src/LaTeX.cpp

   1 /**
   2  * \file LaTeX.cpp
   3  * This file is part of LyX, the document processor.
   4  * Licence details can be found in the file COPYING.
   5  *
   6  * \author Alfredo Braunstein
   7  * \author Lars Gullik Bjønnes
   8  * \author Jean-Marc Lasgouttes
   9  * \author Angus Leeming
  10  * \author Dekel Tsur
  11  * \author Jürgen Spitzmüller
  12  *
  13  * Full author contact details are available in file CREDITS.
  14  */
  15
  16 #include <config.h>
  17
  18 #include "Buffer.h"
  19 #include "BufferList.h"
  20 #include "BufferParams.h"
  21 #include "LaTeX.h"
  22 #include "LyXRC.h"
  23 #include "LyX.h"
  24 #include "DepTable.h"
  25 #include "Encoding.h"
  26
  27 #include "support/debug.h"
  28 #include "support/convert.h"
  29 #include "support/FileName.h"
  30 #include "support/filetools.h"
  31 #include "support/gettext.h"
  32 #include "support/lstrings.h"
  33 #include "support/Systemcall.h"
  34 #include "support/os.h"
  35
  36 #include "support/regex.h"
  37
  38 #include <fstream>
  39 #include <stack>
  40
  41
  42 using namespace std;
  43 using namespace lyx::support;
  44
  45 namespace lyx {
  46
  47 namespace os = support::os;
  48
  49 // TODO: in no particular order
  50 // - get rid of the call to
  51 //   BufferList::updateIncludedTeXfiles, this should either
  52 //   be done before calling LaTeX::funcs or in a completely
  53 //   different way.
  54 // - the makeindex style files should be taken care of with
  55 //   the dependency mechanism.
  56
  57 namespace {
  58
  59 docstring runMessage(unsigned int count)
  60 {
  61         return bformat(_("Waiting for LaTeX run number %1$d"), count);
  62 }
  63
  64 } // namespace
  65
  66 /*
  67  * CLASS TEXERRORS
  68  */
  69
  70 void TeXErrors::insertError(int line, docstring const & error_desc,
  71                             docstring const & error_text,
  72                             string const & child_name)
  73 {
  74         Error newerr(line, error_desc, error_text, child_name);
  75         errors.push_back(newerr);
  76 }
  77
  78
  79 void TeXErrors::insertRef(int line, docstring const & error_desc,
  80                             docstring const & error_text,
  81                             string const & child_name)
  82 {
  83         Error newerr(line, error_desc, error_text, child_name);
  84         undef_ref.push_back(newerr);
  85 }
  86
  87
  88 bool operator==(AuxInfo const & a, AuxInfo const & o)
  89 {
  90         return a.aux_file == o.aux_file
  91                 && a.citations == o.citations
  92                 && a.databases == o.databases
  93                 && a.styles == o.styles;
  94 }
  95
  96
  97 bool operator!=(AuxInfo const & a, AuxInfo const & o)
  98 {
  99         return !(a == o);
 100 }
 101
 102
 103 /*
 104  * CLASS LaTeX
 105  */
 106
 107 LaTeX::LaTeX(string const & latex, OutputParams const & rp,
 108              FileName const & f, string const & p, string const & lp,
 109              bool allow_cancellation, bool const clean_start)
 110         : cmd(latex), file(f), path(p), lpath(lp), runparams(rp), biber(false),
 111           allow_cancel(allow_cancellation)
 112 {
 113         num_errors = 0;
 114         // lualatex can still produce a DVI with --output-format=dvi. However,
 115         // we do not use that internally (we use the "dvilualatex" command) so
 116         // it would only happen from a custom converter. Thus, it is better to
 117         // guess that lualatex produces a PDF than to guess a DVI.
 118         // FIXME we should base the extension on the output format, which we should
 119         // get in a robust way, e.g. from the converter.
 120         if (prefixIs(cmd, "pdf") || prefixIs(cmd, "lualatex") || prefixIs(cmd, "xelatex")) {
 121                 depfile = FileName(file.absFileName() + ".dep-pdf");
 122                 output_file =
 123                         FileName(changeExtension(file.absFileName(), ".pdf"));
 124         } else {
 125                 depfile = FileName(file.absFileName() + ".dep");
 126                 output_file =
 127                         FileName(changeExtension(file.absFileName(), ".dvi"));
 128         }
 129         if (clean_start)
 130                 removeAuxiliaryFiles();
 131 }
 132
 133
 134 void LaTeX::removeAuxiliaryFiles() const
 135 {
 136         // Note that we do not always call this function when there is an error.
 137         // For example, if there is an error but an output file is produced we
 138         // still would like to output (export/view) the file.
 139
 140         // What files do we have to delete?
 141
 142         // This will at least make latex do all the runs
 143         depfile.removeFile();
 144
 145         // but the reason for the error might be in a generated file...
 146
 147         // bibtex file
 148         FileName const bbl(changeExtension(file.absFileName(), ".bbl"));
 149         bbl.removeFile();
 150
 151         // biber file
 152         FileName const bcf(changeExtension(file.absFileName(), ".bcf"));
 153         bcf.removeFile();
 154
 155         // makeindex file
 156         FileName const ind(changeExtension(file.absFileName(), ".ind"));
 157         ind.removeFile();
 158
 159         // nomencl file
 160         FileName const nls(changeExtension(file.absFileName(), ".nls"));
 161         nls.removeFile();
 162
 163         // nomencl file (old version of the package)
 164         FileName const gls(changeExtension(file.absFileName(), ".gls"));
 165         gls.removeFile();
 166
 167         // endnotes file
 168         FileName const ent(changeExtension(file.absFileName(), ".ent"));
 169         ent.removeFile();
 170
 171         // Also remove the aux file
 172         FileName const aux(changeExtension(file.absFileName(), ".aux"));
 173         aux.removeFile();
 174
 175         // Also remove the .out file (e.g. hyperref bookmarks) (#9963)
 176         FileName const out(changeExtension(file.absFileName(), ".out"));
 177         out.removeFile();
 178
 179         // Remove the output file, which is often generated even if error
 180         output_file.removeFile();
 181 }
 182
 183
 184 int LaTeX::run(TeXErrors & terr)
 185         // We know that this function will only be run if the lyx buffer
 186         // has been changed. We also know that a newly written .tex file
 187         // is always different from the previous one because of the date
 188         // in it. However it seems safe to run latex (at least) one time
 189         // each time the .tex file changes.
 190 {
 191         int scanres = NO_ERRORS;
 192         int bscanres = NO_ERRORS;
 193         int iscanres = NO_ERRORS;
 194         unsigned int count = 0; // number of times run
 195         num_errors = 0; // just to make sure.
 196         unsigned int const MAX_RUN = 6;
 197         DepTable head; // empty head
 198         bool rerun = false; // rerun requested
 199
 200         // The class LaTeX does not know the temp path.
 201         theBufferList().updateIncludedTeXfiles(FileName::getcwd().absFileName(),
 202                 runparams);
 203
 204         // 0
 205         // first check if the file dependencies exist:
 206         //     ->If it does exist
 207         //             check if any of the files mentioned in it have
 208         //             changed (done using a checksum).
 209         //                 -> if changed:
 210         //                        run latex once and
 211         //                        remake the dependency file
 212         //                 -> if not changed:
 213         //                        just return there is nothing to do for us.
 214         //     ->if it doesn't exist
 215         //             make it and
 216         //             run latex once (we need to run latex once anyway) and
 217         //             remake the dependency file.
 218         //
 219
 220         bool had_depfile = depfile.exists();
 221         bool run_bibtex = false;
 222         FileName const aux_file(changeExtension(file.absFileName(), ".aux"));
 223
 224         if (had_depfile) {
 225                 LYXERR(Debug::DEPEND, "Dependency file exists");
 226                 // Read the dep file:
 227                 had_depfile = head.read(depfile);
 228         }
 229
 230         if (had_depfile) {
 231                 if (runparams.includeall) {
 232                         // On an "includeall" call (whose purpose is to set up/maintain counters and references
 233                         // for includeonly), we remove the master from the dependency list since
 234                         // (1) it will be checked anyway on the subsequent includeonly run
 235                         // (2) the master is always changed (due to the \includeonly line), and this alone would
 236                         //     trigger a complete, expensive run each time
 237                         head.remove_file(file);
 238                         // Also remove all children which are included
 239                         Buffer const * buf = theBufferList().getBufferFromTmp(file.absFileName());
 240                         if (buf && buf->params().maintain_unincluded_children == BufferParams::CM_Mostly) {
 241                                 for (auto const & incfile : buf->params().getIncludedChildren()) {
 242                                         string const incm =
 243                                                 DocFileName(changeExtension(makeAbsPath(incfile, path)
 244                                                                             .absFileName(), ".tex")).mangledFileName();
 245                                         FileName inc = makeAbsPath(incm, file.onlyPath().realPath());
 246                                         head.remove_file(inc);
 247                                 }
 248                         }
 249                 }
 250                 // Update the checksums
 251                 head.update();
 252                 // Can't just check if anything has changed because it might
 253                 // have aborted on error last time... in which cas we need
 254                 // to re-run latex and collect the error messages
 255                 // (even if they are the same).
