2 * \file tex2lyx/text.cpp
3 * This file is part of LyX, the document processor.
4 * Licence details can be found in the file COPYING.
7 * \author Jean-Marc Lasgouttes
10 * Full author contact details are available in file CREDITS.
21 #include "FloatList.h"
25 #include "support/lassert.h"
26 #include "support/convert.h"
27 #include "support/FileName.h"
28 #include "support/filetools.h"
29 #include "support/lstrings.h"
38 using namespace lyx::support;
43 void parse_text_in_inset(Parser & p, ostream & os, unsigned flags, bool outer,
44 Context const & context, InsetLayout const * layout)
46 bool const forcePlainLayout =
47 layout ? layout->forcePlainLayout() : false;
48 Context newcontext(true, context.textclass);
50 newcontext.layout = &context.textclass.plainLayout();
52 newcontext.font = context.font;
53 parse_text(p, os, flags, outer, newcontext);
54 newcontext.check_end_layout(os);
60 void parse_text_in_inset(Parser & p, ostream & os, unsigned flags, bool outer,
61 Context const & context, string const & name)
63 InsetLayout const * layout = 0;
64 DocumentClass::InsetLayouts::const_iterator it =
65 context.textclass.insetLayouts().find(from_ascii(name));
66 if (it != context.textclass.insetLayouts().end())
67 layout = &(it->second);
68 parse_text_in_inset(p, os, flags, outer, context, layout);
71 /// parses a paragraph snippet, useful for example for \\emph{...}
72 void parse_text_snippet(Parser & p, ostream & os, unsigned flags, bool outer,
75 Context newcontext(context);
76 // Don't inherit the paragraph-level extra stuff
77 newcontext.par_extra_stuff.clear();
78 parse_text(p, os, flags, outer, newcontext);
79 // Make sure that we don't create invalid .lyx files
80 context.need_layout = newcontext.need_layout;
81 context.need_end_layout = newcontext.need_end_layout;
86 * Thin wrapper around parse_text_snippet() using a string.
88 * We completely ignore \c context.need_layout and \c context.need_end_layout,
89 * because our return value is not used directly (otherwise the stream version
90 * of parse_text_snippet() could be used). That means that the caller needs
91 * to do layout management manually.
92 * This is intended to parse text that does not create any layout changes.
94 string parse_text_snippet(Parser & p, unsigned flags, const bool outer,
97 Context newcontext(context);
98 newcontext.need_layout = false;
99 newcontext.need_end_layout = false;
100 newcontext.new_layout_allowed = false;
101 // Avoid warning by Context::~Context()
102 newcontext.par_extra_stuff.clear();
104 parse_text_snippet(p, os, flags, outer, newcontext);
109 char const * const known_ref_commands[] = { "ref", "pageref", "vref",
110 "vpageref", "prettyref", "eqref", 0 };
114 * The starred forms are also known except for "citefullauthor",
115 * "citeyear" and "citeyearpar".
117 char const * const known_natbib_commands[] = { "cite", "citet", "citep",
118 "citealt", "citealp", "citeauthor", "citeyear", "citeyearpar",
119 "citefullauthor", "Citet", "Citep", "Citealt", "Citealp", "Citeauthor", 0 };
123 * No starred form other than "cite*" known.
125 char const * const known_jurabib_commands[] = { "cite", "citet", "citep",
126 "citealt", "citealp", "citeauthor", "citeyear", "citeyearpar",
127 // jurabib commands not (yet) supported by LyX:
129 // "footcite", "footcitet", "footcitep", "footcitealt", "footcitealp",
130 // "footciteauthor", "footciteyear", "footciteyearpar",
131 "citefield", "citetitle", 0 };
133 /// LaTeX names for quotes
134 char const * const known_quotes[] = { "dq", "guillemotleft", "flqq", "og",
135 "guillemotright", "frqq", "fg", "glq", "glqq", "textquoteleft", "grq", "grqq",
136 "quotedblbase", "textquotedblleft", "quotesinglbase", "textquoteright", "flq",
137 "guilsinglleft", "frq", "guilsinglright", 0};
139 /// the same as known_quotes with .lyx names
140 char const * const known_coded_quotes[] = { "prd", "ard", "ard", "ard",
141 "ald", "ald", "ald", "gls", "gld", "els", "els", "grd",
142 "gld", "grd", "gls", "ers", "fls",
143 "fls", "frs", "frs", 0};
145 /// LaTeX names for font sizes
146 char const * const known_sizes[] = { "tiny", "scriptsize", "footnotesize",
147 "small", "normalsize", "large", "Large", "LARGE", "huge", "Huge", 0};
149 /// the same as known_sizes with .lyx names
150 char const * const known_coded_sizes[] = { "tiny", "scriptsize", "footnotesize",
151 "small", "normal", "large", "larger", "largest", "huge", "giant", 0};
153 /// LaTeX 2.09 names for font families
154 char const * const known_old_font_families[] = { "rm", "sf", "tt", 0};
156 /// LaTeX names for font families
157 char const * const known_font_families[] = { "rmfamily", "sffamily",
160 /// the same as known_old_font_families and known_font_families with .lyx names
161 char const * const known_coded_font_families[] = { "roman", "sans",
164 /// LaTeX 2.09 names for font series
165 char const * const known_old_font_series[] = { "bf", 0};
167 /// LaTeX names for font series
168 char const * const known_font_series[] = { "bfseries", "mdseries", 0};
170 /// the same as known_old_font_series and known_font_series with .lyx names
171 char const * const known_coded_font_series[] = { "bold", "medium", 0};
173 /// LaTeX 2.09 names for font shapes
174 char const * const known_old_font_shapes[] = { "it", "sl", "sc", 0};
176 /// LaTeX names for font shapes
177 char const * const known_font_shapes[] = { "itshape", "slshape", "scshape",
180 /// the same as known_old_font_shapes and known_font_shapes with .lyx names
181 char const * const known_coded_font_shapes[] = { "italic", "slanted",
182 "smallcaps", "up", 0};
185 * Graphics file extensions known by the dvips driver of the graphics package.
