src/insets/InsetCitation.cpp

   1 /**
   2  * \file InsetCitation.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  *
   9  * Full author contact details are available in file CREDITS.
  10  */
  11
  12 #include <config.h>
  13
  14 #include "InsetCitation.h"
  15
  16 #include "Buffer.h"
  17 #include "BufferParams.h"
  18 #include "debug.h"
  19 #include "DispatchResult.h"
  20 #include "FuncRequest.h"
  21 #include "LaTeXFeatures.h"
  22
  23 #include "support/fs_extras.h"
  24 #include "support/lstrings.h"
  25
  26 #include <algorithm>
  27
  28 #include <boost/filesystem/operations.hpp>
  29 #include <boost/filesystem/exception.hpp>
  30
  31
  32 namespace lyx {
  33
  34 using support::FileName;
  35 using support::getStringFromVector;
  36 using support::getVectorFromString;
  37 using support::ltrim;
  38 using support::prefixIs;
  39 using support::rtrim;
  40 using support::split;
  41 using support::tokenPos;
  42
  43 using std::endl;
  44 using std::string;
  45 using std::vector;
  46
  47 namespace fs = boost::filesystem;
  48
  49
  50 namespace {
  51
  52 vector<string> const init_possible_cite_commands()
  53 {
  54         char const * const possible[] = {
  55                 "cite", "citet", "citep", "citealt", "citealp",
  56                 "citeauthor", "citeyear", "citeyearpar",
  57                 "citet*", "citep*", "citealt*", "citealp*", "citeauthor*",
  58                 "Citet",  "Citep",  "Citealt",  "Citealp",  "Citeauthor",
  59                 "Citet*", "Citep*", "Citealt*", "Citealp*", "Citeauthor*",
  60                 "fullcite",
  61                 "footcite", "footcitet", "footcitep", "footcitealt",
  62                 "footcitealp", "footciteauthor", "footciteyear", "footciteyearpar",
  63                 "citefield", "citetitle", "cite*"
  64         };
  65         size_t const size_possible = sizeof(possible) / sizeof(possible[0]);
  66
  67         return vector<string>(possible, possible + size_possible);
  68 }
  69
  70
  71 vector<string> const & possible_cite_commands()
  72 {
  73         static vector<string> const possible = init_possible_cite_commands();
  74         return possible;
  75 }
  76
  77
  78 bool is_possible_cite_command(string const & input)
  79 {
  80         vector<string> const & possibles = possible_cite_commands();
  81         vector<string>::const_iterator const end = possibles.end();
  82         return std::find(possibles.begin(), end, input) != end;
  83 }
  84
  85
  86 string const default_cite_command(biblio::CiteEngine engine)
  87 {
  88         string str;
  89         switch (engine) {
  90                 case biblio::ENGINE_BASIC:
  91                         str = "cite";
  92                         break;
  93                 case biblio::ENGINE_NATBIB_AUTHORYEAR:
  94                         str = "citet";
  95                         break;
  96                 case biblio::ENGINE_NATBIB_NUMERICAL:
  97                         str = "citep";
  98                         break;
  99                 case biblio::ENGINE_JURABIB:
 100                         str = "cite";
 101                         break;
 102         }
 103         return str;
 104 }
 105
 106
 107 string const
 108                 asValidLatexCommand(string const & input, biblio::CiteEngine const engine)
 109 {
 110         string const default_str = default_cite_command(engine);
 111         if (!is_possible_cite_command(input))
 112                 return default_str;
 113
 114         string output;
 115         switch (engine) {
 116                 case biblio::ENGINE_BASIC:
 117                         output = default_str;
 118                         break;
 119
 120                 case biblio::ENGINE_NATBIB_AUTHORYEAR:
 121                 case biblio::ENGINE_NATBIB_NUMERICAL:
 122                         if (input == "cite" || input == "citefield" ||
 123                                                         input == "citetitle" || input == "cite*")
 124                                 output = default_str;
 125                         else if (prefixIs(input, "foot"))
 126                                 output = input.substr(4);
 127                         else
 128                                 output = input;
 129                         break;
 130
 131                         case biblio::ENGINE_JURABIB: {
 132                 // Jurabib does not support the 'uppercase' natbib style.
 133                                 if (input[0] == 'C')
 134                                         output = string(1, 'c') + input.substr(1);
 135                                 else
 136                                         output = input;
 137
 138                 // Jurabib does not support the 'full' natbib style.
 139                                 string::size_type const n = output.size() - 1;
 140                                 if (output != "cite*" && output[n] == '*')
 141                                         output = output.substr(0, n);
 142
 143                                 break;
 144                         }
 145         }
 146
 147         return output;
 148 }
 149
 150
 151 docstring const getNatbibLabel(Buffer const & buffer,
 152                             string const & citeType, docstring const & keyList,
 153                             docstring const & before, docstring const & after,
 154                             biblio::CiteEngine engine)
 155 {
 156         // Only start the process off after the buffer is loaded from file.
