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/filesystem/path.hpp>
  34 #include <boost/regex.hpp>
  35
  36 #include <fstream>
  37
  38
  39 namespace lyx {
  40
  41 using support::absolutePath;
  42 using support::bformat;
  43 using support::changeExtension;
  44 using support::contains;
  45 using support::FileName;
  46 using support::findtexfile;
  47 using support::getcwd;
  48 using support::makeAbsPath;
  49 using support::onlyFilename;
  50 using support::prefixIs;
  51 using support::quoteName;
  52 using support::removeExtension;
  53 using support::rtrim;
  54 using support::rsplit;
  55 using support::split;
  56 using support::subst;
  57 using support::suffixIs;
  58 using support::Systemcall;
  59 using support::unlink;
  60 using support::trim;
  61
  62 namespace os = support::os;
  63 namespace fs = boost::filesystem;
  64
  65 using boost::regex;
  66 using boost::smatch;
  67
  68
  69 #ifndef CXX_GLOBAL_CSTD
  70 using std::sscanf;
  71 #endif
  72
  73 using std::endl;
  74 using std::getline;
  75 using std::string;
  76 using std::ifstream;
  77 using std::set;
  78 using std::vector;
  79
  80 // TODO: in no particular order
  81 // - get rid of the call to
  82 //   BufferList::updateIncludedTeXfiles, this should either
  83 //   be done before calling LaTeX::funcs or in a completely
  84 //   different way.
  85 // - the makeindex style files should be taken care of with
  86 //   the dependency mechanism.
  87 // - makeindex commandline options should be supported
  88 // - somewhere support viewing of bibtex and makeindex log files.
  89 // - we should perhaps also scan the bibtex log file
  90
  91 namespace {
  92
  93 docstring runMessage(unsigned int count)
  94 {
  95         return bformat(_("Waiting for LaTeX run number %1$d"), count);
  96 }
  97
  98 } // anon namespace
  99
 100 /*
 101  * CLASS TEXERRORS
 102  */
 103
 104 void TeXErrors::insertError(int line, docstring const & error_desc,
 105                             docstring const & error_text)
 106 {
 107         Error newerr(line, error_desc, error_text);
 108         errors.push_back(newerr);
 109 }
 110
 111
 112 bool operator==(Aux_Info const & a, Aux_Info const & o)
 113 {
 114         return a.aux_file == o.aux_file &&
 115                 a.citations == o.citations &&
 116                 a.databases == o.databases &&
 117                 a.styles == o.styles;
 118 }
 119
 120
 121 bool operator!=(Aux_Info const & a, Aux_Info const & o)
 122 {
 123         return !(a == o);
 124 }
 125
 126
 127 /*
 128  * CLASS LaTeX
 129  */
 130
 131 LaTeX::LaTeX(string const & latex, OutputParams const & rp,
 132              FileName const & f)
 133         : cmd(latex), file(f), runparams(rp)
 134 {
 135         num_errors = 0;
 136         if (prefixIs(cmd, "pdf")) { // Do we use pdflatex ?
 137                 depfile = FileName(file.absFilename() + ".dep-pdf");
 138                 output_file =
 139                         FileName(changeExtension(file.absFilename(), ".pdf"));
 140         } else {
 141                 depfile = FileName(file.absFilename() + ".dep");
 142                 output_file =
 143                         FileName(changeExtension(file.absFilename(), ".dvi"));
 144         }
 145 }
 146
 147
 148 void LaTeX::deleteFilesOnError() const
 149 {
 150         // currently just a dummy function.
 151
 152         // What files do we have to delete?
 153
 154         // This will at least make latex do all the runs
 155         unlink(depfile);
 156
 157         // but the reason for the error might be in a generated file...
 158
 159         // bibtex file
 160         FileName const bbl(changeExtension(file.absFilename(), ".bbl"));
 161         unlink(bbl);
 162
 163         // makeindex file
 164         FileName const ind(changeExtension(file.absFilename(), ".ind"));
 165         unlink(ind);
 166
 167         // nomencl file
 168         FileName const nls(changeExtension(file.absFilename(), ".nls"));
 169         unlink(nls);
 170
 171         // Also remove the aux file
 172         FileName const aux(changeExtension(file.absFilename(), ".aux"));
 173         unlink(aux);
 174 }
 175
 176
 177 int LaTeX::run(TeXErrors & terr)
 178         // We know that this function will only be run if the lyx buffer
 179         // has been changed. We also know that a newly written .tex file
 180         // is always different from the previous one because of the date
 181         // in it. However it seems safe to run latex (at least) on time
 182         // each time the .tex file changes.
 183 {
 184         int scanres = NO_ERRORS;
 185         unsigned int count = 0; // number of times run
 186         num_errors = 0; // just to make sure.
 187         unsigned int const MAX_RUN = 6;
 188         DepTable head; // empty head
 189         bool rerun = false; // rerun requested
 190
 191         // The class LaTeX does not know the temp path.
 192         theBufferList().updateIncludedTeXfiles(getcwd().absFilename(),
 193                 runparams);
 194
 195         // Never write the depfile if an error was encountered.
