src/LaTeX.C

   1 /**
   2  * \file LaTeX.C
   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 "gettext.h"
  21 #include "lyxrc.h"
  22 #include "debug.h"
  23 #include "DepTable.h"
  24
  25 #include "support/filetools.h"
  26 #include "support/convert.h"
  27 #include "support/lstrings.h"
  28 #include "support/lyxlib.h"
  29 #include "support/systemcall.h"
  30 #include "support/os.h"
  31
  32 #include <boost/filesystem/operations.hpp>
  33 #include <boost/regex.hpp>
  34
  35 #include <fstream>
  36
  37
  38 namespace lyx {
  39
  40 using support::absolutePath;
  41 using support::bformat;
  42 using support::changeExtension;
  43 using support::contains;
  44 using support::FileName;
  45 using support::findtexfile;
  46 using support::getcwd;
  47 using support::makeAbsPath;
  48 using support::onlyFilename;
  49 using support::prefixIs;
  50 using support::quoteName;
  51 using support::removeExtension;
  52 using support::rtrim;
  53 using support::rsplit;
  54 using support::split;
  55 using support::subst;
  56 using support::suffixIs;
  57 using support::Systemcall;
  58 using support::unlink;
  59 using support::trim;
  60
  61 namespace os = support::os;
  62 namespace fs = boost::filesystem;
  63
  64 using boost::regex;
  65 using boost::smatch;
  66
  67
  68 #ifndef CXX_GLOBAL_CSTD
  69 using std::sscanf;
  70 #endif
  71
  72 using std::endl;
  73 using std::getline;
  74 using std::string;
  75 using std::ifstream;
  76 using std::set;
  77 using std::vector;
  78
  79 // TODO: in no particular order
  80 // - get rid of the call to
  81 //   BufferList::updateIncludedTeXfiles, this should either
  82 //   be done before calling LaTeX::funcs or in a completely
  83 //   different way.
  84 // - the makeindex style files should be taken care of with
  85 //   the dependency mechanism.
  86 // - makeindex commandline options should be supported
  87 // - somewhere support viewing of bibtex and makeindex log files.
  88 // - we should perhaps also scan the bibtex log file
  89
  90 namespace {
  91
  92 docstring runMessage(unsigned int count)
  93 {
  94         return bformat(_("Waiting for LaTeX run number %1$d"), count);
  95 }
  96
  97 } // anon namespace
  98
  99 /*
 100  * CLASS TEXERRORS
 101  */
 102
 103 void TeXErrors::insertError(int line, docstring const & error_desc,
 104                             docstring const & error_text)
 105 {
 106         Error newerr(line, error_desc, error_text);
 107         errors.push_back(newerr);
 108 }
 109
 110
 111 bool operator==(Aux_Info const & a, Aux_Info const & o)
 112 {
 113         return a.aux_file == o.aux_file &&
 114                 a.citations == o.citations &&
 115                 a.databases == o.databases &&
 116                 a.styles == o.styles;
 117 }
 118
 119
 120 bool operator!=(Aux_Info const & a, Aux_Info const & o)
 121 {
 122         return !(a == o);
 123 }
 124
 125
 126 /*
 127  * CLASS LaTeX
 128  */
 129
 130 LaTeX::LaTeX(string const & latex, OutputParams const & rp,
 131              FileName const & f)
 132         : cmd(latex), file(f), runparams(rp)
 133 {
 134         num_errors = 0;
 135         if (prefixIs(cmd, "pdf")) { // Do we use pdflatex ?
 136                 depfile = FileName(file.absFilename() + ".dep-pdf");
 137                 output_file =
 138                         FileName(changeExtension(file.absFilename(), ".pdf"));
 139         } else {
 140                 depfile = FileName(file.absFilename() + ".dep");
 141                 output_file =
 142                         FileName(changeExtension(file.absFilename(), ".dvi"));
 143         }
 144 }
 145
 146
 147 void LaTeX::deleteFilesOnError() const
 148 {
 149         // currently just a dummy function.
 150
 151         // What files do we have to delete?
 152
 153         // This will at least make latex do all the runs
 154         unlink(depfile);
 155
 156         // but the reason for the error might be in a generated file...
 157
 158         // bibtex file
 159         FileName const bbl(changeExtension(file.absFilename(), ".bbl"));
 160         unlink(bbl);
 161
 162         // makeindex file
 163         FileName const ind(changeExtension(file.absFilename(), ".ind"));
 164         unlink(ind);
 165
 166         // nomencl file
 167         FileName const nls(changeExtension(file.absFilename(), ".nls"));
 168         unlink(nls);
 169
 170         // Also remove the aux file
 171         FileName const aux(changeExtension(file.absFilename(), ".aux"));
 172         unlink(aux);
 173 }
 174
 175
 176 int LaTeX::run(TeXErrors & terr)
 177         // We know that this function will only be run if the lyx buffer
 178         // has been changed. We also know that a newly written .tex file
 179         // is always different from the previous one because of the date
 180         // in it. However it seems safe to run latex (at least) on time
 181         // each time the .tex file changes.
