2 * \file InsetBibtex.cpp
3 * This file is part of LyX, the document processor.
4 * Licence details can be found in the file COPYING.
6 * \author Alejandro Aguilar Sierra
8 * Full author contact details are available in file CREDITS.
13 #include "InsetBibtex.h"
16 #include "BufferParams.h"
17 #include "DispatchResult.h"
20 #include "FuncRequest.h"
22 #include "LaTeXFeatures.h"
23 #include "MetricsInfo.h"
24 #include "OutputParams.h"
26 #include "frontends/Alert.h"
28 #include "support/filetools.h"
29 #include "support/lstrings.h"
30 #include "support/lyxlib.h"
31 #include "support/os.h"
32 #include "support/Path.h"
33 #include "support/textutils.h"
35 #include <boost/tokenizer.hpp>
40 using support::absolutePath;
41 using support::ascii_lowercase;
42 using support::changeExtension;
43 using support::contains;
45 using support::DocFileName;
46 using support::FileName;
47 using support::findtexfile;
48 using support::isFileReadable;
49 using support::latex_path;
51 using support::makeAbsPath;
52 using support::makeRelPath;
53 using support::prefixIs;
54 using support::removeExtension;
58 using support::tokenPos;
60 using support::lowercase;
62 namespace Alert = frontend::Alert;
63 namespace os = support::os;
74 InsetBibtex::InsetBibtex(InsetCommandParams const & p)
75 : InsetCommand(p, "bibtex")
79 std::auto_ptr<InsetBase> InsetBibtex::doClone() const
81 return std::auto_ptr<InsetBase>(new InsetBibtex(*this));
85 void InsetBibtex::doDispatch(Cursor & cur, FuncRequest & cmd)
89 case LFUN_INSET_MODIFY: {
90 InsetCommandParams p("bibtex");
91 InsetCommandMailer::string2params("bibtex", to_utf8(cmd.argument()), p);
92 if (!p.getCmdName().empty()) {
94 cur.buffer().updateBibfilesCache();
101 InsetCommand::doDispatch(cur, cmd);
107 docstring const InsetBibtex::getScreenLabel(Buffer const &) const
109 return _("BibTeX Generated Bibliography");
115 string normalize_name(Buffer const & buffer, OutputParams const & runparams,
116 string const & name, string const & ext)
118 string const fname = makeAbsPath(name, buffer.filePath()).absFilename();
119 if (absolutePath(name) || !isFileReadable(FileName(fname + ext)))
121 else if (!runparams.nice)
125 return to_utf8(makeRelPath(from_utf8(fname),
126 from_utf8(buffer.getMasterBuffer()->filePath())));
132 int InsetBibtex::latex(Buffer const & buffer, odocstream & os,
133 OutputParams const & runparams) const
135 // the sequence of the commands:
136 // 1. \bibliographystyle{style}
137 // 2. \addcontentsline{...} - if option bibtotoc set
138 // 3. \bibliography{database}
139 // and with bibtopic:
140 // 1. \bibliographystyle{style}
141 // 2. \begin{btSect}{database}
142 // 3. \btPrint{Cited|NotCited|All}
146 // If we are processing the LaTeX file in a temp directory then
147 // copy the .bib databases to this temp directory, mangling their
148 // names in the process. Store this mangled name in the list of
150 // (We need to do all this because BibTeX *really*, *really*
151 // can't handle "files with spaces" and Windows users tend to
152 // use such filenames.)
153 // Otherwise, store the (maybe absolute) path to the original,
154 // unmangled database name.
155 typedef boost::char_separator<char_type> Separator;
156 typedef boost::tokenizer<Separator, docstring::const_iterator, docstring> Tokenizer;
158 Separator const separator(from_ascii(",").c_str());
159 // The tokenizer must not be called with temporary strings, since
160 // it does not make a copy and uses iterators of the string further
161 // down. getParam returns a reference, so this is OK.
