything until we
 209                         // find an unescaped $, at which point we break out.
 210                         if (scanning_math) {
 211                                 if (escaped)
 212                                         escaped = false;
 213                                 else if (ch == '\\')
 214                                         escaped = true;
 215                                 else if (ch == '$')
 216                                         scanning_math = false;
 217                                 ret += ch;
 218                                 val = val.substr(1);
 219                                 continue;
 220                         }
 221
 222                         // if we're scanning a command name, then we just
 223                         // discard characters until we hit something that
 224                         // isn't alpha.
 225                         if (scanning_cmd) {
 226                                 if (isAlphaASCII(ch)) {
 227                                         val = val.substr(1);
 228                                         escaped = false;
 229                                         continue;
 230                                 }
 231                                 // so we're done with this command.
 232                                 // now we fall through and check this character.
 233                                 scanning_cmd = false;
 234                         }
 235
 236                         // was the last character a \? If so, then this is something like: \\,
 237                         // or \$, so we'll just output it. That's probably not always right...
 238                         if (escaped) {
 239                                 ret += ch;
 240                                 val = val.substr(1);
 241                                 escaped = false;
 242                                 continue;
 243                         }
 244
 245                         if (ch == '$') {
 246                                 ret += ch;
 247                                 val = val.substr(1);
 248                                 scanning_math = true;
 249                                 continue;
 250                         }
 251
 252                         // we just ignore braces
 253                         if (ch == '{' || ch == '}') {
 254                                 val = val.substr(1);
 255                                 continue;
 256                         }
 257
 258                         // we're going to check things that look like commands, so if
 259                         // this doesn't, just output it.
 260                         if (ch != '\\') {
 261                                 ret += ch;
 262                                 val = val.substr(1);
 263                                 continue;
 264                         }
 265
 266                         // ok, could be a command of some sort
 267                         // let's see if it corresponds to some unicode
 268                         // unicodesymbols has things in the form: \"{u},
 269                         // whereas we may see things like: \"u. So we'll
 270                         // look for that and change it, if necessary.
 271                         static boost::regex const reg("^\\\\\\W\\w");
 272                         if (boost::regex_search(to_utf8(val), reg)) {
 273                                 val.insert(3, from_ascii("}"));
 274                                 val.insert(2, from_ascii("{"));
 275                         }
 276                         docstring rem;
 277                         docstring const cnvtd = Encodings::fromLaTeXCommand(val, rem);
 278                         if (!cnvtd.empty()) {
 279                                 // it did, so we'll take that bit and proceed with what's left
 280                                 ret += cnvtd;
 281                                 val = rem;
 282                                 continue;
 283                         }
 284                         // it's a command of some sort
 285                         scanning_cmd = true;
 286                         escaped = true;
 287                         val = val.substr(1);
 288                 }
 289                 return ret;
 290         }
 291
 292 } // anon namespace
 293
 294
 295 docstring const & BibTeXInfo::getInfo(BibTeXInfo const * const xref) const
 296 {
 297         if (!info_.empty())
 298                 return info_;
 299
 300         if (!is_bibtex_) {
 301                 BibTeXInfo::const_iterator it = find(from_ascii("ref"));
 302                 info_ = it->second;
 303                 return info_;
 304         }
 305
 306         // FIXME
 307         // This could be made a lot better using the entry_type_
 308         // field to customize the output based upon entry type.
 309
 310         // Search for all possible "required" fields
 311         docstring author = getValueForKey("author", xref);
 312         if (author.empty())
 313                 author = getValueForKey("editor", xref);
 314
 315         docstring year   = getValueForKey("year", xref);
 316         docstring title  = getValueForKey("title", xref);
 317         docstring docLoc = getValueForKey("pages", xref);
 318         if (docLoc.empty()) {
 319                 docLoc = getValueForKey("chapter", xref);
 320                 if (!docLoc.empty())
 321                         docLoc = _("Ch. ") + docLoc;
 322         }       else {
 323                 docLoc = _("pp. ") + docLoc;
 324         }
 325
 326         docstring media = getValueForKey("journal", xref);
 327         if (media.empty()) {
 328                 media = getValueForKey("publisher", xref);
 329                 if (media.empty()) {
 330                         media = getValueForKey("school", xref);
 331                         if (media.empty())
 332                                 media = getValueForKey("institution");
 333                 }
 334         }
 335         docstring volume = getValueForKey("volume", xref);
 336
 337         odocstringstream result;
 338         if (!author.empty())
 339                 result << author << ", ";
 340         if (!title.empty())
 341                 result << title;
 342         if (!media.empty())
 343                 result << ", " << media;
 344         if (!year.empty())
 345                 result << " (" << year << ")";
 346         if (!docLoc.empty())
 347                 result << ", " << docLoc;
 348
 349         docstring const result_str = rtrim(result.str());
 350         if (!result_str.empty()) {
 351                 info_ = convertLaTeXCommands(result_str);
 352                 return info_;
 353         }
 354
 355         // This should never happen (or at least be very unusual!)
