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  *
  12  * Full author contact details are available in file CREDITS.
  13  */
  14
  15 #include <config.h>
  16
  17 #include "bufferlist.h"
  18 #include "LaTeX.h"
  19 #include "gettext.h"
  20 #include "lyxrc.h"
  21 #include "debug.h"
  22 #include "DepTable.h"
  23
  24 #include "support/filetools.h"
  25 #include "support/convert.h"
  26 #include "support/lstrings.h"
  27 #include "support/lyxlib.h"
  28 #include "support/systemcall.h"
  29 #include "support/os.h"
  30
  31 #include <boost/filesystem/operations.hpp>
  32 #include <boost/regex.hpp>
  33
  34 #include <fstream>
  35
  36
  37 namespace lyx {
  38
  39 using support::absolutePath;
  40 using support::bformat;
  41 using support::changeExtension;
  42 using support::contains;
  43 using support::FileName;
  44 using support::findtexfile;
  45 using support::getcwd;
  46 using support::makeAbsPath;
  47 using support::onlyFilename;
  48 using support::prefixIs;
  49 using support::quoteName;
  50 using support::removeExtension;
  51 using support::rtrim;
  52 using support::rsplit;
  53 using support::split;
  54 using support::subst;
  55 using support::suffixIs;
  56 using support::Systemcall;
  57 using support::unlink;
  58 using support::trim;
  59
  60 namespace os = support::os;
  61 namespace fs = boost::filesystem;
  62
  63 using boost::regex;
  64 using boost::smatch;
  65
  66
  67 #ifndef CXX_GLOBAL_CSTD
  68 using std::sscanf;
  69 #endif
  70
  71 using std::endl;
  72 using std::getline;
  73 using std::string;
  74 using std::ifstream;
  75 using std::set;
  76 using std::vector;
  77
  78 // TODO: in no particular order
  79 // - get rid of the call to
  80 //   BufferList::updateIncludedTeXfiles, this should either
  81 //   be done before calling LaTeX::funcs or in a completely
  82 //   different way.
  83 // - the makeindex style files should be taken care of with
  84 //   the dependency mechanism.
  85 // - makeindex commandline options should be supported
  86 // - somewhere support viewing of bibtex and makeindex log files.
  87 // - we should perhaps also scan the bibtex log file
  88
  89 namespace {
  90
  91 docstring runMessage(unsigned int count)
  92 {
  93         return bformat(_("Waiting for LaTeX run number %1$d"), count);
  94 }
  95
  96 } // anon namespace
  97
  98 /*
  99  * CLASS TEXERRORS
 100  */
 101
 102 void TeXErrors::insertError(int line, docstring const & error_desc,
 103                             docstring const & error_text)
 104 {
 105         Error newerr(line, error_desc, error_text);
 106         errors.push_back(newerr);
 107 }
 108
 109
 110 bool operator==(Aux_Info const & a, Aux_Info const & o)
 111 {
 112         return a.aux_file == o.aux_file &&
 113                 a.citations == o.citations &&
 114                 a.databases == o.databases &&
 115                 a.styles == o.styles;
 116 }
 117
 118
 119 bool operator!=(Aux_Info const & a, Aux_Info const & o)
 120 {
 121         return !(a == o);
 122 }
 123
 124
 125 /*
 126  * CLASS LaTeX
 127  */
 128
 129 LaTeX::LaTeX(string const & latex, OutputParams const & rp,
 130              FileName const & f)
 131         : cmd(latex), file(f), runparams(rp)
 132 {
 133         num_errors = 0;
 134         if (prefixIs(cmd, "pdf")) { // Do we use pdflatex ?
 135                 depfile = FileName(file.absFilename() + ".dep-pdf");
 136                 output_file = FileName(changeExtension(file.absFilename(), ".pdf"));
 137         } else {
 138                 depfile = FileName(file.absFilename() + ".dep");
 139                 output_file = FileName(changeExtension(file.absFilename(), ".dvi"));
 140         }
 141 }
 142
 143
 144 void LaTeX::deleteFilesOnError() const
 145 {
 146         // currently just a dummy function.
 147
 148         // What files do we have to delete?
 149
 150         // This will at least make latex do all the runs
 151         unlink(depfile);
 152
 153         // but the reason for the error might be in a generated file...
 154
 155         // bibtex file
 156         FileName const bbl(changeExtension(file.absFilename(), ".bbl"));
 157         unlink(bbl);
 158
 159         // makeindex file
 160         FileName const ind(changeExtension(file.absFilename(), ".ind"));
 161         unlink(ind);
 162
 163         // nomencl file
 164         FileName const nls(changeExtension(file.absFilename(), ".nls"));
 165         unlink(nls);
 166
 167         // Also remove the aux file
 168         FileName const aux(changeExtension(file.absFilename(), ".aux"));
 169         unlink(aux);
 170 }
 171
 172
 173 int LaTeX::run(TeXErrors & terr)
 174         // We know that this function will only be run if the lyx buffer
 175         // has been changed. We also know that a newly written .tex file
 176         // is always different from the previous one because of the date
 177         // in it. However it seems safe to run latex (at least) on time each
 178         // time the .tex file changes.
