3 * This file is part of LyX, the document processor.
4 * Licence details can be found in the file COPYING.
6 * \author Alfredo Braunstein
7 * \author Lars Gullik Bjønnes
8 * \author Jean-Marc Lasgouttes
9 * \author Angus Leeming
11 * \author Jürgen Spitzmüller
13 * Full author contact details are available in file CREDITS.
18 #include "BufferList.h"
23 #include "support/debug.h"
24 #include "support/convert.h"
25 #include "support/FileName.h"
26 #include "support/filetools.h"
27 #include "support/gettext.h"
28 #include "support/lstrings.h"
29 #include "support/Systemcall.h"
30 #include "support/os.h"
32 #include "support/regex.h"
39 using namespace lyx::support;
43 namespace os = support::os;
45 // TODO: in no particular order
46 // - get rid of the call to
47 // BufferList::updateIncludedTeXfiles, this should either
48 // be done before calling LaTeX::funcs or in a completely
50 // - the makeindex style files should be taken care of with
51 // the dependency mechanism.
52 // - we should perhaps also scan the bibtex log file
56 docstring runMessage(unsigned int count)
58 return bformat(_("Waiting for LaTeX run number %1$d"), count);
67 void TeXErrors::insertError(int line, docstring const & error_desc,
68 docstring const & error_text,
69 string const & child_name)
71 Error newerr(line, error_desc, error_text, child_name);
72 errors.push_back(newerr);
76 bool operator==(AuxInfo const & a, AuxInfo const & o)
78 return a.aux_file == o.aux_file
79 && a.citations == o.citations
80 && a.databases == o.databases
81 && a.styles == o.styles;
85 bool operator!=(AuxInfo const & a, AuxInfo const & o)
95 LaTeX::LaTeX(string const & latex, OutputParams const & rp,
96 FileName const & f, string const & p)
97 : cmd(latex), file(f), path(p), runparams(rp), biber(false)
100 if (prefixIs(cmd, "pdf")) { // Do we use pdflatex ?
101 depfile = FileName(file.absFileName() + ".dep-pdf");
103 FileName(changeExtension(file.absFileName(), ".pdf"));
105 depfile = FileName(file.absFileName() + ".dep");
107 FileName(changeExtension(file.absFileName(), ".dvi"));
112 void LaTeX::deleteFilesOnError() const
114 // currently just a dummy function.
116 // What files do we have to delete?
118 // This will at least make latex do all the runs
119 depfile.removeFile();
121 // but the reason for the error might be in a generated file...
124 FileName const bbl(changeExtension(file.absFileName(), ".bbl"));
128 FileName const bcf(changeExtension(file.absFileName(), ".bcf"));
132 FileName const ind(changeExtension(file.absFileName(), ".ind"));
136 FileName const nls(changeExtension(file.absFileName(), ".nls"));
139 // nomencl file (old version of the package)
140 FileName const gls(changeExtension(file.absFileName(), ".gls"));
143 // Also remove the aux file
144 FileName const aux(changeExtension(file.absFileName(), ".aux"));
149 int LaTeX::run(TeXErrors & terr)
150 // We know that this function will only be run if the lyx buffer
151 // has been changed. We also know that a newly written .tex file
152 // is always different from the previous one because of the date
153 // in it. However it seems safe to run latex (at least) one time
154 // each time the .tex file changes.
156 int scanres = NO_ERRORS;
157 unsigned int count = 0; // number of times run
158 num_errors = 0; // just to make sure.
159 unsigned int const MAX_RUN = 6;
160 DepTable head; // empty head
161 bool rerun = false; // rerun requested
163 // The class LaTeX does not know the temp path.
164 theBufferList().updateIncludedTeXfiles(FileName::getcwd().absFileName(),
167 // Never write the depfile if an error was encountered.
170 // first check if the file dependencies exist:
171 // ->If it does exist
172 // check if any of the files mentioned in it have
173 // changed (done using a checksum).
175 // run latex once and
176 // remake the dependency file
177 // -> if not changed:
178 // just return there is nothing to do for us.
179 // ->if it doesn't exist
181 // run latex once (we need to run latex once anyway) and
182 // remake the dependency file.
185 bool had_depfile = depfile.exists();
186 bool run_bibtex = false;
187 FileName const aux_file(changeExtension(file.absFileName(), ".aux"));
190 LYXERR(Debug::DEPEND, "Dependency file exists");
191 // Read the dep file:
192 had_depfile = head.read(depfile);
196 // Update the checksums
198 // Can't just check if anything has changed because it might
199 // have aborted on error last time... in which cas we need
200 // to re-run latex and collect the error messages
201 // (even if they are the same).