 196
 197         // 0
 198         // first check if the file dependencies exist:
 199         //     ->If it does exist
 200         //             check if any of the files mentioned in it have
 201         //             changed (done using a checksum).
 202         //                 -> if changed:
 203         //                        run latex once and
 204         //                        remake the dependency file
 205         //                 -> if not changed:
 206         //                        just return there is nothing to do for us.
 207         //     ->if it doesn't exist
 208         //             make it and
 209         //             run latex once (we need to run latex once anyway) and
 210         //             remake the dependency file.
 211         //
 212
 213         bool had_depfile = fs::exists(depfile.toFilesystemEncoding());
 214         bool run_bibtex = false;
 215         FileName const aux_file(changeExtension(file.absFilename(), "aux"));
 216
 217         if (had_depfile) {
 218                 lyxerr[Debug::DEPEND] << "Dependency file exists" << endl;
 219                 // Read the dep file:
 220                 had_depfile = head.read(depfile);
 221         }
 222
 223         if (had_depfile) {
 224                 // Update the checksums
 225                 head.update();
 226                 // Can't just check if anything has changed because it might
 227                 // have aborted on error last time... in which cas we need
 228                 // to re-run latex and collect the error messages
 229                 // (even if they are the same).
 230                 if (!fs::exists(output_file.toFilesystemEncoding())) {
 231                         lyxerr[Debug::DEPEND]
 232                                 << "re-running LaTeX because output file doesn't exist."
 233                                 << endl;
 234                 } else if (!head.sumchange()) {
 235                         lyxerr[Debug::DEPEND] << "return no_change" << endl;
 236                         return NO_CHANGE;
 237                 } else {
 238                         lyxerr[Debug::DEPEND]
 239                                 << "Dependency file has changed" << endl;
 240                 }
 241
 242                 if (head.extchanged(".bib") || head.extchanged(".bst"))
 243                         run_bibtex = true;
 244         } else
 245                 lyxerr[Debug::DEPEND]
 246                         << "Dependency file does not exist, or has wrong format"
 247                         << endl;
 248
 249         /// We scan the aux file even when had_depfile = false,
 250         /// because we can run pdflatex on the file after running latex on it,
 251         /// in which case we will not need to run bibtex again.
 252         vector<Aux_Info> bibtex_info_old;
 253         if (!run_bibtex)
 254                 bibtex_info_old = scanAuxFiles(aux_file);
 255
 256         ++count;
 257         lyxerr[Debug::LATEX] << "Run #" << count << endl;
 258         message(runMessage(count));
 259
 260         startscript();
 261         scanres = scanLogFile(terr);
 262         if (scanres & ERROR_RERUN) {
 263                 lyxerr[Debug::LATEX] << "Rerunning LaTeX" << endl;
 264                 startscript();
 265                 scanres = scanLogFile(terr);
 266         }
 267
 268         if (scanres & ERRORS) {
 269                 deleteFilesOnError();
 270                 return scanres; // return on error
 271         }
 272
 273         vector<Aux_Info> const bibtex_info = scanAuxFiles(aux_file);
 274         if (!run_bibtex && bibtex_info_old != bibtex_info)
 275                 run_bibtex = true;
 276
 277         // update the dependencies.
 278         deplog(head); // reads the latex log
 279         head.update();
 280
 281         // 0.5
 282         // At this point we must run external programs if needed.
 283         // makeindex will be run if a .idx file changed or was generated.
 284         // And if there were undefined citations or changes in references
 285         // the .aux file is checked for signs of bibtex. Bibtex is then run
 286         // if needed.
 287
 288         // memoir (at least) writes an empty *idx file in the first place.
 289         // A second latex run is needed.
 290         FileName const idxfile(changeExtension(file.absFilename(), ".idx"));
 291         rerun = fs::exists(idxfile.toFilesystemEncoding()) &&
 292                 fs::is_empty(idxfile.toFilesystemEncoding());
 293
 294         // run makeindex
 295         if (head.haschanged(idxfile)) {
 296                 // no checks for now
 297                 lyxerr[Debug::LATEX] << "Running MakeIndex." << endl;
 298                 message(_("Running MakeIndex."));
 299                 // onlyFilename() is needed for cygwin
 300                 rerun |= runMakeIndex(onlyFilename(idxfile.absFilename()),
 301                                 runparams);
 302         }
 303         if (head.haschanged(FileName(changeExtension(file.absFilename(), ".nlo")))) {
 304                 lyxerr[Debug::LATEX]
 305                         << "Running MakeIndex for nomencl."