 179 {
 180         int scanres = NO_ERRORS;
 181         unsigned int count = 0; // number of times run
 182         num_errors = 0; // just to make sure.
 183         unsigned int const MAX_RUN = 6;
 184         DepTable head; // empty head
 185         bool rerun = false; // rerun requested
 186
 187         // The class LaTeX does not know the temp path.
 188         theBufferList().updateIncludedTeXfiles(getcwd().absFilename(), runparams);
 189
 190         // Never write the depfile if an error was encountered.
 191
 192         // 0
 193         // first check if the file dependencies exist:
 194         //     ->If it does exist
 195         //             check if any of the files mentioned in it have
 196         //             changed (done using a checksum).
 197         //                 -> if changed:
 198         //                        run latex once and
 199         //                        remake the dependency file
 200         //                 -> if not changed:
 201         //                        just return there is nothing to do for us.
 202         //     ->if it doesn't exist
 203         //             make it and
 204         //             run latex once (we need to run latex once anyway) and
 205         //             remake the dependency file.
 206         //
 207
 208         bool had_depfile = fs::exists(depfile.toFilesystemEncoding());
 209         bool run_bibtex = false;
 210         FileName const aux_file(changeExtension(file.absFilename(), "aux"));
 211
 212         if (had_depfile) {
 213                 lyxerr[Debug::DEPEND] << "Dependency file exists" << endl;
 214                 // Read the dep file:
 215                 had_depfile = head.read(depfile);
 216         }
 217
 218         if (had_depfile) {
 219                 // Update the checksums
 220                 head.update();
 221                 // Can't just check if anything has changed because it might have aborted
 222                 // on error last time... in which cas we need to re-run latex
 223                 // and collect the error messages (even if they are the same).
 224                 if (!fs::exists(output_file.toFilesystemEncoding())) {
 225                         lyxerr[Debug::DEPEND]
 226                                 << "re-running LaTeX because output file doesn't exist." << endl;
 227                 } else if (!head.sumchange()) {
 228                         lyxerr[Debug::DEPEND] << "return no_change" << endl;
 229                         return NO_CHANGE;
 230                 } else {
 231                         lyxerr[Debug::DEPEND]
 232                                 << "Dependency file has changed" << endl;
 233                 }
 234
 235                 if (head.extchanged(".bib") || head.extchanged(".bst"))
 236                         run_bibtex = true;
 237         } else
 238                 lyxerr[Debug::DEPEND]
 239                         << "Dependency file does not exist, or has wrong format" << endl;
 240
 241         /// We scan the aux file even when had_depfile = false,
 242         /// because we can run pdflatex on the file after running latex on it,
 243         /// in which case we will not need to run bibtex again.
 244         vector<Aux_Info> bibtex_info_old;
 245         if (!run_bibtex)
 246                 bibtex_info_old = scanAuxFiles(aux_file);
 247
 248         ++count;
 249         lyxerr[Debug::LATEX] << "Run #" << count << endl;
 250         message(runMessage(count));
 251
 252         startscript();
 253         scanres = scanLogFile(terr);
 254         if (scanres & ERROR_RERUN) {
 255                 lyxerr[Debug::LATEX] << "Rerunning LaTeX" << endl;
 256                 startscript();
 257                 scanres = scanLogFile(terr);
 258         }
 259
 260         if (scanres & ERRORS) {
 261                 deleteFilesOnError();
 262                 return scanres; // return on error
 263         }
 264
 265         vector<Aux_Info> const bibtex_info = scanAuxFiles(aux_file);
 266         if (!run_bibtex && bibtex_info_old != bibtex_info)
 267                 run_bibtex = true;
 268
 269         // update the dependencies.
 270         deplog(head); // reads the latex log
 271         head.update();
 272
 273         // 0.5
 274         // At this point we must run external programs if needed.
 275         // makeindex will be run if a .idx file changed or was generated.
 276         // And if there were undefined citations or changes in references
 277         // the .aux file is checked for signs of bibtex. Bibtex is then run
 278         // if needed.
 279
 280         // memoir (at least) writes an empty *idx file in the first place.
 281         // A second latex run is needed.