 182 {
 183         int scanres = NO_ERRORS;
 184         unsigned int count = 0; // number of times run
 185         num_errors = 0; // just to make sure.
 186         unsigned int const MAX_RUN = 6;
 187         DepTable head; // empty head
 188         bool rerun = false; // rerun requested
 189
 190         // The class LaTeX does not know the temp path.
 191         theBufferList().updateIncludedTeXfiles(getcwd().absFilename(),
 192                 runparams);
 193
 194         // Never write the depfile if an error was encountered.
 195
 196         // 0
 197         // first check if the file dependencies exist:
 198         //     ->If it does exist
 199         //             check if any of the files mentioned in it have
 200         //             changed (done using a checksum).
 201         //                 -> if changed:
 202         //                        run latex once and
 203         //                        remake the dependency file
 204         //                 -> if not changed:
 205         //                        just return there is nothing to do for us.
 206         //     ->if it doesn't exist
 207         //             make it and
 208         //             run latex once (we need to run latex once anyway) and
 209         //             remake the dependency file.
 210         //
 211
 212         bool had_depfile = fs::exists(depfile.toFilesystemEncoding());
 213         bool run_bibtex = false;
 214         FileName const aux_file(changeExtension(file.absFilename(), "aux"));
 215
 216         if (had_depfile) {
 217                 lyxerr[Debug::DEPEND] << "Dependency file exists" << endl;
 218                 // Read the dep file:
 219                 had_depfile = head.read(depfile);
 220         }
 221
 222         if (had_depfile) {
 223                 // Update the checksums
 224                 head.update();
 225                 // Can't just check if anything has changed because it might
 226                 // have aborted on error last time... in which cas we need
 227                 // to re-run latex and collect the error messages
 228                 // (even if they are the same).
 229                 if (!fs::exists(output_file.toFilesystemEncoding())) {
 230                         lyxerr[Debug::DEPEND]
 231                                 << "re-running LaTeX because output file doesn't exist."
 232                                 << endl;
 233                 } else if (!head.sumchange()) {
 234                         lyxerr[Debug::DEPEND] << "return no_change" << endl;
 235                         return NO_CHANGE;
 236                 } else {
 237                         lyxerr[Debug::DEPEND]
 238                                 << "Dependency file has changed" << endl;
 239                 }
 240
 241                 if (head.extchanged(".bib") || head.extchanged(".bst"))
 242                         run_bibtex = true;
 243         } else
 244                 lyxerr[Debug::DEPEND]
 245                         << "Dependency file does not exist, or has wrong format"
 246                         << endl;
 247
 248         /// We scan the aux file even when had_depfile = false,
 249         /// because we can run pdflatex on the file after running latex on it,
 250         /// in which case we will not need to run bibtex again.
 251         vector<Aux_Info> bibtex_info_old;
 252         if (!run_bibtex)
 253                 bibtex_info_old = scanAuxFiles(aux_file);
 254
 255         ++count;
 256         lyxerr[Debug::LATEX] << "Run #" << count << endl;
 257         message(runMessage(count));
 258
 259         startscript();
 260         scanres = scanLogFile(terr);
 261         if (scanres & ERROR_RERUN) {
 262                 lyxerr[Debug::LATEX] << "Rerunning LaTeX" << endl;
 263                 startscript();
 264                 scanres = scanLogFile(terr);
 265         }
 266
 267         if (scanres & ERRORS) {
 268                 deleteFilesOnError();
 269                 return scanres; // return on error
 270         }
 271
 272         vector<Aux_Info> const bibtex_info = scanAuxFiles(aux_file);
 273         if (!run_bibtex && bibtex_info_old != bibtex_info)
 274                 run_bibtex = true;
 275
 276         // update the dependencies.
 277         deplog(head); // reads the latex log
 278         head.update();
 279
 280         // 0.5
 281         // At this point we must run external programs if needed.
 282         // makeindex will be run if a .idx file changed or was generated.
 283         // And if there were undefined citations or changes in references
 284         // the .aux file is checked for signs of bibtex. Bibtex is then run
 285         // if needed.
 286
 287         // memoir (at least) writes an empty *idx file in the first place.
 288         // A second latex run is needed.
 289         FileName const idxfile(changeExtension(file.absFilename(), ".idx"));
 290         rerun = fs::exists(idxfile.toFilesystemEncoding()) &&
 291                 fs::is_empty(idxfile.toFilesystemEncoding());
 292
 293         // run makeindex
 294         if (head.haschanged(idxfile)) {
 295                 // no checks for now
 296                 lyxerr[Debug::LATEX] << "Running MakeIndex." << endl;
 297                 message(_("Running MakeIndex."));
 298                 // onlyFilename() is needed for cygwin
 299                 rerun |= runMakeIndex(onlyFilename(idxfile.absFilename()),
 300                                 runparams);
 301         }
 302         if (head.haschanged(FileName(changeExtension(file.absFilename(), ".nlo")))) {
 303                 lyxerr[Debug::LATEX]
 304                         << "Running MakeIndex for nomencl."