202 if (!output_file.exists()) {
203 LYXERR(Debug::DEPEND,
204 "re-running LaTeX because output file doesn't exist.");
205 } else if (!head.sumchange()) {
206 LYXERR(Debug::DEPEND, "return no_change");
209 LYXERR(Debug::DEPEND, "Dependency file has changed");
212 if (head.extchanged(".bib") || head.extchanged(".bst"))
215 LYXERR(Debug::DEPEND,
216 "Dependency file does not exist, or has wrong format");
218 /// We scan the aux file even when had_depfile = false,
219 /// because we can run pdflatex on the file after running latex on it,
220 /// in which case we will not need to run bibtex again.
221 vector<AuxInfo> bibtex_info_old;
223 bibtex_info_old = scanAuxFiles(aux_file);
226 LYXERR(Debug::LATEX, "Run #" << count);
227 message(runMessage(count));
230 scanres = scanLogFile(terr);
231 if (scanres & ERROR_RERUN) {
232 LYXERR(Debug::LATEX, "Rerunning LaTeX");
234 scanres = scanLogFile(terr);
237 if (scanres & ERRORS) {
238 deleteFilesOnError();
239 return scanres; // return on error
242 vector<AuxInfo> const bibtex_info = scanAuxFiles(aux_file);
243 if (!run_bibtex && bibtex_info_old != bibtex_info)
246 // update the dependencies.
247 deplog(head); // reads the latex log
251 // At this point we must run external programs if needed.
252 // makeindex will be run if a .idx file changed or was generated.
253 // And if there were undefined citations or changes in references
254 // the .aux file is checked for signs of bibtex. Bibtex is then run
257 // memoir (at least) writes an empty *idx file in the first place.
258 // A second latex run is needed.
259 FileName const idxfile(changeExtension(file.absFileName(), ".idx"));
260 rerun = idxfile.exists() && idxfile.isFileEmpty();
263 if (head.haschanged(idxfile)) {
265 LYXERR(Debug::LATEX, "Running MakeIndex.");
266 message(_("Running Index Processor."));
267 // onlyFileName() is needed for cygwin
268 rerun |= runMakeIndex(onlyFileName(idxfile.absFileName()),
271 FileName const nlofile(changeExtension(file.absFileName(), ".nlo"));
272 if (head.haschanged(nlofile))
273 rerun |= runMakeIndexNomencl(file, ".nlo", ".nls");
274 FileName const glofile(changeExtension(file.absFileName(), ".glo"));
275 if (head.haschanged(glofile))
276 rerun |= runMakeIndexNomencl(file, ".glo", ".gls");
278 // check if we're using biber instead of bibtex
279 // biber writes no info to the aux file, so we just check
280 // if a bcf file exists (and if it was updated)
281 FileName const bcffile(changeExtension(file.absFileName(), ".bcf"));
282 biber |= head.exist(bcffile);
285 // if (scanres & UNDEF_CIT || scanres & RERUN || run_bibtex)
286 if (scanres & UNDEF_CIT || run_bibtex) {
287 // Here we must scan the .aux file and look for
288 // "\bibdata" and/or "\bibstyle". If one of those
289 // tags is found -> run bibtex and set rerun = true;
291 LYXERR(Debug::LATEX, "Running BibTeX.");
292 message(_("Running BibTeX."));
293 updateBibtexDependencies(head, bibtex_info);
294 rerun |= runBibTeX(bibtex_info, runparams);
296 // since biber writes no info to the aux file, we have
297 // to parse the blg file (which only exists after biber
299 FileName const blgfile(changeExtension(file.absFileName(), ".blg"));
300 if (blgfile.exists())
303 } else if (!had_depfile) {
304 /// If we run pdflatex on the file after running latex on it,
305 /// then we do not need to run bibtex, but we do need to
306 /// insert the .bib and .bst files into the .dep-pdf file.
307 updateBibtexDependencies(head, bibtex_info);
311 // we know on this point that latex has been run once (or we just
312 // returned) and the question now is to decide if we need to run
313 // it any more. This is done by asking if any of the files in the
314 // dependency file has changed. (remember that the checksum for
315 // a given file is reported to have changed if it just was created)
316 // -> if changed or rerun == true:
317 // run latex once more and
318 // update the dependency structure
319 // -> if not changed:
320 // we do nothing at this point
322 if (rerun || head.sumchange()) {
325 LYXERR(Debug::DEPEND, "Dep. file has changed or rerun requested");
326 LYXERR(Debug::LATEX, "Run #" << count);
327 message(runMessage(count));
329 scanres = scanLogFile(terr);
330 if (scanres & ERRORS) {
331 deleteFilesOnError();
332 return scanres; // return on error
335 // update the depedencies
336 deplog(head); // reads the latex log
339 LYXERR(Debug::DEPEND, "Dep. file has NOT changed");
344 // Complex bibliography packages such as Biblatex require
345 // an additional bibtex cycle sometimes.
