src/LaTeX.cpp

   1 /**
   2  * \file LaTeX.cpp
   3  * This file is part of LyX, the document processor.
   4  * Licence details can be found in the file COPYING.
   5  *
   6  * \author Alfredo Braunstein
   7  * \author Lars Gullik Bjønnes
   8  * \author Jean-Marc Lasgouttes
   9  * \author Angus Leeming
  10  * \author Dekel Tsur
  11  * \author Jürgen Spitzmüller
  12  *
  13  * Full author contact details are available in file CREDITS.
  14  */
  15
  16 #include <config.h>
  17
  18 #include "BufferList.h"
  19 #include "LaTeX.h"
  20 #include "LyXRC.h"
  21 #include "DepTable.h"
  22
  23 #include "support/debug.h"
  24 #include "support/convert.h"
  25 #include "support/FileName.h"
  26 #include "support/filetools.h"
  27 #include "support/gettext.h"
  28 #include "support/lstrings.h"
  29 #include "support/Systemcall.h"
  30 #include "support/os.h"
  31
  32 #include "support/regex.h"
  33
  34 #include <fstream>
  35 #include <stack>
  36
  37
  38 using namespace std;
  39 using namespace lyx::support;
  40
  41 namespace lyx {
  42
  43 namespace os = support::os;
  44
  45 // TODO: in no particular order
  46 // - get rid of the call to
  47 //   BufferList::updateIncludedTeXfiles, this should either
  48 //   be done before calling LaTeX::funcs or in a completely
  49 //   different way.
  50 // - the makeindex style files should be taken care of with
  51 //   the dependency mechanism.
  52 // - we should perhaps also scan the bibtex log file
  53
  54 namespace {
  55
  56 docstring runMessage(unsigned int count)
  57 {
  58         return bformat(_("Waiting for LaTeX run number %1$d"), count);
  59 }
  60
  61 } // anon namespace
  62
  63 /*
  64  * CLASS TEXERRORS
  65  */
  66
  67 void TeXErrors::insertError(int line, docstring const & error_desc,
  68                             docstring const & error_text,
  69                             string const & child_name)
  70 {
  71         Error newerr(line, error_desc, error_text, child_name);
  72         errors.push_back(newerr);
  73 }
  74
  75
  76 bool operator==(AuxInfo const & a, AuxInfo const & o)
  77 {
  78         return a.aux_file == o.aux_file
  79                 && a.citations == o.citations
  80                 && a.databases == o.databases
  81                 && a.styles == o.styles;
  82 }
  83
  84
  85 bool operator!=(AuxInfo const & a, AuxInfo const & o)
  86 {
  87         return !(a == o);
  88 }
  89
  90
  91 /*
  92  * CLASS LaTeX
  93  */
  94
  95 LaTeX::LaTeX(string const & latex, OutputParams const & rp,
  96              FileName const & f, string const & p)
  97         : cmd(latex), file(f), path(p), runparams(rp)
  98 {
  99         num_errors = 0;
 100         if (prefixIs(cmd, "pdf")) { // Do we use pdflatex ?
 101                 depfile = FileName(file.absFileName() + ".dep-pdf");
 102                 output_file =
 103                         FileName(changeExtension(file.absFileName(), ".pdf"));
 104         } else {
 105                 depfile = FileName(file.absFileName() + ".dep");
 106                 output_file =
 107                         FileName(changeExtension(file.absFileName(), ".dvi"));
 108         }
 109 }
 110
 111
 112 void LaTeX::deleteFilesOnError() const
 113 {
 114         // currently just a dummy function.
 115
 116         // What files do we have to delete?
 117
 118         // This will at least make latex do all the runs
 119         depfile.removeFile();
 120
 121         // but the reason for the error might be in a generated file...
 122
 123         // bibtex file
 124         FileName const bbl(changeExtension(file.absFileName(), ".bbl"));
 125         bbl.removeFile();
 126
 127         // biber file
 128         FileName const bcf(changeExtension(file.absFileName(), ".bcf"));
 129         bcf.removeFile();
 130
 131         // makeindex file
 132         FileName const ind(changeExtension(file.absFileName(), ".ind"));
 133         ind.removeFile();
 134
 135         // nomencl file
 136         FileName const nls(changeExtension(file.absFileName(), ".nls"));
 137         nls.removeFile();
 138
 139         // nomencl file (old version of the package)
 140         FileName const gls(changeExtension(file.absFileName(), ".gls"));
 141         gls.removeFile();
 142
 143         // Also remove the aux file
 144         FileName const aux(changeExtension(file.absFileName(), ".aux"));
 145         aux.removeFile();
 146 }
 147
 148
 149 int LaTeX::run(TeXErrors & terr)
 150         // We know that this function will only be run if the lyx buffer
 151         // has been changed. We also know that a newly written .tex file
 152         // is always different from the previous one because of the date
 153         // in it. However it seems safe to run latex (at least) one time
 154         // each time the .tex file changes.
 155 {
 156         int scanres = NO_ERRORS;
 157         unsigned int count = 0; // number of times run
 158         num_errors = 0; // just to make sure.
 159         unsigned int const MAX_RUN = 6;
 160         DepTable head; // empty head
 161         bool rerun = false; // rerun requested
 162
 163         // The class LaTeX does not know the temp path.
 164         theBufferList().updateIncludedTeXfiles(FileName::getcwd().absFileName(),
 165                 runparams);
 166
 167         // Never write the depfile if an error was encountered.
 168
 169         // 0
 170         // first check if the file dependencies exist:
 171         //     ->If it does exist
 172         //             check if any of the files mentioned in it have
 173         //             changed (done using a checksum).
 174         //                 -> if changed:
 175         //                        run latex once and
 176         //                        remake the dependency file
 177         //                 -> if not changed:
 178         //                        just return there is nothing to do for us.
 179         //     ->if it doesn't exist
 180         //             make it and
 181         //             run latex once (we need to run latex once anyway) and
 182         //             remake the dependency file.