 157         if (!buffer.fully_loaded())
 158                 return docstring();
 159
 160         // Cache the labels
 161         typedef std::map<Buffer const *, BiblioInfo> CachedMap;
 162         static CachedMap cached_keys;
 163
 164         // and cache the timestamp of the bibliography files.
 165         static std::map<FileName, time_t> bibfileStatus;
 166
 167         BiblioInfo biblist;
 168
 169         vector<FileName> const & bibfilesCache = buffer.getBibfilesCache();
 170         // compare the cached timestamps with the actual ones.
 171         bool changed = false;
 172         for (vector<FileName>::const_iterator it = bibfilesCache.begin();
 173                         it != bibfilesCache.end(); ++ it) {
 174                 FileName const f = *it;
 175                 try {
 176                         std::time_t lastw = fs::last_write_time(f.toFilesystemEncoding());
 177                         if (lastw != bibfileStatus[f]) {
 178                                 changed = true;
 179                                 bibfileStatus[f] = lastw;
 180                         }
 181                 }
 182                 catch (fs::filesystem_error & fserr) {
 183                         changed = true;
 184                         lyxerr << "Couldn't find or read bibtex file "
 185                                << f << endl;
 186                         LYXERR(Debug::DEBUG) << "Fs error: "
 187                                              << fserr.what() << endl;
 188                 }
 189         }
 190
 191         // build the list only if the bibfiles have been changed
 192         if (cached_keys[&buffer].empty() || bibfileStatus.empty() || changed) {
 193                 biblist.fillWithBibKeys(&buffer);
 194                 cached_keys[&buffer] = biblist;
 195         } else
 196                 // use the cached keys
 197                 biblist = cached_keys[&buffer];
 198
 199         if (biblist.empty())
 200                 return docstring();
 201
 202         // the natbib citation-styles
 203         // CITET:       author (year)
 204         // CITEP:       (author,year)
 205         // CITEALT:     author year
 206         // CITEALP:     author, year
 207         // CITEAUTHOR:  author
 208         // CITEYEAR:    year
 209         // CITEYEARPAR: (year)
 210         // jurabib supports these plus
 211         // CITE:        author/<before field>
 212
 213         // We don't currently use the full or forceUCase fields.
 214         string cite_type = asValidLatexCommand(citeType, engine);
 215         if (cite_type[0] == 'C')
 216                 //If we were going to use them, this would mean ForceUCase
 217                 cite_type = string(1, 'c') + cite_type.substr(1);
 218         if (cite_type[cite_type.size() - 1] == '*')
 219                 //and this would mean FULL
 220                 cite_type = cite_type.substr(0, cite_type.size() - 1);
 221
 222         docstring before_str;
 223         if (!before.empty()) {
 224                 // In CITET and CITEALT mode, the "before" string is
 225                 // attached to the label associated with each and every key.
 226                 // In CITEP, CITEALP and CITEYEARPAR mode, it is attached
 227                 // to the front of the whole only.
 228                 // In other modes, it is not used at all.
 229                 if (cite_type == "citet" ||
 230                     cite_type == "citealt" ||
 231                     cite_type == "citep" ||
 232                     cite_type == "citealp" ||
 233                     cite_type == "citeyearpar")
 234                         before_str = before + ' ';
 235                 // In CITE (jurabib), the "before" string is used to attach
 236                 // the annotator (of legal texts) to the author(s) of the
 237                 // first reference.
 238                 else if (cite_type == "cite")
 239                         before_str = '/' + before;
 240         }
 241
 242         docstring after_str;
 243         if (!after.empty()) {
 244                 // The "after" key is appended only to the end of the whole.
 245                 after_str = ", " + after;
 246         }
 247
 248         // One day, these might be tunable (as they are in BibTeX).
 249         char const op  = '('; // opening parenthesis.
 250         char const cp  = ')'; // closing parenthesis.
 251         // puctuation mark separating citation entries.
 252         char const * const sep = ";";
 253
 254         docstring const op_str(' ' + docstring(1, op));
 255         docstring const cp_str(docstring(1, cp) + ' ');
 256         docstring const sep_str(from_ascii(sep) + ' ');
 257
 258         docstring label;
 259         vector<docstring> keys = getVectorFromString(keyList);
 260         vector<docstring>::const_iterator it  = keys.begin();
 261         vector<docstring>::const_iterator end = keys.end();
 262         for (; it != end; ++it) {
 263                 // get the bibdata corresponding to the key
 264                 docstring const author(biblist.getAbbreviatedAuthor(*it));
 265                 docstring const year(biblist.getYear(*it));
 266
 267                 // Something isn't right. Fail safely.