346 if (scanres & UNDEF_CIT) {
347 // Here we must scan the .aux file and look for
348 // "\bibdata" and/or "\bibstyle". If one of those
349 // tags is found -> run bibtex and set rerun = true;
351 LYXERR(Debug::LATEX, "Running BibTeX.");
352 message(_("Running BibTeX."));
353 updateBibtexDependencies(head, bibtex_info);
354 rerun |= runBibTeX(bibtex_info, runparams);
358 // The inclusion of files generated by external programs such as
359 // makeindex or bibtex might have done changes to pagenumbering,
360 // etc. And because of this we must run the external programs
361 // again to make sure everything is redone correctly.
362 // Also there should be no need to run the external programs any
365 // run makeindex if the <file>.idx has changed or was generated.
366 if (head.haschanged(idxfile)) {
368 LYXERR(Debug::LATEX, "Running MakeIndex.");
369 message(_("Running Index Processor."));
370 // onlyFileName() is needed for cygwin
371 rerun = runMakeIndex(onlyFileName(changeExtension(
372 file.absFileName(), ".idx")), runparams);
375 // I am not pretty sure if need this twice.
376 if (head.haschanged(nlofile))
377 rerun |= runMakeIndexNomencl(file, ".nlo", ".nls");
378 if (head.haschanged(glofile))
379 rerun |= runMakeIndexNomencl(file, ".glo", ".gls");
382 // we will only run latex more if the log file asks for it.
383 // or if the sumchange() is true.
384 // -> rerun asked for:
386 // remake the dependency file
387 // goto 2 or return if max runs are reached.
388 // -> rerun not asked for:
389 // just return (fall out of bottom of func)
391 while ((head.sumchange() || rerun || (scanres & RERUN))
392 && count < MAX_RUN) {
393 // Yes rerun until message goes away, or until
394 // MAX_RUNS are reached.
397 LYXERR(Debug::LATEX, "Run #" << count);
398 message(runMessage(count));
400 scanres = scanLogFile(terr);
401 if (scanres & ERRORS) {
402 deleteFilesOnError();
403 return scanres; // return on error
410 // Write the dependencies to file.
412 LYXERR(Debug::LATEX, "Done.");
417 int LaTeX::startscript()
419 // onlyFileName() is needed for cygwin
420 string tmp = cmd + ' '
421 + quoteName(onlyFileName(file.toFilesystemEncoding()))
422 + " > " + os::nulldev();
424 return one.startscript(Systemcall::Wait, tmp, path);
428 bool LaTeX::runMakeIndex(string const & f, OutputParams const & runparams,
429 string const & params)
431 string tmp = runparams.use_japanese ?
432 lyxrc.jindex_command : lyxrc.index_command;
434 if (!runparams.index_command.empty())
435 tmp = runparams.index_command;
438 "idx file has been made, running index processor ("
439 << tmp << ") on file " << f);
441 tmp = subst(tmp, "$$lang", runparams.document_language);
442 if (runparams.use_indices) {
443 tmp = lyxrc.splitindex_command + " -m " + quoteName(tmp);
445 "Multiple indices. Using splitindex command: " << tmp);
451 one.startscript(Systemcall::Wait, tmp, path);
456 bool LaTeX::runMakeIndexNomencl(FileName const & file,
457 string const & nlo, string const & nls)
459 LYXERR(Debug::LATEX, "Running MakeIndex for nomencl.");
460 message(_("Running MakeIndex for nomencl."));
461 string tmp = lyxrc.nomencl_command + ' ';
462 // onlyFileName() is needed for cygwin
463 tmp += quoteName(onlyFileName(changeExtension(file.absFileName(), nlo)));
465 + onlyFileName(changeExtension(file.toFilesystemEncoding(), nls));
467 one.startscript(Systemcall::Wait, tmp, path);
472 vector<AuxInfo> const
473 LaTeX::scanAuxFiles(FileName const & file)
475 vector<AuxInfo> result;
477 result.push_back(scanAuxFile(file));
479 string const basename = removeExtension(file.absFileName());
480 for (int i = 1; i < 1000; ++i) {
481 FileName const file2(basename
482 + '.' + convert<string>(i)
486 result.push_back(scanAuxFile(file2));
492 AuxInfo const LaTeX::scanAuxFile(FileName const & file)
495 result.aux_file = file;
496 scanAuxFile(file, result);
501 void LaTeX::scanAuxFile(FileName const & file, AuxInfo & aux_info)
503 LYXERR(Debug::LATEX, "Scanning aux file: " << file);
505 ifstream ifs(file.toFilesystemEncoding().c_str());
507 static regex const reg1("\\\\citation\\{([^}]+)\\}");
508 static regex const reg2("\\\\bibdata\\{([^}]+)\\}");
509 static regex const reg3("\\\\bibstyle\\{([^}]+)\\}");
510 static regex const reg4("\\\\@input\\{([^}]+)\\}");
512 while (getline(ifs, token)) {
513 token = rtrim(token, "\r");
515 // FIXME UNICODE: We assume that citation keys and filenames
516 // in the aux file are in the file system encoding.