 356         static docstring e = docstring();
 357         return e;
 358 }
 359
 360
 361 //////////////////////////////////////////////////////////////////////
 362 //
 363 // BiblioInfo
 364 //
 365 //////////////////////////////////////////////////////////////////////
 366
 367 namespace {
 368 // A functor for use with sort, leading to case insensitive sorting
 369         class compareNoCase: public binary_function<docstring, docstring, bool>
 370         {
 371                 public:
 372                         bool operator()(docstring const & s1, docstring const & s2) const {
 373                                 return compare_no_case(s1, s2) < 0;
 374                         }
 375         };
 376 } // namespace anon
 377
 378
 379 vector<docstring> const BiblioInfo::getKeys() const
 380 {
 381         vector<docstring> bibkeys;
 382         BiblioInfo::const_iterator it  = begin();
 383         for (; it != end(); ++it)
 384                 bibkeys.push_back(it->first);
 385         sort(bibkeys.begin(), bibkeys.end(), compareNoCase());
 386         return bibkeys;
 387 }
 388
 389
 390 vector<docstring> const BiblioInfo::getFields() const
 391 {
 392         vector<docstring> bibfields;
 393         set<docstring>::const_iterator it = field_names_.begin();
 394         set<docstring>::const_iterator end = field_names_.end();
 395         for (; it != end; ++it)
 396                 bibfields.push_back(*it);
 397         sort(bibfields.begin(), bibfields.end());
 398         return bibfields;
 399 }
 400
 401
 402 vector<docstring> const BiblioInfo::getEntries() const
 403 {
 404         vector<docstring> bibentries;
 405         set<docstring>::const_iterator it = entry_types_.begin();
 406         set<docstring>::const_iterator end = entry_types_.end();
 407         for (; it != end; ++it)
 408                 bibentries.push_back(*it);
 409         sort(bibentries.begin(), bibentries.end());
 410         return bibentries;
 411 }
 412
 413
 414 docstring const BiblioInfo::getAbbreviatedAuthor(docstring const & key) const
 415 {
 416         BiblioInfo::const_iterator it = find(key);
 417         if (it == end())
 418                 return docstring();
 419         BibTeXInfo const & data = it->second;
 420         return data.getAbbreviatedAuthor();
 421 }
 422
 423
 424 docstring const BiblioInfo::getYear(docstring const & key) const
 425 {
 426         BiblioInfo::const_iterator it = find(key);
 427         if (it == end())
 428                 return docstring();
 429         BibTeXInfo const & data = it->second;
 430         docstring year = data.getYear();
 431         if (!year.empty())
 432                 return year;
 433         // let's try the crossref
 434         docstring const xref = data.getXRef();
 435         if (xref.empty())
 436                 return year; // no luck
 437         BiblioInfo::const_iterator const xrefit = find(xref);
 438         if (xrefit == end())
 439                 return year; // no luck again
 440         BibTeXInfo const & xref_data = xrefit->second;
 441         return xref_data.getYear();
 442         return data.getYear();
 443 }
 444
 445
 446 docstring const BiblioInfo::getInfo(docstring const & key) const
 447 {
 448         BiblioInfo::const_iterator it = find(key);
 449         if (it == end())
 450                 return docstring();
 451         BibTeXInfo const & data = it->second;
 452         BibTeXInfo const * xrefptr = 0;
 453         docstring const xref = data.getXRef();
 454         if (!xref.empty()) {
 455                 BiblioInfo::const_iterator const xrefit = find(xref);
 456                 if (xrefit != end())
 457                         xrefptr = &(xrefit->second);
 458         }
 459         return data.getInfo(xrefptr);
 460 }
 461
 462
 463 vector<docstring> const BiblioInfo::getCiteStrings(
 464         docstring const & key, Buffer const & buf) const
 465 {
 466         CiteEngine const engine = buf.params().citeEngine();
 467         if (engine == ENGINE_BASIC || engine == ENGINE_NATBIB_NUMERICAL)
 468                 return getNumericalStrings(key, buf);
 469         else
 470                 return getAuthorYearStrings(key, buf);
 471 }