 306                         << endl;
 307                 message(_("Running MakeIndex for nomencl."));
 308                 // onlyFilename() is needed for cygwin
 309                 string const nomenclstr = " -s nomencl.ist -o "
 310                         + onlyFilename(changeExtension(
 311                                 file.toFilesystemEncoding(), ".nls"));
 312                 rerun |= runMakeIndex(onlyFilename(changeExtension(
 313                                 file.absFilename(), ".nlo")),
 314                                 runparams,
 315                                 nomenclstr);
 316         }
 317
 318         // run bibtex
 319         // if (scanres & UNDEF_CIT || scanres & RERUN || run_bibtex)
 320         if (scanres & UNDEF_CIT || run_bibtex) {
 321                 // Here we must scan the .aux file and look for
 322                 // "\bibdata" and/or "\bibstyle". If one of those
 323                 // tags is found -> run bibtex and set rerun = true;
 324                 // no checks for now
 325                 lyxerr[Debug::LATEX] << "Running BibTeX." << endl;
 326                 message(_("Running BibTeX."));
 327                 updateBibtexDependencies(head, bibtex_info);
 328                 rerun |= runBibTeX(bibtex_info);
 329         } else if (!had_depfile) {
 330                 /// If we run pdflatex on the file after running latex on it,
 331                 /// then we do not need to run bibtex, but we do need to
 332                 /// insert the .bib and .bst files into the .dep-pdf file.
 333                 updateBibtexDependencies(head, bibtex_info);
 334         }
 335
 336         // 1
 337         // we know on this point that latex has been run once (or we just
 338         // returned) and the question now is to decide if we need to run
 339         // it any more. This is done by asking if any of the files in the
 340         // dependency file has changed. (remember that the checksum for
 341         // a given file is reported to have changed if it just was created)
 342         //     -> if changed or rerun == true:
 343         //             run latex once more and
 344         //             update the dependency structure
 345         //     -> if not changed:
 346         //             we does nothing at this point
 347         //
 348         if (rerun || head.sumchange()) {
 349                 rerun = false;
 350                 ++count;
 351                 lyxerr[Debug::DEPEND]
 352                         << "Dep. file has changed or rerun requested"
 353                         << endl;
 354                 lyxerr[Debug::LATEX]
 355                         << "Run #" << count << endl;
 356                 message(runMessage(count));
 357                 startscript();
 358                 scanres = scanLogFile(terr);
 359                 if (scanres & ERRORS) {
 360                         deleteFilesOnError();
 361                         return scanres; // return on error
 362                 }
 363
 364                 // update the depedencies
 365                 deplog(head); // reads the latex log
 366                 head.update();
 367         } else {
 368                 lyxerr[Debug::DEPEND]
 369                         << "Dep. file has NOT changed"
 370                         << endl;
 371         }
 372
 373         // 1.5
 374         // The inclusion of files generated by external programs like
 375         // makeindex or bibtex might have done changes to pagenumbering,
 376         // etc. And because of this we must run the external programs
 377         // again to make sure everything is redone correctly.
 378         // Also there should be no need to run the external programs any
 379         // more after this.
 380
 381         // run makeindex if the <file>.idx has changed or was generated.
 382         if (head.haschanged(FileName(changeExtension(file.absFilename(), ".idx")))) {
 383                 // no checks for now
 384                 lyxerr[Debug::LATEX] << "Running MakeIndex." << endl;
 385                 message(_("Running MakeIndex."));
 386                 // onlyFilename() is needed for cygwin
 387                 rerun = runMakeIndex(onlyFilename(changeExtension(
 388                                 file.absFilename(), ".idx")), runparams);
 389         }
 390
 391         // I am not pretty sure if need this twice.
 392         if (head.haschanged(FileName(changeExtension(file.absFilename(), ".nlo")))) {
 393                 lyxerr[Debug::LATEX]
 394                         << "Running MakeIndex for nomencl."
 395                         << endl;
 396                 message(_("Running MakeIndex for nomencl."));
 397                 // onlyFilename() is needed for cygwin
 398                 string nomenclstr = " -s nomencl.ist -o "
 399                         + onlyFilename(changeExtension(
 400                                 file.toFilesystemEncoding(), ".nls"));
 401                 rerun |= runMakeIndex(onlyFilename(changeExtension(
 402                                 file.absFilename(), ".nlo")),
 403                                 runparams,
 404                                 nomenclstr);
 405         }
 406
 407         // 2
 408         // we will only run latex more if the log file asks for it.
 409         // or if the sumchange() is true.
 410         //     -> rerun asked for:
 411         //             run latex and
 412         //             remake the dependency file
 413         //             goto 2 or return if max runs are reached.
 414         //     -> rerun not asked for:
 415         //             just return (fall out of bottom of func)
 416         //
 417         while ((head.sumchange() || rerun || (scanres & RERUN))
 418                && count < MAX_RUN) {
 419                 // Yes rerun until message goes away, or until
 420                 // MAX_RUNS are reached.
 421                 rerun = false;
 422                 ++count;
 423                 lyxerr[Debug::LATEX] << "Run #" << count << endl;
 424                 message(runMessage(count));
 425                 startscript();
 426                 scanres = scanLogFile(terr);
 427                 if (scanres & ERRORS) {
 428                         deleteFilesOnError();
 429                         return scanres; // return on error
 430                 }
 431
 432                 // keep this updated
 433                 head.update();
 434         }
 435
 436         // Write the dependencies to file.