 282         FileName const idxfile(changeExtension(file.absFilename(), ".idx"));
 283         rerun = fs::exists(idxfile.toFilesystemEncoding()) &&
 284                 fs::is_empty(idxfile.toFilesystemEncoding());
 285
 286         // run makeindex
 287         if (head.haschanged(idxfile)) {
 288                 // no checks for now
 289                 lyxerr[Debug::LATEX] << "Running MakeIndex." << endl;
 290                 message(_("Running MakeIndex."));
 291                 // onlyFilename() is needed for cygwin
 292                 rerun |= runMakeIndex(onlyFilename(idxfile.absFilename()), runparams);
 293         }
 294         if (head.haschanged(FileName(changeExtension(file.absFilename(), ".nlo")))) {
 295                 lyxerr[Debug::LATEX] << "Running MakeIndex for nomencl." << endl;
 296                 message(_("Running MakeIndex for nomencl."));
 297                 // onlyFilename() is needed for cygwin
 298                 string const nomenclstr = " -s nomencl.ist -o " + onlyFilename(changeExtension(file.toFilesystemEncoding(), ".nls"));
 299                 rerun |= runMakeIndex(onlyFilename(changeExtension(file.absFilename(), ".nlo")), runparams, nomenclstr);
 300         }
 301
 302         // run bibtex
 303         // if (scanres & UNDEF_CIT || scanres & RERUN || run_bibtex)
 304         if (scanres & UNDEF_CIT || run_bibtex) {
 305                 // Here we must scan the .aux file and look for
 306                 // "\bibdata" and/or "\bibstyle". If one of those
 307                 // tags is found -> run bibtex and set rerun = true;
 308                 // no checks for now
 309                 lyxerr[Debug::LATEX] << "Running BibTeX." << endl;
 310                 message(_("Running BibTeX."));
 311                 updateBibtexDependencies(head, bibtex_info);
 312                 rerun |= runBibTeX(bibtex_info);
 313         } else if (!had_depfile) {
 314                 /// If we run pdflatex on the file after running latex on it,
 315                 /// then we do not need to run bibtex, but we do need to
 316                 /// insert the .bib and .bst files into the .dep-pdf file.
 317                 updateBibtexDependencies(head, bibtex_info);
 318         }
 319
 320         // 1
 321         // we know on this point that latex has been run once (or we just
 322         // returned) and the question now is to decide if we need to run
 323         // it any more. This is done by asking if any of the files in the
 324         // dependency file has changed. (remember that the checksum for
 325         // a given file is reported to have changed if it just was created)
 326         //     -> if changed or rerun == true:
 327         //             run latex once more and
 328         //             update the dependency structure
 329         //     -> if not changed:
 330         //             we does nothing at this point
 331         //
 332         if (rerun || head.sumchange()) {
 333                 rerun = false;
 334                 ++count;
 335                 lyxerr[Debug::DEPEND]
 336                         << "Dep. file has changed or rerun requested" << endl;
 337                 lyxerr[Debug::LATEX]
 338                         << "Run #" << count << endl;
 339                 message(runMessage(count));
 340                 startscript();
 341                 scanres = scanLogFile(terr);
 342                 if (scanres & ERRORS) {
 343                         deleteFilesOnError();
 344                         return scanres; // return on error
 345                 }
 346
 347                 // update the depedencies
 348                 deplog(head); // reads the latex log
 349                 head.update();
 350         } else {
 351                 lyxerr[Debug::DEPEND] << "Dep. file has NOT changed" << endl;
 352         }
 353
 354         // 1.5
 355         // The inclusion of files generated by external programs like
 356         // makeindex or bibtex might have done changes to pagenumbering,
 357         // etc. And because of this we must run the external programs
 358         // again to make sure everything is redone correctly.
 359         // Also there should be no need to run the external programs any
 360         // more after this.
 361
 362         // run makeindex if the <file>.idx has changed or was generated.
 363         if (head.haschanged(FileName(changeExtension(file.absFilename(), ".idx")))) {
 364                 // no checks for now
 365                 lyxerr[Debug::LATEX] << "Running MakeIndex." << endl;
 366                 message(_("Running MakeIndex."));
 367                 // onlyFilename() is needed for cygwin
 368                 rerun = runMakeIndex(onlyFilename(changeExtension(file.absFilename(), ".idx")), runparams);
 369         }
 370
 371         // I am not pretty sure if need this twice.
 372         if (head.haschanged(FileName(changeExtension(file.absFilename(), ".nlo")))) {
 373                 lyxerr[Debug::LATEX] << "Running MakeIndex for nomencl." << endl;
 374                 message(_("Running MakeIndex for nomencl."));
 375                 // onlyFilename() is needed for cygwin
 376                 string nomenclstr = " -s nomencl.ist -o " + onlyFilename(changeExtension(file.toFilesystemEncoding(), ".nls"));
 377                 rerun |= runMakeIndex(onlyFilename(changeExtension(file.absFilename(), ".nlo")), runparams, nomenclstr);
 378         }
 379
 380         // 2
 381         // we will only run latex more if the log file asks for it.
 382         // or if the sumchange() is true.
 383         //     -> rerun asked for:
 384         //             run latex and
 385         //             remake the dependency file
 386         //             goto 2 or return if max runs are reached.
 387         //     -> rerun not asked for:
 388         //             just return (fall out of bottom of func)
 389         //
 390         while ((head.sumchange() || rerun || (scanres & RERUN))
 391                && count < MAX_RUN) {
 392                 // Yes rerun until message goes away, or until
 393                 // MAX_RUNS are reached.
 394                 rerun = false;
 395                 ++count;
 396                 lyxerr[Debug::LATEX] << "Run #" << count << endl;
 397                 message(runMessage(count));
 398                 startscript();
 399                 scanres = scanLogFile(terr);
 400                 if (scanres & ERRORS) {
 401                         deleteFilesOnError();
 402                         return scanres; // return on error
 403                 }
 404
 405                 // keep this updated
 406                 head.update();
 407         }
 408
 409         // Write the dependencies to file.