186 * These extensions are used to complete the filename of an included
187 * graphics file if it does not contain an extension.
188 * The order must be the same that latex uses to find a file, because we
189 * will use the first extension that matches.
190 * This is only an approximation for the common cases. If we would want to
191 * do it right in all cases, we would need to know which graphics driver is
192 * used and know the extensions of every driver of the graphics package.
194 char const * const known_dvips_graphics_formats[] = {"eps", "ps", "eps.gz",
195 "ps.gz", "eps.Z", "ps.Z", 0};
198 * Graphics file extensions known by the pdftex driver of the graphics package.
199 * \sa known_dvips_graphics_formats
201 char const * const known_pdftex_graphics_formats[] = {"png", "pdf", "jpg",
205 * Known file extensions for TeX files as used by \\include.
207 char const * const known_tex_extensions[] = {"tex", 0};
209 /// spaces known by InsetSpace
210 char const * const known_spaces[] = { " ", "space", ",", "thinspace", "quad",
211 "qquad", "enspace", "enskip", "negthinspace", "hfill", "dotfill", "hrulefill",
212 "leftarrowfill", "rightarrowfill", "upbracefill", "downbracefill", 0};
214 /// the same as known_spaces with .lyx names
215 char const * const known_coded_spaces[] = { "space{}", "space{}",
216 "thinspace{}", "thinspace{}", "quad{}", "qquad{}", "enspace{}", "enskip{}",
217 "negthinspace{}", "hfill{}", "dotfill{}", "hrulefill{}", "leftarrowfill{}",
218 "rightarrowfill{}", "upbracefill{}", "downbracefill{}", 0};
220 /// These are translated by LyX to commands like "\\LyX{}", so we have to put
221 /// them in ERT. "LaTeXe" must come before "LaTeX"!
222 char const * const known_phrases[] = {"LyX", "TeX", "LaTeXe", "LaTeX", 0};
223 char const * const known_coded_phrases[] = {"LyX", "TeX", "LaTeX2e", "LaTeX", 0};
224 int const known_phrase_lengths[] = {3, 5, 7, 0};
227 /// splits "x=z, y=b" into a map and an ordered keyword vector
228 void split_map(string const & s, map<string, string> & res, vector<string> & keys)
233 keys.resize(v.size());
234 for (size_t i = 0; i < v.size(); ++i) {
235 size_t const pos = v[i].find('=');
236 string const index = trim(v[i].substr(0, pos));
237 string const value = trim(v[i].substr(pos + 1, string::npos));
245 * Split a LaTeX length into value and unit.
246 * The latter can be a real unit like "pt", or a latex length variable
247 * like "\textwidth". The unit may contain additional stuff like glue
248 * lengths, but we don't care, because such lengths are ERT anyway.
249 * \returns true if \p value and \p unit are valid.
251 bool splitLatexLength(string const & len, string & value, string & unit)
255 const string::size_type i = len.find_first_not_of(" -+0123456789.,");
256 //'4,5' is a valid LaTeX length number. Change it to '4.5'
257 string const length = subst(len, ',', '.');
258 if (i == string::npos)
261 if (len[0] == '\\') {
262 // We had something like \textwidth without a factor
268 value = trim(string(length, 0, i));
272 // 'cM' is a valid LaTeX length unit. Change it to 'cm'
273 if (contains(len, '\\'))
274 unit = trim(string(len, i));
276 unit = ascii_lowercase(trim(string(len, i)));
281 /// A simple function to translate a latex length to something LyX can
282 /// understand. Not perfect, but rather best-effort.