 437         head.write(depfile);
 438         lyxerr[Debug::LATEX] << "Done." << endl;
 439         return scanres;
 440 }
 441
 442
 443 int LaTeX::startscript()
 444 {
 445         // onlyFilename() is needed for cygwin
 446         string tmp = cmd + ' '
 447                      + quoteName(onlyFilename(file.toFilesystemEncoding()))
 448                      + " > " + os::nulldev();
 449         Systemcall one;
 450         return one.startscript(Systemcall::Wait, tmp);
 451 }
 452
 453
 454 bool LaTeX::runMakeIndex(string const & f, OutputParams const & runparams,
 455                          string const & params)
 456 {
 457         lyxerr[Debug::LATEX]
 458                 << "idx file has been made, running makeindex on file "
 459                 << f << endl;
 460         string tmp = lyxrc.index_command + ' ';
 461
 462         tmp = subst(tmp, "$$lang", runparams.document_language);
 463         tmp += quoteName(f);
 464         tmp += params;
 465         Systemcall one;
 466         one.startscript(Systemcall::Wait, tmp);
 467         return true;
 468 }
 469
 470
 471 vector<Aux_Info> const
 472 LaTeX::scanAuxFiles(FileName const & file)
 473 {
 474         vector<Aux_Info> result;
 475
 476         result.push_back(scanAuxFile(file));
 477
 478         string const basename = removeExtension(file.absFilename());
 479         for (int i = 1; i < 1000; ++i) {
 480                 FileName const file2(basename
 481                         + '.' + convert<string>(i)
 482                         + ".aux");
 483                 if (!fs::exists(file2.toFilesystemEncoding()))
 484                         break;
 485                 result.push_back(scanAuxFile(file2));
 486         }
 487         return result;
 488 }
 489
 490
 491 Aux_Info const LaTeX::scanAuxFile(FileName const & file)
 492 {
 493         Aux_Info result;
 494         result.aux_file = file;
 495         scanAuxFile(file, result);
 496         return result;
 497 }
 498
 499
 500 void LaTeX::scanAuxFile(FileName const & file, Aux_Info & aux_info)
 501 {
 502         lyxerr[Debug::LATEX] << "Scanning aux file: " << file << endl;
 503
 504         ifstream ifs(file.toFilesystemEncoding().c_str());
 505         string token;
 506         static regex const reg1("\\\\citation\\{([^}]+)\\}");
 507         static regex const reg2("\\\\bibdata\\{([^}]+)\\}");
 508         static regex const reg3("\\\\bibstyle\\{([^}]+)\\}");
 509         static regex const reg4("\\\\@input\\{([^}]+)\\}");
 510
 511         while (getline(ifs, token)) {
 512                 token = rtrim(token, "\r");
 513                 smatch sub;
 514                 // FIXME UNICODE: We assume that citation keys and filenames
 515                 // in the aux file are in the file system encoding.
 516                 token = to_utf8(from_filesystem8bit(token));
 517                 if (regex_match(token, sub, reg1)) {
 518                         string data = sub.str(1);
 519                         while (!data.empty()) {
 520                                 string citation;
 521                                 data = split(data, citation, ',');
 522                                 lyxerr[Debug::LATEX] << "Citation: "
 523                                                      << citation << endl;
 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: `"
 535                                                      << database << '\'' << endl;
 536                                 aux_info.databases.insert(database);
 537                         }
 538                 } else if (regex_match(token, sub, reg3)) {
 539                         string style = sub.str(1);
 540                         // token is now the style file
 541                         // pass it to the helper
 542                         style = changeExtension(style, "bst");
 543                         lyxerr[Debug::LATEX] << "BibTeX style: `"
 544                                              << style << '\'' << endl;
 545                         aux_info.styles.insert(style);
 546                 } else if (regex_match(token, sub, reg4)) {
 547                         string const file2 = sub.str(1);
 548                         scanAuxFile(makeAbsPath(file2), aux_info);
 549                 }
 550         }
 551 }
 552
 553
 554 void LaTeX::updateBibtexDependencies(DepTable & dep,
 555                                      vector<Aux_Info> const & bibtex_info)
 556 {
 557         // Since a run of Bibtex mandates more latex runs it is ok to
 558         // remove all ".bib" and ".bst" files.
