]> git.lyx.org Git - lyx.git/blob - src/tex2lyx/preamble.cpp
235ff4579945e68f26948bf5adc350566f3e184e
[lyx.git] / src / tex2lyx / preamble.cpp
1 /**
2  * \file preamble.cpp
3  * This file is part of LyX, the document processor.
4  * Licence details can be found in the file COPYING.
5  *
6  * \author André Pönitz
7  * \author Uwe Stöhr
8  *
9  * Full author contact details are available in file CREDITS.
10  */
11
12 // {[(
13
14 #include <config.h>
15
16 #include "tex2lyx.h"
17
18 #include "Layout.h"
19 #include "Lexer.h"
20 #include "TextClass.h"
21 #include "support/convert.h"
22 #include "support/filetools.h"
23 #include "support/lstrings.h"
24
25 #include <algorithm>
26 #include <iostream>
27 #include <sstream>
28 #include <string>
29 #include <vector>
30 #include <map>
31
32 using namespace std;
33
34 namespace lyx {
35
36 using support::FileName;
37 using support::libFileSearch;
38 using support::isStrDbl;
39
40 // special columntypes
41 extern std::map<char, int> special_columns;
42
43 std::map<string, vector<string> > used_packages;
44
45 // needed to handle encodings with babel
46 bool one_language = true;
47
48 // to avoid that the babel options overwrite the documentclass options
49 bool documentclass_language;
50
51 namespace {
52
53 const char * const known_languages[] = { "afrikaans", "american", "arabic",
54 "austrian", "bahasa", "basque", "belarusian", "brazil", "breton", "british",
55 "bulgarian", "canadian", "canadien", "catalan", "croatian", "czech", "danish",
56 "dutch", "english", "esperanto", "estonian", "finnish", "francais", "french",
57 "frenchb", "frenchle", "frenchpro", "galician", "german", "germanb", "greek",
58 "hebrew", "icelandic", "irish", "italian", "lsorbian", "magyar", "naustrian",
59 "ngerman", "ngermanb", "norsk", "nynorsk", "polish", "portuges", "romanian",
60 "russian", "russianb", "scottish", "serbian", "slovak", "slovene", "spanish",
61 "swedish", "thai", "turkish", "ukraineb", "ukrainian", "usorbian", "welsh", 0};
62
63 //note this when updating to lyxformat 305:
64 //bahasai, indonesian, and indon = equal to bahasa
65 //malay, and meyalu = equal to bahasam
66
67 const char * const known_french_languages[] = {"french", "frenchb", "francais",
68                                                 "frenchle", "frenchpro", 0};
69 const char * const known_german_languages[] = {"german", "germanb", 0};
70 const char * const known_ngerman_languages[] = {"ngerman", "ngermanb", 0};
71 const char * const known_russian_languages[] = {"russian", "russianb", 0};
72 const char * const known_ukrainian_languages[] = {"ukrainian", "ukraineb", 0};
73
74 char const * const known_fontsizes[] = { "10pt", "11pt", "12pt", 0 };
75
76 const char * const known_roman_fonts[] = { "ae", "bookman", "charter",
77 "cmr", "fourier", "lmodern", "mathpazo", "mathptmx", "newcent", 0};
78
79 const char * const known_sans_fonts[] = { "avant", "berasans", "cmbr", "cmss",
80 "helvet", "lmss", 0};
81
82 const char * const known_typewriter_fonts[] = { "beramono", "cmtl", "cmtt",
83 "courier", "lmtt", "luximono", "fourier", "lmodern", "mathpazo", "mathptmx",
84 "newcent", 0};
85
86 // some ugly stuff
87 ostringstream h_preamble;
88 string h_textclass               = "article";
89 string h_options                 = string();
90 string h_language                = "english";
91 string h_inputencoding           = "auto";
92 string h_font_roman              = "default";
93 string h_font_sans               = "default";
94 string h_font_typewriter         = "default";
95 string h_font_default_family     = "default";
96 string h_font_sc                 = "false";
97 string h_font_osf                = "false";
98 string h_font_sf_scale           = "100";
99 string h_font_tt_scale           = "100";
100 string h_graphics                = "default";
101 string h_paperfontsize           = "default";
102 string h_spacing                 = "single";
103 string h_papersize               = "default";
104 string h_use_geometry            = "false";
105 string h_use_amsmath             = "0";
106 string h_cite_engine             = "basic";
107 string h_use_bibtopic            = "false";
108 string h_paperorientation        = "portrait";
109 string h_secnumdepth             = "3";
110 string h_tocdepth                = "3";
111 string h_paragraph_separation    = "indent";
112 string h_defskip                 = "medskip";