 183         //
 184
 185         bool had_depfile = depfile.exists();
 186         bool run_bibtex = false;
 187         FileName const aux_file(changeExtension(file.absFileName(), "aux"));
 188
 189         if (had_depfile) {
 190                 LYXERR(Debug::DEPEND, "Dependency file exists");
 191                 // Read the dep file:
 192                 had_depfile = head.read(depfile);
 193         }
 194
 195         if (had_depfile) {
 196                 // Update the checksums
 197                 head.update();
 198                 // Can't just check if anything has changed because it might
 199                 // have aborted on error last time... in which cas we need
 200                 // to re-run latex and collect the error messages
 201                 // (even if they are the same).
 202                 if (!output_file.exists()) {
 203                         LYXERR(Debug::DEPEND,
 204                                 "re-running LaTeX because output file doesn't exist.");
 205                 } else if (!head.sumchange()) {
 206                         LYXERR(Debug::DEPEND, "return no_change");
 207                         return NO_CHANGE;
 208                 } else {
 209                         LYXERR(Debug::DEPEND, "Dependency file has changed");
 210                 }
 211
 212                 if (head.extchanged(".bib") || head.extchanged(".bst"))
 213                         run_bibtex = true;
 214         } else
 215                 LYXERR(Debug::DEPEND,
 216                         "Dependency file does not exist, or has wrong format");
 217
 218         /// We scan the aux file even when had_depfile = false,
 219         /// because we can run pdflatex on the file after running latex on it,
 220         /// in which case we will not need to run bibtex again.
 221         vector<AuxInfo> bibtex_info_old;
 222         if (!run_bibtex)
 223                 bibtex_info_old = scanAuxFiles(aux_file);
 224
 225         ++count;
 226         LYXERR(Debug::LATEX, "Run #" << count);
 227         message(runMessage(count));
 228
 229         startscript();
 230         scanres = scanLogFile(terr);
 231         if (scanres & ERROR_RERUN) {
 232                 LYXERR(Debug::LATEX, "Rerunning LaTeX");
 233                 startscript();
 234                 scanres = scanLogFile(terr);
 235         }
 236
 237         if (scanres & ERRORS) {
 238                 deleteFilesOnError();
 239                 return scanres; // return on error
 240         }
 241
 242         vector<AuxInfo> const bibtex_info = scanAuxFiles(aux_file);
 243         if (!run_bibtex && bibtex_info_old != bibtex_info)
 244                 run_bibtex = true;
 245
 246         // update the dependencies.
 247         deplog(head); // reads the latex log
 248         head.update();
 249
 250         // 1
 251         // At this point we must run external programs if needed.
 252         // makeindex will be run if a .idx file changed or was generated.
 253         // And if there were undefined citations or changes in references
 254         // the .aux file is checked for signs of bibtex. Bibtex is then run
 255         // if needed.
 256
 257         // memoir (at least) writes an empty *idx file in the first place.
 258         // A second latex run is needed.
 259         FileName const idxfile(changeExtension(file.absFileName(), ".idx"));
 260         rerun = idxfile.exists() && idxfile.isFileEmpty();
 261
 262         // run makeindex
 263         if (head.haschanged(idxfile)) {
 264                 // no checks for now
 265                 LYXERR(Debug::LATEX, "Running MakeIndex.");
 266                 message(_("Running Index Processor."));
 267                 // onlyFileName() is needed for cygwin
 268                 rerun |= runMakeIndex(onlyFileName(idxfile.absFileName()),
 269                                 runparams);
 270         }
 271         FileName const nlofile(changeExtension(file.absFileName(), ".nlo"));
 272         if (head.haschanged(nlofile))
 273                 rerun |= runMakeIndexNomencl(file, ".nlo", ".nls");
 274         FileName const glofile(changeExtension(file.absFileName(), ".glo"));
 275         if (head.haschanged(glofile))
 276                 rerun |= runMakeIndexNomencl(file, ".glo", ".gls");
 277
 278         // check if we're using biber instead of bibtex
 279         // biber writes no info to the aux file, so we just check
 280         // if a bcf file exists (and if it was updated)
 281         FileName const bcffile(changeExtension(file.absFileName(), ".bcf"));
 282         bool const biber = head.exist(bcffile);
 283
 284         // run bibtex
 285         // if (scanres & UNDEF_CIT || scanres & RERUN || run_bibtex)
 286         if (scanres & UNDEF_CIT || run_bibtex) {
 287                 // Here we must scan the .aux file and look for
 288                 // "\bibdata" and/or "\bibstyle". If one of those
 289                 // tags is found -> run bibtex and set rerun = true;
 290                 // no checks for now
 291                 LYXERR(Debug::LATEX, "Running BibTeX.");
 292                 message(_("Running BibTeX."));
 293                 updateBibtexDependencies(head, bibtex_info, biber);
 294                 rerun |= runBibTeX(bibtex_info, runparams, biber);
 295                 if (biber) {
 296                         // since biber writes no info to the aux file, we have
 297                         // to parse the blg file (which only exists after biber
 298                         // was first issued)
 299                         FileName const blgfile(changeExtension(file.absFileName(), ".blg"));
 300                         if (blgfile.exists())
 301                                 scanBlgFile(head);
 302                 }
 303         } else if (!had_depfile) {
 304                 /// If we run pdflatex on the file after running latex on it,
 305                 /// then we do not need to run bibtex, but we do need to
 306                 /// insert the .bib and .bst files into the .dep-pdf file.
 307                 updateBibtexDependencies(head, bibtex_info, biber);
 308         }
 309
 310         // 2
 311         // we know on this point that latex has been run once (or we just
 312         // returned) and the question now is to decide if we need to run
 313         // it any more. This is done by asking if any of the files in the
 314         // dependency file has changed. (remember that the checksum for
 315         // a given file is reported to have changed if it just was created)
 316         //     -> if changed or rerun == true:
 317         //             run latex once more and
 318         //             update the dependency structure
 319         //     -> if not changed:
 320         //             we do nothing at this point
 321         //
 322         if (rerun || head.sumchange()) {
 323                 rerun = false;
 324                 ++count;
 325                 LYXERR(Debug::DEPEND, "Dep. file has changed or rerun requested");
 326                 LYXERR(Debug::LATEX, "Run #" << count);
 327                 message(runMessage(count));
 328                 startscript();
 329                 scanres = scanLogFile(terr);
 330                 if (scanres & ERRORS) {
 331                         deleteFilesOnError();
 332                         return scanres; // return on error
 333                 }
 334
 335                 // update the depedencies
 336                 deplog(head); // reads the latex log
 337                 head.update();
 338         } else {
 339                 LYXERR(Debug::DEPEND, "Dep. file has NOT changed");
 340         }
 341
 342         // 3
 343         // rerun bibtex?