 559         dep.remove_files_with_extension(".bib");
 560         dep.remove_files_with_extension(".bst");
 561         //string aux = OnlyFilename(ChangeExtension(file, ".aux"));
 562
 563         for (vector<Aux_Info>::const_iterator it = bibtex_info.begin();
 564              it != bibtex_info.end(); ++it) {
 565                 for (set<string>::const_iterator it2 = it->databases.begin();
 566                      it2 != it->databases.end(); ++it2) {
 567                         FileName const file = findtexfile(*it2, "bib");
 568                         if (!file.empty())
 569                                 dep.insert(file, true);
 570                 }
 571
 572                 for (set<string>::const_iterator it2 = it->styles.begin();
 573                      it2 != it->styles.end(); ++it2) {
 574                         FileName const file = findtexfile(*it2, "bst");
 575                         if (!file.empty())
 576                                 dep.insert(file, true);
 577                 }
 578         }
 579 }
 580
 581
 582 bool LaTeX::runBibTeX(vector<Aux_Info> const & bibtex_info)
 583 {
 584         bool result = false;
 585         for (vector<Aux_Info>::const_iterator it = bibtex_info.begin();
 586              it != bibtex_info.end(); ++it) {
 587                 if (it->databases.empty())
 588                         continue;
 589                 result = true;
 590
 591                 string tmp = lyxrc.bibtex_command + " ";
 592                 // onlyFilename() is needed for cygwin
 593                 tmp += quoteName(onlyFilename(removeExtension(
 594                                 it->aux_file.absFilename())));
 595                 Systemcall one;
 596                 one.startscript(Systemcall::Wait, tmp);
 597         }
 598         // Return whether bibtex was run
 599         return result;
 600 }
 601
 602
 603 int LaTeX::scanLogFile(TeXErrors & terr)
 604 {
 605         int last_line = -1;
 606         int line_count = 1;
 607         int retval = NO_ERRORS;
 608         string tmp =
 609                 onlyFilename(changeExtension(file.absFilename(), ".log"));
 610         lyxerr[Debug::LATEX] << "Log file: " << tmp << endl;
 611         FileName const fn = FileName(makeAbsPath(tmp));
 612         ifstream ifs(fn.toFilesystemEncoding().c_str());
 613
 614         string token;
 615         while (getline(ifs, token)) {
 616                 // MikTeX sometimes inserts \0 in the log file. They can't be
 617                 // removed directly with the existing string utility
 618                 // functions, so convert them first to \r, and remove all
 619                 // \r's afterwards, since we need to remove them anyway.
 620                 token = subst(token, '\0', '\r');
 621                 token = subst(token, "\r", "");
 622
 623                 lyxerr[Debug::LATEX] << "Log line: " << token << endl;
 624
 625                 if (token.empty())
 626                         continue;
 627
 628                 if (prefixIs(token, "LaTeX Warning:") ||
 629                     prefixIs(token, "! pdfTeX warning")) {
 630                         // Here shall we handle different
 631                         // types of warnings
 632                         retval |= LATEX_WARNING;
 633                         lyxerr[Debug::LATEX] << "LaTeX Warning." << endl;
 634                         if (contains(token, "Rerun to get cross-references")) {
 635                                 retval |= RERUN;
 636                                 lyxerr[Debug::LATEX]
 637                                         << "We should rerun." << endl;
 638                         } else if (contains(token, "Citation")
 639                                    && contains(token, "on page")
 640                                    && contains(token, "undefined")) {
 641                                 retval |= UNDEF_CIT;
 642                         }
 643                 } else if (prefixIs(token, "Package")) {
 644                         // Package warnings
 645                         retval |= PACKAGE_WARNING;
 646                         if (contains(token, "natbib Warning:")) {
 647                                 // Natbib warnings
 648                                 if (contains(token, "Citation")
 649                                     && contains(token, "on page")
 650                                     && contains(token, "undefined")) {
 651                                         retval |= UNDEF_CIT;
 652                                 }
 653                         } else if (contains(token, "run BibTeX")) {
 654                                 retval |= UNDEF_CIT;
 655                         } else if (contains(token, "Rerun LaTeX") ||
 656                                    contains(token, "Rerun to get")) {
 657                                 // at least longtable.sty and bibtopic.sty
 658                                 // might use this.
 659                                 lyxerr[Debug::LATEX]
 660                                         << "We should rerun." << endl;
 661                                 retval |= RERUN;
 662                         }
 663                 } else if (token[0] == '(') {
 664                         if (contains(token, "Rerun LaTeX") ||
 665                             contains(token, "Rerun to get")) {
 666                                 // Used by natbib
 667                                 lyxerr[Debug::LATEX]
 668                                         << "We should rerun." << endl;
 669                                 retval |= RERUN;
 670                         }
 671                 } else if (prefixIs(token, "! ")) {
 672                         // Ok, we have something that looks like a TeX Error
 673                         // but what do we really have.
 674
 675                         // Just get the error description:
 676                         string desc(token, 2);
 677                         if (contains(token, "LaTeX Error:"))
 678                                 retval |= LATEX_ERROR;
 679                         // get the next line
 680                         string tmp;
 681                         int count = 0;
 682                         do {
 683                                 if (!getline(ifs, tmp))
 684                                         break;
 685                                 if (++count > 10)
 686                                         break;
 687                         } while (!prefixIs(tmp, "l."));
 688                         if (prefixIs(tmp, "l.")) {
 689                                 // we have a latex error
 690                                 retval |=  TEX_ERROR;
 691                                 if (contains(desc,
 692                                     "Package babel Error: You haven't defined the language"))
 693                                         retval |= ERROR_RERUN;
 694                                 // get the line number:
 695                                 int line = 0;
 696                                 sscanf(tmp.c_str(), "l.%d", &line);
 697                                 // get the rest of the message:
 698                                 string errstr(tmp, tmp.find(' '));
 699                                 errstr += '\n';
 700                                 getline(ifs, tmp);
 701                                 while (!contains(errstr, "l.")