 410         head.write(depfile);
 411         lyxerr[Debug::LATEX] << "Done." << endl;
 412         return scanres;
 413 }
 414
 415
 416 int LaTeX::startscript()
 417 {
 418         // onlyFilename() is needed for cygwin
 419         string tmp = cmd + ' ' + quoteName(onlyFilename(file.toFilesystemEncoding())) + " > " + os::nulldev();
 420         Systemcall one;
 421         return one.startscript(Systemcall::Wait, tmp);
 422 }
 423
 424
 425 bool LaTeX::runMakeIndex(string const & f, OutputParams const & runparams,
 426                          string const & params)
 427 {
 428         lyxerr[Debug::LATEX]
 429                 << "idx file has been made, running makeindex on file "
 430                 << f << endl;
 431         string tmp = lyxrc.index_command + ' ';
 432
 433         tmp = subst(tmp, "$$lang", runparams.document_language);
 434         tmp += quoteName(f);
 435         tmp += params;
 436         Systemcall one;
 437         one.startscript(Systemcall::Wait, tmp);
 438         return true;
 439 }
 440
 441
 442 vector<Aux_Info> const
 443 LaTeX::scanAuxFiles(FileName const & file)
 444 {
 445         vector<Aux_Info> result;
 446
 447         result.push_back(scanAuxFile(file));
 448
 449         string const basename = removeExtension(file.absFilename());
 450         for (int i = 1; i < 1000; ++i) {
 451                 FileName const file2(basename
 452                         + '.' + convert<string>(i)
 453                         + ".aux");
 454                 if (!fs::exists(file2.toFilesystemEncoding()))
 455                         break;
 456                 result.push_back(scanAuxFile(file2));
 457         }
 458         return result;
 459 }
 460
 461
 462 Aux_Info const LaTeX::scanAuxFile(FileName const & file)
 463 {
 464         Aux_Info result;
 465         result.aux_file = file;
 466         scanAuxFile(file, result);
 467         return result;
 468 }
 469
 470
 471 void LaTeX::scanAuxFile(FileName const & file, Aux_Info & aux_info)
 472 {
 473         lyxerr[Debug::LATEX] << "Scanning aux file: " << file << endl;
 474
 475         ifstream ifs(file.toFilesystemEncoding().c_str());
 476         string token;
 477         static regex const reg1("\\\\citation\\{([^}]+)\\}");
 478         static regex const reg2("\\\\bibdata\\{([^}]+)\\}");
 479         static regex const reg3("\\\\bibstyle\\{([^}]+)\\}");
 480         static regex const reg4("\\\\@input\\{([^}]+)\\}");
 481
 482         while (getline(ifs, token)) {
 483                 token = rtrim(token, "\r");
 484                 smatch sub;
 485                 // FIXME UNICODE: We assume that citation keys and filenames
 486                 // in the aux file are in the file system encoding.
 487                 token = to_utf8(from_filesystem8bit(token));
 488                 if (regex_match(token, sub, reg1)) {
 489                         string data = sub.str(1);
 490                         while (!data.empty()) {
 491                                 string citation;
 492                                 data = split(data, citation, ',');
 493                                 lyxerr[Debug::LATEX] << "Citation: "
 494                                                      << citation << endl;
 495                                 aux_info.citations.insert(citation);
 496                         }
 497                 } else if (regex_match(token, sub, reg2)) {
 498                         string data = sub.str(1);
 499                         // data is now all the bib files separated by ','
 500                         // get them one by one and pass them to the helper
 501                         while (!data.empty()) {
 502                                 string database;
 503                                 data = split(data, database, ',');
 504                                 database = changeExtension(database, "bib");
 505                                 lyxerr[Debug::LATEX] << "BibTeX database: `"
 506                                                      << database << '\'' << endl;
 507                                 aux_info.databases.insert(database);
 508                         }
 509                 } else if (regex_match(token, sub, reg3)) {
 510                         string style = sub.str(1);
 511                         // token is now the style file
 512                         // pass it to the helper
 513                         style = changeExtension(style, "bst");
 514                         lyxerr[Debug::LATEX] << "BibTeX style: `"
 515                                              << style << '\'' << endl;
 516                         aux_info.styles.insert(style);
 517                 } else if (regex_match(token, sub, reg4)) {
 518                         string const file2 = sub.str(1);
 519                         scanAuxFile(makeAbsPath(file2), aux_info);
 520                 }
 521         }
 522 }
 523
 524
 525 void LaTeX::updateBibtexDependencies(DepTable & dep,
 526                                      vector<Aux_Info> const & bibtex_info)
 527 {
 528         // Since a run of Bibtex mandates more latex runs it is ok to
 529         // remove all ".bib" and ".bst" files.