 344         // Complex bibliography packages such as Biblatex require
 345         // an additional bibtex cycle sometimes.
 346         if (scanres & UNDEF_CIT) {
 347                 // Here we must scan the .aux file and look for
 348                 // "\bibdata" and/or "\bibstyle". If one of those
 349                 // tags is found -> run bibtex and set rerun = true;
 350                 // no checks for now
 351                 LYXERR(Debug::LATEX, "Running BibTeX.");
 352                 message(_("Running BibTeX."));
 353                 updateBibtexDependencies(head, bibtex_info, biber);
 354                 rerun |= runBibTeX(bibtex_info, runparams, biber);
 355         }
 356
 357         // 4
 358         // The inclusion of files generated by external programs such as
 359         // makeindex or bibtex might have done changes to pagenumbering,
 360         // etc. And because of this we must run the external programs
 361         // again to make sure everything is redone correctly.
 362         // Also there should be no need to run the external programs any
 363         // more after this.
 364
 365         // run makeindex if the <file>.idx has changed or was generated.
 366         if (head.haschanged(idxfile)) {
 367                 // no checks for now
 368                 LYXERR(Debug::LATEX, "Running MakeIndex.");
 369                 message(_("Running Index Processor."));
 370                 // onlyFileName() is needed for cygwin
 371                 rerun = runMakeIndex(onlyFileName(changeExtension(
 372                                 file.absFileName(), ".idx")), runparams);
 373         }
 374
 375         // I am not pretty sure if need this twice.
 376         if (head.haschanged(nlofile))
 377                 rerun |= runMakeIndexNomencl(file, ".nlo", ".nls");
 378         if (head.haschanged(glofile))
 379                 rerun |= runMakeIndexNomencl(file, ".glo", ".gls");
 380
 381         // 5
 382         // we will only run latex more if the log file asks for it.
 383         // or if the sumchange() is true.
 384         //     -> rerun asked for:
 385         //             run latex and
 386         //             remake the dependency file
 387         //             goto 2 or return if max runs are reached.
 388         //     -> rerun not asked for:
 389         //             just return (fall out of bottom of func)
 390         //
 391         while ((head.sumchange() || rerun || (scanres & RERUN))
 392                && count < MAX_RUN) {
 393                 // Yes rerun until message goes away, or until
 394                 // MAX_RUNS are reached.
 395                 rerun = false;
 396                 ++count;
 397                 LYXERR(Debug::LATEX, "Run #" << count);
 398                 message(runMessage(count));
 399                 startscript();
 400                 scanres = scanLogFile(terr);
 401                 if (scanres & ERRORS) {
 402                         deleteFilesOnError();
 403                         return scanres; // return on error
 404                 }
 405
 406                 // keep this updated
 407                 head.update();
 408         }
 409
 410         // Write the dependencies to file.
 411         head.write(depfile);
 412         LYXERR(Debug::LATEX, "Done.");
 413         return scanres;
 414 }
 415
 416
 417 int LaTeX::startscript()
 418 {
 419         // onlyFileName() is needed for cygwin
 420         string tmp = cmd + ' '
 421                      + quoteName(onlyFileName(file.toFilesystemEncoding()))
 422                      + " > " + os::nulldev();
 423         Systemcall one;
 424         return one.startscript(Systemcall::Wait, tmp, path);
 425 }
 426
 427
 428 bool LaTeX::runMakeIndex(string const & f, OutputParams const & runparams,
 429                          string const & params)
 430 {
 431         string tmp = runparams.use_japanese ?
 432                 lyxrc.jindex_command : lyxrc.index_command;
 433
 434         if (!runparams.index_command.empty())
 435                 tmp = runparams.index_command;
 436
 437         LYXERR(Debug::LATEX,
 438                 "idx file has been made, running index processor ("
 439                 << tmp << ") on file " << f);
 440
 441         tmp = subst(tmp, "$$lang", runparams.document_language);
 442         if (runparams.use_indices) {
 443                 tmp = lyxrc.splitindex_command + " -m " + quoteName(tmp);
 444                 LYXERR(Debug::LATEX,
 445                 "Multiple indices. Using splitindex command: " << tmp);
 446         }
 447         tmp += ' ';
 448         tmp += quoteName(f);
 449         tmp += params;
 450         Systemcall one;
 451         one.startscript(Systemcall::Wait, tmp, path);
 452         return true;
 453 }
 454
 455
 456 bool LaTeX::runMakeIndexNomencl(FileName const & file,
 457                 string const & nlo, string const & nls)
 458 {
 459         LYXERR(Debug::LATEX, "Running MakeIndex for nomencl.");
 460         message(_("Running MakeIndex for nomencl."));
 461         string tmp = lyxrc.nomencl_command + ' ';
 462         // onlyFileName() is needed for cygwin
 463         tmp += quoteName(onlyFileName(changeExtension(file.absFileName(), nlo)));
 464         tmp += " -o "
 465                 + onlyFileName(changeExtension(file.toFilesystemEncoding(), nls));
 466         Systemcall one;
 467         one.startscript(Systemcall::Wait, tmp, path);
 468         return true;
 469 }
 470
 471
 472 vector<AuxInfo> const
 473 LaTeX::scanAuxFiles(FileName const & file)
 474 {
 475         vector<AuxInfo> result;
 476
 477         result.push_back(scanAuxFile(file));
 478
 479         string const basename = removeExtension(file.absFileName());
 480         for (int i = 1; i < 1000; ++i) {
 481                 FileName const file2(basename
 482                         + '.' + convert<string>(i)
 483                         + ".aux");
 484                 if (!file2.exists())
 485                         break;
 486                 result.push_back(scanAuxFile(file2));
 487         }
 488         return result;
 489 }
 490
 491
 492 AuxInfo const LaTeX::scanAuxFile(FileName const & file)
 493 {
 494         AuxInfo result;
 495         result.aux_file = file;
 496         scanAuxFile(file, result);
 497         return result;
 498 }
 499
 500
 501 void LaTeX::scanAuxFile(FileName const & file, AuxInfo & aux_info)
 502 {
 503         LYXERR(Debug::LATEX, "Scanning aux file: " << file);
 504
 505         ifstream ifs(file.toFilesystemEncoding().c_str());
 506         string token;
 507         static regex const reg1("\\\\citation\\{([^}]+)\\}");
 508         static regex const reg2("\\\\bibdata\\{([^}]+)\\}");
 509         static regex const reg3("\\\\bibstyle\\{([^}]+)\\}");
 510         static regex const reg4("\\\\@input\\{([^}]+)\\}");
 511
 512         while (getline(ifs, token)) {
 513                 token = rtrim(token, "\r");
 514                 smatch sub;
 515                 // FIXME UNICODE: We assume that citation keys and filenames
 516                 // in the aux file are in the file system encoding.