 702                                        && !tmp.empty()
 703                                        && !prefixIs(tmp, "! ")
 704                                        && !contains(tmp, "(job aborted")) {
 705                                         errstr += tmp;
 706                                         errstr += "\n";
 707                                         getline(ifs, tmp);
 708                                 }
 709                                 lyxerr[Debug::LATEX]
 710                                         << "line: " << line << '\n'
 711                                         << "Desc: " << desc << '\n'
 712                                         << "Text: " << errstr << endl;
 713                                 if (line == last_line)
 714                                         ++line_count;
 715                                 else {
 716                                         line_count = 1;
 717                                         last_line = line;
 718                                 }
 719                                 if (line_count <= 5) {
 720                                         // FIXME UNICODE
 721                                         // We have no idea what the encoding of
 722                                         // the log file is.
 723                                         // It seems that the output from the
 724                                         // latex compiler itself is pure ASCII,
 725                                         // but it can include bits from the
 726                                         // document, so whatever encoding we
 727                                         // assume here it can be wrong.
 728                                         terr.insertError(line,
 729                                                          from_local8bit(desc),
 730                                                          from_local8bit(errstr));
 731                                         ++num_errors;
 732                                 }
 733                         }
 734                 } else {
 735                         // information messages, TeX warnings and other
 736                         // warnings we have not caught earlier.
 737                         if (prefixIs(token, "Overfull ")) {
 738                                 retval |= TEX_WARNING;
 739                         } else if (prefixIs(token, "Underfull ")) {
 740                                 retval |= TEX_WARNING;
 741                         } else if (contains(token, "Rerun to get citations")) {
 742                                 // Natbib seems to use this.
 743                                 retval |= UNDEF_CIT;
 744                         } else if (contains(token, "No pages of output")) {
 745                                 // A dvi file was not created
 746                                 retval |= NO_OUTPUT;
 747                         } else if (contains(token, "That makes 100 errors")) {
 748                                 // More than 100 errors were reprted
 749                                 retval |= TOO_MANY_ERRORS;
 750                         }
 751                 }
 752         }
 753         lyxerr[Debug::LATEX] << "Log line: " << token << endl;
 754         return retval;
 755 }
 756
 757
 758 namespace {
 759
 760 /**
 761  * Wrapper around fs::exists that can handle invalid file names.
 762  * In theory we could test with fs::native whether a filename is valid
 763  * before calling fs::exists, but in practice it is unusable: On windows it
 764  * does not allow spaces, and on unix it does not allow absolute file names.
 765  * This function has the disadvantage that it catches also other errors than
 766  * invalid names, but for dependency checking we can live with that.
 767  */
 768 bool exists(FileName const & possible_name) {
 769         try {
 770                 return fs::exists(possible_name.toFilesystemEncoding());
 771         }
 772         catch (fs::filesystem_error const & fe) {
 773                 lyxerr[Debug::DEPEND] << "Got error `" << fe.what()
 774                         << "' while checking whether file `" << possible_name
 775                         << "' exists." << endl;
 776                 return false;
 777         }
 778 }
 779
 780
 781 bool insertIfExists(FileName const & absname, DepTable & head)
 782 {
 783         if (exists(absname) &&
 784             !fs::is_directory(absname.toFilesystemEncoding())) {
 785                 head.insert(absname, true);
 786                 return true;
 787         }
 788         return false;
 789 }
 790
 791
 792 bool handleFoundFile(string const & ff, DepTable & head)
 793 {
 794         // convert from native os path to unix path
 795         string foundfile = os::internal_path(trim(ff));
 796
 797         lyxerr[Debug::DEPEND] << "Found file: " << foundfile << endl;
 798
 799         // Ok now we found a file.
 800         // Now we should make sure that this is a file that we can
 801         // access through the normal paths.
 802         // We will not try any fancy search methods to
 803         // find the file.
 804
 805         // (1) foundfile is an
 806         //     absolute path and should
 807         //     be inserted.
 808         if (absolutePath(foundfile)) {
 809                 lyxerr[Debug::DEPEND] << "AbsolutePath file: "
 810                                       << foundfile << endl;
 811                 // On initial insert we want to do the update at once
 812                 // since this file cannot be a file generated by
 813                 // the latex run.
 814                 FileName absname(foundfile);
 815                 if (!insertIfExists(absname, head)) {
 816                         // check for spaces
 817                         string strippedfile = foundfile;
 818                         while (contains(strippedfile, " ")) {
 819                                 // files with spaces are often enclosed in quotation
 820                                 // marks; those have to be removed
 821                                 string unquoted = subst(strippedfile, "\"", "");
 822                                 absname.set(unquoted);
 823                                 if (insertIfExists(absname, head))
 824                                         return true;
 825                                 // strip off part after last space and try again
 826                                 string tmp = strippedfile;
 827                                 string const stripoff =
 828                                         rsplit(tmp, strippedfile, ' ');
 829                                 absname.set(strippedfile);
 830                                 if (insertIfExists(absname, head))
 831                                         return true;
 832                         }
 833                 }
 834         }
 835
 836         string onlyfile = onlyFilename(foundfile);
 837         FileName absname(makeAbsPath(onlyfile));
 838
 839         // check for spaces
 840         while (contains(foundfile, ' ')) {
 841                 if (exists(absname))
 842                         // everything o.k.