 530         dep.remove_files_with_extension(".bib");
 531         dep.remove_files_with_extension(".bst");
 532         //string aux = OnlyFilename(ChangeExtension(file, ".aux"));
 533
 534         for (vector<Aux_Info>::const_iterator it = bibtex_info.begin();
 535              it != bibtex_info.end(); ++it) {
 536                 for (set<string>::const_iterator it2 = it->databases.begin();
 537                      it2 != it->databases.end(); ++it2) {
 538                         FileName const file = findtexfile(*it2, "bib");
 539                         if (!file.empty())
 540                                 dep.insert(file, true);
 541                 }
 542
 543                 for (set<string>::const_iterator it2 = it->styles.begin();
 544                      it2 != it->styles.end(); ++it2) {
 545                         FileName const file = findtexfile(*it2, "bst");
 546                         if (!file.empty())
 547                                 dep.insert(file, true);
 548                 }
 549         }
 550 }
 551
 552
 553 bool LaTeX::runBibTeX(vector<Aux_Info> const & bibtex_info)
 554 {
 555         bool result = false;
 556         for (vector<Aux_Info>::const_iterator it = bibtex_info.begin();
 557              it != bibtex_info.end(); ++it) {
 558                 if (it->databases.empty())
 559                         continue;
 560                 result = true;
 561
 562                 string tmp = lyxrc.bibtex_command + " ";
 563                 // onlyFilename() is needed for cygwin
 564                 tmp += quoteName(onlyFilename(removeExtension(it->aux_file.absFilename())));
 565                 Systemcall one;
 566                 one.startscript(Systemcall::Wait, tmp);
 567         }
 568         // Return whether bibtex was run
 569         return result;
 570 }
 571
 572
 573 int LaTeX::scanLogFile(TeXErrors & terr)
 574 {
 575         int last_line = -1;
 576         int line_count = 1;
 577         int retval = NO_ERRORS;
 578         string tmp = onlyFilename(changeExtension(file.absFilename(), ".log"));
 579         lyxerr[Debug::LATEX] << "Log file: " << tmp << endl;
 580         FileName const fn = FileName(makeAbsPath(tmp));
 581         ifstream ifs(fn.toFilesystemEncoding().c_str());
 582
 583         string token;
 584         while (getline(ifs, token)) {
 585                 // MikTeX sometimes inserts \0 in the log file. They can't be
 586                 // removed directly with the existing string utility
 587                 // functions, so convert them first to \r, and remove all
 588                 // \r's afterwards, since we need to remove them anyway.
 589                 token = subst(token, '\0', '\r');
 590                 token = subst(token, "\r", "");
 591
 592                 lyxerr[Debug::LATEX] << "Log line: " << token << endl;
 593
 594                 if (token.empty())
 595                         continue;
 596
 597                 if (prefixIs(token, "LaTeX Warning:") ||
 598                     prefixIs(token, "! pdfTeX warning")) {
 599                         // Here shall we handle different
 600                         // types of warnings
 601                         retval |= LATEX_WARNING;
 602                         lyxerr[Debug::LATEX] << "LaTeX Warning." << endl;
 603                         if (contains(token, "Rerun to get cross-references")) {
 604                                 retval |= RERUN;
 605                                 lyxerr[Debug::LATEX]
 606                                         << "We should rerun." << endl;
 607                         } else if (contains(token, "Citation")
 608                                    && contains(token, "on page")
 609                                    && contains(token, "undefined")) {
 610                                 retval |= UNDEF_CIT;
 611                         }
 612                 } else if (prefixIs(token, "Package")) {
 613                         // Package warnings
 614                         retval |= PACKAGE_WARNING;
 615                         if (contains(token, "natbib Warning:")) {
 616                                 // Natbib warnings
 617                                 if (contains(token, "Citation")
 618                                     && contains(token, "on page")
 619                                     && contains(token, "undefined")) {
 620                                         retval |= UNDEF_CIT;
 621                                 }
 622                         } else if (contains(token, "run BibTeX")) {
 623                                 retval |= UNDEF_CIT;
 624                         } else if (contains(token, "Rerun LaTeX") ||
 625                                    contains(token, "Rerun to get")) {
 626                                 // at least longtable.sty and bibtopic.sty
 627                                 // might use this.
 628                                 lyxerr[Debug::LATEX]
 629                                         << "We should rerun." << endl;
 630                                 retval |= RERUN;
 631                         }
 632                 } else if (token[0] == '(') {
 633                         if (contains(token, "Rerun LaTeX") ||
 634                             contains(token, "Rerun to get")) {
 635                                 // Used by natbib
 636                                 lyxerr[Debug::LATEX]
 637                                         << "We should rerun." << endl;
 638                                 retval |= RERUN;
 639                         }
 640                 } else if (prefixIs(token, "! ")) {
 641                         // Ok, we have something that looks like a TeX Error
 642                         // but what do we really have.