517 token = to_utf8(from_filesystem8bit(token));
518 if (regex_match(token, sub, reg1)) {
519 string data = sub.str(1);
520 while (!data.empty()) {
522 data = split(data, citation, ',');
523 LYXERR(Debug::LATEX, "Citation: " << citation);
524 aux_info.citations.insert(citation);
526 } else if (regex_match(token, sub, reg2)) {
527 string data = sub.str(1);
528 // data is now all the bib files separated by ','
529 // get them one by one and pass them to the helper
530 while (!data.empty()) {
532 data = split(data, database, ',');
533 database = changeExtension(database, "bib");
534 LYXERR(Debug::LATEX, "BibTeX database: `" << database << '\'');
535 aux_info.databases.insert(database);
537 } else if (regex_match(token, sub, reg3)) {
538 string style = sub.str(1);
539 // token is now the style file
540 // pass it to the helper
541 style = changeExtension(style, "bst");
542 LYXERR(Debug::LATEX, "BibTeX style: `" << style << '\'');
543 aux_info.styles.insert(style);
544 } else if (regex_match(token, sub, reg4)) {
545 string const file2 = sub.str(1);
546 scanAuxFile(makeAbsPath(file2), aux_info);
552 void LaTeX::updateBibtexDependencies(DepTable & dep,
553 vector<AuxInfo> const & bibtex_info)
555 // Since a run of Bibtex mandates more latex runs it is ok to
556 // remove all ".bib" and ".bst" files.
557 dep.remove_files_with_extension(".bib");
558 dep.remove_files_with_extension(".bst");
559 //string aux = OnlyFileName(ChangeExtension(file, ".aux"));
561 for (vector<AuxInfo>::const_iterator it = bibtex_info.begin();
562 it != bibtex_info.end(); ++it) {
563 for (set<string>::const_iterator it2 = it->databases.begin();
564 it2 != it->databases.end(); ++it2) {
565 FileName const file = findtexfile(*it2, "bib");
567 dep.insert(file, true);
570 for (set<string>::const_iterator it2 = it->styles.begin();
571 it2 != it->styles.end(); ++it2) {
572 FileName const file = findtexfile(*it2, "bst");
574 dep.insert(file, true);
578 // biber writes nothing into the aux file.
579 // Instead, we have to scan the blg file
586 bool LaTeX::runBibTeX(vector<AuxInfo> const & bibtex_info,
587 OutputParams const & runparams)
590 for (vector<AuxInfo>::const_iterator it = bibtex_info.begin();
591 it != bibtex_info.end(); ++it) {
592 if (!biber && it->databases.empty())
596 string tmp = runparams.use_japanese ?
597 lyxrc.jbibtex_command : lyxrc.bibtex_command;
599 if (!runparams.bibtex_command.empty())
600 tmp = runparams.bibtex_command;
602 // onlyFileName() is needed for cygwin
603 tmp += quoteName(onlyFileName(removeExtension(
604 it->aux_file.absFileName())));
606 one.startscript(Systemcall::Wait, tmp, path);
608 // Return whether bibtex was run
613 int LaTeX::scanLogFile(TeXErrors & terr)
617 int retval = NO_ERRORS;
619 onlyFileName(changeExtension(file.absFileName(), ".log"));
620 LYXERR(Debug::LATEX, "Log file: " << tmp);
621 FileName const fn = FileName(makeAbsPath(tmp));
622 ifstream ifs(fn.toFilesystemEncoding().c_str());
623 bool fle_style = false;
624 static regex file_line_error(".+\\.\\D+:[0-9]+: (.+)");
625 static regex child_file(".*([0-9]+[A-Za-z]*_.+\\.tex).*");
626 // Flag for 'File ended while scanning' message.
627 // We need to wait for subsequent processing.
628 string wait_for_error;
631 stack <pair<string, int> > child;
634 while (getline(ifs, token)) {
635 // MikTeX sometimes inserts \0 in the log file. They can't be
636 // removed directly with the existing string utility
637 // functions, so convert them first to \r, and remove all
638 // \r's afterwards, since we need to remove them anyway.