 305                         << endl;
 306                 message(_("Running MakeIndex for nomencl."));
 307                 // onlyFilename() is needed for cygwin
 308                 string const nomenclstr = " -s nomencl.ist -o "
 309                         + onlyFilename(changeExtension(
 310                                 file.toFilesystemEncoding(), ".nls"));
 311                 rerun |= runMakeIndex(onlyFilename(changeExtension(
 312                                 file.absFilename(), ".nlo")),
 313                                 runparams,
 314                                 nomenclstr);
 315         }
 316
 317         // run bibtex
 318         // if (scanres & UNDEF_CIT || scanres & RERUN || run_bibtex)
 319         if (scanres & UNDEF_CIT || run_bibtex) {
 320                 // Here we must scan the .aux file and look for
 321                 // "\bibdata" and/or "\bibstyle". If one of those
 322                 // tags is found -> run bibtex and set rerun = true;
 323                 // no checks for now
 324                 lyxerr[Debug::LATEX] << "Running BibTeX." << endl;
 325                 message(_("Running BibTeX."));
 326                 updateBibtexDependencies(head, bibtex_info);
 327                 rerun |= runBibTeX(bibtex_info);
 328         } else if (!had_depfile) {
 329                 /// If we run pdflatex on the file after running latex on it,
 330                 /// then we do not need to run bibtex, but we do need to
 331                 /// insert the .bib and .bst files into the .dep-pdf file.
 332                 updateBibtexDependencies(head, bibtex_info);
 333         }
 334
 335         // 1
 336         // we know on this point that latex has been run once (or we just
 337         // returned) and the question now is to decide if we need to run
 338         // it any more. This is done by asking if any of the files in the
 339         // dependency file has changed. (remember that the checksum for
 340         // a given file is reported to have changed if it just was created)
 341         //     -> if changed or rerun == true:
 342         //             run latex once more and
 343         //             update the dependency structure
 344         //     -> if not changed:
 345         //             we does nothing at this point
 346         //
 347         if (rerun || head.sumchange()) {
 348                 rerun = false;
 349                 ++count;
 350                 lyxerr[Debug::DEPEND]
 351                         << "Dep. file has changed or rerun requested"
 352                         << endl;
 353                 lyxerr[Debug::LATEX]
 354                         << "Run #" << count << endl;
 355                 message(runMessage(count));
 356                 startscript();
 357                 scanres = scanLogFile(terr);
 358                 if (scanres & ERRORS) {
 359                         deleteFilesOnError();
 360                         return scanres; // return on error
 361                 }
 362
 363                 // update the depedencies
 364                 deplog(head); // reads the latex log
 365                 head.update();
 366         } else {
 367                 lyxerr[Debug::DEPEND]
 368                         << "Dep. file has NOT changed"
 369                         << endl;
 370         }
 371
 372         // 1.5
 373         // The inclusion of files generated by external programs like
 374         // makeindex or bibtex might have done changes to pagenumbering,
 375         // etc. And because of this we must run the external programs
 376         // again to make sure everything is redone correctly.
 377         // Also there should be no need to run the external programs any
 378         // more after this.
 379
 380         // run makeindex if the <file>.idx has changed or was generated.
 381         if (head.haschanged(FileName(changeExtension(file.absFilename(), ".idx")))) {
 382                 // no checks for now
 383                 lyxerr[Debug::LATEX] << "Running MakeIndex." << endl;
 384                 message(_("Running MakeIndex."));
 385                 // onlyFilename() is needed for cygwin
 386                 rerun = runMakeIndex(onlyFilename(changeExtension(
 387                                 file.absFilename(), ".idx")), runparams);
 388         }
 389
 390         // I am not pretty sure if need this twice.
 391         if (head.haschanged(FileName(changeExtension(file.absFilename(), ".nlo")))) {
 392                 lyxerr[Debug::LATEX]
 393                         << "Running MakeIndex for nomencl."
 394                         << endl;
 395                 message(_("Running MakeIndex for nomencl."));
 396                 // onlyFilename() is needed for cygwin
 397                 string nomenclstr = " -s nomencl.ist -o "
 398                         + onlyFilename(changeExtension(
 399                                 file.toFilesystemEncoding(), ".nls"));
 400                 rerun |= runMakeIndex(onlyFilename(changeExtension(
 401                                 file.absFilename(), ".nlo")),
 402                                 runparams,
 403                                 nomenclstr);
 404         }
 405
 406         // 2
 407         // we will only run latex more if the log file asks for it.
 408         // or if the sumchange() is true.
 409         //     -> rerun asked for:
 410         //             run latex and
 411         //             remake the dependency file
 412         //             goto 2 or return if max runs are reached.