 472
 473
 474 vector<docstring> const BiblioInfo::getNumericalStrings(
 475         docstring const & key, Buffer const & buf) const
 476 {
 477         if (empty())
 478                 return vector<docstring>();
 479
 480         docstring const author = getAbbreviatedAuthor(key);
 481         docstring const year   = getYear(key);
 482         if (author.empty() || year.empty())
 483                 return vector<docstring>();
 484
 485         vector<CiteStyle> const & styles = citeStyles(buf.params().citeEngine());
 486
 487         vector<docstring> vec(styles.size());
 488         for (size_t i = 0; i != vec.size(); ++i) {
 489                 docstring str;
 490
 491                 switch (styles[i]) {
 492                         case CITE:
 493                         case CITEP:
 494                                 str = from_ascii("[#ID]");
 495                                 break;
 496
 497                         case NOCITE:
 498                                 str = _("Add to bibliography only.");
 499                                 break;
 500
 501                         case CITET:
 502                                 str = author + " [#ID]";
 503                                 break;
 504
 505                         case CITEALT:
 506                                 str = author + " #ID";
 507                                 break;
 508
 509                         case CITEALP:
 510                                 str = from_ascii("#ID");
 511                                 break;
 512
 513                         case CITEAUTHOR:
 514                                 str = author;
 515                                 break;
 516
 517                         case CITEYEAR:
 518                                 str = year;
 519                                 break;
 520
 521                         case CITEYEARPAR:
 522                                 str = '(' + year + ')';
 523                                 break;
 524                 }
 525
 526                 vec[i] = str;
 527         }
 528
 529         return vec;
 530 }
 531
 532
 533 vector<docstring> const BiblioInfo::getAuthorYearStrings(
 534         docstring const & key, Buffer const & buf) const
 535 {
 536         if (empty())
 537                 return vector<docstring>();
 538
 539         docstring const author = getAbbreviatedAuthor(key);
 540         docstring const year   = getYear(key);
 541         if (author.empty() || year.empty())
 542                 return vector<docstring>();
 543
 544         vector<CiteStyle> const & styles = citeStyles(buf.params().citeEngine());
 545
 546         vector<docstring> vec(styles.size());
 547         for (size_t i = 0; i != vec.size(); ++i) {
 548                 docstring str;
 549
 550                 switch (styles[i]) {
 551                         case CITE:
 552                 // jurabib only: Author/Annotator
 553                 // (i.e. the "before" field, 2nd opt arg)
 554                                 str = author + "/<" + _("before") + '>';
 555                                 break;
 556
 557                         case NOCITE:
 558                                 str = _("Add to bibliography only.");
 559                                 break;
 560
 561                         case CITET:
 562                                 str = author + " (" + year + ')';
 563                                 break;
 564
 565                         case CITEP:
 566                                 str = '(' + author + ", " + year + ')';
 567                                 break;
 568
 569                         case CITEALT:
 570                                 str = author + ' ' + year ;
 571                                 break;
 572
 573                         case CITEALP:
 574                                 str = author + ", " + year ;
 575                                 break;
 576
 577                         case CITEAUTHOR:
 578                                 str = author;
 579                                 break;
 580
 581                         case CITEYEAR:
 582                                 str = year;
 583                                 break;
 584