 256                 if (!output_file.exists()) {
 257                         LYXERR(Debug::DEPEND,
 258                                 "re-running LaTeX because output file doesn't exist.");
 259                 } else if (!head.sumchange()) {
 260                         LYXERR(Debug::DEPEND, "return no_change");
 261                         return NO_CHANGE;
 262                 } else {
 263                         LYXERR(Debug::DEPEND, "Dependency file has changed");
 264                 }
 265
 266                 if (head.extchanged(".bib") || head.extchanged(".bst"))
 267                         run_bibtex = true;
 268         } else
 269                 LYXERR(Debug::DEPEND,
 270                         "Dependency file does not exist, or has wrong format");
 271
 272         /// We scan the aux file even when had_depfile = false,
 273         /// because we can run pdflatex on the file after running latex on it,
 274         /// in which case we will not need to run bibtex again.
 275         vector<AuxInfo> bibtex_info_old;
 276         if (!run_bibtex)
 277                 bibtex_info_old = scanAuxFiles(aux_file, runparams.only_childbibs);
 278
 279         ++count;
 280         LYXERR(Debug::LATEX, "Run #" << count);
 281         message(runMessage(count));
 282
 283         int exit_code = startscript();
 284         if (exit_code == Systemcall::KILLED)
 285                 return Systemcall::KILLED;
 286
 287         scanres = scanLogFile(terr);
 288         if (scanres & ERROR_RERUN) {
 289                 LYXERR(Debug::LATEX, "Rerunning LaTeX");
 290                 terr.clearErrors();
 291                 exit_code = startscript();
 292                 if (exit_code == Systemcall::KILLED)
 293                         return Systemcall::KILLED;
 294                 scanres = scanLogFile(terr);
 295         }
 296
 297         vector<AuxInfo> const bibtex_info = scanAuxFiles(aux_file, runparams.only_childbibs);
 298         if (!run_bibtex && bibtex_info_old != bibtex_info)
 299                 run_bibtex = true;
 300
 301         // update the dependencies.
 302         deplog(head); // reads the latex log
 303         head.update();
 304
 305         // 1
 306         // At this point we must run external programs if needed.
 307         // makeindex will be run if a .idx file changed or was generated.
 308         // And if there were undefined citations or changes in references
 309         // the .aux file is checked for signs of bibtex. Bibtex is then run
 310         // if needed.
 311
 312         // memoir (at least) writes an empty *idx file in the first place.
 313         // A second latex run is needed.
 314         FileName const idxfile(changeExtension(file.absFileName(), ".idx"));
 315         rerun = idxfile.exists() && idxfile.isFileEmpty();
 316
 317         // run makeindex
 318         if (head.haschanged(idxfile)) {
 319                 // no checks for now
 320                 LYXERR(Debug::LATEX, "Running MakeIndex.");
 321                 message(_("Running Index Processor."));
 322                 // onlyFileName() is needed for cygwin
 323                 int const ret =
 324                                 runMakeIndex(onlyFileName(idxfile.absFileName()), runparams);
 325                 if (ret == Systemcall::KILLED)
 326                         return Systemcall::KILLED;
 327                 FileName const ilgfile(changeExtension(file.absFileName(), ".ilg"));
 328                 if (ilgfile.exists())
 329                         iscanres = scanIlgFile(terr);
 330                 rerun = true;
 331         }
 332
 333         FileName const nlofile(changeExtension(file.absFileName(), ".nlo"));
 334         // If all nomencl entries are removed, nomencl writes an empty nlo file.
 335         // DepTable::hasChanged() returns false in this case, since it does not
 336         // distinguish empty files from non-existing files. This is why we need
 337         // the extra checks here (to trigger a rerun). Cf. discussions in #8905.
 338         // FIXME: Sort out the real problem in DepTable.
 339         if (head.haschanged(nlofile) || (nlofile.exists() && nlofile.isFileEmpty())) {
 340                 int const ret = runMakeIndexNomencl(file, ".nlo", ".nls");
 341                 if (ret == Systemcall::KILLED)
 342                         return Systemcall::KILLED;
 343                 rerun = true;
 344         }
 345
 346         FileName const glofile(changeExtension(file.absFileName(), ".glo"));
 347         if (head.haschanged(glofile)) {
 348                 int const ret = runMakeIndexNomencl(file, ".glo", ".gls");
 349                 if (ret)
 350                         return ret;
 351                 rerun = true;
 352         }
 353
 354
 355         // check if we're using biber instead of bibtex
 356         // biber writes no info to the aux file, so we just check
 357         // if a bcf file exists (and if it was updated)
 358         FileName const bcffile(changeExtension(file.absFileName(), ".bcf"));
 359         biber |= head.exist(bcffile);
 360
 361         // run bibtex
 362         // if (scanres & UNDEF_CIT || scanres & RERUN || run_bibtex)
 363         // We do not run bibtex/biber on an "includeall" call (whose purpose is
 364         // to set up/maintain counters and references for includeonly) since
 365         // (1) bibliographic references will be updated on the subsequent includeonly run
 366         // (2) this would trigger a complete run each time (as references in non-included
 367         //     children are removed on subsequent includeonly runs)
 368         if (!runparams.includeall && (scanres & UNDEF_CIT || run_bibtex)) {
 369                 // Here we must scan the .aux file and look for
 370                 // "\bibdata" and/or "\bibstyle". If one of those
 371                 // tags is found -> run bibtex and set rerun = true;
 372                 // no checks for now
 373                 LYXERR(Debug::LATEX, "Running BibTeX.");
 374                 message(_("Running BibTeX."));
 375                 updateBibtexDependencies(head, bibtex_info);
 376                 int exit_code;
 377                 rerun |= runBibTeX(bibtex_info, runparams, exit_code);
 378                 if (exit_code == Systemcall::KILLED)
 379                         return Systemcall::KILLED;
 380                 FileName const blgfile(changeExtension(file.absFileName(), ".blg"));
 381                 if (blgfile.exists())
 382                         bscanres = scanBlgFile(head, terr);
 383         } else if (!had_depfile) {
 384                 /// If we run pdflatex on the file after running latex on it,
 385                 /// then we do not need to run bibtex, but we do need to
 386                 /// insert the .bib and .bst files into the .dep-pdf file.
 387                 updateBibtexDependencies(head, bibtex_info);
 388         }
 389
 390         // 2
 391         // we know on this point that latex has been run once (or we just
 392         // returned) and the question now is to decide if we need to run
 393         // it any more. This is done by asking if any of the files in the
 394         // dependency file has changed. (remember that the checksum for
 395         // a given file is reported to have changed if it just was created)
 396         //     -> if changed or rerun == true:
 397         //             run latex once more and
 398         //             update the dependency structure
 399         //     -> if not changed:
 400         //             we do nothing at this point
 401         //
 402         if (rerun || head.sumchange()) {
 403                 rerun = false;
 404                 ++count;
 405                 LYXERR(Debug::DEPEND, "Dep. file has changed or rerun requested");
 406                 LYXERR(Debug::LATEX, "Run #" << count);
 407                 message(runMessage(count));
 408                 int exit_code = startscript();
 409                 if (exit_code == Systemcall::KILLED)
 410                         return Systemcall::KILLED;
 411                 scanres = scanLogFile(terr);
 412
 413                 // update the depedencies
 414                 deplog(head); // reads the latex log
 415                 head.update();
 416         } else {
 417                 LYXERR(Debug::DEPEND, "Dep. file has NOT changed");
 418         }
 419
 420         // 3
 421         // rerun bibtex?
 422         // Complex bibliography packages such as Biblatex require
 423         // an additional bibtex cycle sometimes.