 413         //     -> rerun not asked for:
 414         //             just return (fall out of bottom of func)
 415         //
 416         while ((head.sumchange() || rerun || (scanres & RERUN))
 417                && count < MAX_RUN) {
 418                 // Yes rerun until message goes away, or until
 419                 // MAX_RUNS are reached.
 420                 rerun = false;
 421                 ++count;
 422                 lyxerr[Debug::LATEX] << "Run #" << count << endl;
 423                 message(runMessage(count));
 424                 startscript();
 425                 scanres = scanLogFile(terr);
 426                 if (scanres & ERRORS) {
 427                         deleteFilesOnError();
 428                         return scanres; // return on error
 429                 }
 430
 431                 // keep this updated
 432                 head.update();
 433         }
 434
 435         // Write the dependencies to file.
 436         head.write(depfile);
 437         lyxerr[Debug::LATEX] << "Done." << endl;
 438         return scanres;
 439 }
 440
 441
 442 int LaTeX::startscript()
 443 {
 444         // onlyFilename() is needed for cygwin
 445         string tmp = cmd + ' '
 446                      + quoteName(onlyFilename(file.toFilesystemEncoding()))
 447                      + " > " + os::nulldev();
 448         Systemcall one;
 449         return one.startscript(Systemcall::Wait, tmp);
 450 }
 451
 452
 453 bool LaTeX::runMakeIndex(string const & f, OutputParams const & runparams,
 454                          string const & params)
 455 {
 456         lyxerr[Debug::LATEX]
 457                 << "idx file has been made, running makeindex on file "
 458                 << f << endl;
 459         string tmp = lyxrc.index_command + ' ';
 460
 461         tmp = subst(tmp, "$$lang", runparams.document_language);
 462         tmp += quoteName(f);
 463         tmp += params;
 464         Systemcall one;
 465         one.startscript(Systemcall::Wait, tmp);
 466         return true;
 467 }
 468
 469
 470 vector<Aux_Info> const
 471 LaTeX::scanAuxFiles(FileName const & file)
 472 {
 473         vector<Aux_Info> result;
 474
 475         result.push_back(scanAuxFile(file));
 476
 477         string const basename = removeExtension(file.absFilename());
 478         for (int i = 1; i < 1000; ++i) {
 479                 FileName const file2(basename
 480                         + '.' + convert<string>(i)
 481                         + ".aux");
 482                 if (!fs::exists(file2.toFilesystemEncoding()))
 483                         break;
 484                 result.push_back(scanAuxFile(file2));
 485         }
 486         return result;
 487 }
 488
 489
 490 Aux_Info const LaTeX::scanAuxFile(FileName const & file)
 491 {
 492         Aux_Info result;
 493         result.aux_file = file;
 494         scanAuxFile(file, result);
 495         return result;
 496 }
 497
 498
 499 void LaTeX::scanAuxFile(FileName const & file, Aux_Info & aux_info)
 500 {
 501         lyxerr[Debug::LATEX] << "Scanning aux file: " << file << endl;
 502
 503         ifstream ifs(file.toFilesystemEncoding().c_str());
 504         string token;
 505         static regex const reg1("\\\\citation\\{([^}]+)\\}");
 506         static regex const reg2("\\\\bibdata\\{([^}]+)\\}");
 507         static regex const reg3("\\\\bibstyle\\{([^}]+)\\}");
 508         static regex const reg4("\\\\@input\\{([^}]+)\\}");
 509
 510         while (getline(ifs, token)) {
 511                 token = rtrim(token, "\r");
 512                 smatch sub;
 513                 // FIXME UNICODE: We assume that citation keys and filenames
 514                 // in the aux file are in the file system encoding.