 517                 token = to_utf8(from_filesystem8bit(token));
 518                 if (regex_match(token, sub, reg1)) {
 519                         string data = sub.str(1);
 520                         while (!data.empty()) {
 521                                 string citation;
 522                                 data = split(data, citation, ',');
 523                                 LYXERR(Debug::LATEX, "Citation: " << citation);
 524                                 aux_info.citations.insert(citation);
 525                         }
 526                 } else if (regex_match(token, sub, reg2)) {
 527                         string data = sub.str(1);
 528                         // data is now all the bib files separated by ','
 529                         // get them one by one and pass them to the helper
 530                         while (!data.empty()) {
 531                                 string database;
 532                                 data = split(data, database, ',');
 533                                 database = changeExtension(database, "bib");
 534                                 LYXERR(Debug::LATEX, "BibTeX database: `" << database << '\'');
 535                                 aux_info.databases.insert(database);
 536                         }
 537                 } else if (regex_match(token, sub, reg3)) {
 538                         string style = sub.str(1);
 539                         // token is now the style file
 540                         // pass it to the helper
 541                         style = changeExtension(style, "bst");
 542                         LYXERR(Debug::LATEX, "BibTeX style: `" << style << '\'');
 543                         aux_info.styles.insert(style);
 544                 } else if (regex_match(token, sub, reg4)) {
 545                         string const file2 = sub.str(1);
 546                         scanAuxFile(makeAbsPath(file2), aux_info);
 547                 }
 548         }
 549 }
 550
 551
 552 void LaTeX::updateBibtexDependencies(DepTable & dep,
 553                                      vector<AuxInfo> const & bibtex_info,
 554                                      bool biber)
 555 {
 556         // Since a run of Bibtex mandates more latex runs it is ok to
 557         // remove all ".bib" and ".bst" files.
 558         dep.remove_files_with_extension(".bib");
 559         dep.remove_files_with_extension(".bst");
 560         //string aux = OnlyFileName(ChangeExtension(file, ".aux"));
 561
 562         for (vector<AuxInfo>::const_iterator it = bibtex_info.begin();
 563              it != bibtex_info.end(); ++it) {
 564                 for (set<string>::const_iterator it2 = it->databases.begin();
 565                      it2 != it->databases.end(); ++it2) {
 566                         FileName const file = findtexfile(*it2, "bib");
 567                         if (!file.empty())
 568                                 dep.insert(file, true);
 569                 }
 570
 571                 for (set<string>::const_iterator it2 = it->styles.begin();
 572                      it2 != it->styles.end(); ++it2) {
 573                         FileName const file = findtexfile(*it2, "bst");
 574                         if (!file.empty())
 575                                 dep.insert(file, true);
 576                 }
 577         }
 578
 579         // biber writes nothing into the aux file.
 580         // Instead, we have to scan the blg file
 581         if (biber) {
 582                 scanBlgFile(dep);
 583         }
 584 }
 585
 586
 587 bool LaTeX::runBibTeX(vector<AuxInfo> const & bibtex_info,
 588                       OutputParams const & runparams, bool biber)
 589 {
 590         bool result = false;
 591         for (vector<AuxInfo>::const_iterator it = bibtex_info.begin();
 592              it != bibtex_info.end(); ++it) {
 593                 if (!biber && it->databases.empty())
 594                         continue;
 595                 result = true;
 596
 597                 string tmp = runparams.use_japanese ?
 598                         lyxrc.jbibtex_command : lyxrc.bibtex_command;
 599
 600                 if (!runparams.bibtex_command.empty())
 601                         tmp = runparams.bibtex_command;
 602                 tmp += " ";
 603                 // onlyFileName() is needed for cygwin
 604                 tmp += quoteName(onlyFileName(removeExtension(
 605                                 it->aux_file.absFileName())));
 606                 Systemcall one;
 607                 one.startscript(Systemcall::Wait, tmp, path);
 608         }
 609         // Return whether bibtex was run
 610         return result;
 611 }
 612
 613
 614 int LaTeX::scanLogFile(TeXErrors & terr)
 615 {
 616         int last_line = -1;
 617         int line_count = 1;
 618         int retval = NO_ERRORS;
 619         string tmp =
 620                 onlyFileName(changeExtension(file.absFileName(), ".log"));
 621         LYXERR(Debug::LATEX, "Log file: " << tmp);
 622         FileName const fn = FileName(makeAbsPath(tmp));
 623         ifstream ifs(fn.toFilesystemEncoding().c_str());
 624         bool fle_style = false;
 625         static regex file_line_error(".+\\.\\D+:[0-9]+: (.+)");
 626         static regex child_file(".*([0-9]+[A-Za-z]*_.+\\.tex).*");
 627         // Flag for 'File ended while scanning' message.