 585                         case CITEYEARPAR:
 586                                 str = '(' + year + ')';
 587                                 break;
 588                 }
 589                 vec[i] = str;
 590         }
 591         return vec;
 592 }
 593
 594
 595 void BiblioInfo::mergeBiblioInfo(BiblioInfo const & info)
 596 {
 597         bimap_.insert(info.begin(), info.end());
 598 }
 599
 600
 601 //////////////////////////////////////////////////////////////////////
 602 //
 603 // CitationStyle
 604 //
 605 //////////////////////////////////////////////////////////////////////
 606
 607 namespace {
 608
 609
 610 char const * const citeCommands[] = {
 611         "cite", "citet", "citep", "citealt", "citealp",
 612         "citeauthor", "citeyear", "citeyearpar", "nocite" };
 613
 614 unsigned int const nCiteCommands =
 615                 sizeof(citeCommands) / sizeof(char *);
 616
 617 CiteStyle const citeStylesArray[] = {
 618         CITE, CITET, CITEP, CITEALT, CITEALP,
 619         CITEAUTHOR, CITEYEAR, CITEYEARPAR, NOCITE };
 620
 621 unsigned int const nCiteStyles =
 622                 sizeof(citeStylesArray) / sizeof(CiteStyle);
 623
 624 CiteStyle const citeStylesFull[] = {
 625         CITET, CITEP, CITEALT, CITEALP, CITEAUTHOR };
 626
 627 unsigned int const nCiteStylesFull =
 628                 sizeof(citeStylesFull) / sizeof(CiteStyle);
 629
 630 CiteStyle const citeStylesUCase[] = {
 631         CITET, CITEP, CITEALT, CITEALP, CITEAUTHOR };
 632
 633 unsigned int const nCiteStylesUCase =
 634         sizeof(citeStylesUCase) / sizeof(CiteStyle);
 635
 636 } // namespace anon
 637
 638
 639 CitationStyle citationStyleFromString(string const & command)
 640 {
 641         CitationStyle s;
 642         if (command.empty())
 643                 return s;
 644
 645         string cmd = command;
 646         if (cmd[0] == 'C') {
 647                 s.forceUpperCase = true;
 648                 cmd[0] = 'c';
 649         }
 650
 651         size_t const n = cmd.size() - 1;
 652         if (cmd != "cite" && cmd[n] == '*') {
 653                 s.full = true;
 654                 cmd = cmd.substr(0, n);
 655         }
 656
 657         char const * const * const last = citeCommands + nCiteCommands;
 658         char const * const * const ptr = find(citeCommands, last, cmd);
 659
 660         if (ptr != last) {
 661                 size_t idx = ptr - citeCommands;
 662                 s.style = citeStylesArray[idx];
 663         }
 664         return s;
 665 }
 666
 667
 668 string citationStyleToString(const CitationStyle & s)
 669 {
 670         string cite = citeCommands[s.style];
 671         if (s.full) {
 672                 CiteStyle const * last = citeStylesFull + nCiteStylesFull;
 673                 if (find(citeStylesFull, last, s.style) != last)
 674                         cite += '*';
 675         }
 676
 677         if (s.forceUpperCase) {
 678                 CiteStyle const * last = citeStylesUCase + nCiteStylesUCase;
 679                 if (find(citeStylesUCase, last, s.style) != last)
 680                         cite[0] = 'C';
 681         }
 682
 683         return cite;
 684 }
 685
 686 vector<CiteStyle> citeStyles(CiteEngine engine)
 687 {
 688         unsigned int nStyles = 0;
 689         unsigned int start = 0;
 690
 691         switch (engine) {
 692                 case ENGINE_BASIC:
 693                         nStyles = 2;
 694                         start = 0;
 695                         break;
 696                 case ENGINE_NATBIB_AUTHORYEAR:
 697                 case ENGINE_NATBIB_NUMERICAL:
 698                         nStyles = nCiteStyles - 1;
 699                         start = 1;
 700                         break;
 701                 case ENGINE_JURABIB:
 702                         nStyles = nCiteStyles;
 703                         start = 0;
 704                         break;
 705         }
 706
 707         vector<CiteStyle> styles(nStyles);
 708         size_t i = 0;
 709         int j = start;
 710         for (; i != styles.size(); ++i, ++j)
 711                 styles[i] = citeStylesArray[j];
 712
 713         return styles;
 714 }
 715
 716 } // namespace lyx
 717