 843                         break;
 844                 else {
 845                         // files with spaces are often enclosed in quotation
 846                         // marks; those have to be removed
 847                         string unquoted = subst(foundfile, "\"", "");
 848                         absname = makeAbsPath(unquoted);
 849                         if (exists(absname))
 850                                 break;
 851                         // strip off part after last space and try again
 852                         string strippedfile;
 853                         string const stripoff =
 854                                 rsplit(foundfile, strippedfile, ' ');
 855                         foundfile = strippedfile;
 856                         onlyfile = onlyFilename(strippedfile);
 857                         absname = makeAbsPath(onlyfile);
 858                 }
 859         }
 860
 861         // (2) foundfile is in the tmpdir
 862         //     insert it into head
 863         if (exists(absname) &&
 864             !fs::is_directory(absname.toFilesystemEncoding())) {
 865                 static regex unwanted("^.*\\.(aux|log|dvi|bbl|ind|glo)$");
 866                 if (regex_match(onlyfile, unwanted)) {
 867                         lyxerr[Debug::DEPEND]
 868                                 << "We don't want "
 869                                 << onlyfile
 870                                 << " in the dep file"
 871                                 << endl;
 872                 } else if (suffixIs(onlyfile, ".tex")) {
 873                         // This is a tex file generated by LyX
 874                         // and latex is not likely to change this
 875                         // during its runs.
 876                         lyxerr[Debug::DEPEND]
 877                                 << "Tmpdir TeX file: "
 878                                 << onlyfile
 879                                 << endl;
 880                         head.insert(absname, true);
 881                 } else {
 882                         lyxerr[Debug::DEPEND]
 883                                 << "In tmpdir file:"
 884                                 << onlyfile
 885                                 << endl;
 886                         head.insert(absname);
 887                 }
 888                 return true;
 889         } else {
 890                 lyxerr[Debug::DEPEND]
 891                         << "Not a file or we are unable to find it."
 892                         << endl;
 893                 return false;
 894         }
 895 }
 896
 897
 898 bool checkLineBreak(string const & ff, DepTable & head)
 899 {
 900         if (contains(ff, '.'))
 901                 // if we have a dot, we let handleFoundFile decide
 902                 return handleFoundFile(ff, head);
 903         else
 904                 // else, we suspect a line break
 905                 return false;
 906 }
 907
 908 } // anon namespace
 909
 910
 911 void LaTeX::deplog(DepTable & head)
 912 {
 913         // This function reads the LaTeX log file end extracts all the
 914         // external files used by the LaTeX run. The files are then
 915         // entered into the dependency file.
 916
 917         string const logfile =
 918                 onlyFilename(changeExtension(file.absFilename(), ".log"));
 919
 920         static regex reg1("File: (.+).*");
 921         static regex reg2("No file (.+)(.).*");
 922         static regex reg3("\\\\openout[0-9]+.*=.*`(.+)(..).*");
 923         // If an index should be created, MikTex does not write a line like
 924         //    \openout# = 'sample.idx'.
 925         // but instead only a line like this into the log:
 926         //   Writing index file sample.idx
 927         static regex reg4("Writing index file (.+).*");
 928         // files also can be enclosed in <...>
 929         static regex reg5("<([^>]+)(.).*");
 930         static regex regnomencl("Writing nomenclature file (.+).*");
 931         // If a toc should be created, MikTex does not write a line like
 932         //    \openout# = `sample.toc'.
 933         // but only a line like this into the log:
 934         //    \tf@toc=\write#
 935         // This line is also written by tetex.
 936         // This line is not present if no toc should be created.
 937         static regex miktexTocReg("\\\\tf@toc=\\\\write.*");
 938         static regex reg6(".*\\([^)]+.*");
 939
 940         FileName const fn(makeAbsPath(logfile));
 941         ifstream ifs(fn.toFilesystemEncoding().c_str());
 942         string lastline;
 943         while (ifs) {
 944                 // Ok, the scanning of files here is not sufficient.
 945                 // Sometimes files are named by "File: xxx" only
 946                 // So I think we should use some regexps to find files instead.
 947                 // Note: all file names and paths might contains spaces.
 948                 bool found_file = false;
 949                 string token;
 950                 getline(ifs, token);
 951                 // MikTeX sometimes inserts \0 in the log file. They can't be
 952                 // removed directly with the existing string utility
 953                 // functions, so convert them first to \r, and remove all
 954                 // \r's afterwards, since we need to remove them anyway.
 955                 token = subst(token, '\0', '\r');
 956                 token = subst(token, "\r", "");
 957                 if (token.empty() || token == ")") {
 958                         lastline = string();
 959                         continue;
 960                 }
 961
 962                 // Sometimes, filenames are broken across lines.
 963                 // We care for that and save suspicious lines.
 964                 // Here we exclude some cases where we are sure
 965                 // that there is no continued filename
 966                 if (!lastline.empty()) {
 967                         static regex package_info("Package \\w+ Info: .*");
 968                         static regex package_warning("Package \\w+ Warning: .*");
 969                         if (prefixIs(token, "File:") || prefixIs(token, "(Font)")
 970                             || prefixIs(token, "Package:")
 971                             || prefixIs(token, "Language:")
 972                             || prefixIs(token, "LaTeX Info:")
 973                             || prefixIs(token, "LaTeX Font Info:")
 974                             || prefixIs(token, "\\openout[")
 975                             || prefixIs(token, "))")
 976                             || regex_match(token, package_info)
 977                             || regex_match(token, package_warning))
 978                                 lastline = string();
 979                 }
 980
 981                 if (!lastline.empty())
 982                         // probably a continued filename from last line
 983                         token = lastline + token;
 984                 if (token.length() > 255) {
 985                         // string too long. Cut off.