113 string h_quotes_language         = "english";
114 string h_papercolumns            = "1";
115 string h_papersides              = string();
116 string h_paperpagestyle          = "default";
117 string h_tracking_changes        = "false";
118 string h_output_changes          = "false";
119
120
121 void handle_opt(vector<string> & opts, char const * const * what, string & target)
122 {
123         if (opts.empty())
124                 return;
125
126         // the last language option is the document language (for babel and LyX)
127         // the last size option is the document font size
128         vector<string>::iterator it;
129         vector<string>::iterator position = opts.begin();
130         for (; *what; ++what) {
131                 it = find(opts.begin(), opts.end(), *what);
132                 if (it != opts.end()) {
133                         documentclass_language = true;
134                         if (it >= position) {
135                                 target = *what;
136                                 position = it;
137                         }
138                 }
139         }
140 }
141
142
143 void delete_opt(vector<string> & opts, char const * const * what)
144 {
145         if (opts.empty())
146                 return;
147
148         // remove found options from the list
149         // do this after handle_opt to avoid potential memory leaks and to be able
150         // to find in every case the last language option
151         vector<string>::iterator it;
152         for (; *what; ++what) {
153                 it = find(opts.begin(), opts.end(), *what);
154                 if (it != opts.end())
155                         opts.erase(it);
156         }
157 }
158
159
160 /*!
161  * Split a package options string (keyval format) into a vector.
162  * Example input:
163  *   authorformat=smallcaps,
164  *   commabeforerest,
165  *   titleformat=colonsep,
166  *   bibformat={tabular,ibidem,numbered}
167  */
168 vector<string> split_options(string const & input)
169 {
170         vector<string> options;
171         string option;
172         Parser p(input);
173         while (p.good()) {
174                 Token const & t = p.get_token();
175                 if (t.asInput() == ",") {
176                         options.push_back(trim(option));
177                         option.erase();
178                 } else if (t.asInput() == "=") {
179                         option += '=';
180                         p.skip_spaces(true);
181                         if (p.next_token().asInput() == "{")
182                                 option += '{' + p.getArg('{', '}') + '}';
183                 } else if (t.cat() != catSpace)
184                         option += t.asInput();
185         }
186
187         if (!option.empty())
188                 options.push_back(trim(option));
189
190         return options;
191 }
192
193
194 /*!
195  * Add package \p name with options \p options to used_packages.
196  * Remove options from \p options that we don't want to output.
197  */
198 void add_package(string const & name, vector<string> & options)
199 {
200         // every package inherits the global options
201         if (used_packages.find(name) == used_packages.end())
202                 used_packages[name] = split_options(h_options);
203
204         vector<string> & v = used_packages[name];
205         v.insert(v.end(), options.begin(), options.end());
206         if (name == "jurabib") {
207                 // Don't output the order argument (see the cite command
208                 // handling code in text.cpp).
209                 vector<string>::iterator end =
210                         remove(options.begin(), options.end(), "natbiborder");
211                 end = remove(options.begin(), end, "jurabiborder");
212                 options.erase(end, options.end());
213         }
214 }
215
216
217 // Given is a string like "scaled=0.9", return 0.9 * 100
218 string const scale_as_percentage(string const & scale)
219 {
220         string::size_type pos = scale.find('=');
221         if (pos != string::npos) {
222                 string value = scale.substr(pos + 1);
223                 if (isStrDbl(value))
224                         return convert<string>(100 * convert<double>(value));
225         }
226         // If the input string didn't match our expectations.
