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;
54 using support::prefixIs;
55 using support::removeExtension;
59 using support::tokenPos;
61 using support::lowercase;
63 namespace Alert = frontend::Alert;
64 namespace os = support::os;
75 InsetBibtex::InsetBibtex(InsetCommandParams const & p)
76 : InsetCommand(p, "bibtex")
80 std::auto_ptr<InsetBase> InsetBibtex::doClone() const
82 return std::auto_ptr<InsetBase>(new InsetBibtex(*this));
86 void InsetBibtex::doDispatch(LCursor & cur, FuncRequest & cmd)
90 case LFUN_INSET_MODIFY: {
91 InsetCommandParams p("bibtex");
92 InsetCommandMailer::string2params("bibtex", to_utf8(cmd.argument()), p);
93 if (!p.getCmdName().empty()) {
95 cur.buffer().updateBibfilesCache();
102 InsetCommand::doDispatch(cur, cmd);
108 docstring const InsetBibtex::getScreenLabel(Buffer const &) const
110 return _("BibTeX Generated Bibliography");
116 string normalize_name(Buffer const & buffer, OutputParams const & runparams,
117 string const & name, string const & ext)
119 string const fname = makeAbsPath(name, buffer.filePath()).absFilename();
120 if (absolutePath(name) || !isFileReadable(FileName(fname + ext)))
122 else if (!runparams.nice)
126 return to_utf8(makeRelPath(from_utf8(fname),
127 from_utf8(buffer.getMasterBuffer()->filePath())));
133 int InsetBibtex::latex(Buffer const & buffer, odocstream & os,
134 OutputParams const & runparams) const
136 // the sequence of the commands:
137 // 1. \bibliographystyle{style}
138 // 2. \addcontentsline{...} - if option bibtotoc set
139 // 3. \bibliography{database}
140 // and with bibtopic:
141 // 1. \bibliographystyle{style}
142 // 2. \begin{btSect}{database}
143 // 3. \btPrint{Cited|NotCited|All}
147 // If we are processing the LaTeX file in a temp directory then
148 // copy the .bib databases to this temp directory, mangling their
149 // names in the process. Store this mangled name in the list of
151 // (We need to do all this because BibTeX *really*, *really*
152 // can't handle "files with spaces" and Windows users tend to
153 // use such filenames.)
154 // Otherwise, store the (maybe absolute) path to the original,
155 // unmangled database name.
156 typedef boost::char_separator<char_type> Separator;
157 typedef boost::tokenizer<Separator, docstring::const_iterator, docstring> Tokenizer;
159 Separator const separator(from_ascii(",").c_str());
160 // The tokenizer must not be called with temporary strings, since
161 // it does not make a copy and uses iterators of the string further
162 // down. getParam returns a reference, so this is OK.
163 Tokenizer const tokens(getParam("bibfiles"), separator);
164 Tokenizer::const_iterator const begin = tokens.begin();
165 Tokenizer::const_iterator const end = tokens.end();
167 odocstringstream dbs;
168 for (Tokenizer::const_iterator it = begin; it != end; ++it) {
169 docstring const input = trim(*it);
171 string utf8input(to_utf8(input));
173 normalize_name(buffer, runparams, utf8input, ".bib");
174 FileName const try_in_file(makeAbsPath(database + ".bib", buffer.filePath()));
175 bool const not_from_texmf = isFileReadable(try_in_file);
177 if (!runparams.inComment && !runparams.dryrun && !runparams.nice &&
180 // mangledFilename() needs the extension
181 DocFileName const in_file = DocFileName(try_in_file);
182 database = removeExtension(in_file.mangledFilename());
183 FileName const out_file(makeAbsPath(database + ".bib",
184 buffer.getMasterBuffer()->temppath()));
186 bool const success = copy(in_file, out_file);
188 lyxerr << "Failed to copy '" << in_file
189 << "' to '" << out_file << "'"
197 dbs << from_utf8(latex_path(database));
199 docstring const db_out = dbs.str();
201 // Post this warning only once.
202 static bool warned_about_spaces = false;
203 if (!warned_about_spaces &&
204 runparams.nice && db_out.find(' ') != docstring::npos) {
205 warned_about_spaces = true;
207 Alert::warning(_("Export Warning!"),
208 _("There are spaces in the paths to your BibTeX databases.\n"
209 "BibTeX will be unable to find them."));
214 string style = to_utf8(getParam("options")); // maybe empty! and with bibtotoc
216 if (prefixIs(style, "bibtotoc")) {
217 bibtotoc = "bibtotoc";
218 if (contains(style, ',')) {
219 style = split(style, bibtotoc, ',');
226 if (!style.empty()) {
228 normalize_name(buffer, runparams, style, ".bst");
229 FileName const try_in_file(makeAbsPath(base + ".bst", buffer.filePath()));
230 bool const not_from_texmf = isFileReadable(try_in_file);
231 // If this style does not come from texmf and we are not
232 // exporting to .tex copy it to the tmp directory.