 515                 token = to_utf8(from_filesystem8bit(token));
 516                 if (regex_match(token, sub, reg1)) {
 517                         string data = sub.str(1);
 518                         while (!data.empty()) {
 519                                 string citation;
 520                                 data = split(data, citation, ',');
 521                                 lyxerr[Debug::LATEX] << "Citation: "
 522                                                      << citation << endl;
 523                                 aux_info.citations.insert(citation);
 524                         }
 525                 } else if (regex_match(token, sub, reg2)) {
 526                         string data = sub.str(1);
 527                         // data is now all the bib files separated by ','
 528                         // get them one by one and pass them to the helper
 529                         while (!data.empty()) {
 530                                 string database;
 531                                 data = split(data, database, ',');
 532                                 database = changeExtension(database, "bib");
 533                                 lyxerr[Debug::LATEX] << "BibTeX database: `"
 534                                                      << database << '\'' << endl;
 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: `"
 543                                              << style << '\'' << endl;
 544                         aux_info.styles.insert(style);
 545                 } else if (regex_match(token, sub, reg4)) {
 546                         string const file2 = sub.str(1);
 547                         scanAuxFile(makeAbsPath(file2), aux_info);
 548                 }
 549         }
 550 }
 551
 552
 553 void LaTeX::updateBibtexDependencies(DepTable & dep,
 554                                      vector<Aux_Info> const & bibtex_info)
 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<Aux_Info>::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
 580
 581 bool LaTeX::runBibTeX(vector<Aux_Info> const & bibtex_info)
 582 {
 583         bool result = false;
 584         for (vector<Aux_Info>::const_iterator it = bibtex_info.begin();
 585              it != bibtex_info.end(); ++it) {
 586                 if (it->databases.empty())
 587                         continue;
 588                 result = true;
 589
 590                 string tmp = lyxrc.bibtex_command + " ";
 591                 // onlyFilename() is needed for cygwin
 592                 tmp += quoteName(onlyFilename(removeExtension(
 593                                 it->aux_file.absFilename())));
 594                 Systemcall one;
 595                 one.startscript(Systemcall::Wait, tmp);
 596         }
 597         // Return whether bibtex was run
 598         return result;
 599 }
 600
 601
 602 int LaTeX::scanLogFile(TeXErrors & terr)
 603 {
 604         int last_line = -1;
 605         int line_count = 1;
 606         int retval = NO_ERRORS;
 607         string tmp =
 608                 onlyFilename(changeExtension(file.absFilename(), ".log"));
 609         lyxerr[Debug::LATEX] << "Log file: " << tmp << endl;
 610         FileName const fn = FileName(makeAbsPath(tmp));
 611         ifstream ifs(fn.toFilesystemEncoding().c_str());
 612
 613         string token;
 614         while (getline(ifs, token)) {
 615                 // MikTeX sometimes inserts \0 in the log file. They can't be
 616                 // removed directly with the existing string utility
 617                 // functions, so convert them first to \r, and remove all
 618                 // \r's afterwards, since we need to remove them anyway.
 619                 token = subst(token, '\0', '\r');
 620                 token = subst(token, "\r", "");
 621
 622                 lyxerr[Debug::LATEX] << "Log line: " << token << endl;
 623
 624                 if (token.empty())
 625                         continue;
 626
 627                 if (prefixIs(token, "LaTeX Warning:") ||
 628                     prefixIs(token, "! pdfTeX warning")) {
 629                         // Here shall we handle different
 630                         // types of warnings
 631                         retval |= LATEX_WARNING;
 632                         lyxerr[Debug::LATEX] << "LaTeX Warning." << endl;
 633                         if (contains(token, "Rerun to get cross-references")) {
 634                                 retval |= RERUN;
 635                                 lyxerr[Debug::LATEX]
 636                                         << "We should rerun." << endl;
 637                         } else if (contains(token, "Citation")
 638                                    && contains(token, "on page")
 639                                    && contains(token, "undefined")) {
 640                                 retval |= UNDEF_CIT;
 641                         }
 642                 } else if (prefixIs(token, "Package")) {
 643                         // Package warnings
 644                         retval |= PACKAGE_WARNING;
 645                         if (contains(token, "natbib Warning:")) {
 646                                 // Natbib warnings
 647                                 if (contains(token, "Citation")
 648                                     && contains(token, "on page")
 649                                     && contains(token, "undefined")) {
 650                                         retval |= UNDEF_CIT;
 651                                 }
 652                         } else if (contains(token, "run BibTeX")) {
 653                                 retval |= UNDEF_CIT;
 654                         } else if (contains(token, "Rerun LaTeX") ||
 655                                    contains(token, "Rerun to get")) {
 656                                 // at least longtable.sty and bibtopic.sty
 657                                 // might use this.
 658                                 lyxerr[Debug::LATEX]
 659                                         << "We should rerun." << endl;
 660                                 retval |= RERUN;
 661                         }
 662                 } else if (token[0] == '(') {
 663                         if (contains(token, "Rerun LaTeX") ||
 664                             contains(token, "Rerun to get")) {
 665                                 // Used by natbib
 666                                 lyxerr[Debug::LATEX]
 667                                         << "We should rerun." << endl;
 668                                 retval |= RERUN;
 669                         }
 670                 } else if (prefixIs(token, "! ")) {
 671                         // Ok, we have something that looks like a TeX Error
 672                         // but what do we really have.
 673
 674                         // Just get the error description:
 675                         string desc(token, 2);
 676                         if (contains(token, "LaTeX Error:"))
 677                                 retval |= LATEX_ERROR;
 678                         // get the next line
 679                         string tmp;
 680                         int count = 0;
 681                         do {
 682                                 if (!getline(ifs, tmp))
 683                                         break;
 684                                 if (++count > 10)
 685                                         break;
 686                         } while (!prefixIs(tmp, "l."));
 687                         if (prefixIs(tmp, "l.")) {
 688                                 // we have a latex error
 689                                 retval |=  TEX_ERROR;
 690                                 if (contains(desc,
 691                                     "Package babel Error: You haven't defined the language"))
 692                                         retval |= ERROR_RERUN;
 693                                 // get the line number:
 694                                 int line = 0;
 695                                 sscanf(tmp.c_str(), "l.%d", &line);
 696                                 // get the rest of the message:
 697                                 string errstr(tmp, tmp.find(' '));
 698                                 errstr += '\n';
 699                                 getline(ifs, tmp);
 700                                 while (!contains(errstr, "l.")