 628         // We need to wait for subsequent processing.
 629         string wait_for_error;
 630         string child_name;
 631         int pnest = 0;
 632         stack <pair<string, int> > child;
 633
 634         string token;
 635         while (getline(ifs, token)) {
 636                 // MikTeX sometimes inserts \0 in the log file. They can't be
 637                 // removed directly with the existing string utility
 638                 // functions, so convert them first to \r, and remove all
 639                 // \r's afterwards, since we need to remove them anyway.
 640                 token = subst(token, '\0', '\r');
 641                 token = subst(token, "\r", "");
 642                 smatch sub;
 643
 644                 LYXERR(Debug::LATEX, "Log line: " << token);
 645
 646                 if (token.empty())
 647                         continue;
 648
 649                 // Track child documents
 650                 for (size_t i = 0; i < token.length(); ++i) {
 651                         if (token[i] == '(') {
 652                                 ++pnest;
 653                                 size_t j = token.find('(', i + 1);
 654                                 size_t len = j == string::npos
 655                                                 ? token.substr(i + 1).length()
 656                                                 : j - i - 1;
 657                                 if (regex_match(token.substr(i + 1, len),
 658                                                         sub, child_file)) {
 659                                         string const name = sub.str(1);
 660                                         child.push(make_pair(name, pnest));
 661                                         i += len;
 662                                 }
 663                         } else if (token[i] == ')') {
 664                                 if (!child.empty()
 665                                     && child.top().second == pnest)
 666                                         child.pop();
 667                                 --pnest;
 668                         }
 669                 }
 670                 child_name = child.empty() ? empty_string() : child.top().first;
 671
 672                 if (contains(token, "file:line:error style messages enabled"))
 673                         fle_style = true;
 674
 675                 if (prefixIs(token, "LaTeX Warning:") ||
 676                     prefixIs(token, "! pdfTeX warning")) {
 677                         // Here shall we handle different
 678                         // types of warnings
 679                         retval |= LATEX_WARNING;
 680                         LYXERR(Debug::LATEX, "LaTeX Warning.");
 681                         if (contains(token, "Rerun to get cross-references")) {
 682                                 retval |= RERUN;
 683                                 LYXERR(Debug::LATEX, "We should rerun.");
 684                         // package clefval needs 2 latex runs before bibtex
 685                         } else if (contains(token, "Value of")
 686                                    && contains(token, "on page")
 687                                    && contains(token, "undefined")) {
 688                                 retval |= ERROR_RERUN;
 689                                 LYXERR(Debug::LATEX, "Force rerun.");
 690                         // package etaremune
 691                         } else if (contains(token, "Etaremune labels have changed")) {
 692                                 retval |= ERROR_RERUN;
 693                                 LYXERR(Debug::LATEX, "Force rerun.");
 694                         } else if (contains(token, "Citation")
 695                                    && contains(token, "on page")
 696                                    && contains(token, "undefined")) {
 697                                 retval |= UNDEF_CIT;
 698                         }
 699                 } else if (prefixIs(token, "Package")) {
 700                         // Package warnings
 701                         retval |= PACKAGE_WARNING;
 702                         if (contains(token, "natbib Warning:")) {
 703                                 // Natbib warnings
 704                                 if (contains(token, "Citation")
 705                                     && contains(token, "on page")
 706                                     && contains(token, "undefined")) {
 707                                         retval |= UNDEF_CIT;
 708                                 }
 709                         } else if (contains(token, "run BibTeX")) {
 710                                 retval |= UNDEF_CIT;
 711                         } else if (contains(token, "Rerun LaTeX") ||
 712                                    contains(token, "Please rerun LaTeX") ||
 713                                    contains(token, "Rerun to get")) {
 714                                 // at least longtable.sty and bibtopic.sty
 715                                 // might use this.
 716                                 LYXERR(Debug::LATEX, "We should rerun.");
 717                                 retval |= RERUN;
 718                         }
 719                 } else if (prefixIs(token, "LETTRE WARNING:")) {
 720                         if (contains(token, "veuillez recompiler")) {
 721                                 // lettre.cls
 722                                 LYXERR(Debug::LATEX, "We should rerun.");
 723                                 retval |= RERUN;
 724                         }
 725                 } else if (token[0] == '(') {
 726                         if (contains(token, "Rerun LaTeX") ||
 727                             contains(token, "Rerun to get")) {
 728                                 // Used by natbib
 729                                 LYXERR(Debug::LATEX, "We should rerun.");
 730                                 retval |= RERUN;
 731                         }
 732                 } else if (prefixIs(token, "! ")
 733                             || (fle_style
 734                                 && regex_match(token, sub, file_line_error)
 735                                 && !contains(token, "pdfTeX warning"))) {
 736                            // Ok, we have something that looks like a TeX Error
 737                            // but what do we really have.
 738
 739                         // Just get the error description:
 740                         string desc;
 741                         if (prefixIs(token, "! "))
 742                                 desc = string(token, 2);
 743                         else if (fle_style)
 744                                 desc = sub.str();
 745                         if (contains(token, "LaTeX Error:"))
 746                                 retval |= LATEX_ERROR;
 747
 748                         if (prefixIs(token, "! File ended while scanning")){
 749                                 if (prefixIs(token, "! File ended while scanning use of \\Hy@setref@link.")){
 750                                         // bug 7344. We must rerun LaTeX if hyperref has been toggled.
 751                                         retval |= ERROR_RERUN;
 752                                         LYXERR(Debug::LATEX, "Force rerun.");
 753                                 } else {
 754                                         // bug 6445. At this point its not clear we finish with error.
 755                                         wait_for_error = desc;
 756                                         continue;
 757                                 }
 758                         }
 759
 760                         if (prefixIs(token, "! Paragraph ended before \\Hy@setref@link was complete.")){
 761                                         // bug 7344. We must rerun LaTeX if hyperref has been toggled.