 643
 644                         // Just get the error description:
 645                         string desc(token, 2);
 646                         if (contains(token, "LaTeX Error:"))
 647                                 retval |= LATEX_ERROR;
 648                         // get the next line
 649                         string tmp;
 650                         int count = 0;
 651                         do {
 652                                 if (!getline(ifs, tmp))
 653                                         break;
 654                                 if (++count > 10)
 655                                         break;
 656                         } while (!prefixIs(tmp, "l."));
 657                         if (prefixIs(tmp, "l.")) {
 658                                 // we have a latex error
 659                                 retval |=  TEX_ERROR;
 660                                 if (contains(desc, "Package babel Error: You haven't defined the language"))
 661                                         retval |= ERROR_RERUN;
 662                                 // get the line number:
 663                                 int line = 0;
 664                                 sscanf(tmp.c_str(), "l.%d", &line);
 665                                 // get the rest of the message:
 666                                 string errstr(tmp, tmp.find(' '));
 667                                 errstr += '\n';
 668                                 getline(ifs, tmp);
 669                                 while (!contains(errstr, "l.")
 670                                        && !tmp.empty()
 671                                        && !prefixIs(tmp, "! ")
 672                                        && !contains(tmp, "(job aborted")) {
 673                                         errstr += tmp;
 674                                         errstr += "\n";
 675                                         getline(ifs, tmp);
 676                                 }
 677                                 lyxerr[Debug::LATEX]
 678                                         << "line: " << line << '\n'
 679                                         << "Desc: " << desc << '\n'
 680                                         << "Text: " << errstr << endl;
 681                                 if (line == last_line)
 682                                         ++line_count;
 683                                 else {
 684                                         line_count = 1;
 685                                         last_line = line;
 686                                 }
 687                                 if (line_count <= 5) {
 688                                         // FIXME UNICODE
 689                                         // We have no idea what the encoding of
 690                                         // the log file is.
 691                                         // It seems that the output from the
 692                                         // latex compiler itself is pure ASCII,
 693                                         // but it can include bits from the
 694                                         // document, so whatever encoding we
 695                                         // assume here it can be wrong.
 696                                         terr.insertError(line, from_local8bit(desc), from_local8bit(errstr));
 697                                         ++num_errors;
 698                                 }
 699                         }
 700                 } else {
 701                         // information messages, TeX warnings and other
 702                         // warnings we have not caught earlier.
 703                         if (prefixIs(token, "Overfull ")) {
 704                                 retval |= TEX_WARNING;
 705                         } else if (prefixIs(token, "Underfull ")) {
 706                                 retval |= TEX_WARNING;
 707                         } else if (contains(token, "Rerun to get citations")) {
 708                                 // Natbib seems to use this.
 709                                 retval |= UNDEF_CIT;
 710                         } else if (contains(token, "No pages of output")) {
 711                                 // A dvi file was not created
 712                                 retval |= NO_OUTPUT;
 713                         } else if (contains(token, "That makes 100 errors")) {
 714                                 // More than 100 errors were reprted
 715                                 retval |= TOO_MANY_ERRORS;
 716                         }
 717                 }
 718         }
 719         lyxerr[Debug::LATEX] << "Log line: " << token << endl;
 720         return retval;
 721 }
 722
 723
 724 namespace {
 725
 726 bool handleFoundFile(string const & ff, DepTable & head)
 727 {
 728         // convert from native os path to unix path
 729         string foundfile = os::internal_path(trim(ff));
 730
 731         lyxerr[Debug::DEPEND] << "Found file: " << foundfile << endl;
 732
 733         // Ok now we found a file.
 734         // Now we should make sure that this is a file that we can
 735         // access through the normal paths.
 736         // We will not try any fancy search methods to
 737         // find the file.
 738
 739         // (1) foundfile is an
 740         //     absolute path and should
 741         //     be inserted.
 742         if (absolutePath(foundfile)) {
 743                 lyxerr[Debug::DEPEND] << "AbsolutePath file: "
 744                                       << foundfile << endl;
 745                 // On initial insert we want to do the update at once
 746                 // since this file cannot be a file generated by
 747                 // the latex run.
 748                 FileName absname(foundfile);
 749                 if (fs::exists(absname.toFilesystemEncoding()) &&
 750                     !fs::is_directory(absname.toFilesystemEncoding())) {
 751                                 head.insert(absname, true);
 752                                 return true;
 753                 } else {
 754                         // check for spaces
 755                         string strippedfile = foundfile;
 756                         while (contains(strippedfile, " ")) {
 757                                 // strip off part after last space and try again
 758                                 string tmp = strippedfile;
 759                                 string const stripoff =
 760                                         rsplit(tmp, strippedfile, ' ');
 761                                 absname.set(strippedfile);
 762                                 if (fs::exists(absname.toFilesystemEncoding()) &&
 763                                     !fs::is_directory(absname.toFilesystemEncoding())) {
 764                                         head.insert(absname, true);
 765                                         return true;
 766                                 }
 767                         }
 768                 }
 769         }
 770
 771         string onlyfile = onlyFilename(foundfile);
 772         FileName absname(makeAbsPath(onlyfile));
 773
 774         // check for spaces
 775         while (contains(foundfile, " ")) {
 776                 if (fs::exists(absname.toFilesystemEncoding()))
 777                         // everything o.k.