 701                                        && !tmp.empty()
 702                                        && !prefixIs(tmp, "! ")
 703                                        && !contains(tmp, "(job aborted")) {
 704                                         errstr += tmp;
 705                                         errstr += "\n";
 706                                         getline(ifs, tmp);
 707                                 }
 708                                 lyxerr[Debug::LATEX]
 709                                         << "line: " << line << '\n'
 710                                         << "Desc: " << desc << '\n'
 711                                         << "Text: " << errstr << endl;
 712                                 if (line == last_line)
 713                                         ++line_count;
 714                                 else {
 715                                         line_count = 1;
 716                                         last_line = line;
 717                                 }
 718                                 if (line_count <= 5) {
 719                                         // FIXME UNICODE
 720                                         // We have no idea what the encoding of
 721                                         // the log file is.
 722                                         // It seems that the output from the
 723                                         // latex compiler itself is pure ASCII,
 724                                         // but it can include bits from the
 725                                         // document, so whatever encoding we
 726                                         // assume here it can be wrong.
 727                                         terr.insertError(line,
 728                                                          from_local8bit(desc),
 729                                                          from_local8bit(errstr));
 730                                         ++num_errors;
 731                                 }
 732                         }
 733                 } else {
 734                         // information messages, TeX warnings and other
 735                         // warnings we have not caught earlier.
 736                         if (prefixIs(token, "Overfull ")) {
 737                                 retval |= TEX_WARNING;
 738                         } else if (prefixIs(token, "Underfull ")) {
 739                                 retval |= TEX_WARNING;
 740                         } else if (contains(token, "Rerun to get citations")) {
 741                                 // Natbib seems to use this.
 742                                 retval |= UNDEF_CIT;
 743                         } else if (contains(token, "No pages of output")) {
 744                                 // A dvi file was not created
 745                                 retval |= NO_OUTPUT;
 746                         } else if (contains(token, "That makes 100 errors")) {
 747                                 // More than 100 errors were reprted
 748                                 retval |= TOO_MANY_ERRORS;
 749                         }
 750                 }
 751         }
 752         lyxerr[Debug::LATEX] << "Log line: " << token << endl;
 753         return retval;
 754 }
 755
 756
 757 namespace {
 758
 759 bool insertIfExists(FileName const & absname, DepTable & head)
 760 {
 761         fs::path const path = absname.toFilesystemEncoding();
 762         if (fs::exists(path) && !fs::is_directory(path)) {
 763                         head.insert(absname, true);
 764                         return true;
 765         }
 766         return false;
 767 }
 768
 769
 770 bool handleFoundFile(string const & ff, DepTable & head)
 771 {
 772         // convert from native os path to unix path
 773         string foundfile = os::internal_path(trim(ff));
 774
 775         lyxerr[Debug::DEPEND] << "Found file: " << foundfile << endl;
 776
 777         // Ok now we found a file.
 778         // Now we should make sure that this is a file that we can
 779         // access through the normal paths.
 780         // We will not try any fancy search methods to
 781         // find the file.
 782
 783         // (1) foundfile is an
 784         //     absolute path and should
 785         //     be inserted.
 786         if (absolutePath(foundfile)) {
 787                 lyxerr[Debug::DEPEND] << "AbsolutePath file: "
 788                                       << foundfile << endl;
 789                 // On initial insert we want to do the update at once
 790                 // since this file cannot be a file generated by
 791                 // the latex run.
 792                 FileName absname(foundfile);
 793                 if (!insertIfExists(absname, head)) {
 794                         // check for spaces
 795                         string strippedfile = foundfile;
 796                         while (contains(strippedfile, " ")) {
 797                                 // strip off part after last space and try again
 798                                 string tmp = strippedfile;
 799                                 string const stripoff =
 800                                         rsplit(tmp, strippedfile, ' ');
 801                                 absname.set(strippedfile);
 802                                 if (insertIfExists(absname, head))
 803                                         return true;
 804                         }
 805                 }
 806         }
 807
 808         string onlyfile = onlyFilename(foundfile);
 809         FileName absname(makeAbsPath(onlyfile));
 810
 811         // check for spaces
 812         while (contains(foundfile, " ")) {
 813                 if (fs::exists(absname.toFilesystemEncoding()))
 814                         // everything o.k.