 268                 if (author.empty() || year.empty())
 269                         return docstring();
 270
 271                 // authors1/<before>;  ... ;
 272                 //  authors_last, <after>
 273                 if (cite_type == "cite" && engine == biblio::ENGINE_JURABIB) {
 274                         if (it == keys.begin())
 275                                 label += author + before_str + sep_str;
 276                         else
 277                                 label += author + sep_str;
 278
 279                 // (authors1 (<before> year);  ... ;
 280                 //  authors_last (<before> year, <after>)
 281                 } else if (cite_type == "citet") {
 282                         switch (engine) {
 283                         case biblio::ENGINE_NATBIB_AUTHORYEAR:
 284                                 label += author + op_str + before_str +
 285                                         year + cp + sep_str;
 286                                 break;
 287                         case biblio::ENGINE_NATBIB_NUMERICAL:
 288                                 label += author + op_str + before_str + '#' + *it + cp + sep_str;
 289                                 break;
 290                         case biblio::ENGINE_JURABIB:
 291                                 label += before_str + author + op_str +
 292                                         year + cp + sep_str;
 293                                 break;
 294                         case biblio::ENGINE_BASIC:
 295                                 break;
 296                         }
 297
 298                 // author, year; author, year; ...
 299                 } else if (cite_type == "citep" ||
 300                            cite_type == "citealp") {
 301                         if (engine == biblio::ENGINE_NATBIB_NUMERICAL) {
 302                                 label += *it + sep_str;
 303                         } else {
 304                                 label += author + ", " + year + sep_str;
 305                         }
 306
 307                 // (authors1 <before> year;
 308                 //  authors_last <before> year, <after>)
 309                 } else if (cite_type == "citealt") {
 310                         switch (engine) {
 311                         case biblio::ENGINE_NATBIB_AUTHORYEAR:
 312                                 label += author + ' ' + before_str +
 313                                         year + sep_str;
 314                                 break;
 315                         case biblio::ENGINE_NATBIB_NUMERICAL:
 316                                 label += author + ' ' + before_str + '#' + *it + sep_str;
 317                                 break;
 318                         case biblio::ENGINE_JURABIB:
 319                                 label += before_str + author + ' ' +
 320                                         year + sep_str;
 321                                 break;
 322                         case biblio::ENGINE_BASIC:
 323                                 break;
 324                         }
 325
 326                 // author; author; ...
 327                 } else if (cite_type == "citeauthor") {
 328                         label += author + sep_str;
 329
 330                 // year; year; ...
 331                 } else if (cite_type == "citeyear" ||
 332                            cite_type == "citeyearpar") {
 333                         label += year + sep_str;
 334                 }
 335         }
 336         label = rtrim(rtrim(label), sep);
 337
 338         if (!after_str.empty()) {
 339                 if (cite_type == "citet") {
 340                         // insert "after" before last ')'
 341                         label.insert(label.size() - 1, after_str);
 342                 } else {
 343                         bool const add =
 344                                 !(engine == biblio::ENGINE_NATBIB_NUMERICAL &&
 345                                   (cite_type == "citeauthor" ||
 346                                    cite_type == "citeyear"));
 347                         if (add)
 348                                 label += after_str;
 349                 }
 350         }
 351
 352         if (!before_str.empty() && (cite_type == "citep" ||
 353                                     cite_type == "citealp" ||
 354                                     cite_type == "citeyearpar")) {
 355                 label = before_str + label;
 356         }
 357
 358         if (cite_type == "citep" || cite_type == "citeyearpar")
 359                 label = op + label + cp;
 360
 361         return label;
 362 }
 363
 364
 365 docstring const getBasicLabel(docstring const & keyList, docstring const & after)
 366 {
 367         using support::contains;
 368
 369         docstring keys(keyList);
 370         docstring label;
 371
 372         if (contains(keys, ',')) {
 373                 // Final comma allows while loop to cover all keys
 374                 keys = ltrim(split(keys, label, ',')) + ',';
 375                 while (contains(keys, ',')) {
 376                         docstring key;
 377                         keys = ltrim(split(keys, key, ','));
 378                         label += ", " + key;
 379                 }
 380         } else
 381                 label = keys;
 382
 383         if (!after.empty())
 384                 label += ", " + after;
 385
 386         return '[' + label + ']';
 387 }
 388
 389 } // anon namespace
 390
 391
 392 InsetCitation::InsetCitation(InsetCommandParams const & p)
 393         : InsetCommand(p, "citation")
 394 {}
 395
 396
 397 docstring const InsetCitation::generateLabel(Buffer const & buffer) const
 398 {
 399         docstring const before = getParam("before");
 400         docstring const after  = getParam("after");
 401
 402         docstring label;
 403         biblio::CiteEngine const engine = buffer.params().getEngine();
 404         if (engine != biblio::ENGINE_BASIC) {
 405                 label = getNatbibLabel(buffer, getCmdName(), getParam("key"),
 406                                        before, after, engine);
 407         }
 408
 409         // Fallback to fail-safe
 410         if (label.empty()) {
 411                 label = getBasicLabel(getParam("key"), after);
 412         }
 413
 414         return label;
 415 }
 416
 417
 418 docstring const InsetCitation::getScreenLabel(Buffer const & buffer) const
 419 {
 420         biblio::CiteEngine const engine = buffer.params().getEngine();
 421         if (cache.params == params() && cache.engine == engine)
 422                 return cache.screen_label;
 423
 424         // The label has changed, so we have to re-create it.