162 Tokenizer const tokens(getParam("bibfiles"), separator);
163 Tokenizer::const_iterator const begin = tokens.begin();
164 Tokenizer::const_iterator const end = tokens.end();
166 odocstringstream dbs;
167 for (Tokenizer::const_iterator it = begin; it != end; ++it) {
168 docstring const input = trim(*it);
170 string utf8input(to_utf8(input));
172 normalize_name(buffer, runparams, utf8input, ".bib");
173 FileName const try_in_file(makeAbsPath(database + ".bib", buffer.filePath()));
174 bool const not_from_texmf = isFileReadable(try_in_file);
176 if (!runparams.inComment && !runparams.dryrun && !runparams.nice &&
179 // mangledFilename() needs the extension
180 DocFileName const in_file = DocFileName(try_in_file);
181 database = removeExtension(in_file.mangledFilename());
182 FileName const out_file(makeAbsPath(database + ".bib",
183 buffer.getMasterBuffer()->temppath()));
185 bool const success = copy(in_file, out_file);
187 lyxerr << "Failed to copy '" << in_file
188 << "' to '" << out_file << "'"
196 dbs << from_utf8(latex_path(database));
198 docstring const db_out = dbs.str();
200 // Post this warning only once.
201 static bool warned_about_spaces = false;
202 if (!warned_about_spaces &&
203 runparams.nice && db_out.find(' ') != docstring::npos) {
204 warned_about_spaces = true;
206 Alert::warning(_("Export Warning!"),
207 _("There are spaces in the paths to your BibTeX databases.\n"
208 "BibTeX will be unable to find them."));
213 string style = to_utf8(getParam("options")); // maybe empty! and with bibtotoc
215 if (prefixIs(style, "bibtotoc")) {
216 bibtotoc = "bibtotoc";
217 if (contains(style, ',')) {
218 style = split(style, bibtotoc, ',');
225 if (!style.empty()) {
227 normalize_name(buffer, runparams, style, ".bst");
228 FileName const try_in_file(makeAbsPath(base + ".bst", buffer.filePath()));
229 bool const not_from_texmf = isFileReadable(try_in_file);
230 // If this style does not come from texmf and we are not
231 // exporting to .tex copy it to the tmp directory.
232 // This prevents problems with spaces and 8bit charcaters
234 if (!runparams.inComment && !runparams.dryrun && !runparams.nice &&
236 // use new style name
237 DocFileName const in_file = DocFileName(try_in_file);
238 base = removeExtension(in_file.mangledFilename());
239 FileName const out_file(makeAbsPath(base + ".bst",
240 buffer.getMasterBuffer()->temppath()));
241 bool const success = copy(in_file, out_file);
243 lyxerr << "Failed to copy '" << in_file
244 << "' to '" << out_file << "'"
249 os << "\\bibliographystyle{"
250 << from_utf8(latex_path(normalize_name(buffer, runparams, base, ".bst")))
255 // Post this warning only once.
256 static bool warned_about_bst_spaces = false;
257 if (!warned_about_bst_spaces && runparams.nice && contains(style, ' ')) {
258 warned_about_bst_spaces = true;
259 Alert::warning(_("Export Warning!"),
260 _("There are spaces in the path to your BibTeX style file.\n"
261 "BibTeX will be unable to find it."));
264 if (!db_out.empty() && buffer.params().use_bibtopic){
265 os << "\\begin{btSect}{" << db_out << "}\n";
266 docstring btprint = getParam("btprint");
269 btprint = from_ascii("btPrintCited");
270 os << "\\" << btprint << "\n"
271 << "\\end{btSect}\n";
276 if (!bibtotoc.empty() && !buffer.params().use_bibtopic) {
277 // maybe a problem when a textclass has no "art" as
278 // part of its name, because it's than book.