639 token = subst(token, '\0', '\r');
640 token = subst(token, "\r", "");
643 LYXERR(Debug::LATEX, "Log line: " << token);
648 // Track child documents
649 for (size_t i = 0; i < token.length(); ++i) {
650 if (token[i] == '(') {
652 size_t j = token.find('(', i + 1);
653 size_t len = j == string::npos
654 ? token.substr(i + 1).length()
656 if (regex_match(token.substr(i + 1, len),
658 string const name = sub.str(1);
659 child.push(make_pair(name, pnest));
662 } else if (token[i] == ')') {
664 && child.top().second == pnest)
669 child_name = child.empty() ? empty_string() : child.top().first;
671 if (contains(token, "file:line:error style messages enabled"))
674 if (prefixIs(token, "LaTeX Warning:") ||
675 prefixIs(token, "! pdfTeX warning")) {
676 // Here shall we handle different
678 retval |= LATEX_WARNING;
679 LYXERR(Debug::LATEX, "LaTeX Warning.");
680 if (contains(token, "Rerun to get cross-references")) {
682 LYXERR(Debug::LATEX, "We should rerun.");
683 // package clefval needs 2 latex runs before bibtex
684 } else if (contains(token, "Value of")
685 && contains(token, "on page")
686 && contains(token, "undefined")) {
687 retval |= ERROR_RERUN;
688 LYXERR(Debug::LATEX, "Force rerun.");
690 } else if (contains(token, "Etaremune labels have changed")) {
691 retval |= ERROR_RERUN;
692 LYXERR(Debug::LATEX, "Force rerun.");
693 } else if (contains(token, "Citation")
694 && contains(token, "on page")
695 && contains(token, "undefined")) {
697 } else if (contains(token, "Citation")
698 && contains(token, "on input line")
699 && contains(token, "undefined")) {
702 } else if (prefixIs(token, "Package")) {
704 retval |= PACKAGE_WARNING;
705 if (contains(token, "natbib Warning:")) {
707 if (contains(token, "Citation")
708 && contains(token, "on page")
709 && contains(token, "undefined")) {
712 } else if (contains(token, "run BibTeX")) {
714 } else if (contains(token, "run Biber")) {
717 } else if (contains(token, "Rerun LaTeX") ||
718 contains(token, "Please rerun LaTeX") ||
719 contains(token, "Rerun to get")) {
720 // at least longtable.sty and bibtopic.sty
722 LYXERR(Debug::LATEX, "We should rerun.");
725 } else if (prefixIs(token, "LETTRE WARNING:")) {
726 if (contains(token, "veuillez recompiler")) {
728 LYXERR(Debug::LATEX, "We should rerun.");
731 } else if (token[0] == '(') {
732 if (contains(token, "Rerun LaTeX") ||
733 contains(token, "Rerun to get")) {
735 LYXERR(Debug::LATEX, "We should rerun.");
738 } else if (prefixIs(token, "! ")
740 && regex_match(token, sub, file_line_error)
741 && !contains(token, "pdfTeX warning"))) {
742 // Ok, we have something that looks like a TeX Error
743 // but what do we really have.
745 // Just get the error description:
747 if (prefixIs(token, "! "))
748 desc = string(token, 2);
751 if (contains(token, "LaTeX Error:"))
752 retval |= LATEX_ERROR;
754 if (prefixIs(token, "! File ended while scanning")){
755 if (prefixIs(token, "! File ended while scanning use of \\Hy@setref@link.")){
756 // bug 7344. We must rerun LaTeX if hyperref has been toggled.
757 retval |= ERROR_RERUN;
758 LYXERR(Debug::LATEX, "Force rerun.");
760 // bug 6445. At this point its not clear we finish with error.
761 wait_for_error = desc;
766 if (prefixIs(token, "! Paragraph ended before \\Hy@setref@link was complete.")){
767 // bug 7344. We must rerun LaTeX if hyperref has been toggled.
768 retval |= ERROR_RERUN;
769 LYXERR(Debug::LATEX, "Force rerun.");
772 if (!wait_for_error.empty() && prefixIs(token, "! Emergency stop.")){
773 retval |= LATEX_ERROR;
776 errstr = wait_for_error;
778 if (!getline(ifs, tmp))
780 tmp = rtrim(tmp, "\r");
781 errstr += "\n" + tmp;
784 } while (!contains(tmp, "(job aborted"));
787 from_local8bit("Emergency stop"),
788 from_local8bit(errstr),
796 if (!getline(ifs, tmp))
798 tmp = rtrim(tmp, "\r");
801 } while (!prefixIs(tmp, "l."));
802 if (prefixIs(tmp, "l.")) {
803 // we have a latex error
806 "Package babel Error: You haven't defined the language") ||
808 "Package babel Error: You haven't loaded the option"))
809 retval |= ERROR_RERUN;
810 // get the line number:
812 sscanf(tmp.c_str(), "l.%d", &line);
813 // get the rest of the message:
814 string errstr(tmp, tmp.find(' '));
817 tmp = rtrim(tmp, "\r");
818 while (!contains(errstr, "l.")