 424         // We do not run bibtex/biber on an "includeall" call (whose purpose is
 425         // to set up/maintain counters and references for includeonly) since
 426         // (1) bibliographic references will be updated on the subsequent includeonly run
 427         // (2) this would trigger a complete run each time (as references in non-included
 428         //     children are removed on subsequent includeonly runs)
 429         if (!runparams.includeall && scanres & UNDEF_CIT) {
 430                 // Here we must scan the .aux file and look for
 431                 // "\bibdata" and/or "\bibstyle". If one of those
 432                 // tags is found -> run bibtex and set rerun = true;
 433                 // no checks for now
 434                 LYXERR(Debug::LATEX, "Running BibTeX.");
 435                 message(_("Running BibTeX."));
 436                 updateBibtexDependencies(head, bibtex_info);
 437                 int exit_code;
 438                 rerun |= runBibTeX(bibtex_info, runparams, exit_code);
 439                 if (exit_code == Systemcall::KILLED)
 440                         return Systemcall::KILLED;
 441                 FileName const blgfile(changeExtension(file.absFileName(), ".blg"));
 442                 if (blgfile.exists())
 443                         bscanres = scanBlgFile(head, terr);
 444         }
 445
 446         // 4
 447         // The inclusion of files generated by external programs such as
 448         // makeindex or bibtex might have done changes to pagenumbering,
 449         // etc. And because of this we must run the external programs
 450         // again to make sure everything is redone correctly.
 451         // Also there should be no need to run the external programs any
 452         // more after this.
 453
 454         // run makeindex if the <file>.idx has changed or was generated.
 455         if (head.haschanged(idxfile)) {
 456                 // no checks for now
 457                 LYXERR(Debug::LATEX, "Running MakeIndex.");
 458                 message(_("Running Index Processor."));
 459                 // onlyFileName() is needed for cygwin
 460                 int const ret = runMakeIndex(onlyFileName(changeExtension(
 461                                 file.absFileName(), ".idx")), runparams);
 462                 if (ret == Systemcall::KILLED)
 463                         return Systemcall::KILLED;
 464                 FileName const ilgfile(changeExtension(file.absFileName(), ".ilg"));
 465                 if (ilgfile.exists())
 466                         iscanres = scanIlgFile(terr);
 467                 rerun = true;
 468         }
 469
 470         // MSVC complains that bool |= int is unsafe. Not sure why.
 471         if (head.haschanged(nlofile))
 472                 rerun |= (runMakeIndexNomencl(file, ".nlo", ".nls") != 0);
 473         if (head.haschanged(glofile))
 474                 rerun |= (runMakeIndexNomencl(file, ".glo", ".gls") != 0);
 475
 476         // 5
 477         // we will only run latex more if the log file asks for it.
 478         // or if the sumchange() is true.
 479         //     -> rerun asked for:
 480         //             run latex and
 481         //             remake the dependency file
 482         //             goto 2 or return if max runs are reached.
 483         //     -> rerun not asked for:
 484         //             just return (fall out of bottom of func)
 485         //
 486         while ((head.sumchange() || rerun || (scanres & RERUN))
 487                && count < MAX_RUN) {
 488                 // Yes rerun until message goes away, or until
 489                 // MAX_RUNS are reached.
 490                 rerun = false;
 491                 ++count;
 492                 LYXERR(Debug::LATEX, "Run #" << count);
 493                 message(runMessage(count));
 494                 startscript();
 495                 scanres = scanLogFile(terr);
 496
 497                 // keep this updated
 498                 head.update();
 499         }
 500
 501         // Write the dependencies to file.
 502         head.write(depfile);
 503
 504         if (exit_code) {
 505                 // add flag here, just before return, instead of when exit_code
 506                 // is defined because scanres is sometimes overwritten above
 507                 // (e.g. rerun)
 508                 scanres |= NONZERO_ERROR;
 509         }
 510
 511         LYXERR(Debug::LATEX, "Done.");
 512
 513         if (bscanres & ERRORS)
 514                 return bscanres; // return on error
 515
 516         if (iscanres & ERRORS)
 517                 return iscanres; // return on error
 518
 519         return scanres;
 520 }
 521
 522
 523 int LaTeX::startscript()
 524 {
 525         // onlyFileName() is needed for cygwin
 526         string tmp = cmd + ' '
 527                      + quoteName(onlyFileName(file.toFilesystemEncoding()))
 528                      + " > " + os::nulldev();
 529         Systemcall one;
 530         Systemcall::Starttype const starttype =
 531                 allow_cancel ? Systemcall::WaitLoop : Systemcall::Wait;
 532         return one.startscript(starttype, tmp, path, lpath, true);
 533 }
 534
 535
 536 int LaTeX::runMakeIndex(string const & f, OutputParams const & rp,
 537                          string const & params)
 538 {
 539         string tmp = rp.use_japanese ?
 540                 lyxrc.jindex_command : lyxrc.index_command;
 541
 542         if (!rp.index_command.empty())
 543                 tmp = rp.index_command;
 544
 545         if (contains(tmp, "$$x")) {
 546                 // This adds appropriate [te]xindy options
 547                 // such as language and codepage (for the
 548                 // main document language/encoding) as well
 549                 // as input markup (latex or xelatex)
 550                 string xdyopts = rp.xindy_language;
 551                 if (!xdyopts.empty())
 552                         xdyopts = "-L " + xdyopts;
 553                 if (rp.isFullUnicode() && rp.encoding->package() == Encoding::none) {
 554                         if (!xdyopts.empty())
 555                                 xdyopts += " ";
 556                         // xelatex includes lualatex
 557                         xdyopts += "-I xelatex";
 558                 }
 559                 else if (rp.encoding->iconvName() == "UTF-8") {
 560                         if (!xdyopts.empty())
 561                                 xdyopts += " ";
 562                         // -I not really needed for texindy, but for xindy
 563                         xdyopts += "-C utf8 -I latex";
 564                 }
 565                 else {
 566                         if (!xdyopts.empty())
 567                                 xdyopts += " ";
 568                         // not really needed for texindy, but for xindy
 569                         xdyopts += "-I latex";
 570                 }
 571                 tmp = subst(tmp, "$$x", xdyopts);
 572         }
 573
 574         if (contains(tmp, "$$b")) {
 575                 // advise xindy to write a log file
 576                 tmp = subst(tmp, "$$b", removeExtension(f));
 577         }
 578
 579         LYXERR(Debug::LATEX,
 580                 "idx file has been made, running index processor ("
 581                 << tmp << ") on file " << f);
 582
 583         tmp = subst(tmp, "$$lang", rp.document_language);
 584         if (rp.use_indices) {
 585                 tmp = lyxrc.splitindex_command + " -m " + quoteName(tmp);
 586                 LYXERR(Debug::LATEX,
 587                 "Multiple indices. Using splitindex command: " << tmp);
 588         }
 589         tmp += ' ';
 590         tmp += quoteName(f);
 591         tmp += params;
 592         Systemcall one;
 593         Systemcall::Starttype const starttype =
 594                 allow_cancel ? Systemcall::WaitLoop : Systemcall::Wait;
 595         return one.startscript(starttype, tmp, path, lpath, true);
 596 }
 597
 598
 599 int LaTeX::runMakeIndexNomencl(FileName const & fname,
 600                 string const & nlo, string const & nls)
 601 {
 602         LYXERR(Debug::LATEX, "Running MakeIndex for nomencl.");
 603         message(_("Running MakeIndex for nomencl."));
 604         string tmp = lyxrc.nomencl_command + ' ';
 605         // onlyFileName() is needed for cygwin
 606         tmp += quoteName(onlyFileName(changeExtension(fname.absFileName(), nlo)));
 607         tmp += " -o "
 608                 + onlyFileName(changeExtension(fname.toFilesystemEncoding(), nls));
 609         Systemcall one;
 610         Systemcall::Starttype const starttype =
 611                 allow_cancel ? Systemcall::WaitLoop : Systemcall::Wait;
 612         return one.startscript(starttype, tmp, path, lpath, true);
 613 }
 614
 615
 616 vector<AuxInfo> const
 617 LaTeX::scanAuxFiles(FileName const & fname, bool const only_childbibs)
 618 {
 619         vector<AuxInfo> result;
 620
 621         // With chapterbib, we have to bibtex all children's aux files
 622         // but _not_ the master's!
 623         if (only_childbibs) {
 624                 for (string const &s: children) {
 625                         FileName fn =
 626                                 makeAbsPath(s, fname.onlyPath().realPath());
 627                         fn.changeExtension("aux");
 628                         if (fn.exists())
 629                                 result.push_back(scanAuxFile(fn));
 630                 }
 631                 return result;
 632         }
 633
 634         result.push_back(scanAuxFile(fname));
 635
 636         // This is for bibtopic
 637         string const basename = removeExtension(fname.absFileName());
 638         for (int i = 1; i < 1000; ++i) {
 639                 FileName const file2(basename
 640                         + '.' + convert<string>(i)
 641                         + ".aux");
 642                 if (!file2.exists())
 643                         break;
 644                 result.push_back(scanAuxFile(file2));
 645         }
 646         return result;
 647 }
 648
 649
 650 AuxInfo const LaTeX::scanAuxFile(FileName const & fname)
 651 {
 652         AuxInfo result;
 653         result.aux_file = fname;
 654         scanAuxFile(fname, result);
 655         return result;
 656 }
 657
 658
 659 void LaTeX::scanAuxFile(FileName const & fname, AuxInfo & aux_info)
 660 {
 661         LYXERR(Debug::LATEX, "Scanning aux file: " << fname);
 662
 663         ifstream ifs(fname.toFilesystemEncoding().c_str());
 664         string token;
 665         static regex const reg1("\\\\citation\\{([^}]+)\\}");
 666         static regex const reg2("\\\\bibdata\\{([^}]+)\\}");
 667         static regex const reg3("\\\\bibstyle\\{([^}]+)\\}");
 668         static regex const reg4("\\\\@input\\{([^}]+)\\}");
 669
 670         while (getline(ifs, token)) {
 671                 token = rtrim(token, "\r");
 672                 smatch sub;
 673                 // FIXME UNICODE: We assume that citation keys and filenames
 674                 // in the aux file are in the file system encoding.