279 // For the "official" lyx-layouts it's no problem to support
281 if (!contains(buffer.params().getLyXTextClass().name(),
283 if (buffer.params().sides == LyXTextClass::OneSide) {
288 os << "\\cleardoublepage";
292 os << "\\addcontentsline{toc}{chapter}{\\bibname}";
296 os << "\\addcontentsline{toc}{section}{\\refname}";
300 if (!db_out.empty() && !buffer.params().use_bibtopic){
301 os << "\\bibliography{" << db_out << "}\n";
309 vector<FileName> const InsetBibtex::getFiles(Buffer const & buffer) const
311 FileName path(buffer.filePath());
312 support::Path p(path);
314 vector<FileName> vec;
318 string bibfiles = to_utf8(getParam("bibfiles"));
319 bibfiles = split(bibfiles, tmp, ',');
320 while (!tmp.empty()) {
321 FileName const file = findtexfile(changeExtension(tmp, "bib"), "bib");
322 LYXERR(Debug::LATEX) << "Bibfile: " << file << endl;
324 // If we didn't find a matching file name just fail silently
328 // Get next file name
329 bibfiles = split(bibfiles, tmp, ',');
337 // methods for parsing bibtex files
339 typedef map<docstring, docstring> VarMap;
341 /// remove whitespace characters, optionally a single comma,
342 /// and further whitespace characters from the stream.
343 /// @return true if a comma was found, false otherwise
345 bool removeWSAndComma(idocfstream & ifs) {
354 } while (ifs && isSpace(ch));
367 } while (ifs && isSpace(ch));
382 /// remove whitespace characters, read characer sequence
383 /// not containing whitespace characters or characters in
384 /// delimChars, and remove further whitespace characters.
386 /// @return true if a string of length > 0 could be read.
388 bool readTypeOrKey(docstring & val, idocfstream & ifs,
389 docstring const & delimChars, charCase chCase) {
401 } while (ifs && isSpace(ch));
407 while (ifs && !isSpace(ch) && delimChars.find(ch) == docstring::npos) {
408 if (chCase == makeLowerCase) {
409 val += lowercase(ch);
417 while (ifs && isSpace(ch)) {
425 return val.length() > 0;
428 /// read subsequent bibtex values that are delimited with a #-character.
429 /// Concatenate all parts and replace names with the associated string in
430 /// the variable strings.
431 /// @return true if reading was successfull (all single parts were delimited
433 bool readValue(docstring & val, idocfstream & ifs, const VarMap & strings) {
446 } while (ifs && isSpace(ch));
451 // check for field type
454 // read integer value
458 } while (ifs && isDigit(ch));
463 } else if (ch == '"' || ch == '{') {
465 // read delimited text - set end delimiter
466 char_type delim = ch == '"'? '"': '}';
468 // inside this delimited text braces must match.
469 // Thus we can have a closing delimiter only
470 // when nestLevel == 0
474 while (ifs && (nestLevel > 0 || ch != delim)) {
477 // update nesting level
484 if (nestLevel < 0) return false;
501 // reading a string name
504 while (ifs && !isSpace(ch) && ch != '#' && ch != ',' && ch != '}' && ch != ')') {
505 strName += lowercase(ch);
512 // replace the string with its assigned value or
513 // discard it if it's not assigned
514 if (strName.length()) {
515 VarMap::const_iterator pos = strings.find(strName);
516 if (pos != strings.end()) {
523 while (ifs && isSpace(ch)) {
530 // continue reading next value on concatenate with '#'
540 // This method returns a comma separated list of Bibtex entries
541 void InsetBibtex::fillWithBibKeys(Buffer const & buffer,
542 std::vector<std::pair<string, docstring> > & keys) const
544 vector<FileName> const files = getFiles(buffer);
545 for (vector<FileName>::const_iterator it = files.begin();
546 it != files.end(); ++ it) {
547 // This bibtex parser is a first step to parse bibtex files
550 // - it reads the whole bibtex entry and does a syntax check
551 // (matching delimiters, missing commas,...