227         // return the default value "100"
228         return "100";
229 }
230
231
232 void handle_package(string const & name, string const & opts)
233 {
234         vector<string> options = split_options(opts);
235         add_package(name, options);
236         string scale;
237
238         // roman fonts
239         if (is_known(name, known_roman_fonts))
240                 h_font_roman = name;
241         if (name == "fourier") {
242                 h_font_roman = "utopia";
243                 // when font uses real small capitals
244                 if (opts == "expert")
245                         h_font_sc = "true";
246         }
247         if (name == "mathpazo")
248                 h_font_roman = "palatino";
249         if (name == "mathptmx")
250                 h_font_roman = "times";
251         // sansserif fonts
252         if (is_known(name, known_sans_fonts)) {
253                 h_font_sans = name;
254                 if (!opts.empty()) {
255                         scale = opts;
256                         h_font_sf_scale = scale_as_percentage(scale);
257                 }
258         }
259         // typewriter fonts
260         if (is_known(name, known_typewriter_fonts)) {
261                 h_font_typewriter = name;
262                 if (!opts.empty()) {
263                         scale = opts;
264                         h_font_tt_scale = scale_as_percentage(scale);
265                 }
266         }
267         // font uses old-style figure
268         if (name == "eco")
269                 h_font_osf = "true";
270
271         else if (name == "amsmath" || name == "amssymb")
272                 h_use_amsmath = "1";
273         else if (name == "babel" && !opts.empty()) {
274                 // check if more than one option was used - used later for inputenc
275                 // in case inputenc is parsed before babel, set the encoding to auto
276                 if (options.begin() != options.end() - 1) {
277                         one_language = false;
278                         h_inputencoding = "auto";
279                 }
280                 // only set the document language when there was not already one set
281                 // via the documentclass options
282                 // babel takes the the last language given in the documentclass options
283                 // as document language. If there is no such language option, the last
284                 // option of its \usepackage call is used.
285                 if (documentclass_language == false) {
286                         handle_opt(options, known_languages, h_language);
287                         delete_opt(options, known_languages);
288                         if (is_known(h_language, known_french_languages))
289                                 h_language = "french";
290                         else if (is_known(h_language, known_german_languages))
291                                 h_language = "german";
292                         else if (is_known(h_language, known_ngerman_languages))
293                                 h_language = "ngerman";
294                         else if (is_known(h_language, known_russian_languages))
295                                 h_language = "russian";
296                         else if (is_known(h_language, known_ukrainian_languages))
297                                 h_language = "ukrainian";
298                         h_quotes_language = h_language;
299                 }
300         }
301         else if (name == "fontenc")
302                 ; // ignore this
303         else if (name == "inputenc") {
304                 // only set when there is not more than one inputenc option
305                 // therefore check for the "," character
306                 // also only set when there is not more then one babel language option
307                 if (opts.find(",") == string::npos && one_language == true)
308                         h_inputencoding = opts;
309                 options.clear();
310         } else if (name == "makeidx")
311                 ; // ignore this
312         else if (name == "verbatim")
313                 ; // ignore this
314         else if (name == "graphicx")
315                 ; // ignore this
316         else if (is_known(name, known_languages)) {
317                 if (is_known(name, known_french_languages))
318                         h_language = "french";
319                 else if (is_known(name, known_german_languages))
320                         h_language = "german";
321                 else if (is_known(name, known_ngerman_languages))
322                         h_language = "ngerman";
323                 else if (is_known(name, known_russian_languages))
324                         h_language = "russian";
325                 else if (is_known(name, known_ukrainian_languages))
326                         h_language = "ukrainian";
327                 else
328                         h_language = name;
329                 h_quotes_language = h_language;
330
331         } else if (name == "natbib") {
332                 h_cite_engine = "natbib_authoryear";
333                 vector<string>::iterator it =
334                         find(options.begin(), options.end(), "authoryear");
335                 if (it != options.end())
336                         options.erase(it);
337                 else {
338                         it = find(options.begin(), options.end(), "numbers");
339                         if (it != options.end()) {
340                                 h_cite_engine = "natbib_numerical";
341                                 options.erase(it);
342                         }
343                 }
344         } else if (name == "jurabib") {
345                 h_cite_engine = "jurabib";
346         } else if (options.empty())
347                 h_preamble << "\\usepackage{" << name << "}\n";
348         else {
349                 h_preamble << "\\usepackage[" << opts << "]{" << name << "}\n";
350                 options.clear();
351         }
352
353         // We need to do something with the options...