820 && !prefixIs(tmp, "! ")
821 && !contains(tmp, "(job aborted")) {
825 tmp = rtrim(tmp, "\r");
827 LYXERR(Debug::LATEX, "line: " << line << '\n'
828 << "Desc: " << desc << '\n' << "Text: " << errstr);
829 if (line == last_line)
835 if (line_count <= 5) {
837 // We have no idea what the encoding of
839 // It seems that the output from the
840 // latex compiler itself is pure ASCII,
841 // but it can include bits from the
842 // document, so whatever encoding we
843 // assume here it can be wrong.
844 terr.insertError(line,
845 from_local8bit(desc),
846 from_local8bit(errstr),
852 // information messages, TeX warnings and other
853 // warnings we have not caught earlier.
854 if (prefixIs(token, "Overfull ")) {
855 retval |= TEX_WARNING;
856 } else if (prefixIs(token, "Underfull ")) {
857 retval |= TEX_WARNING;
858 } else if (contains(token, "Rerun to get citations")) {
859 // Natbib seems to use this.
861 } else if (contains(token, "No pages of output")) {
862 // A dvi file was not created
864 } else if (contains(token, "That makes 100 errors")) {
865 // More than 100 errors were reprted
866 retval |= TOO_MANY_ERRORS;
867 } else if (prefixIs(token, "!pdfTeX error:")){
868 // otherwise we dont catch e.g.:
869 // !pdfTeX error: pdflatex (file feyn10): Font feyn10 at 600 not found
872 from_local8bit("pdfTeX Error"),
873 from_local8bit(token),
878 LYXERR(Debug::LATEX, "Log line: " << token);
885 bool insertIfExists(FileName const & absname, DepTable & head)
887 if (absname.exists() && !absname.isDirectory()) {
888 head.insert(absname, true);
895 bool handleFoundFile(string const & ff, DepTable & head)
897 // convert from native os path to unix path
898 string foundfile = os::internal_path(trim(ff));
900 LYXERR(Debug::DEPEND, "Found file: " << foundfile);
902 // Ok now we found a file.
903 // Now we should make sure that this is a file that we can
904 // access through the normal paths.
905 // We will not try any fancy search methods to
908 // (1) foundfile is an
909 // absolute path and should
912 if (FileName::isAbsolute(foundfile)) {
913 LYXERR(Debug::DEPEND, "AbsolutePath file: " << foundfile);
914 // On initial insert we want to do the update at once
915 // since this file cannot be a file generated by
917 absname.set(foundfile);
918 if (!insertIfExists(absname, head)) {
920 string strippedfile = foundfile;
921 while (contains(strippedfile, " ")) {
922 // files with spaces are often enclosed in quotation
923 // marks; those have to be removed
924 string unquoted = subst(strippedfile, "\"", "");
925 absname.set(unquoted);
926 if (insertIfExists(absname, head))
928 // strip off part after last space and try again
929 string tmp = strippedfile;
930 string const stripoff =
931 rsplit(tmp, strippedfile, ' ');
932 absname.set(strippedfile);
933 if (insertIfExists(absname, head))
939 string onlyfile = onlyFileName(foundfile);
940 absname = makeAbsPath(onlyfile);
943 while (contains(foundfile, ' ')) {
944 if (absname.exists())
948 // files with spaces are often enclosed in quotation
949 // marks; those have to be removed
950 string unquoted = subst(foundfile, "\"", "");
951 absname = makeAbsPath(unquoted);
952 if (absname.exists())
954 // strip off part after last space and try again
956 string const stripoff =
957 rsplit(foundfile, strippedfile, ' ');
958 foundfile = strippedfile;
959 onlyfile = onlyFileName(strippedfile);
960 absname = makeAbsPath(onlyfile);
964 // (2) foundfile is in the tmpdir
965 // insert it into head
966 if (absname.exists() && !absname.isDirectory()) {
967 // FIXME: This regex contained glo, but glo is used by the old
968 // version of nomencl.sty. Do we need to put it back?
969 static regex const unwanted("^.*\\.(aux|log|dvi|bbl|ind)$");
970 if (regex_match(onlyfile, unwanted)) {
971 LYXERR(Debug::DEPEND, "We don't want " << onlyfile
972 << " in the dep file");
973 } else if (suffixIs(onlyfile, ".tex")) {
974 // This is a tex file generated by LyX
975 // and latex is not likely to change this
977 LYXERR(Debug::DEPEND, "Tmpdir TeX file: " << onlyfile);
978 head.insert(absname, true);
980 LYXERR(Debug::DEPEND, "In tmpdir file:" << onlyfile);
981 head.insert(absname);
985 LYXERR(Debug::DEPEND, "Not a file or we are unable to find it.");
991 bool completeFilename(string const & ff, DepTable & head)
993 // If we do not find a dot, we suspect
994 // a fragmental file name
995 if (!contains(ff, '.'))