 815                         break;
 816                 else {
 817                         // strip off part after last space and try again
 818                         string strippedfile;
 819                         string const stripoff =
 820                                 rsplit(foundfile, strippedfile, ' ');
 821                         foundfile = strippedfile;
 822                         onlyfile = onlyFilename(strippedfile);
 823                         absname = makeAbsPath(onlyfile);
 824                 }
 825         }
 826
 827         // (2) foundfile is in the tmpdir
 828         //     insert it into head
 829         fs::path const path = absname.toFilesystemEncoding();
 830         if (fs::exists(path) && !fs::is_directory(path)) {
 831                 static regex unwanted("^.*\\.(aux|log|dvi|bbl|ind|glo)$");
 832                 if (regex_match(onlyfile, unwanted)) {
 833                         lyxerr[Debug::DEPEND]
 834                                 << "We don't want "
 835                                 << onlyfile
 836                                 << " in the dep file"
 837                                 << endl;
 838                 } else if (suffixIs(onlyfile, ".tex")) {
 839                         // This is a tex file generated by LyX
 840                         // and latex is not likely to change this
 841                         // during its runs.
 842                         lyxerr[Debug::DEPEND]
 843                                 << "Tmpdir TeX file: "
 844                                 << onlyfile
 845                                 << endl;
 846                         head.insert(absname, true);
 847                 } else {
 848                         lyxerr[Debug::DEPEND]
 849                                 << "In tmpdir file:"
 850                                 << onlyfile
 851                                 << endl;
 852                         head.insert(absname);
 853                 }
 854                 return true;
 855         } else {
 856                 lyxerr[Debug::DEPEND]
 857                         << "Not a file or we are unable to find it."
 858                         << endl;
 859                 return false;
 860         }
 861 }
 862
 863
 864 bool checkLineBreak(string const & ff, DepTable & head)
 865 {
 866         if (contains(ff, '.'))
 867                 // if we have a dot, we let handleFoundFile decide
 868                 return handleFoundFile(ff, head);
 869         else
 870                 // else, we suspect a line break
 871                 return false;
 872 }
 873
 874 } // anon namespace
 875
 876
 877 void LaTeX::deplog(DepTable & head)
 878 {
 879         // This function reads the LaTeX log file end extracts all the
 880         // external files used by the LaTeX run. The files are then
 881         // entered into the dependency file.
 882
 883         string const logfile =
 884                 onlyFilename(changeExtension(file.absFilename(), ".log"));
 885
 886         static regex reg1("File: (.+).*");
 887         static regex reg2("No file (.+)(.).*");
 888         static regex reg3("\\\\openout[0-9]+.*=.*`(.+)(..).*");
 889         // If an index should be created, MikTex does not write a line like
 890         //    \openout# = 'sample.idx'.
 891         // but instead only a line like this into the log:
 892         //   Writing index file sample.idx
 893         static regex reg4("Writing index file (.+).*");
 894         // files also can be enclosed in <...>
 895         static regex reg5("<([^>]+)(.).*");
 896         static regex regnomencl("Writing nomenclature file (.+).*");
 897         // If a toc should be created, MikTex does not write a line like
 898         //    \openout# = `sample.toc'.
 899         // but only a line like this into the log:
 900         //    \tf@toc=\write#
 901         // This line is also written by tetex.
 902         // This line is not present if no toc should be created.
 903         static regex miktexTocReg("\\\\tf@toc=\\\\write.*");
 904         static regex reg6(".*\\([^)]+.*");
 905
 906         FileName const fn(makeAbsPath(logfile));
 907         ifstream ifs(fn.toFilesystemEncoding().c_str());
 908         string lastline;
 909         while (ifs) {
 910                 // Ok, the scanning of files here is not sufficient.
 911                 // Sometimes files are named by "File: xxx" only
 912                 // So I think we should use some regexps to find files instead.
 913                 // Note: all file names and paths might contains spaces.
 914                 bool found_file = false;
 915                 string token;
 916                 getline(ifs, token);
 917                 // MikTeX sometimes inserts \0 in the log file. They can't be
 918                 // removed directly with the existing string utility
 919                 // functions, so convert them first to \r, and remove all
 920                 // \r's afterwards, since we need to remove them anyway.
 921                 token = subst(token, '\0', '\r');
 922                 token = subst(token, "\r", "");
 923                 if (token.empty() || token == ")") {
 924                         lastline = string();
 925                         continue;
 926                 }
 927
 928                 // Sometimes, filenames are broken across lines.
 929                 // We care for that and save suspicious lines.
 930                 // Here we exclude some cases where we are sure
 931                 // that there is no continued filename
 932                 if (!lastline.empty()) {
 933                         static regex package_info("Package \\w+ Info: .*");
 934                         static regex package_warning("Package \\w+ Warning: .*");
 935                         if (prefixIs(token, "File:") || prefixIs(token, "(Font)")
 936                             || prefixIs(token, "Package:")
 937                             || prefixIs(token, "Language:")
 938                             || prefixIs(token, "LaTeX Info:")
 939                             || prefixIs(token, "LaTeX Font Info:")
 940                             || prefixIs(token, "\\openout[")
 941                             || prefixIs(token, "))")
 942                             || regex_match(token, package_info)
 943                             || regex_match(token, package_warning))
 944                                 lastline = string();
 945                 }
 946
 947                 if (!lastline.empty())
 948                         // probably a continued filename from last line
 949                         token = lastline + token;
 950                 if (token.length() > 255) {
 951                         // string too long. Cut off.