354         if (!options.empty())
355                 cerr << "Ignoring options '" << join(options, ",")
356                      << "' of package " << name << '.' << endl;
357 }
358
359
360
361 void end_preamble(ostream & os, TextClass const & /*textclass*/)
362 {
363         os << "#LyX file created by tex2lyx " << PACKAGE_VERSION << "\n"
364            << "\\lyxformat 247\n"
365            << "\\begin_document\n"
366            << "\\begin_header\n"
367            << "\\textclass " << h_textclass << "\n";
368         if (!h_preamble.str().empty())
369                 os << "\\begin_preamble\n" << h_preamble.str() << "\n\\end_preamble\n";
370         if (!h_options.empty())
371                 os << "\\options " << h_options << "\n";
372         os << "\\language " << h_language << "\n"
373            << "\\inputencoding " << h_inputencoding << "\n"
374            << "\\font_roman " << h_font_roman << "\n"
375            << "\\font_sans " << h_font_sans << "\n"
376            << "\\font_typewriter " << h_font_typewriter << "\n"
377            << "\\font_default_family " << h_font_default_family << "\n"
378            << "\\font_sc " << h_font_sc << "\n"
379            << "\\font_osf " << h_font_osf << "\n"
380            << "\\font_sf_scale " << h_font_sf_scale << "\n"
381            << "\\font_tt_scale " << h_font_tt_scale << "\n"
382            << "\\graphics " << h_graphics << "\n"
383            << "\\paperfontsize " << h_paperfontsize << "\n"
384            << "\\spacing " << h_spacing << "\n"
385            << "\\papersize " << h_papersize << "\n"
386            << "\\use_geometry " << h_use_geometry << "\n"
387            << "\\use_amsmath " << h_use_amsmath << "\n"
388            << "\\cite_engine " << h_cite_engine << "\n"
389            << "\\use_bibtopic " << h_use_bibtopic << "\n"
390            << "\\paperorientation " << h_paperorientation << "\n"
391            << "\\secnumdepth " << h_secnumdepth << "\n"
392            << "\\tocdepth " << h_tocdepth << "\n"
393            << "\\paragraph_separation " << h_paragraph_separation << "\n"
394            << "\\defskip " << h_defskip << "\n"
395            << "\\quotes_language " << h_quotes_language << "\n"
396            << "\\papercolumns " << h_papercolumns << "\n"
397            << "\\papersides " << h_papersides << "\n"
398            << "\\paperpagestyle " << h_paperpagestyle << "\n"
399            << "\\tracking_changes " << h_tracking_changes << "\n"
400            << "\\output_changes " << h_output_changes << "\n"
401            << "\\end_header\n\n"
402            << "\\begin_body\n";
403         // clear preamble for subdocuments
404         h_preamble.str("");
405 }
406
407 } // anonymous namespace
408
409 TextClass const parse_preamble(Parser & p, ostream & os, string const & forceclass)
410 {
411         // initialize fixed types
412         special_columns['D'] = 3;
413         bool is_full_document = false;
414
415         // determine whether this is a full document or a fragment for inclusion
416         while (p.good()) {
417                 Token const & t = p.get_token();
418
419                 if (t.cat() == catEscape && t.cs() == "documentclass") {
420                         is_full_document = true;
421                         break;
422                 }
423         }
424         p.reset();
425
426         while (is_full_document && p.good()) {
427                 Token const & t = p.get_token();
428
429 #ifdef FILEDEBUG
430                 cerr << "t: " << t << "\n";
431 #endif
432
433                 //
434                 // cat codes
435                 //
436                 if (t.cat() == catLetter ||
437                           t.cat() == catSuper ||
438                           t.cat() == catSub ||
439                           t.cat() == catOther ||
440                           t.cat() == catMath ||
441                           t.cat() == catActive ||
442                           t.cat() == catBegin ||
443                           t.cat() == catEnd ||
444                           t.cat() == catAlign ||
445                           t.cat() == catParameter)
446                 h_preamble << t.character();
447
448                 else if (t.cat() == catSpace || t.cat() == catNewline)
449                         h_preamble << t.asInput();
450
451                 else if (t.cat() == catComment)
452                         h_preamble << t.asInput();
453
454                 else if (t.cs() == "pagestyle")
455                         h_paperpagestyle = p.verbatim_item();
456
457                 else if (t.cs() == "makeatletter") {
458                         p.setCatCode('@', catLetter);
459                 }
460
461                 else if (t.cs() == "makeatother") {
462                         p.setCatCode('@', catOther);
463                 }
464
465                 else if (t.cs() == "newcommand" || t.cs() == "renewcommand"
466                             || t.cs() == "providecommand") {
467                         bool star = false;
468                         if (p.next_token().character() == '*') {
469                                 p.get_token();
470                                 star = true;
471                         }
472                         string const name = p.verbatim_item();