 986                         token.erase(0, token.length() - 251);
 987                 }
 988
 989                 smatch sub;
 990
 991                 // FIXME UNICODE: We assume that the file names in the log
 992                 // file are in the file system encoding.
 993                 token = to_utf8(from_filesystem8bit(token));
 994
 995                 // (1) "File: file.ext"
 996                 if (regex_match(token, sub, reg1)) {
 997                         // check for dot
 998                         found_file = checkLineBreak(sub.str(1), head);
 999                         // However, ...
1000                         if (suffixIs(token, ")"))
1001                                 // no line break for sure
1002                                 // pretend we've been succesfully searching
1003                                 found_file = true;
1004                 // (2) "No file file.ext"
1005                 } else if (regex_match(token, sub, reg2)) {
1006                         // file names must contains a dot, line ends with dot
1007                         if (contains(sub.str(1), '.') && sub.str(2) == ".")
1008                                 found_file = handleFoundFile(sub.str(1), head);
1009                         else
1010                                 // we suspect a line break
1011                                 found_file = false;
1012                 // (3) "\openout<nr> = `file.ext'."
1013                 } else if (regex_match(token, sub, reg3)) {
1014                         // search for closing '. at the end of the line
1015                         if (sub.str(2) == "\'.")
1016                                 found_file = handleFoundFile(sub.str(1), head);
1017                         else
1018                                 // probable line break
1019                                 found_file = false;
1020                 // (4) "Writing index file file.ext"
1021                 } else if (regex_match(token, sub, reg4))
1022                         // check for dot
1023                         found_file = checkLineBreak(sub.str(1), head);
1024                 // (5) "<file.ext>"
1025                 else if (regex_match(token, sub, reg5)) {
1026                         // search for closing '>' and dot ('*.*>') at the eol
1027                         if (contains(sub.str(1), '.') && sub.str(2) == ">")
1028                                 found_file = handleFoundFile(sub.str(1), head);
1029                         else
1030                                 // probable line break
1031                                 found_file = false;
1032                 // (6) "Writing nomenclature file file.ext"
1033                 } else if (regex_match(token, sub, regnomencl))
1034                         // check for dot
1035                         found_file = checkLineBreak(sub.str(1), head);
1036                 // (7) "\tf@toc=\write<nr>" (for MikTeX)
1037                 else if (regex_match(token, sub, miktexTocReg))
1038                         found_file = handleFoundFile(onlyFilename(changeExtension(
1039                                                 file.absFilename(), ".toc")), head);
1040                 else
1041                         // not found, but we won't check further
1042                         // pretend we've been succesfully searching
1043                         found_file = true;
1044
1045                 // (8) "(file.ext"
1046                 // note that we can have several of these on one line
1047                 // this must be queried separated, because of
1048                 // cases such as "File: file.ext (type eps)"
1049                 // where "File: file.ext" would be skipped
1050                 if (regex_match(token, sub, reg6)) {
1051                         // search for strings in (...)
1052                         static regex reg6_1("\\(([^()]+)(.).*");
1053                         smatch what;
1054                         string::const_iterator first = token.begin();
1055                         string::const_iterator end = token.end();
1056
1057                         while (regex_search(first, end, what, reg6_1)) {
1058                                 // if we have a dot, try to handle as file
1059                                 if (contains(what.str(1), '.')) {
1060                                         first = what[0].second;
1061                                         if (what.str(2) == ")") {
1062                                                 handleFoundFile(what.str(1), head);
1063                                                 // since we had a closing bracket,
1064                                                 // do not investigate further
1065                                                 found_file = true;
1066                                         } else
1067                                                 // if we have no closing bracket,
1068                                                 // try to handle as file nevertheless
1069                                                 found_file = handleFoundFile(
1070                                                         what.str(1) + what.str(2), head);
1071                                 }
1072                                 // if we do not have a dot, check if the line has
1073                                 // a closing bracket (else, we suspect a line break)
1074                                 else if (what.str(2) != ")") {
1075                                         first = what[0].second;
1076                                         found_file = false;
1077                                 } else {
1078                                         // we have a closing bracket, so the content
1079                                         // is not a file name.
1080                                         // no need to investigate further
1081                                         // pretend we've been succesfully searching
1082                                         first = what[0].second;
1083                                         found_file = true;
1084                                 }
1085                         }
1086                 }
1087
1088                 if (!found_file)
1089                         // probable linebreak:
1090                         // save this line
1091                         lastline = token;
1092                 else
1093                         // no linebreak: reset
1094                         lastline = string();
1095         }
1096
1097         // Make sure that the main .tex file is in the dependency file.
1098         head.insert(file, true);
1099 }
1100
1101
1102 } // namespace lyx