 675                 token = to_utf8(from_filesystem8bit(token));
 676                 if (regex_match(token, sub, reg1)) {
 677                         string data = sub.str(1);
 678                         while (!data.empty()) {
 679                                 string citation;
 680                                 data = split(data, citation, ',');
 681                                 LYXERR(Debug::LATEX, "Citation: " << citation);
 682                                 aux_info.citations.insert(citation);
 683                         }
 684                 } else if (regex_match(token, sub, reg2)) {
 685                         string data = sub.str(1);
 686                         // data is now all the bib files separated by ','
 687                         // get them one by one and pass them to the helper
 688                         while (!data.empty()) {
 689                                 string database;
 690                                 data = split(data, database, ',');
 691                                 database = changeExtension(database, "bib");
 692                                 LYXERR(Debug::LATEX, "BibTeX database: `" << database << '\'');
 693                                 aux_info.databases.insert(database);
 694                         }
 695                 } else if (regex_match(token, sub, reg3)) {
 696                         string style = sub.str(1);
 697                         // token is now the style file
 698                         // pass it to the helper
 699                         style = changeExtension(style, "bst");
 700                         LYXERR(Debug::LATEX, "BibTeX style: `" << style << '\'');
 701                         aux_info.styles.insert(style);
 702                 } else if (regex_match(token, sub, reg4)) {
 703                         string const file2 = sub.str(1);
 704                         scanAuxFile(makeAbsPath(file2), aux_info);
 705                 }
 706         }
 707 }
 708
 709
 710 void LaTeX::updateBibtexDependencies(DepTable & dep,
 711                                      vector<AuxInfo> const & bibtex_info)
 712 {
 713         // Since a run of Bibtex mandates more latex runs it is ok to
 714         // remove all ".bib" and ".bst" files.
 715         dep.remove_files_with_extension(".bib");
 716         dep.remove_files_with_extension(".bst");
 717         //string aux = OnlyFileName(ChangeExtension(file, ".aux"));
 718
 719         for (vector<AuxInfo>::const_iterator it = bibtex_info.begin();
 720              it != bibtex_info.end(); ++it) {
 721                 for (set<string>::const_iterator it2 = it->databases.begin();
 722                      it2 != it->databases.end(); ++it2) {
 723                         FileName const file = findtexfile(*it2, "bib");
 724                         if (!file.empty())
 725                                 dep.insert(file, true);
 726                 }
 727
 728                 for (set<string>::const_iterator it2 = it->styles.begin();
 729                      it2 != it->styles.end(); ++it2) {
 730                         FileName const file = findtexfile(*it2, "bst");
 731                         if (!file.empty())
 732                                 dep.insert(file, true);
 733                 }
 734         }
 735
 736         // biber writes nothing into the aux file.
 737         // Instead, we have to scan the blg file
 738         if (biber) {
 739                 TeXErrors terr;
 740                 scanBlgFile(dep, terr);
 741         }
 742 }
 743
 744
 745 bool LaTeX::runBibTeX(vector<AuxInfo> const & bibtex_info,
 746                       OutputParams const & rp, int & exit_code)
 747 {
 748         bool result = false;
 749         exit_code = 0;
 750         for (vector<AuxInfo>::const_iterator it = bibtex_info.begin();
 751              it != bibtex_info.end(); ++it) {
 752                 if (!biber && it->databases.empty())
 753                         continue;
 754                 result = true;
 755
 756                 string tmp = rp.bibtex_command;
 757                 tmp += " ";
 758                 // onlyFileName() is needed for cygwin
 759                 tmp += quoteName(onlyFileName(removeExtension(
 760                                 it->aux_file.absFileName())));
 761                 Systemcall one;
 762                 Systemcall::Starttype const starttype =
 763                 allow_cancel ? Systemcall::WaitLoop : Systemcall::Wait;
 764                 exit_code = one.startscript(starttype, tmp, path, lpath, true);
 765                 if (exit_code) {
 766                         return result;
 767                 }
 768         }
 769         // Return whether bibtex was run
 770         return result;
 771 }
 772
 773
 774 //helper func for scanLogFile; gets line number X from strings "... on input line X ..."
 775 //returns 0 if none is found
 776 int getLineNumber(const string &token){
 777         string l = support::token(token, ' ', tokenPos(token,' ',"line") + 1);
 778         return l.empty() ? 0 : convert<int>(l);
 779 }
 780
 781
 782 int LaTeX::scanLogFile(TeXErrors & terr)
 783 {
 784         int last_line = -1;
 785         int line_count = 1;
 786         int retval = NO_ERRORS;
 787         string tmp =
 788                 onlyFileName(changeExtension(file.absFileName(), ".log"));
 789         LYXERR(Debug::LATEX, "Log file: " << tmp);
 790         FileName const fn = FileName(makeAbsPath(tmp));
 791         // FIXME we should use an ifdocstream here and a docstring for token
 792         // below. The encoding of the log file depends on the _output_ (font)
 793         // encoding of the TeX file (T1, TU etc.). See #10728.
 794         ifstream ifs(fn.toFilesystemEncoding().c_str());
 795         bool fle_style = false;
 796         static regex const file_line_error(".+\\.\\D+:[0-9]+: (.+)");
 797         static regex const child_file("[^0-9]*([0-9]+[A-Za-z]*_.+\\.tex).*");
 798         static regex const undef_ref(".*Reference `(\\w+)\\' on page.*");
 799         // Flag for 'File ended while scanning' message.
 800         // We need to wait for subsequent processing.
 801         string wait_for_error;
 802         string child_name;
 803         int pnest = 0;
 804         stack <pair<string, int> > child;
 805         children.clear();
 806
 807         terr.clearRefs();
 808
 809         string token;
 810         string ml_token;
 811         while (getline(ifs, token)) {
 812                 // MikTeX sometimes inserts \0 in the log file. They can't be
 813                 // removed directly with the existing string utility
 814                 // functions, so convert them first to \r, and remove all
 815                 // \r's afterwards, since we need to remove them anyway.
 816                 token = subst(token, '\0', '\r');
 817                 token = subst(token, "\r", "");
 818                 smatch sub;
 819
 820                 LYXERR(Debug::LATEX, "Log line: " << token);
 821
 822                 if (token.empty())
 823                         continue;
 824
 825                 if (!ml_token.empty())
 826                         ml_token += token;
 827
 828                 // Track child documents
 829                 for (size_t i = 0; i < token.length(); ++i) {
 830                         if (token[i] == '(') {
 831                                 ++pnest;
 832                                 size_t j = token.find('(', i + 1);
 833                                 size_t len = j == string::npos
 834                                                 ? token.substr(i + 1).length()
 835                                                 : j - i - 1;
 836                                 string const substr = token.substr(i + 1, len);
 837                                 if (regex_match(substr, sub, child_file)) {
 838                                         string const name = sub.str(1);
 839                                         // Sometimes also masters have a name that matches
 840                                         // (if their name starts with a number and _)
 841                                         if (name != file.onlyFileName()) {
 842                                                 child.push(make_pair(name, pnest));
 843                                                 children.push_back(name);
 844                                         }
 845                                         i += len;
 846                                 }
 847                         } else if (token[i] == ')') {
 848                                 if (!child.empty()
 849                                     && child.top().second == pnest)
 850                                         child.pop();
 851                                 --pnest;
 852                         }
 853                 }
 854                 child_name = child.empty() ? empty_string() : child.top().first;
 855
 856                 if (contains(token, "file:line:error style messages enabled"))
 857                         fle_style = true;
 858
 859                 //Handles both "LaTeX Warning:" & "Package natbib Warning:"
 860                 //Various handlers for missing citations below won't catch the problem if citation
 861                 //key is long (>~25chars), because pdflatex splits output at line length 80.
 862                 //TODO: TL 2020 engines will contain new commandline switch --cnf-line which we
 863                 //can use to set max_print_line variable for appropriate length and detect all
 864                 //errors correctly.