 778                         break;
 779                 else {
 780                         // strip off part after last space and try again
 781                         string strippedfile;
 782                         string const stripoff = rsplit(foundfile, strippedfile, ' ');
 783                         foundfile = strippedfile;
 784                         onlyfile = onlyFilename(strippedfile);
 785                         absname = makeAbsPath(onlyfile);
 786                 }
 787         }
 788
 789         // (2) foundfile is in the tmpdir
 790         //     insert it into head
 791         if (fs::exists(absname.toFilesystemEncoding()) &&
 792             !fs::is_directory(absname.toFilesystemEncoding())) {
 793                 static regex unwanted("^.*\\.(aux|log|dvi|bbl|ind|glo)$");
 794                 if (regex_match(onlyfile, unwanted)) {
 795                         lyxerr[Debug::DEPEND]
 796                                 << "We don't want "
 797                                 << onlyfile
 798                                 << " in the dep file"
 799                                 << endl;
 800                 } else if (suffixIs(onlyfile, ".tex")) {
 801                         // This is a tex file generated by LyX
 802                         // and latex is not likely to change this
 803                         // during its runs.
 804                         lyxerr[Debug::DEPEND]
 805                                 << "Tmpdir TeX file: "
 806                                 << onlyfile
 807                                 << endl;
 808                         head.insert(absname, true);
 809                 } else {
 810                         lyxerr[Debug::DEPEND]
 811                                 << "In tmpdir file:"
 812                                 << onlyfile
 813                                 << endl;
 814                         head.insert(absname);
 815                 }
 816                 return true;
 817         } else {
 818                 lyxerr[Debug::DEPEND]
 819                         << "Not a file or we are unable to find it."
 820                         << endl;
 821                 return false;
 822         }
 823 }
 824
 825
 826 bool checkLineBreak(string const & ff, DepTable & head)
 827 {
 828         if (contains(ff, '.'))
 829                 // if we have a dot, we let handleFoundFile decide
 830                 return handleFoundFile(ff, head);
 831         else
 832                 // else, we suspect a line break
 833                 return false;
 834 }
 835
 836 } // anon namespace
 837
 838
 839 void LaTeX::deplog(DepTable & head)
 840 {
 841         // This function reads the LaTeX log file end extracts all the external
 842         // files used by the LaTeX run. The files are then entered into the
 843         // dependency file.
 844
 845         string const logfile = onlyFilename(changeExtension(file.absFilename(), ".log"));
 846
 847         static regex reg1("File: (.+).*");
 848         static regex reg2("No file (.+)(.).*");
 849         static regex reg3("\\\\openout[0-9]+.*=.*`(.+)(..).*");
 850         // If an index should be created, MikTex does not write a line like
 851         //    \openout# = 'sample.idx'.
 852         // but instead only a line like this into the log:
 853         //   Writing index file sample.idx
 854         static regex reg4("Writing index file (.+).*");
 855         // files also can be enclosed in <...>
 856         static regex reg5("<([^>]+)(.).*");
 857         static regex regnomencl("Writing nomenclature file (.+).*");
 858         // If a toc should be created, MikTex does not write a line like
 859         //    \openout# = `sample.toc'.
 860         // but only a line like this into the log:
 861         //    \tf@toc=\write#
 862         // This line is also written by tetex.
 863         // This line is not present if no toc should be created.
 864         static regex miktexTocReg("\\\\tf@toc=\\\\write.*");
 865         static regex reg6(".*\\([^)]+.*");
 866
 867         FileName const fn(makeAbsPath(logfile));
 868         ifstream ifs(fn.toFilesystemEncoding().c_str());
 869         string lastline;
 870         while (ifs) {
 871                 // Ok, the scanning of files here is not sufficient.
 872                 // Sometimes files are named by "File: xxx" only
 873                 // So I think we should use some regexps to find files instead.
 874                 // Note: all file names and paths might contains spaces.
 875
 876                 bool found_file = false;
 877                 string token;
 878                 getline(ifs, token);
 879                 // MikTeX sometimes inserts \0 in the log file. They can't be
 880                 // removed directly with the existing string utility
 881                 // functions, so convert them first to \r, and remove all
 882                 // \r's afterwards, since we need to remove them anyway.
 883                 token = subst(token, '\0', '\r');
 884                 token = subst(token, "\r", "");
 885                 if (token.empty() || token == ")") {
 886                         lastline = string();
 887                         continue;
 888                 }
 889
 890                 // Sometimes, filenames are broken across lines.
 891                 // We care for that and save suspicious lines.
 892                 // Here we exclude some cases where we are sure
 893                 // that there is no continued filename
 894                 if (!lastline.empty()) {
 895                         static regex package_info("Package \\w+ Info: .*");
 896                         static regex package_warning("Package \\w+ Warning: .*");
 897                         if (prefixIs(token, "File:") || prefixIs(token, "(Font)")
 898                         || prefixIs(token, "Package:")
 899                         || prefixIs(token, "Language:")
 900                         || prefixIs(token, "LaTeX Info:")
 901                         || prefixIs(token, "LaTeX Font Info:")
 902                         || prefixIs(token, "\\openout[")
 903                         || prefixIs(token, "))")
 904                         || regex_match(token, package_info)
 905                         || regex_match(token, package_warning))
 906                                 lastline = string();
 907                 }
 908
 909                 if (!lastline.empty())
 910                         // probably a continued filename from last line
 911                         token = lastline + token;
 912                 if (token.length() > 255) {
 913                         // string too long. Cut off.