998 // if we have a dot, we let handleFoundFile decide
999 return handleFoundFile(ff, head);
1003 int iterateLine(string const token, regex const reg, string const & closing,
1004 int fragment_pos, DepTable & head)
1007 string::const_iterator first = token.begin();
1008 string::const_iterator end = token.end();
1009 bool fragment = false;
1012 while (regex_search(first, end, what, reg)) {
1013 // if we have a dot, try to handle as file
1014 if (contains(what.str(1), '.')) {
1015 first = what[0].second;
1016 if (what.str(2) == closing) {
1017 handleFoundFile(what.str(1), head);
1018 // since we had a closing bracket,
1019 // do not investigate further
1022 // if we have no closing bracket,
1023 // try to handle as file nevertheless
1024 fragment = !handleFoundFile(
1025 what.str(1) + what.str(2), head);
1027 // if we do not have a dot, check if the line has
1028 // a closing bracket (else, we suspect a line break)
1029 else if (what.str(2) != closing) {
1030 first = what[0].second;
1033 // we have a closing bracket, so the content
1034 // is not a file name.
1035 // no need to investigate further
1036 first = what[0].second;
1039 last_match = what.str(1);
1042 // We need to consider the result from previous line iterations:
1043 // We might not find a fragment here, but another one might follow
1044 // E.g.: (filename.ext) <filenam
1045 // Vice versa, we consider the search completed if a real match
1046 // follows a potential fragment from a previous iteration.
1047 // E.g. <some text we considered a fragment (filename.ext)
1048 // result = -1 means we did not find a fragment!
1050 int last_match_pos = -1;
1051 if (!last_match.empty() && token.find(last_match) != string::npos)
1052 last_match_pos = int(token.find(last_match));
1054 if (last_match_pos > fragment_pos)
1055 result = last_match_pos;
1057 result = fragment_pos;
1059 if (last_match_pos < fragment_pos)
1060 result = fragment_pos;
1068 void LaTeX::deplog(DepTable & head)
1070 // This function reads the LaTeX log file end extracts all the
1071 // external files used by the LaTeX run. The files are then
1072 // entered into the dependency file.
1074 string const logfile =
1075 onlyFileName(changeExtension(file.absFileName(), ".log"));
1077 static regex const reg1("File: (.+).*");
1078 static regex const reg2("No file (.+)(.).*");
1079 static regex const reg3("\\\\openout[0-9]+.*=.*`(.+)(..).*");
1080 // If an index should be created, MikTex does not write a line like
1081 // \openout# = 'sample.idx'.
1082 // but instead only a line like this into the log:
1083 // Writing index file sample.idx
1084 static regex const reg4("Writing index file (.+).*");
1085 static regex const regoldnomencl("Writing glossary file (.+).*");
1086 static regex const regnomencl("Writing nomenclature file (.+).*");
1087 // If a toc should be created, MikTex does not write a line like
1088 // \openout# = `sample.toc'.
1089 // but only a line like this into the log:
1091 // This line is also written by tetex.
1092 // This line is not present if no toc should be created.
1093 static regex const miktexTocReg("\\\\tf@toc=\\\\write.*");
1094 // file names can be enclosed in <...> (anywhere on the line)
1095 static regex const reg5(".*<[^>]+.*");
1096 // and also (...) anywhere on the line
1097 static regex const reg6(".*\\([^)]+.*");
1099 FileName const fn = makeAbsPath(logfile);
1100 ifstream ifs(fn.toFilesystemEncoding().c_str());
1103 // Ok, the scanning of files here is not sufficient.
1104 // Sometimes files are named by "File: xxx" only
1105 // Therefore we use some regexps to find files instead.
1106 // Note: all file names and paths might contains spaces.
1107 // Also, file names might be broken across lines. Therefore
1108 // we mark (potential) fragments and merge those lines.
1109 bool fragment = false;
1111 getline(ifs, token);
1112 // MikTeX sometimes inserts \0 in the log file. They can't be
1113 // removed directly with the existing string utility
1114 // functions, so convert them first to \r, and remove all
1115 // \r's afterwards, since we need to remove them anyway.
1116 token = subst(token, '\0', '\r');
1117 token = subst(token, "\r", "");
1118 if (token.empty() || token == ")") {
1119 lastline = string();
1123 // FIXME UNICODE: We assume that the file names in the log
1124 // file are in the file system encoding.