 865                 if (!runparams.includeall && (contains(token, "There were undefined citations.") ||
 866                     prefixIs(token, "Package biblatex Warning: The following entry could not be found")))
 867                         retval |= UNDEF_CIT;
 868
 869                 if (prefixIs(token, "LaTeX Warning:")
 870                     || prefixIs(token, "! pdfTeX warning")
 871                     || prefixIs(ml_token, "LaTeX Warning:")
 872                     || prefixIs(ml_token, "! pdfTeX warning")) {
 873                         // Here shall we handle different
 874                         // types of warnings
 875                         retval |= LATEX_WARNING;
 876                         LYXERR(Debug::LATEX, "LaTeX Warning.");
 877                         if (contains(token, "Rerun to get cross-references")) {
 878                                 retval |= RERUN;
 879                                 LYXERR(Debug::LATEX, "We should rerun.");
 880                         // package clefval needs 2 latex runs before bibtex
 881                         } else if (contains(token, "Value of")
 882                                    && contains(token, "on page")
 883                                    && contains(token, "undefined")) {
 884                                 retval |= ERROR_RERUN;
 885                                 LYXERR(Debug::LATEX, "Force rerun.");
 886                         // package etaremune
 887                         } else if (contains(token, "Etaremune labels have changed")) {
 888                                 retval |= ERROR_RERUN;
 889                                 LYXERR(Debug::LATEX, "Force rerun.");
 890                         // package enotez
 891                         } else if (contains(token, "Endnotes may have changed. Rerun")) {
 892                                 retval |= RERUN;
 893                                 LYXERR(Debug::LATEX, "We should rerun.");
 894                         //"Citation `cit' on page X undefined on input line X."
 895                         } else if (!runparams.includeall && contains(token, "Citation")
 896                                    //&& contains(token, "on input line") //often split to newline
 897                                    && contains(token, "undefined")) {
 898                                 retval |= UNDEF_CIT;
 899                                 terr.insertRef(getLineNumber(token), from_ascii("Citation undefined"),
 900                                         from_utf8(token), child_name);
 901                         //"Reference `X' on page Y undefined on input line Z."
 902                         // This warning might be broken accross multiple lines with long labels.
 903                         // Thus we check that
 904                         } else if (contains(token, "Reference `") && !contains(token, "on input line")) {
 905                                 // Rest of warning in next line(s)
 906                                 // Save to ml_token
 907                                 ml_token = token;
 908                         } else if (!ml_token.empty() && contains(ml_token, "Reference `")
 909                                    && !contains(ml_token, "on input line")) {
 910                                 // not finished yet. Continue with next line.
 911                                 continue;
 912                         } else if (!ml_token.empty() && contains(ml_token, "Reference `")
 913                                    && contains(ml_token, "on input line")) {
 914                                 // We have collected the whole warning now.
 915                                 if (!contains(ml_token, "undefined")) {
 916                                         // Not the warning we are looking for
 917                                         ml_token.clear();
 918                                         continue;
 919                                 }
 920                                 if (regex_match(ml_token, sub, undef_ref)) {
 921                                         string const ref = sub.str(1);
 922                                         Buffer const * buf = theBufferList().getBufferFromTmp(file.absFileName());
 923                                         if (!buf || !buf->masterBuffer()->activeLabel(from_utf8(ref))) {
 924                                                 terr.insertRef(getLineNumber(ml_token), from_ascii("Reference undefined"),
 925                                                         from_utf8(ml_token), child_name);
 926                                                 retval |= UNDEF_UNKNOWN_REF;
 927                                         }
 928                                 }
 929                                 ml_token.clear();
 930                                 retval |= UNDEF_REF;
 931                         } else if (contains(token, "Reference `")
 932                                    && contains(token, "on input line")
 933                                    && contains(token, "undefined")) {
 934                                 if (regex_match(token, sub, undef_ref)) {
 935                                         string const ref = sub.str(1);
 936                                         Buffer const * buf = theBufferList().getBufferFromTmp(file.absFileName());
 937                                         if (!buf || !buf->masterBuffer()->activeLabel(from_utf8(ref))) {
 938                                                 terr.insertRef(getLineNumber(token), from_ascii("Reference undefined"),
 939                                                         from_utf8(token), child_name);
 940                                                 retval |= UNDEF_UNKNOWN_REF;
 941                                         }
 942                                 }
 943                                 retval |= UNDEF_REF;
 944                         // In case the above checks fail we catch at least this generic statement
 945                         // occuring for both CIT & REF.
 946                         } else if (!runparams.includeall && contains(token, "There were undefined references.")) {
 947                                 if (!(retval & UNDEF_CIT)) //if not handled already
 948                                         retval |= UNDEF_REF;
 949                         }
 950
 951                 } else if (prefixIs(token, "Package")) {
 952                         // Package warnings
 953                         retval |= PACKAGE_WARNING;
 954                         if (contains(token, "natbib Warning:")) {
 955                                 // Natbib warnings
 956                                 if (!runparams.includeall
 957                                     && contains(token, "Citation")
 958                                     && contains(token, "on page")
 959                                     && contains(token, "undefined")) {
 960                                         retval |= UNDEF_CIT;
 961                                         //Unf only keys up to ~6 chars will make it due to line splits
 962                                         terr.insertRef(getLineNumber(token), from_ascii("Citation undefined"),
 963                                                 from_utf8(token), child_name);
 964                                 }
 965                         } else if (!runparams.includeall && contains(token, "run BibTeX")) {
 966                                 retval |= UNDEF_CIT;
 967                         } else if (!runparams.includeall && contains(token, "run Biber")) {
 968                                 retval |= UNDEF_CIT;
 969                                 biber = true;
 970                         } else if (contains(token, "Rerun LaTeX") ||
 971                                    contains(token, "Please rerun LaTeX") ||
 972                                    contains(token, "Rerun to get")) {
 973                                 // at least longtable.sty and bibtopic.sty
 974                                 // might use this.
 975                                 LYXERR(Debug::LATEX, "We should rerun.");
 976                                 retval |= RERUN;
 977                         }
 978                 } else if (prefixIs(token, "LETTRE WARNING:")) {
 979                         if (contains(token, "veuillez recompiler")) {
 980                                 // lettre.cls
 981                                 LYXERR(Debug::LATEX, "We should rerun.");
 982                                 retval |= RERUN;
 983                         }
 984                 } else if (token[0] == '(') {
 985                         if (contains(token, "Rerun LaTeX") ||
 986                             contains(token, "Rerun to get")) {
 987                                 // Used by natbib
 988                                 LYXERR(Debug::LATEX, "We should rerun.");
 989                                 retval |= RERUN;
 990                         }
 991                 } else if (prefixIs(token, "! ")
 992                             || (fle_style
 993                                 && regex_match(token, sub, file_line_error)
 994                                 && !contains(token, "pdfTeX warning"))) {
 995                            // Ok, we have something that looks like a TeX Error
 996                            // but what do we really have.
 997
 998                         // Just get the error description:
 999                         string desc;
1000                         if (prefixIs(token, "! "))
1001                                 desc = string(token, 2);
1002                         else if (fle_style)
1003                                 desc = sub.str();
1004                         if (contains(token, "LaTeX Error:"))
1005                                 retval |= LATEX_ERROR;
1006
1007                         if (prefixIs(token, "! File ended while scanning")) {
1008                                 if (prefixIs(token, "! File ended while scanning use of \\Hy@setref@link.")){
1009                                         // bug 7344. We must rerun LaTeX if hyperref has been toggled.
1010                                         retval |= ERROR_RERUN;
1011                                         LYXERR(Debug::LATEX, "Force rerun.");
1012                                 } else {
1013                                         // bug 6445. At this point its not clear we finish with error.
1014                                         wait_for_error = desc;
1015                                         continue;
1016                                 }
1017                         }
1018
1019                         if (prefixIs(token, "! Incomplete \\if")) {
1020                                 // bug 10666. At this point its not clear we finish with error.
1021                                 wait_for_error = desc;
1022                                 continue;
1023                         }
1024
1025                         if (prefixIs(token, "! Paragraph ended before \\Hy@setref@link was complete.")){
1026                                         // bug 7344. We must rerun LaTeX if hyperref has been toggled.