 425         docstring const glabel = generateLabel(buffer);
 426
 427         unsigned int const maxLabelChars = 45;
 428
 429         docstring label = glabel;
 430         if (label.size() > maxLabelChars) {
 431                 label.erase(maxLabelChars-3);
 432                 label += "...";
 433         }
 434
 435         cache.engine  = engine;
 436         cache.params = params();
 437         cache.generated_label = glabel;
 438         cache.screen_label = label;
 439
 440         return label;
 441 }
 442
 443
 444 int InsetCitation::plaintext(Buffer const & buffer, odocstream & os,
 445                              OutputParams const &) const
 446 {
 447         docstring str;
 448
 449         if (cache.params == params() &&
 450             cache.engine == buffer.params().getEngine())
 451                 str = cache.generated_label;
 452         else
 453                 str = generateLabel(buffer);
 454
 455         os << str;
 456         return str.size();
 457 }
 458
 459
 460 static docstring const cleanupWhitespace(docstring const & citelist)
 461 {
 462         docstring::const_iterator it  = citelist.begin();
 463         docstring::const_iterator end = citelist.end();
 464         // Paranoia check: make sure that there is no whitespace in here
 465         // -- at least not behind commas or at the beginning
 466         docstring result;
 467         char_type last = ',';
 468         for (; it != end; ++it) {
 469                 if (*it != ' ')
 470                         last = *it;
 471                 if (*it != ' ' || last != ',')
 472                         result += *it;
 473         }
 474         return result;
 475 }
 476
 477
 478 int InsetCitation::docbook(Buffer const &, odocstream & os,
 479                            OutputParams const &) const
 480 {
 481         os << "<citation>"
 482            << cleanupWhitespace(getParam("key"))
 483            << "</citation>";
 484         return 0;
 485 }
 486
 487
 488 int InsetCitation::textString(Buffer const & buf, odocstream & os,
 489                        OutputParams const & op) const
 490 {
 491         return plaintext(buf, os, op);
 492 }
 493
 494
 495 // Have to overwrite the default InsetCommand method in order to check that
 496 // the \cite command is valid. Eg, the user has natbib enabled, inputs some
 497 // citations and then changes his mind, turning natbib support off. The output
 498 // should revert to \cite[]{}
 499 int InsetCitation::latex(Buffer const & buffer, odocstream & os,
 500                          OutputParams const &) const
 501 {
 502         biblio::CiteEngine cite_engine = buffer.params().getEngine();
 503         // FIXME UNICODE
 504         docstring const cite_str = from_utf8(
 505                 asValidLatexCommand(getCmdName(), cite_engine));
 506
 507         os << "\\" << cite_str;
 508
 509         docstring const & before = getParam("before");
 510         docstring const & after  = getParam("after");
 511         if (!before.empty() && cite_engine != biblio::ENGINE_BASIC)
 512                 os << '[' << before << "][" << after << ']';
 513         else if (!after.empty())
 514                 os << '[' << after << ']';
 515
 516         os << '{' << cleanupWhitespace(getParam("key")) << '}';
 517
 518         return 0;
 519 }
 520
 521
 522 void InsetCitation::validate(LaTeXFeatures & features) const
 523 {
 524         switch (features.bufferParams().getEngine()) {
 525         case biblio::ENGINE_BASIC:
 526                 break;
 527         case biblio::ENGINE_NATBIB_AUTHORYEAR:
 528         case biblio::ENGINE_NATBIB_NUMERICAL:
 529                 features.require("natbib");
 530                 break;
 531         case biblio::ENGINE_JURABIB:
 532                 features.require("jurabib");
 533                 break;
 534         }
 535 }
 536
 537
 538 void InsetCitation::replaceContents(string const & from, string const & to)
 539 {
 540         if (tokenPos(getContents(), ',', from) != -1) {
 541                 vector<string> items = getVectorFromString(getContents());
 542                 std::replace(items.begin(), items.end(), from, to);
 543                 setContents(getStringFromVector(items));
 544         }
 545 }
 546
 547
 548 } // namespace lyx