 914                         int r = token.length() - 250;
 915                         string ntoken = token.substr(r, token.length());
 916                         token = ntoken;
 917                 }
 918
 919                 smatch sub;
 920
 921                 // FIXME UNICODE: We assume that the file names in the log
 922                 // file are in the file system encoding.
 923                 token = to_utf8(from_filesystem8bit(token));
 924
 925                 // (1) "File: file.ext"
 926                 if (regex_match(token, sub, reg1)) {
 927                         // check for dot
 928                         found_file = checkLineBreak(sub.str(1), head);
 929                         // However, ...
 930                         if (suffixIs(token, ")"))
 931                                 // no line break for sure
 932                                 // pretend we've been succesfully searching
 933                                 found_file = true;
 934                 // (2) "No file file.ext"
 935                 } else if (regex_match(token, sub, reg2)) {
 936                         // file names must contains a dot, line ends with dot
 937                         if (contains(sub.str(1), '.') && sub.str(2) == ".")
 938                                 found_file = handleFoundFile(sub.str(1), head);
 939                         else
 940                                 // we suspect a line break
 941                                 found_file = false;
 942                 // (3) "\openout<nr> = `file.ext'."
 943                 } else if (regex_match(token, sub, reg3)) {
 944                         // search for closing '. at the end of the line
 945                         if (sub.str(2) == "\'.")
 946                                 found_file = handleFoundFile(sub.str(1), head);
 947                         else
 948                                 // probable line break
 949                                 found_file = false;
 950                 // (4) "Writing index file file.ext"
 951                 } else if (regex_match(token, sub, reg4))
 952                         // check for dot
 953                         found_file = checkLineBreak(sub.str(1), head);
 954                 // (5) "<file.ext>"
 955                 else if (regex_match(token, sub, reg5)) {
 956                         // search for closing '>' and dot ('*.*>') at the eol
 957                         if (contains(sub.str(1), '.') && sub.str(2) == ">")
 958                                 found_file = handleFoundFile(sub.str(1), head);
 959                         else
 960                                 // probable line break
 961                                 found_file = false;
 962                 // (6) "Writing nomenclature file file.ext"
 963                 } else if (regex_match(token, sub, regnomencl))
 964                         // check for dot
 965                         found_file = checkLineBreak(sub.str(1), head);
 966                 // (7) "\tf@toc=\write<nr>" (for MikTeX)
 967                 else if (regex_match(token, sub, miktexTocReg))
 968                         found_file = handleFoundFile(onlyFilename(changeExtension(
 969                                                 file.absFilename(), ".toc")), head);
 970                 else
 971                         // not found, but we won't check further
 972                         // pretend we've been succesfully searching
 973                         found_file = true;
 974
 975                 // (8) "(file.ext"
 976                 // note that we can have several of these on one line
 977                 // this must be queried separated, because of
 978                 // cases such as "File: file.ext (type eps)"
 979                 // where "File: file.ext" would be skipped
 980                 if (regex_match(token, sub, reg6)) {
 981                         // search for strings in (...)
 982                         static regex reg6_1("\\(([^()]+)(.).*");
 983                         smatch what;
 984                         string::const_iterator first = token.begin();
 985                         string::const_iterator end = token.end();
 986
 987                         while (regex_search(first, end, what, reg6_1)) {
 988                                 // if we have a dot, try to handle as file
 989                                 if (contains(what.str(1), '.')) {
 990                                         first = what[0].second;
 991                                         if (what.str(2) == ")") {
 992                                                 handleFoundFile(what.str(1), head);
 993                                                 // since we had a closing bracket,
 994                                                 // do not investigate further
 995                                                 found_file = true;
 996                                         } else
 997                                                 // if we have no closing bracket,
 998                                                 // try to handle as file nevertheless
 999                                                 found_file = handleFoundFile(
1000                                                         what.str(1) + what.str(2), head);
1001                                 }
1002                                 // if we do not have a dot, check if the line has
1003                                 // a closing bracket (else, we suspect a line break)
1004                                 else if (what.str(2) != ")") {
1005                                         first = what[0].second;
1006                                         found_file = false;
1007                                 } else {
1008                                         // we have a closing bracket, so the content
1009                                         // is not a file name.
1010                                         // no need to investigate further
1011                                         // pretend we've been succesfully searching
1012                                         first = what[0].second;
1013                                         found_file = true;
1014                                 }
1015                         }
1016                 }
1017
1018                 if (!found_file)
1019                         // probable linebreak:
1020                         // save this line
1021                         lastline = token;
1022                 else
1023                         // no linebreak: reset
1024                         lastline = string();
1025         }
1026
1027         // Make sure that the main .tex file is in the dependency file.
1028         head.insert(file, true);
1029 }
1030
1031
1032 } // namespace lyx