1125 token = to_utf8(from_filesystem8bit(token));
1127 // Sometimes, filenames are broken across lines.
1128 // We care for that and save suspicious lines.
1129 // Here we exclude some cases where we are sure
1130 // that there is no continued filename
1131 if (!lastline.empty()) {
1132 static regex const package_info("Package \\w+ Info: .*");
1133 static regex const package_warning("Package \\w+ Warning: .*");
1134 if (prefixIs(token, "File:") || prefixIs(token, "(Font)")
1135 || prefixIs(token, "Package:")
1136 || prefixIs(token, "Language:")
1137 || prefixIs(token, "LaTeX Info:")
1138 || prefixIs(token, "LaTeX Font Info:")
1139 || prefixIs(token, "\\openout[")
1140 || prefixIs(token, "))")
1141 || regex_match(token, package_info)
1142 || regex_match(token, package_warning))
1143 lastline = string();
1146 if (!lastline.empty())
1147 // probably a continued filename from last line
1148 token = lastline + token;
1149 if (token.length() > 255) {
1150 // string too long. Cut off.
1151 token.erase(0, token.length() - 251);
1156 // (1) "File: file.ext"
1157 if (regex_match(token, sub, reg1)) {
1158 // is this a fragmental file name?
1159 fragment = !completeFilename(sub.str(1), head);
1161 if (suffixIs(token, ")"))
1162 // no fragment for sure
1164 // (2) "No file file.ext"
1165 } else if (regex_match(token, sub, reg2)) {
1166 // file names must contains a dot, line ends with dot
1167 if (contains(sub.str(1), '.') && sub.str(2) == ".")
1168 fragment = !handleFoundFile(sub.str(1), head);
1170 // we suspect a line break
1172 // (3) "\openout<nr> = `file.ext'."
1173 } else if (regex_match(token, sub, reg3)) {
1174 // search for closing '. at the end of the line
1175 if (sub.str(2) == "\'.")
1176 fragment = !handleFoundFile(sub.str(1), head);
1178 // potential fragment
1180 // (4) "Writing index file file.ext"
1181 } else if (regex_match(token, sub, reg4))
1182 // fragmential file name?
1183 fragment = !completeFilename(sub.str(1), head);
1184 // (5) "Writing nomenclature file file.ext"
1185 else if (regex_match(token, sub, regnomencl) ||
1186 regex_match(token, sub, regoldnomencl))
1187 // fragmental file name?
1188 fragment= !completeFilename(sub.str(1), head);
1189 // (6) "\tf@toc=\write<nr>" (for MikTeX)
1190 else if (regex_match(token, sub, miktexTocReg))
1191 fragment = !handleFoundFile(onlyFileName(changeExtension(
1192 file.absFileName(), ".toc")), head);
1194 // not found, but we won't check further
1197 int fragment_pos = -1;
1199 // We can have several of these on one line
1200 // (and in addition to those above)
1201 if (regex_match(token, sub, reg5)) {
1202 // search for strings in <...>
1203 static regex reg5_1("<([^>]+)(.)");
1204 fragment_pos = iterateLine(token, reg5_1, ">",
1205 fragment_pos, head);
1206 fragment = (fragment_pos != -1);
1210 // We can have several of these on one line
1211 // this must be queried separated, because of
1212 // cases such as "File: file.ext (type eps)"
1213 // where "File: file.ext" would be skipped
1214 if (regex_match(token, sub, reg6)) {
1215 // search for strings in (...)
1216 static regex reg6_1("\\(([^()]+)(.)");
1217 fragment_pos = iterateLine(token, reg6_1, ")",
1218 fragment_pos, head);
1219 fragment = (fragment_pos != -1);
1223 // probable linebreak within file name:
1227 // no linebreak: reset
1228 lastline = string();
1231 // Make sure that the main .tex file is in the dependency file.
1232 head.insert(file, true);
1236 void LaTeX::scanBlgFile(DepTable & dep)
1238 FileName const blg_file(changeExtension(file.absFileName(), "blg"));
1239 LYXERR(Debug::LATEX, "Scanning blg file: " << blg_file);
1241 ifstream ifs(blg_file.toFilesystemEncoding().c_str());
1243 static regex const reg1(".*Found (bibtex|BibTeX) data (file|source) '([^']+).*");
1245 while (getline(ifs, token)) {
1246 token = rtrim(token, "\r");
1248 // FIXME UNICODE: We assume that citation keys and filenames
1249 // in the aux file are in the file system encoding.
1250 token = to_utf8(from_filesystem8bit(token));
1251 if (regex_match(token, sub, reg1)) {
1252 string data = sub.str(3);
1253 if (!data.empty()) {
1254 LYXERR(Debug::LATEX, "Found bib file: " << data);
1255 handleFoundFile(data, dep);