1027                                         retval |= ERROR_RERUN;
1028                                         LYXERR(Debug::LATEX, "Force rerun.");
1029                         }
1030
1031                         if (!wait_for_error.empty() && prefixIs(token, "! Emergency stop.")){
1032                                 retval |= LATEX_ERROR;
1033                                 string errstr;
1034                                 int count = 0;
1035                                 errstr = wait_for_error;
1036                                 wait_for_error.clear();
1037                                 do {
1038                                         if (!getline(ifs, tmp))
1039                                                 break;
1040                                         tmp = rtrim(tmp, "\r");
1041                                         errstr += "\n" + tmp;
1042                                         if (++count > 5)
1043                                                 break;
1044                                 } while (!contains(tmp, "(job aborted"));
1045
1046                                 terr.insertError(0,
1047                                                  from_ascii("Emergency stop"),
1048                                                  from_local8bit(errstr),
1049                                                  child_name);
1050                         }
1051
1052                         // get the next line
1053                         int count = 0;
1054                         // We also collect intermediate lines
1055                         // This is needed for errors in preamble
1056                         string intermediate;
1057                         do {
1058                                 if (!getline(ifs, tmp))
1059                                         break;
1060                                 tmp = rtrim(tmp, "\r");
1061                                 if (!prefixIs(tmp, "l."))
1062                                         intermediate += tmp;
1063                                 // 15 is somewhat arbitrarily chosen, based on practice.
1064                                 // We used 10 for 14 years and increased it to 15 when we
1065                                 // saw one case.
1066                                 if (++count > 15)
1067                                         break;
1068                         } while (!prefixIs(tmp, "l."));
1069                         if (prefixIs(tmp, "l.")) {
1070                                 // we have a latex error
1071                                 retval |=  TEX_ERROR;
1072                                 if (contains(desc,
1073                                         "Package babel Error: You haven't defined the language")
1074                                     || contains(desc,
1075                                         "Package babel Error: You haven't loaded the option")
1076                                     || contains(desc,
1077                                         "Package babel Error: Unknown language"))
1078                                         retval |= ERROR_RERUN;
1079                                 // get the line number:
1080                                 int line = 0;
1081                                 sscanf(tmp.c_str(), "l.%d", &line);
1082                                 // get the rest of the message:
1083                                 string errstr(tmp, tmp.find(' '));
1084                                 bool preamble_error = false;
1085                                 if (suffixIs(errstr, "\\begin{document}")) {
1086                                         // this is an error in preamble
1087                                         // the real error is in the
1088                                         // intermediate lines
1089                                         errstr = intermediate;
1090                                         tmp = intermediate;
1091                                         preamble_error = true;
1092                                 }
1093                                 errstr += '\n';
1094                                 getline(ifs, tmp);
1095                                 tmp = rtrim(tmp, "\r");
1096                                 while (!contains(errstr, "l.")
1097                                        && !tmp.empty()
1098                                        && !prefixIs(tmp, "! ")
1099                                        && !contains(tmp, "(job aborted")) {
1100                                         errstr += tmp;
1101                                         errstr += "\n";
1102                                         getline(ifs, tmp);
1103                                         tmp = rtrim(tmp, "\r");
1104                                 }
1105                                 if (preamble_error)
1106                                         // Add a note that the error is to be found in preamble
1107                                         errstr += "\n" + to_utf8(_("(NOTE: The erroneous command is in the preamble)"));
1108                                 LYXERR(Debug::LATEX, "line: " << line << '\n'
1109                                         << "Desc: " << desc << '\n' << "Text: " << errstr);
1110                                 if (line == last_line)
1111                                         ++line_count;
1112                                 else {
1113                                         line_count = 1;
1114                                         last_line = line;
1115                                 }
1116                                 if (line_count <= 5) {
1117                                         // FIXME UNICODE
1118                                         // We have no idea what the encoding of
1119                                         // the log file is.
1120                                         // It seems that the output from the
1121                                         // latex compiler itself is pure ASCII,
1122                                         // but it can include bits from the
1123                                         // document, so whatever encoding we
1124                                         // assume here it can be wrong.
1125                                         terr.insertError(line,
1126                                                          from_local8bit(desc),
1127                                                          from_local8bit(errstr),
1128                                                          child_name);
1129                                         ++num_errors;
1130                                 }
1131                         }
1132                 } else {
1133                         // information messages, TeX warnings and other
1134                         // warnings we have not caught earlier.
1135                         if (prefixIs(token, "Overfull ")) {
1136                                 retval |= TEX_WARNING;
1137                         } else if (prefixIs(token, "Underfull ")) {
1138                                 retval |= TEX_WARNING;
1139                         } else if (!runparams.includeall && contains(token, "Rerun to get citations")) {
1140                                 // Natbib seems to use this.
1141                                 retval |= UNDEF_CIT;
1142                         } else if (contains(token, "No pages of output")
1143                                 || contains(token, "no pages of output")) {
1144                                 // No output file (e.g. the DVI or PDF) was created
1145                                 retval |= NO_OUTPUT;
1146                         } else if (contains(token, "Error 256 (driver return code)")) {
1147                                 // This is a xdvipdfmx driver error reported by XeTeX.
1148                                 // We have to check whether an output PDF file was created.
1149                                 FileName pdffile = file;
1150                                 pdffile.changeExtension("pdf");
1151                                 if (!pdffile.exists())
1152                                         // No output PDF file was created (see #10076)
1153                                         retval |= NO_OUTPUT;
1154                         } else if (contains(token, "That makes 100 errors")) {
1155                                 // More than 100 errors were reported
1156                                 retval |= TOO_MANY_ERRORS;
1157                         } else if (prefixIs(token, "!pdfTeX error:")) {
1158                                 // otherwise we dont catch e.g.:
1159                                 // !pdfTeX error: pdflatex (file feyn10): Font feyn10 at 600 not found
1160                                 retval |= ERRORS;
1161                                 terr.insertError(0,
1162                                                  from_ascii("pdfTeX Error"),
1163                                                  from_local8bit(token),
1164                                                  child_name);
1165                         } else if (!ignore_missing_glyphs
1166                                    && prefixIs(token, "Missing character: There is no ")
1167                                    && !contains(token, "nullfont")) {
1168                                 // Warning about missing glyph in selected font
1169                                 // may be dataloss (bug 9610)
1170                                 // but can be ignored for 'nullfont' (bug 10394).
1171                                 // as well as for ZERO WIDTH NON-JOINER (0x200C) which is
1172                                 // missing in many fonts and output for ligature break (bug 10727).
1173                                 // Since this error only occurs with utf8 output, we can safely assume
1174                                 // that the log file is utf8-encoded
1175                                 docstring const utoken = from_utf8(token);
1176                                 if (!contains(utoken, 0x200C)) {
1177                                         retval |= LATEX_ERROR;
1178                                         terr.insertError(0,
1179                                                          from_ascii("Missing glyphs!"),
1180                                                          utoken,
1181                                                          child_name);
1182                                 }
1183                         } else if (!wait_for_error.empty()) {
1184                                 // We collect information until we know we have an error.
1185                                 wait_for_error += token + '\n';
1186                         }
1187                 }
1188         }
1189         LYXERR(Debug::LATEX, "Log line: " << token);
1190         return retval;
1191 }
1192
1193
1194 namespace {
1195
1196 bool insertIfExists(FileName const & absname, DepTable & head)
1197 {
1198         if (absname.exists() && !absname.isDirectory()) {
1199                 head.insert(absname, true);
1200                 return true;
1201         }
1202         return false;
1203 }
1204
1205
1206 bool handleFoundFile(string const & ff, DepTable & head)
1207 {
1208         // convert from native os path to unix path
1209         string foundfile = os::internal_path(trim(ff));
1210
1211         LYXERR(Debug::DEPEND, "Found file: " << foundfile);
1212
1213         // Ok now we found a file.
1214         // Now we should make sure that this is a file that we can
1215         // access through the normal paths.
1216         // We will not try any fancy search methods to
1217         // find the file.
1218
1219         // (1) foundfile is an
1220         //     absolute path and should
1221         //     be inserted.
1222         FileName absname;
1223         if (FileName::isAbsolute(foundfile)) {
1224                 LYXERR(Debug::DEPEND, "AbsolutePath file: " << foundfile);
1225                 // On initial insert we want to do the update at once
1226                 // since this file cannot be a file generated by
1227                 // the latex run.
1228                 absname.set(foundfile);
1229                 if (!insertIfExists(absname, head)) {
1230                         // check for spaces
1231                         string strippedfile = foundfile;
1232                         while (contains(strippedfile, " ")) {
1233                                 // files with spaces are often enclosed in quotation
1234                                 // marks; those have to be removed
1235                                 string unquoted = subst(strippedfile, "\"", "");
1236                                 absname.set(unquoted);
1237                                 if (insertIfExists(absname, head))
1238                                         return true;
1239                                 // strip off part after last space and try again
1240                                 string tmp = strippedfile;
1241                                 rsplit(tmp, strippedfile, ' ');
1242                                 absname.set(strippedfile);
1243                                 if (insertIfExists(absname, head))
1244                                         return true;
1245                         }
1246                 }
1247         }
1248
1249         string onlyfile = onlyFileName(foundfile);
1250         absname = makeAbsPath(onlyfile);
1251
1252         // check for spaces
1253         while (contains(foundfile, ' ')) {
1254                 if (absname.exists())
1255                         // everything o.k.