473                         string const opt1 = p.getOpt();
474                         string const opt2 = p.getFullOpt();
475                         string const body = p.verbatim_item();
476                         // font settings
477                         if (name == "\\rmdefault")
478                                 if (is_known(body, known_roman_fonts))
479                                         h_font_roman = body;
480
481                         if (name == "\\sfdefault")
482                                 if (is_known(body, known_sans_fonts))
483                                         h_font_sans = body;
484
485                         if (name == "\\ttdefault")
486                                 if (is_known(body, known_typewriter_fonts))
487                                         h_font_typewriter = body;
488
489                         if (name == "\\familydefault") {
490                                 string family = body;
491                                 // remove leading "\"
492                                 h_font_default_family = family.erase(0,1);
493                         }
494                         // only non-lyxspecific stuff
495                         if (   name != "\\noun"
496                             && name != "\\tabularnewline"
497                             && name != "\\LyX"
498                             && name != "\\lyxline"
499                             && name != "\\lyxaddress"
500                             && name != "\\lyxrightaddress"
501                             && name != "\\lyxdot"
502                             && name != "\\boldsymbol"
503                             && name != "\\lyxarrow"
504                             && name != "\\rmdefault"
505                             && name != "\\sfdefault"
506                             && name != "\\ttdefault"
507                             && name != "\\familydefault") {
508                                 ostringstream ss;
509                                 ss << '\\' << t.cs();
510                                 if (star)
511                                         ss << '*';
512                                 ss << '{' << name << '}' << opt1 << opt2
513                                    << '{' << body << "}";
514                                 h_preamble << ss.str();
515
516                                 // Add the command to the known commands
517                                 add_known_command(name, opt1, !opt2.empty());
518 /*
519                                 ostream & out = in_preamble ? h_preamble : os;
520                                 out << "\\" << t.cs() << "{" << name << "}"
521                                     << opts << "{" << body << "}";
522 */
523                         }
524                 }
525
526                 else if (t.cs() == "documentclass") {
527                         vector<string> opts = split_options(p.getArg('[', ']'));
528                         handle_opt(opts, known_fontsizes, h_paperfontsize);
529                         delete_opt(opts, known_fontsizes);
530                         // delete "pt" at the end
531                         string::size_type i = h_paperfontsize.find("pt");
532                         if (i != string::npos)
533                                 h_paperfontsize.erase(i);
534                         // to avoid that the babel options overwrite the documentclass options
535                         documentclass_language = false;
536                         handle_opt(opts, known_languages, h_language);
537                         delete_opt(opts, known_languages);
538                         if (is_known(h_language, known_french_languages))
539                                 h_language = "french";
540                         else if (is_known(h_language, known_german_languages))
541                                 h_language = "german";
542                         else if (is_known(h_language, known_ngerman_languages))
543                                 h_language = "ngerman";
544                         else if (is_known(h_language, known_russian_languages))
545                                 h_language = "russian";
546                         else if (is_known(h_language, known_ukrainian_languages))
547                                 h_language = "ukrainian";
548                         h_quotes_language = h_language;
549                         h_options = join(opts, ",");
550                         h_textclass = p.getArg('{', '}');
551                 }
552
553                 else if (t.cs() == "usepackage") {
554                         string const options = p.getArg('[', ']');
555                         string const name = p.getArg('{', '}');
556                         if (options.empty() && name.find(',')) {
557                                 vector<string> vecnames;
558                                 split(name, vecnames, ',');
559                                 vector<string>::const_iterator it  = vecnames.begin();
560                                 vector<string>::const_iterator end = vecnames.end();
561                                 for (; it != end; ++it)
562                                         handle_package(trim(*it), string());
563                         } else {