 952                         int r = token.length() - 250;
 953                         string ntoken = token.substr(r, token.length());
 954                         token = ntoken;
 955                 }
 956
 957                 smatch sub;
 958
 959                 // FIXME UNICODE: We assume that the file names in the log
 960                 // file are in the file system encoding.
 961                 token = to_utf8(from_filesystem8bit(token));
 962
 963                 // (1) "File: file.ext"
 964                 if (regex_match(token, sub, reg1)) {
 965                         // check for dot
 966                         found_file = checkLineBreak(sub.str(1), head);
 967                         // However, ...
 968                         if (suffixIs(token, ")"))
 969                                 // no line break for sure
 970                                 // pretend we've been succesfully searching
 971                                 found_file = true;
 972                 // (2) "No file file.ext"
 973                 } else if (regex_match(token, sub, reg2)) {
 974                         // file names must contains a dot, line ends with dot
 975                         if (contains(sub.str(1), '.') && sub.str(2) == ".")
 976                                 found_file = handleFoundFile(sub.str(1), head);
 977                         else
 978                                 // we suspect a line break
 979                                 found_file = false;
 980                 // (3) "\openout<nr> = `file.ext'."
 981                 } else if (regex_match(token, sub, reg3)) {
 982                         // search for closing '. at the end of the line
 983                         if (sub.str(2) == "\'.")
 984                                 found_file = handleFoundFile(sub.str(1), head);
 985                         else
 986                                 // probable line break
 987                                 found_file = false;
 988                 // (4) "Writing index file file.ext"
 989                 } else if (regex_match(token, sub, reg4))
 990                         // check for dot
 991                         found_file = checkLineBreak(sub.str(1), head);
 992                 // (5) "<file.ext>"
 993                 else if (regex_match(token, sub, reg5)) {
 994                         // search for closing '>' and dot ('*.*>') at the eol
 995                         if (contains(sub.str(1), '.') && sub.str(2) == ">")
 996                                 found_file = handleFoundFile(sub.str(1), head);
 997                         else
 998                                 // probable line break
 999                                 found_file = false;
1000                 // (6) "Writing nomenclature file file.ext"
1001                 } else if (regex_match(token, sub, regnomencl))
1002                         // check for dot
1003                         found_file = checkLineBreak(sub.str(1), head);
1004                 // (7) "\tf@toc=\write<nr>" (for MikTeX)
1005                 else if (regex_match(token, sub, miktexTocReg))
1006                         found_file = handleFoundFile(onlyFilename(changeExtension(
1007                                                 file.absFilename(), ".toc")), head);
1008                 else
1009                         // not found, but we won't check further
1010                         // pretend we've been succesfully searching
1011                         found_file = true;
1012
1013                 // (8) "(file.ext"
1014                 // note that we can have several of these on one line
1015                 // this must be queried separated, because of
1016                 // cases such as "File: file.ext (type eps)"
1017                 // where "File: file.ext" would be skipped
1018                 if (regex_match(token, sub, reg6)) {
1019                         // search for strings in (...)
1020                         static regex reg6_1("\\(([^()]+)(.).*");
1021                         smatch what;
1022                         string::const_iterator first = token.begin();
1023                         string::const_iterator end = token.end();
1024
1025                         while (regex_search(first, end, what, reg6_1)) {
1026                                 // if we have a dot, try to handle as file
1027                                 if (contains(what.str(1), '.')) {
1028                                         first = what[0].second;
1029                                         if (what.str(2) == ")") {
1030                                                 handleFoundFile(what.str(1), head);
1031                                                 // since we had a closing bracket,
1032                                                 // do not investigate further
1033                                                 found_file = true;
1034                                         } else
1035                                                 // if we have no closing bracket,
1036                                                 // try to handle as file nevertheless
1037                                                 found_file = handleFoundFile(
1038                                                         what.str(1) + what.str(2), head);
1039                                 }
1040                                 // if we do not have a dot, check if the line has
1041                                 // a closing bracket (else, we suspect a line break)
1042                                 else if (what.str(2) != ")") {
1043                                         first = what[0].second;
1044                                         found_file = false;
1045                                 } else {
1046                                         // we have a closing bracket, so the content
1047                                         // is not a file name.
1048                                         // no need to investigate further
1049                                         // pretend we've been succesfully searching
1050                                         first = what[0].second;
1051                                         found_file = true;
1052                                 }
1053                         }
1054                 }
1055
1056                 if (!found_file)
1057                         // probable linebreak:
1058                         // save this line
1059                         lastline = token;
1060                 else
1061                         // no linebreak: reset
1062                         lastline = string();
1063         }
1064
1065         // Make sure that the main .tex file is in the dependency file.
1066         head.insert(file, true);
1067 }
1068
1069
1070 } // namespace lyx