 762                                         retval |= ERROR_RERUN;
 763                                         LYXERR(Debug::LATEX, "Force rerun.");
 764                         }
 765
 766                         if (!wait_for_error.empty() && prefixIs(token, "! Emergency stop.")){
 767                                 retval |= LATEX_ERROR;
 768                                 string errstr;
 769                                 int count = 0;
 770                                 errstr = wait_for_error;
 771                                 do {
 772                                         if (!getline(ifs, tmp))
 773                                                 break;
 774                                         tmp = rtrim(tmp, "\r");
 775                                         errstr += "\n" + tmp;
 776                                         if (++count > 5)
 777                                                 break;
 778                                 } while (!contains(tmp, "(job aborted"));
 779
 780                                 terr.insertError(0,
 781                                                  from_local8bit("Emergency stop"),
 782                                                  from_local8bit(errstr),
 783                                                  child_name);
 784                         }
 785
 786                         // get the next line
 787                         string tmp;
 788                         int count = 0;
 789                         do {
 790                                 if (!getline(ifs, tmp))
 791                                         break;
 792                                 tmp = rtrim(tmp, "\r");
 793                                 if (++count > 10)
 794                                         break;
 795                         } while (!prefixIs(tmp, "l."));
 796                         if (prefixIs(tmp, "l.")) {
 797                                 // we have a latex error
 798                                 retval |=  TEX_ERROR;
 799                                 if (contains(desc,
 800                                     "Package babel Error: You haven't defined the language") ||
 801                                     contains(desc,
 802                                     "Package babel Error: You haven't loaded the option"))
 803                                         retval |= ERROR_RERUN;
 804                                 // get the line number:
 805                                 int line = 0;
 806                                 sscanf(tmp.c_str(), "l.%d", &line);
 807                                 // get the rest of the message:
 808                                 string errstr(tmp, tmp.find(' '));
 809                                 errstr += '\n';
 810                                 getline(ifs, tmp);
 811                                 tmp = rtrim(tmp, "\r");
 812                                 while (!contains(errstr, "l.")
 813                                        && !tmp.empty()
 814                                        && !prefixIs(tmp, "! ")
 815                                        && !contains(tmp, "(job aborted")) {
 816                                         errstr += tmp;
 817                                         errstr += "\n";
 818                                         getline(ifs, tmp);
 819                                         tmp = rtrim(tmp, "\r");
 820                                 }
 821                                 LYXERR(Debug::LATEX, "line: " << line << '\n'
 822                                         << "Desc: " << desc << '\n' << "Text: " << errstr);
 823                                 if (line == last_line)
 824                                         ++line_count;
 825                                 else {
 826                                         line_count = 1;
 827                                         last_line = line;
 828                                 }
 829                                 if (line_count <= 5) {
 830                                         // FIXME UNICODE
 831                                         // We have no idea what the encoding of
 832                                         // the log file is.
 833                                         // It seems that the output from the
 834                                         // latex compiler itself is pure ASCII,
 835                                         // but it can include bits from the
 836                                         // document, so whatever encoding we
 837                                         // assume here it can be wrong.
 838                                         terr.insertError(line,
 839                                                          from_local8bit(desc),
 840                                                          from_local8bit(errstr),
 841                                                          child_name);
 842                                         ++num_errors;
 843                                 }
 844                         }
 845                 } else {
 846                         // information messages, TeX warnings and other
 847                         // warnings we have not caught earlier.
 848                         if (prefixIs(token, "Overfull ")) {
 849                                 retval |= TEX_WARNING;
 850                         } else if (prefixIs(token, "Underfull ")) {
 851                                 retval |= TEX_WARNING;
 852                         } else if (contains(token, "Rerun to get citations")) {
 853                                 // Natbib seems to use this.
 854                                 retval |= UNDEF_CIT;
 855                         } else if (contains(token, "No pages of output")) {
 856                                 // A dvi file was not created
 857                                 retval |= NO_OUTPUT;
 858                         } else if (contains(token, "That makes 100 errors")) {
 859                                 // More than 100 errors were reprted
 860                                 retval |= TOO_MANY_ERRORS;
 861                         } else if (prefixIs(token, "!pdfTeX error:")){
 862                                 // otherwise we dont catch e.g.:
 863                                 // !pdfTeX error: pdflatex (file feyn10): Font feyn10 at 600 not found
 864                                 retval |= ERRORS;
 865                                         terr.insertError(0,
 866                                                          from_local8bit("pdfTeX Error"),
 867                                                          from_local8bit(token),
 868                                                          child_name);
 869                         }
 870                 }
 871         }
 872         LYXERR(Debug::LATEX, "Log line: " << token);
 873         return retval;
 874 }
 875
 876
 877 namespace {
 878
 879 bool insertIfExists(FileName const & absname, DepTable & head)
 880 {
 881         if (absname.exists() && !absname.isDirectory()) {
 882                 head.insert(absname, true);
 883                 return true;
 884         }
 885         return false;
 886 }
 887
 888
 889 bool handleFoundFile(string const & ff, DepTable & head)
 890 {
 891         // convert from native os path to unix path
 892         string foundfile = os::internal_path(trim(ff));
 893
 894         LYXERR(Debug::DEPEND, "Found file: " << foundfile);
 895
 896         // Ok now we found a file.
 897         // Now we should make sure that this is a file that we can
 898         // access through the normal paths.
 899         // We will not try any fancy search methods to
 900         // find the file.
 901
 902         // (1) foundfile is an
 903         //     absolute path and should
 904         //     be inserted.
 905         FileName absname;
 906         if (FileName::isAbsolute(foundfile)) {
 907                 LYXERR(Debug::DEPEND, "AbsolutePath file: " << foundfile);
 908                 // On initial insert we want to do the update at once
 909                 // since this file cannot be a file generated by
 910                 // the latex run.