552 // - it recovers from errors starting with the next @-character
553 // - it reads @string definitions and replaces them in the
555 // - it accepts more characters in keys or value names than
559 // - the entries are split into name = value pairs by the
560 // parser. These have to be merged again because of the
561 // way lyx treats the entries ( pair<...>(...) ). The citation
562 // mechanism in lyx should be changed such that it can use
563 // the split entries.
564 // - messages on parsing errors can be generated.
567 // Officially bibtex does only support ASCII, but in practice
568 // you can use the encoding of the main document as long as
569 // some elements like keys and names are pure ASCII. Therefore
570 // we convert the file from the buffer encoding.
571 // We don't restrict keys to ASCII in LyX, since our own
572 // InsetBibitem can generate non-ASCII keys, and nonstandard
573 // 8bit clean bibtex forks exist.
574 idocfstream ifs(it->toFilesystemEncoding().c_str(),
576 buffer.params().encoding().iconvName());
592 if (!readTypeOrKey(entryType, ifs, from_ascii("{("), makeLowerCase) || !ifs)
595 if (entryType == from_ascii("comment")) {
597 ifs.ignore(std::numeric_limits<int>::max(), '\n');
605 if ((ch != '(') && (ch != '{')) {
606 // invalid entry delimiter
612 if (entryType == from_ascii("string")) {
614 // read string and add it to the strings map
615 // (or replace it's old value)
619 if (!readTypeOrKey(name, ifs, from_ascii("#=}),"), makeLowerCase) || !ifs)
623 if (!ifs || ch != '=')
626 if (!readValue(value, ifs, strings))
629 strings[name] = value;
631 } else if (entryType == from_ascii("preamble")) {
633 // preamble definitions are discarded.
634 // can they be of any use in lyx?
637 if (!readValue(value, ifs, strings))
642 // Citation entry. Read the key and all name = value pairs
647 docstring commaNewline;
649 if (!readTypeOrKey(key, ifs, from_ascii(",})"), keepCase) || !ifs)
652 // now we have a key, so we will add an entry
653 // (even if it's empty, as bibtex does)
655 // all items must be separated by a comma. If
656 // it is missing the scanning of this entry is
657 // stopped and the next is searched.
658 bool readNext = removeWSAndComma(ifs);
660 while (ifs && readNext) {
663 if (!readTypeOrKey(name, ifs, from_ascii("=}),"), makeLowerCase) || !ifs)
666 // next char must be an equal sign
676 if (!readValue(value, ifs, strings))
679 // append field to the total entry string.
681 // TODO: Here is where the fields can be put in
682 // a more intelligent structure that preserves
683 // the already known parts.
684 fields += commaNewline;
685 fields += name + from_ascii(" = {") + value + '}';
687 if (!commaNewline.length())
688 commaNewline = from_ascii(",\n");
690 readNext = removeWSAndComma(ifs);
694 keys.push_back(pair<string, docstring>(
695 to_utf8(key), fields));
700 } //< for loop over files
705 bool InsetBibtex::addDatabase(string const & db)
708 string bibfiles(to_utf8(getParam("bibfiles")));
709 if (tokenPos(bibfiles, ',', db) == -1) {
710 if (!bibfiles.empty())
712 setParam("bibfiles", from_utf8(bibfiles + db));
719 bool InsetBibtex::delDatabase(string const & db)
722 string bibfiles(to_utf8(getParam("bibfiles")));
723 if (contains(bibfiles, db)) {
724 int const n = tokenPos(bibfiles, ',', db);
727 // this is not the first database
728 string tmp = ',' + bd;
729 setParam("bibfiles", from_utf8(subst(bibfiles, tmp, string())));
731 // this is the first (or only) database
732 setParam("bibfiles", from_utf8(split(bibfiles, bd, ',')));
740 void InsetBibtex::validate(LaTeXFeatures & features) const
742 if (features.bufferParams().use_bibtopic)
743 features.require("bibtopic");