233 // This prevents problems with spaces and 8bit charcaters
235 if (!runparams.inComment && !runparams.dryrun && !runparams.nice &&
237 // use new style name
238 DocFileName const in_file = DocFileName(try_in_file);
239 base = removeExtension(in_file.mangledFilename());
240 FileName const out_file(makeAbsPath(base + ".bst",
241 buffer.getMasterBuffer()->temppath()));
242 bool const success = copy(in_file, out_file);
244 lyxerr << "Failed to copy '" << in_file
245 << "' to '" << out_file << "'"
250 os << "\\bibliographystyle{"
251 << from_utf8(latex_path(normalize_name(buffer, runparams, base, ".bst")))
256 // Post this warning only once.
257 static bool warned_about_bst_spaces = false;
258 if (!warned_about_bst_spaces && runparams.nice && contains(style, ' ')) {
259 warned_about_bst_spaces = true;
260 Alert::warning(_("Export Warning!"),
261 _("There are spaces in the path to your BibTeX style file.\n"
262 "BibTeX will be unable to find it."));
265 if (!db_out.empty() && buffer.params().use_bibtopic){
266 os << "\\begin{btSect}{" << db_out << "}\n";
267 docstring btprint = getParam("btprint");
270 btprint = from_ascii("btPrintCited");
271 os << "\\" << btprint << "\n"
272 << "\\end{btSect}\n";
277 if (!bibtotoc.empty() && !buffer.params().use_bibtopic) {
278 // maybe a problem when a textclass has no "art" as
279 // part of its name, because it's than book.
280 // For the "official" lyx-layouts it's no problem to support
282 if (!contains(buffer.params().getLyXTextClass().name(),
284 if (buffer.params().sides == LyXTextClass::OneSide) {
289 os << "\\cleardoublepage";
293 os << "\\addcontentsline{toc}{chapter}{\\bibname}";
297 os << "\\addcontentsline{toc}{section}{\\refname}";
301 if (!db_out.empty() && !buffer.params().use_bibtopic){
302 os << "\\bibliography{" << db_out << "}\n";
310 vector<FileName> const InsetBibtex::getFiles(Buffer const & buffer) const
312 Path p(FileName(buffer.filePath()));
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));
376 /// remove whitespace characters, read characer sequence
377 /// not containing whitespace characters or characters in
378 /// delimChars, and remove further whitespace characters.
380 /// @return true if a string of length > 0 could be read.
382 bool readTypeOrKey(docstring & val, idocfstream & ifs, docstring const & delimChars) {
394 } while (ifs && isSpace(ch));
400 while (ifs && !isSpace(ch) && delimChars.find(ch) == docstring::npos) {
401 val += lowercase(ch);
406 while (ifs && isSpace(ch)) {
414 return val.length() > 0;
417 /// read subsequent bibtex values that are delimited with a #-character.
418 /// Concatenate all parts and replace names with the associated string in
419 /// the variable strings.
420 /// @return true if reading was successfull (all single parts were delimited
422 bool readValue(docstring & val, idocfstream & ifs, const VarMap & strings) {
435 } while (ifs && isSpace(ch));
440 // check for field type
443 // read integer value
447 } while (ifs && isDigit(ch));
452 } else if (ch == '"' || ch == '{') {
454 // read delimited text - set end delimiter
455 char_type delim = ch == '"'? '"': '}';
457 // inside this delimited text braces must match.
458 // Thus we can have a closing delimiter only
459 // when nestLevel == 0
463 while (ifs && (nestLevel > 0 || ch != delim)) {
466 // update nesting level
473 if (nestLevel < 0) return false;
490 // reading a string name
493 while (ifs && !isSpace(ch) && ch != '#' && ch != ',' && ch != '}' && ch != ')') {
494 strName += lowercase(ch);
501 // replace the string with its assigned value or
502 // discard it if it's not assigned
503 if (strName.length()) {
504 VarMap::const_iterator pos = strings.find(strName);
505 if (pos != strings.end()) {
512 while (ifs && isSpace(ch)) {
519 // continue reading next value on concatenate with '#'
529 // This method returns a comma separated list of Bibtex entries
530 void InsetBibtex::fillWithBibKeys(Buffer const & buffer,
531 std::vector<std::pair<string, docstring> > & keys) const
533 vector<FileName> const files = getFiles(buffer);
534 for (vector<FileName>::const_iterator it = files.begin();
535 it != files.end(); ++ it) {
536 // This bibtex parser is a first step to parse bibtex files
539 // - it reads the whole bibtex entry and does a syntax check
540 // (matching delimiters, missing commas,...