 911                 absname.set(foundfile);
 912                 if (!insertIfExists(absname, head)) {
 913                         // check for spaces
 914                         string strippedfile = foundfile;
 915                         while (contains(strippedfile, " ")) {
 916                                 // files with spaces are often enclosed in quotation
 917                                 // marks; those have to be removed
 918                                 string unquoted = subst(strippedfile, "\"", "");
 919                                 absname.set(unquoted);
 920                                 if (insertIfExists(absname, head))
 921                                         return true;
 922                                 // strip off part after last space and try again
 923                                 string tmp = strippedfile;
 924                                 string const stripoff =
 925                                         rsplit(tmp, strippedfile, ' ');
 926                                 absname.set(strippedfile);
 927                                 if (insertIfExists(absname, head))
 928                                         return true;
 929                         }
 930                 }
 931         }
 932
 933         string onlyfile = onlyFileName(foundfile);
 934         absname = makeAbsPath(onlyfile);
 935
 936         // check for spaces
 937         while (contains(foundfile, ' ')) {
 938                 if (absname.exists())
 939                         // everything o.k.
 940                         break;
 941                 else {
 942                         // files with spaces are often enclosed in quotation
 943                         // marks; those have to be removed
 944                         string unquoted = subst(foundfile, "\"", "");
 945                         absname = makeAbsPath(unquoted);
 946                         if (absname.exists())
 947                                 break;
 948                         // strip off part after last space and try again
 949                         string strippedfile;
 950                         string const stripoff =
 951                                 rsplit(foundfile, strippedfile, ' ');
 952                         foundfile = strippedfile;
 953                         onlyfile = onlyFileName(strippedfile);
 954                         absname = makeAbsPath(onlyfile);
 955                 }
 956         }
 957
 958         // (2) foundfile is in the tmpdir
 959         //     insert it into head
 960         if (absname.exists() && !absname.isDirectory()) {
 961                 // FIXME: This regex contained glo, but glo is used by the old
 962                 // version of nomencl.sty. Do we need to put it back?
 963                 static regex const unwanted("^.*\\.(aux|log|dvi|bbl|ind)$");
 964                 if (regex_match(onlyfile, unwanted)) {
 965                         LYXERR(Debug::DEPEND, "We don't want " << onlyfile
 966                                 << " in the dep file");
 967                 } else if (suffixIs(onlyfile, ".tex")) {
 968                         // This is a tex file generated by LyX
 969                         // and latex is not likely to change this
 970                         // during its runs.
 971                         LYXERR(Debug::DEPEND, "Tmpdir TeX file: " << onlyfile);
 972                         head.insert(absname, true);
 973                 } else {
 974                         LYXERR(Debug::DEPEND, "In tmpdir file:" << onlyfile);
 975                         head.insert(absname);
 976                 }
 977                 return true;
 978         } else {
 979                 LYXERR(Debug::DEPEND, "Not a file or we are unable to find it.");
 980                 return false;
 981         }
 982 }
 983
 984
 985 bool checkLineBreak(string const & ff, DepTable & head)
 986 {
 987         if (!contains(ff, '.'))
 988                 return false;
 989
 990         // if we have a dot, we let handleFoundFile decide
 991         return handleFoundFile(ff, head);
 992 }
 993
 994 } // anon namespace
 995
 996
 997 void LaTeX::deplog(DepTable & head)
 998 {
 999         // This function reads the LaTeX log file end extracts all the
1000         // external files used by the LaTeX run. The files are then
1001         // entered into the dependency file.
1002
1003         string const logfile =
1004                 onlyFileName(changeExtension(file.absFileName(), ".log"));
1005
1006         static regex const reg1("File: (.+).*");
1007         static regex const reg2("No file (.+)(.).*");
1008         static regex const reg3("\\\\openout[0-9]+.*=.*`(.+)(..).*");
1009         // If an index should be created, MikTex does not write a line like
1010         //    \openout# = 'sample.idx'.
1011         // but instead only a line like this into the log:
1012         //   Writing index file sample.idx
1013         static regex const reg4("Writing index file (.+).*");
1014         // files also can be enclosed in <...>
1015         static regex const reg5("<([^>]+)(.).*");
1016         static regex const regoldnomencl("Writing glossary file (.+).*");
1017         static regex const regnomencl("Writing nomenclature file (.+).*");
1018         // If a toc should be created, MikTex does not write a line like
1019         //    \openout# = `sample.toc'.
1020         // but only a line like this into the log:
1021         //    \tf@toc=\write#
1022         // This line is also written by tetex.
1023         // This line is not present if no toc should be created.
1024         static regex const miktexTocReg("\\\\tf@toc=\\\\write.*");
1025         static regex const reg6(".*\\([^)]+.*");
1026
1027         FileName const fn = makeAbsPath(logfile);
1028         ifstream ifs(fn.toFilesystemEncoding().c_str());
1029         string lastline;
1030         while (ifs) {
1031                 // Ok, the scanning of files here is not sufficient.
1032                 // Sometimes files are named by "File:� xxx" only
1033                 // So I think we should use some regexps to find files instead.
1034                 // Note: all file names and paths might contains spaces.
1035                 bool found_file = false;
1036                 string token;
1037                 getline(ifs, token);
1038                 // MikTeX sometimes inserts \0 in the log file. They can't be
1039                 // removed directly with the existing string utility
1040                 // functions, so convert them first to \r, and remove all
1041                 // \r's afterwards, since we need to remove them anyway.
1042                 token = subst(token, '\0', '\r');
1043                 token = subst(token, "\r", "");
1044                 if (token.empty() || token == ")") {
1045                         lastline = string();
1046                         continue;
1047                 }
1048
1049                 // FIXME UNICODE: We assume that the file names in the log
1050                 // file are in the file system encoding.
1051                 token = to_utf8(from_filesystem8bit(token));
1052
1053                 // Sometimes, filenames are broken across lines.
1054                 // We care for that and save suspicious lines.
1055                 // Here we exclude some cases where we are sure
1056                 // that there is no continued filename
1057                 if (!lastline.empty()) {
1058                         static regex const package_info("Package \\w+ Info: .*");
1059                         static regex const package_warning("Package \\w+ Warning: .*");
1060                         if (prefixIs(token, "File:") || prefixIs(token, "(Font)")
1061                             || prefixIs(token, "Package:")
1062                             || prefixIs(token, "Language:")
1063                             || prefixIs(token, "LaTeX Info:")
1064                             || prefixIs(token, "LaTeX Font Info:")
1065                             || prefixIs(token, "\\openout[")
1066                             || prefixIs(token, "))")
1067                             || regex_match(token, package_info)
1068                             || regex_match(token, package_warning))
1069                                 lastline = string();
1070                 }
1071
1072                 if (!lastline.empty())
1073                         // probably a continued filename from last line
1074                         token = lastline + token;
1075                 if (token.length() > 255) {
1076                         // string too long. Cut off.