1256                         break;
1257                 else {
1258                         // files with spaces are often enclosed in quotation
1259                         // marks; those have to be removed
1260                         string unquoted = subst(foundfile, "\"", "");
1261                         absname = makeAbsPath(unquoted);
1262                         if (absname.exists())
1263                                 break;
1264                         // strip off part after last space and try again
1265                         string strippedfile;
1266                         rsplit(foundfile, strippedfile, ' ');
1267                         foundfile = strippedfile;
1268                         onlyfile = onlyFileName(strippedfile);
1269                         absname = makeAbsPath(onlyfile);
1270                 }
1271         }
1272
1273         // (2) foundfile is in the tmpdir
1274         //     insert it into head
1275         if (absname.exists() && !absname.isDirectory()) {
1276                 // FIXME: This regex contained glo, but glo is used by the old
1277                 // version of nomencl.sty. Do we need to put it back?
1278                 static regex const unwanted("^.*\\.(aux|log|dvi|bbl|ind)$");
1279                 if (regex_match(onlyfile, unwanted)) {
1280                         LYXERR(Debug::DEPEND, "We don't want " << onlyfile
1281                                 << " in the dep file");
1282                 } else if (suffixIs(onlyfile, ".tex")) {
1283                         // This is a tex file generated by LyX
1284                         // and latex is not likely to change this
1285                         // during its runs.
1286                         LYXERR(Debug::DEPEND, "Tmpdir TeX file: " << onlyfile);
1287                         head.insert(absname, true);
1288                 } else {
1289                         LYXERR(Debug::DEPEND, "In tmpdir file:" << onlyfile);
1290                         head.insert(absname);
1291                 }
1292                 return true;
1293         } else {
1294                 LYXERR(Debug::DEPEND, "Not a file or we are unable to find it.");
1295                 return false;
1296         }
1297 }
1298
1299
1300 bool completeFilename(string const & ff, DepTable & head)
1301 {
1302         // If we do not find a dot, we suspect
1303         // a fragmental file name
1304         if (!contains(ff, '.'))
1305                 return false;
1306
1307         // if we have a dot, we let handleFoundFile decide
1308         return handleFoundFile(ff, head);
1309 }
1310
1311
1312 int iterateLine(string const & token, regex const & reg, string const & opening,
1313                 string const & closing, int fragment_pos, DepTable & head)
1314 {
1315         smatch what;
1316         string::const_iterator first = token.begin();
1317         string::const_iterator end = token.end();
1318         bool fragment = false;
1319         string last_match;
1320
1321         while (regex_search(first, end, what, reg)) {
1322                 // if we have a dot, try to handle as file
1323                 if (contains(what.str(1), '.')) {
1324                         first = what[0].second;
1325                         if (what.str(2) == closing) {
1326                                 handleFoundFile(what.str(1), head);
1327                                 // since we had a closing bracket,
1328                                 // do not investigate further
1329                                 fragment = false;
1330                         } else if (what.str(2) == opening) {
1331                                 // if we have another opening bracket,
1332                                 // we might have a nested file chain
1333                                 // as is (file.ext (subfile.ext))
1334                                 fragment = !handleFoundFile(rtrim(what.str(1)), head);
1335                                 // decrease first position by one in order to
1336                                 // consider the opening delimiter on next iteration
1337                                 if (first > token.begin())
1338                                         --first;
1339                         } else
1340                                 // if we have no closing bracket,
1341                                 // try to handle as file nevertheless
1342                                 fragment = !handleFoundFile(
1343                                         what.str(1) + what.str(2), head);
1344                 }
1345                 // if we do not have a dot, check if the line has
1346                 // a closing bracket (else, we suspect a line break)
1347                 else if (what.str(2) != closing) {
1348                         first = what[0].second;
1349                         fragment = true;
1350                 } else {
1351                         // we have a closing bracket, so the content
1352                         // is not a file name.
1353                         // no need to investigate further
1354                         first = what[0].second;
1355                         fragment = false;
1356                 }
1357                 last_match = what.str(1);
1358         }
1359
1360         // We need to consider the result from previous line iterations:
1361         // We might not find a fragment here, but another one might follow
1362         // E.g.: (filename.ext) <filenam
1363         // Vice versa, we consider the search completed if a real match
1364         // follows a potential fragment from a previous iteration.
1365         // E.g. <some text we considered a fragment (filename.ext)
1366         // result = -1 means we did not find a fragment!
1367         int result = -1;
1368         int last_match_pos = -1;
1369         if (!last_match.empty() && token.find(last_match) != string::npos)
1370                 last_match_pos = int(token.find(last_match));
1371         if (fragment) {
1372                 if (last_match_pos > fragment_pos)
1373                         result = last_match_pos;
1374                 else
1375                         result = fragment_pos;
1376         } else
1377                 if (last_match_pos < fragment_pos)
1378                         result = fragment_pos;
1379
1380         return result;
1381 }
1382
1383 } // namespace
1384
1385
1386 void LaTeX::deplog(DepTable & head)
1387 {
1388         // This function reads the LaTeX log file end extracts all the
1389         // external files used by the LaTeX run. The files are then
1390         // entered into the dependency file.
1391
1392         string const logfile =
1393                 onlyFileName(changeExtension(file.absFileName(), ".log"));
1394
1395         static regex const reg1("File: (.+).*");
1396         static regex const reg2("No file (.+)(.).*");
1397         static regex const reg3a("\\\\openout[0-9]+.*=.*`(.+)(..).*");
1398         // LuaTeX has a slightly different output
1399         static regex const reg3b("\\\\openout[0-9]+.*=\\s*(.+)");
1400         // If an index should be created, MikTex does not write a line like
1401         //    \openout# = 'sample.idx'.
1402         // but instead only a line like this into the log:
1403         //   Writing index file sample.idx
1404         static regex const reg4("Writing index file (.+).*");
1405         static regex const regoldnomencl("Writing glossary file (.+).*");
1406         static regex const regnomencl(".*Writing nomenclature file (.+).*");
1407         // If a toc should be created, MikTex does not write a line like
1408         //    \openout# = `sample.toc'.
1409         // but only a line like this into the log:
1410         //    \tf@toc=\write#
1411         // This line is also written by tetex.
1412         // This line is not present if no toc should be created.
1413         static regex const miktexTocReg("\\\\tf@toc=\\\\write.*");
1414         // file names can be enclosed in <...> (anywhere on the line)
1415         static regex const reg5(".*<[^>]+.*");
1416         // and also (...) anywhere on the line
1417         static regex const reg6(".*\\([^)]+.*");
1418
1419         FileName const fn = makeAbsPath(logfile);
1420         ifstream ifs(fn.toFilesystemEncoding().c_str());
1421         string lastline;
1422         while (ifs) {
1423                 // Ok, the scanning of files here is not sufficient.
1424                 // Sometimes files are named by "File: xxx" only
1425                 // Therefore we use some regexps to find files instead.
1426                 // Note: all file names and paths might contains spaces.
1427                 // Also, file names might be broken across lines. Therefore
1428                 // we mark (potential) fragments and merge those lines.
1429                 bool fragment = false;
1430                 string token;
1431                 getline(ifs, token);
1432                 // MikTeX sometimes inserts \0 in the log file. They can't be
1433                 // removed directly with the existing string utility
1434                 // functions, so convert them first to \r, and remove all
1435                 // \r's afterwards, since we need to remove them anyway.
1436                 token = subst(token, '\0', '\r');
1437                 token = subst(token, "\r", "");
1438                 if (token.empty() || token == ")") {
1439                         lastline = string();
1440                         continue;
1441                 }
1442
1443                 // FIXME UNICODE: We assume that the file names in the log
1444                 // file are in the file system encoding.
1445                 token = to_utf8(from_filesystem8bit(token));
1446
1447                 // Sometimes, filenames are broken across lines.
1448                 // We care for that and save suspicious lines.
1449                 // Here we exclude some cases where we are sure
1450                 // that there is no continued filename
1451                 if (!lastline.empty()) {
1452                         static regex const package_info("Package \\w+ Info: .*");
1453                         static regex const package_warning("Package \\w+ Warning: .*");
1454                         if (prefixIs(token, "File:") || prefixIs(token, "(Font)")
1455                             || prefixIs(token, "Package:")
1456                             || prefixIs(token, "Language:")
1457                             || prefixIs(token, "LaTeX Info:")
1458                             || prefixIs(token, "LaTeX Font Info:")
1459                             || prefixIs(token, "\\openout[")
1460                             || prefixIs(token, "))")
1461                             || regex_match(token, package_info)
1462                             || regex_match(token, package_warning))
1463                                 lastline = string();
1464                 }
1465
1466                 if (!lastline.empty())
1467                         // probably a continued filename from last line
1468                         token = lastline + token;
1469                 if (token.length() > 255) {
1470                         // string too long. Cut off.