541 // - it recovers from errors starting with the next @-character
542 // - it reads @string definitions and replaces them in the
544 // - it accepts more characters in keys or value names than
548 // - the entries are split into name = value pairs by the
549 // parser. These have to be merged again because of the
550 // way lyx treats the entries ( pair<...>(...) ). The citation
551 // mechanism in lyx should be changed such that it can use
552 // the split entries.
553 // - messages on parsing errors can be generated.
556 // Officially bibtex does only support ASCII, but in practice
557 // you can use the encoding of the main document as long as
558 // some elements like keys and names are pure ASCII. Therefore
559 // we convert the file from the buffer encoding.
560 // We don't restrict keys to ASCII in LyX, since our own
561 // InsetBibitem can generate non-ASCII keys, and nonstandard
562 // 8bit clean bibtex forks exist.
563 idocfstream ifs(it->toFilesystemEncoding().c_str(),
565 buffer.params().encoding().iconvName());
581 if (!readTypeOrKey(entryType, ifs, from_ascii("{(")) || !ifs)
584 if (entryType == from_ascii("comment")) {
586 ifs.ignore(std::numeric_limits<int>::max(), '\n');
590 // check entry delimiter
591 char_type entryDelim;
597 if (ch == '(') entryDelim = ')';
598 else if (ch == '{') entryDelim = ')';
600 // invalid entry delimiter
606 if (entryType == from_ascii("string")) {
608 // read string and add it to the strings map
609 // (or replace it's old value)
613 if (!readTypeOrKey(name, ifs, from_ascii("#=}),")) || !ifs)
617 if (!ifs || ch != '=')
620 if (!readValue(value, ifs, strings))
623 strings[name] = value;
625 } else if (entryType == from_ascii("preamble")) {
627 // preamble definitions are discarded.
628 // can they be of any use in lyx?
631 if (!readValue(value, ifs, strings))
636 // Citation entry. Read the key and all name = value pairs
641 docstring commaNewline;
643 if (!readTypeOrKey(key, ifs, from_ascii(",})")) || !ifs)
646 // now we have a key, so we will add an entry
647 // (even if it's empty, as bibtex does)
649 // all items must be separated by a comma. If
650 // it is missing the scanning of this entry is
651 // stopped and the next is searched.
652 bool readNext = removeWSAndComma(ifs);
654 while (ifs && readNext) {
657 if (!readTypeOrKey(name, ifs, from_ascii("=}),")) || !ifs)
660 // next char must be an equal sign
670 if (!readValue(value, ifs, strings))
673 // append field to the total entry string.
675 // TODO: Here is where the fields can be put in
676 // a more intelligent structure that preserves
677 // the already known parts.
678 fields += commaNewline;
679 fields += name + from_ascii(" = {") + value + '}';
681 if (!commaNewline.length())
682 commaNewline = from_ascii(",\n");
684 readNext = removeWSAndComma(ifs);
688 keys.push_back(pair<string, docstring>(
689 to_utf8(key), fields));
694 } //< for loop over files
699 bool InsetBibtex::addDatabase(string const & db)
702 string bibfiles(to_utf8(getParam("bibfiles")));
703 if (tokenPos(bibfiles, ',', db) == -1) {
704 if (!bibfiles.empty())
706 setParam("bibfiles", from_utf8(bibfiles + db));
713 bool InsetBibtex::delDatabase(string const & db)
716 string bibfiles(to_utf8(getParam("bibfiles")));
717 if (contains(bibfiles, db)) {
718 int const n = tokenPos(bibfiles, ',', db);
721 // this is not the first database
722 string tmp = ',' + bd;
723 setParam("bibfiles", from_utf8(subst(bibfiles, tmp, string())));
725 // this is the first (or only) database
726 setParam("bibfiles", from_utf8(split(bibfiles, bd, ',')));
734 void InsetBibtex::validate(LaTeXFeatures & features) const
736 if (features.bufferParams().use_bibtopic)
737 features.require("bibtopic");