1077                         token.erase(0, token.length() - 251);
1078                 }
1079
1080                 smatch sub;
1081
1082                 // (1) "File: file.ext"
1083                 if (regex_match(token, sub, reg1)) {
1084                         // check for dot
1085                         found_file = checkLineBreak(sub.str(1), head);
1086                         // However, ...
1087                         if (suffixIs(token, ")"))
1088                                 // no line break for sure
1089                                 // pretend we've been successfully searching
1090                                 found_file = true;
1091                 // (2) "No file file.ext"
1092                 } else if (regex_match(token, sub, reg2)) {
1093                         // file names must contains a dot, line ends with dot
1094                         if (contains(sub.str(1), '.') && sub.str(2) == ".")
1095                                 found_file = handleFoundFile(sub.str(1), head);
1096                         else
1097                                 // we suspect a line break
1098                                 found_file = false;
1099                 // (3) "\openout<nr> = `file.ext'."
1100                 } else if (regex_match(token, sub, reg3)) {
1101                         // search for closing '. at the end of the line
1102                         if (sub.str(2) == "\'.")
1103                                 found_file = handleFoundFile(sub.str(1), head);
1104                         else
1105                                 // probable line break
1106                                 found_file = false;
1107                 // (4) "Writing index file file.ext"
1108                 } else if (regex_match(token, sub, reg4))
1109                         // check for dot
1110                         found_file = checkLineBreak(sub.str(1), head);
1111                 // (5) "<file.ext>"
1112                 else if (regex_match(token, sub, reg5)) {
1113                         // search for closing '>' and dot ('*.*>') at the eol
1114                         if (contains(sub.str(1), '.') && sub.str(2) == ">")
1115                                 found_file = handleFoundFile(sub.str(1), head);
1116                         else
1117                                 // probable line break
1118                                 found_file = false;
1119                 // (6) "Writing nomenclature file file.ext"
1120                 } else if (regex_match(token, sub, regnomencl) ||
1121                            regex_match(token, sub, regoldnomencl))
1122                         // check for dot
1123                         found_file = checkLineBreak(sub.str(1), head);
1124                 // (7) "\tf@toc=\write<nr>" (for MikTeX)
1125                 else if (regex_match(token, sub, miktexTocReg))
1126                         found_file = handleFoundFile(onlyFileName(changeExtension(
1127                                                 file.absFileName(), ".toc")), head);
1128                 else
1129                         // not found, but we won't check further
1130                         // pretend we've been successfully searching
1131                         found_file = true;
1132
1133                 // (8) "(file.ext"
1134                 // note that we can have several of these on one line
1135                 // this must be queried separated, because of
1136                 // cases such as "File: file.ext (type eps)"
1137                 // where "File: file.ext" would be skipped
1138                 if (regex_match(token, sub, reg6)) {
1139                         // search for strings in (...)
1140                         static regex reg6_1("\\(([^()]+)(.)");
1141                         smatch what;
1142                         string::const_iterator first = token.begin();
1143                         string::const_iterator end = token.end();
1144
1145                         while (regex_search(first, end, what, reg6_1)) {
1146                                 // if we have a dot, try to handle as file
1147                                 if (contains(what.str(1), '.')) {
1148                                         first = what[0].second;
1149                                         if (what.str(2) == ")") {
1150                                                 handleFoundFile(what.str(1), head);
1151                                                 // since we had a closing bracket,
1152                                                 // do not investigate further
1153                                                 found_file = true;
1154                                         } else
1155                                                 // if we have no closing bracket,
1156                                                 // try to handle as file nevertheless
1157                                                 found_file = handleFoundFile(
1158                                                         what.str(1) + what.str(2), head);
1159                                 }
1160                                 // if we do not have a dot, check if the line has
1161                                 // a closing bracket (else, we suspect a line break)
1162                                 else if (what.str(2) != ")") {
1163                                         first = what[0].second;
1164                                         found_file = false;
1165                                 } else {
1166                                         // we have a closing bracket, so the content
1167                                         // is not a file name.
1168                                         // no need to investigate further
1169                                         // pretend we've been successfully searching
1170                                         first = what[0].second;
1171                                         found_file = true;
1172                                 }
1173                         }
1174                 }
1175
1176                 if (!found_file)
1177                         // probable linebreak:
1178                         // save this line
1179                         lastline = token;
1180                 else
1181                         // no linebreak: reset
1182                         lastline = string();
1183         }
1184
1185         // Make sure that the main .tex file is in the dependency file.
1186         head.insert(file, true);
1187 }
1188
1189
1190 void LaTeX::scanBlgFile(DepTable & dep)
1191 {
1192         FileName const blg_file(changeExtension(file.absFileName(), "blg"));
1193         LYXERR(Debug::LATEX, "Scanning blg file: " << blg_file);
1194
1195         ifstream ifs(blg_file.toFilesystemEncoding().c_str());
1196         string token;
1197         static regex const reg1(".*Found bibtex data file '([^']+).*");
1198
1199         while (getline(ifs, token)) {
1200                 token = rtrim(token, "\r");
1201                 smatch sub;
1202                 // FIXME UNICODE: We assume that citation keys and filenames
1203                 // in the aux file are in the file system encoding.
1204                 token = to_utf8(from_filesystem8bit(token));
1205                 if (regex_match(token, sub, reg1)) {
1206                         string data = sub.str(1);
1207                         if (!data.empty()) {
1208                                 LYXERR(Debug::LATEX, "Found bib file: " << data);
1209                                 handleFoundFile(data, dep);
1210                         }
1211                 }
1212         }
1213 }
1214
1215
1216 } // namespace lyx