564                                 handle_package(name, options);
565                         }
566                 }
567
568                 else if (t.cs() == "newenvironment") {
569                         string const name = p.getArg('{', '}');
570                         ostringstream ss;
571                         ss << "\\newenvironment{" << name << "}";
572                         ss << p.getOpt();
573                         ss << p.getOpt();
574                         ss << '{' << p.verbatim_item() << '}';
575                         ss << '{' << p.verbatim_item() << '}';
576                         if (name != "lyxcode" && name != "lyxlist" &&
577                             name != "lyxrightadress" &&
578                             name != "lyxaddress" && name != "lyxgreyedout")
579                                 h_preamble << ss.str();
580                 }
581
582                 else if (t.cs() == "def") {
583                         string name = p.get_token().cs();
584                         while (p.next_token().cat() != catBegin)
585                                 name += p.get_token().asString();
586                         h_preamble << "\\def\\" << name << '{'
587                                    << p.verbatim_item() << "}";
588                 }
589
590                 else if (t.cs() == "newcolumntype") {
591                         string const name = p.getArg('{', '}');
592                         trim(name);
593                         int nargs = 0;
594                         string opts = p.getOpt();
595                         if (!opts.empty()) {
596                                 istringstream is(string(opts, 1));
597                                 is >> nargs;
598                         }
599                         special_columns[name[0]] = nargs;
600                         h_preamble << "\\newcolumntype{" << name << "}";
601                         if (nargs)
602                                 h_preamble << "[" << nargs << "]";
603                         h_preamble << "{" << p.verbatim_item() << "}";
604                 }
605
606                 else if (t.cs() == "setcounter") {
607                         string const name = p.getArg('{', '}');
608                         string const content = p.getArg('{', '}');
609                         if (name == "secnumdepth")
610                                 h_secnumdepth = content;
611                         else if (name == "tocdepth")
612                                 h_tocdepth = content;
613                         else
614                                 h_preamble << "\\setcounter{" << name << "}{" << content << "}";
615                 }
616
617                 else if (t.cs() == "setlength") {
618                         string const name = p.verbatim_item();
619                         string const content = p.verbatim_item();
620                         // Is this correct?
621                         if (name == "parskip")
622                                 h_paragraph_separation = "skip";
623                         else if (name == "parindent")
624                                 h_paragraph_separation = "skip";
625                         else
626                                 h_preamble << "\\setlength{" << name << "}{" << content << "}";
627                 }
628
629                 else if (t.cs() == "begin") {
630                         string const name = p.getArg('{', '}');
631                         if (name == "document")
632                                 break;
633                         h_preamble << "\\begin{" << name << "}";
634                 }
635
636                 else if (t.cs() == "jurabibsetup") {
637                         vector<string> jurabibsetup =
638                                 split_options(p.getArg('{', '}'));
639                         // add jurabibsetup to the jurabib package options
640                         add_package("jurabib", jurabibsetup);
641                         if (!jurabibsetup.empty()) {
642                                 h_preamble << "\\jurabibsetup{"
643                                            << join(jurabibsetup, ",") << '}';
644                         }
645                 }
646
647                 else if (!t.cs().empty())
648                         h_preamble << '\\' << t.cs();
649         }
650         p.skip_spaces();
651
652         // Force textclass if the user wanted it
653         if (!forceclass.empty())
654                 h_textclass = forceclass;
655         if (noweb_mode && !lyx::support::prefixIs(h_textclass, "literate-"))
656                 h_textclass.insert(0, "literate-");
657         FileName layoutfilename = libFileSearch("layouts", h_textclass, "layout");
658         if (layoutfilename.empty()) {
659                 cerr << "Error: Could not find layout file for textclass \"" << h_textclass << "\"." << endl;
660                 exit(1);
661         }
662         TextClass textclass;
663         textclass.read(layoutfilename);
664         if (h_papersides.empty()) {
665                 ostringstream ss;
666                 ss << textclass.sides();
667                 h_papersides = ss.str();
668         }
669         end_preamble(os, textclass);
670         return textclass;
671 }
672
673 // }])
674
675
676 } // namespace lyx