1471                         token.erase(0, token.length() - 251);
1472                 }
1473
1474                 smatch sub;
1475
1476                 // (1) "File: file.ext"
1477                 if (regex_match(token, sub, reg1)) {
1478                         // is this a fragmental file name?
1479                         fragment = !completeFilename(sub.str(1), head);
1480                         // However, ...
1481                         if (suffixIs(token, ")"))
1482                                 // no fragment for sure
1483                                 fragment = false;
1484                 // (2) "No file file.ext"
1485                 } else if (regex_match(token, sub, reg2)) {
1486                         // file names must contains a dot, line ends with dot
1487                         if (contains(sub.str(1), '.') && sub.str(2) == ".")
1488                                 fragment = !handleFoundFile(sub.str(1), head);
1489                         else
1490                                 // we suspect a line break
1491                                 fragment = true;
1492                 // (3)(a) "\openout<nr> = `file.ext'."
1493                 } else if (regex_match(token, sub, reg3a)) {
1494                         // search for closing '. at the end of the line
1495                         if (sub.str(2) == "\'.")
1496                                 fragment = !handleFoundFile(sub.str(1), head);
1497                         else
1498                                 // potential fragment
1499                                 fragment = true;
1500                 // (3)(b) "\openout<nr> = file.ext" (LuaTeX)
1501                 } else if (regex_match(token, sub, reg3b)) {
1502                         // file names must contains a dot
1503                         if (contains(sub.str(1), '.'))
1504                                 fragment = !handleFoundFile(sub.str(1), head);
1505                         else
1506                                 // potential fragment
1507                                 fragment = true;
1508                 // (4) "Writing index file file.ext"
1509                 } else if (regex_match(token, sub, reg4))
1510                         // fragmential file name?
1511                         fragment = !completeFilename(sub.str(1), head);
1512                 // (5) "Writing nomenclature file file.ext"
1513                 else if (regex_match(token, sub, regnomencl) ||
1514                            regex_match(token, sub, regoldnomencl))
1515                         // fragmental file name?
1516                         fragment= !completeFilename(sub.str(1), head);
1517                 // (6) "\tf@toc=\write<nr>" (for MikTeX)
1518                 else if (regex_match(token, sub, miktexTocReg))
1519                         fragment = !handleFoundFile(onlyFileName(changeExtension(
1520                                                 file.absFileName(), ".toc")), head);
1521                 else
1522                         // not found, but we won't check further
1523                         fragment = false;
1524
1525                 int fragment_pos = -1;
1526                 // (7) "<file.ext>"
1527                 // We can have several of these on one line
1528                 // (and in addition to those above)
1529                 if (regex_match(token, sub, reg5)) {
1530                         // search for strings in <...>
1531                         static regex const reg5_1("<([^>]+)(.)");
1532                         fragment_pos = iterateLine(token, reg5_1, "<", ">",
1533                                                    fragment_pos, head);
1534                         fragment = (fragment_pos != -1);
1535                 }
1536
1537                 // (8) "(file.ext)"
1538                 // We can have several of these on one line
1539                 // this must be queried separated, because of
1540                 // cases such as "File: file.ext (type eps)"
1541                 // where "File: file.ext" would be skipped
1542                 if (regex_match(token, sub, reg6)) {
1543                         // search for strings in (...)
1544                         static regex const reg6_1("\\(([^()]+)(.)");
1545                         fragment_pos = iterateLine(token, reg6_1, "(", ")",
1546                                                    fragment_pos, head);
1547                         fragment = (fragment_pos != -1);
1548                 }
1549
1550                 if (fragment)
1551                         // probable linebreak within file name:
1552                         // save this line
1553                         lastline = token;
1554                 else
1555                         // no linebreak: reset
1556                         lastline = string();
1557         }
1558
1559         // Make sure that the main .tex file is in the dependency file.
1560         head.insert(file, true);
1561 }
1562
1563
1564 int LaTeX::scanBlgFile(DepTable & dep, TeXErrors & terr)
1565 {
1566         FileName const blg_file(changeExtension(file.absFileName(), "blg"));
1567         LYXERR(Debug::LATEX, "Scanning blg file: " << blg_file);
1568
1569         ifstream ifs(blg_file.toFilesystemEncoding().c_str());
1570         string token;
1571         static regex const reg1(".*Found (bibtex|BibTeX) data (file|source) '([^']+).*");
1572         static regex const bibtexError("^(.*---line [0-9]+ of file).*$");
1573         static regex const bibtexError2("^(.*---while reading file).*$");
1574         static regex const bibtexError3("(A bad cross reference---).*");
1575         static regex const bibtexError4("(Sorry---you've exceeded BibTeX's).*");
1576         static regex const bibtexError5("\\*Please notify the BibTeX maintainer\\*");
1577         static regex const biberError("^.*> (FATAL|ERROR) - (.*)$");
1578         int retval = NO_ERRORS;
1579
1580         string prevtoken;
1581         while (getline(ifs, token)) {
1582                 token = rtrim(token, "\r");
1583                 smatch sub;
1584                 // FIXME UNICODE: We assume that citation keys and filenames
1585                 // in the aux file are in the file system encoding.
1586                 token = to_utf8(from_filesystem8bit(token));
1587                 if (regex_match(token, sub, reg1)) {
1588                         string data = sub.str(3);
1589                         if (!data.empty()) {
1590                                 LYXERR(Debug::LATEX, "Found bib file: " << data);
1591                                 handleFoundFile(data, dep);
1592                         }
1593                 }
1594                 else if (regex_match(token, sub, bibtexError)
1595                          || regex_match(token, sub, bibtexError2)
1596                          || regex_match(token, sub, bibtexError4)
1597                          || regex_match(token, sub, bibtexError5)) {
1598                         retval |= BIBTEX_ERROR;
1599                         string errstr = N_("BibTeX error: ") + token;
1600                         string msg;
1601                         if ((prefixIs(token, "while executing---line")
1602                              || prefixIs(token, "---line ")
1603                              || prefixIs(token, "*Please notify the BibTeX"))
1604                             && !prevtoken.empty()) {
1605                                 errstr = N_("BibTeX error: ") + prevtoken;
1606                                 msg = prevtoken + '\n';
1607                         }
1608                         msg += token;
1609                         terr.insertError(0,
1610                                          from_local8bit(errstr),
1611                                          from_local8bit(msg));
1612                 } else if (regex_match(prevtoken, sub, bibtexError3)) {
1613                         retval |= BIBTEX_ERROR;
1614                         string errstr = N_("BibTeX error: ") + prevtoken;
1615                         string msg = prevtoken + '\n' + token;
1616                         terr.insertError(0,
1617                                          from_local8bit(errstr),
1618                                          from_local8bit(msg));
1619                 } else if (regex_match(token, sub, biberError)) {
1620                         retval |= BIBTEX_ERROR;
1621                         string errstr = N_("Biber error: ") + sub.str(2);
1622                         string msg = token;
1623                         terr.insertError(0,
1624                                          from_local8bit(errstr),
1625                                          from_local8bit(msg));
1626                 }
1627                 prevtoken = token;
1628         }
1629         return retval;
1630 }
1631
1632
1633 int LaTeX::scanIlgFile(TeXErrors & terr)
1634 {
1635         FileName const ilg_file(changeExtension(file.absFileName(), "ilg"));
1636         LYXERR(Debug::LATEX, "Scanning ilg file: " << ilg_file);
1637
1638         ifstream ifs(ilg_file.toFilesystemEncoding().c_str());
1639         string token;
1640         int retval = NO_ERRORS;
1641
1642         string prevtoken;
1643         while (getline(ifs, token)) {
1644                 token = rtrim(token, "\r");
1645                 smatch sub;
1646                 if (prefixIs(token, "!! "))
1647                         prevtoken = token;
1648                 else if (!prevtoken.empty()) {
1649                         retval |= INDEX_ERROR;
1650                         string errstr = N_("Makeindex error: ") + prevtoken;
1651                         string msg = prevtoken + '\n';
1652                         msg += token;
1653                         terr.insertError(0,
1654                                          from_local8bit(errstr),
1655                                          from_local8bit(msg));
1656                         prevtoken.clear();
1657                 } else if (prefixIs(token, "ERROR: ")) {
1658                         retval |= BIBTEX_ERROR;
1659                         string errstr = N_("Xindy error: ") + token.substr(6);
1660                         string msg = token;
1661                         terr.insertError(0,
1662                                          from_local8bit(errstr),
1663                                          from_local8bit(msg));
1664                 }
1665         }
1666         return retval;
1667 }
1668
1669
1670
1671 } // namespace lyx