283 bool translate_len(string const & length, string & valstring, string & unit)
285 if (!splitLatexLength(length, valstring, unit))
287 // LyX uses percent values
289 istringstream iss(valstring);
294 string const percentval = oss.str();
296 if (unit.empty() || unit[0] != '\\')
298 string::size_type const i = unit.find(' ');
299 string const endlen = (i == string::npos) ? string() : string(unit, i);
300 if (unit == "\\textwidth") {
301 valstring = percentval;
302 unit = "text%" + endlen;
303 } else if (unit == "\\columnwidth") {
304 valstring = percentval;
305 unit = "col%" + endlen;
306 } else if (unit == "\\paperwidth") {
307 valstring = percentval;
308 unit = "page%" + endlen;
309 } else if (unit == "\\linewidth") {
310 valstring = percentval;
311 unit = "line%" + endlen;
312 } else if (unit == "\\paperheight") {
313 valstring = percentval;
314 unit = "pheight%" + endlen;
315 } else if (unit == "\\textheight") {
316 valstring = percentval;
317 unit = "theight%" + endlen;
325 string translate_len(string const & length)
329 if (translate_len(length, value, unit))
331 // If the input is invalid, return what we have.
339 * Translates a LaTeX length into \p value, \p unit and
340 * \p special parts suitable for a box inset.
341 * The difference from translate_len() is that a box inset knows about
342 * some special "units" that are stored in \p special.
344 void translate_box_len(string const & length, string & value, string & unit, string & special)
346 if (translate_len(length, value, unit)) {
347 if (unit == "\\height" || unit == "\\depth" ||
348 unit == "\\totalheight" || unit == "\\width") {
349 special = unit.substr(1);
350 // The unit is not used, but LyX requires a dummy setting
363 * Find a file with basename \p name in path \p path and an extension
366 string find_file(string const & name, string const & path,
367 char const * const * extensions)
369 for (char const * const * what = extensions; *what; ++what) {
370 string const trial = addExtension(name, *what);
371 if (makeAbsPath(trial, path).exists())
378 void begin_inset(ostream & os, string const & name)
380 os << "\n\\begin_inset " << name;
384 void begin_command_inset(ostream & os, string const & name,
385 string const & latexname)
387 begin_inset(os, "CommandInset ");
388 os << name << "\nLatexCommand " << latexname << '\n';
392 void end_inset(ostream & os)
394 os << "\n\\end_inset\n\n";
398 bool skip_braces(Parser & p)
400 if (p.next_token().cat() != catBegin)
403 if (p.next_token().cat() == catEnd) {
412 void handle_ert(ostream & os, string const & s, Context & context)
414 // We must have a valid layout before outputting the ERT inset.
415 context.check_layout(os);
416 Context newcontext(true, context.textclass);
417 begin_inset(os, "ERT");
418 os << "\nstatus collapsed\n";
419 newcontext.check_layout(os);
420 for (string::const_iterator it = s.begin(), et = s.end(); it != et; ++it) {
422 os << "\n\\backslash\n";
423 else if (*it == '\n') {
424 newcontext.new_paragraph(os);
425 newcontext.check_layout(os);
429 newcontext.check_end_layout(os);
434 void handle_comment(ostream & os, string const & s, Context & context)
436 // TODO: Handle this better
437 Context newcontext(true, context.textclass);
438 begin_inset(os, "ERT");
439 os << "\nstatus collapsed\n";
440 newcontext.check_layout(os);
441 for (string::const_iterator it = s.begin(), et = s.end(); it != et; ++it) {
443 os << "\n\\backslash\n";
447 // make sure that our comment is the last thing on the line
448 newcontext.new_paragraph(os);
449 newcontext.check_layout(os);
450 newcontext.check_end_layout(os);
455 Layout const * findLayout(TextClass const & textclass, string const & name, bool command)
457 Layout const * layout = findLayoutWithoutModule(textclass, name, command);
460 if (checkModule(name, command))
461 return findLayoutWithoutModule(textclass, name, command);
466 InsetLayout const * findInsetLayout(TextClass const & textclass, string const & name, bool command)
468 InsetLayout const * insetlayout = findInsetLayoutWithoutModule(textclass, name, command);
471 if (checkModule(name, command))
472 return findInsetLayoutWithoutModule(textclass, name, command);
477 void eat_whitespace(Parser &, ostream &, Context &, bool);
481 * Skips whitespace and braces.
482 * This should be called after a command has been parsed that is not put into
483 * ERT, and where LyX adds "{}" if needed.
485 void skip_spaces_braces(Parser & p, bool keepws = false)
487 /* The following four examples produce the same typeset output and
488 should be handled by this function:
496 // Unfortunately we need to skip comments, too.
497 // We can't use eat_whitespace since writing them after the {}
498 // results in different output in some cases.
499 bool const skipped_spaces = p.skip_spaces(true);
500 bool const skipped_braces = skip_braces(p);
501 if (keepws && skipped_spaces && !skipped_braces)
502 // put back the space (it is better handled by check_space)
503 p.unskip_spaces(true);
507 void output_command_layout(ostream & os, Parser & p, bool outer,
508 Context & parent_context,
509 Layout const * newlayout)
511 TeXFont const oldFont = parent_context.font;
512 // save the current font size
513 string const size = oldFont.size;
514 // reset the font size to default, because the font size switches
515 // don't affect section headings and the like
516 parent_context.font.size = Context::normalfont.size;
517 // we only need to write the font change if we have an open layout
518 if (!parent_context.atParagraphStart())
519 output_font_change(os, oldFont, parent_context.font);
520 parent_context.check_end_layout(os);
521 Context context(true, parent_context.textclass, newlayout,
522 parent_context.layout, parent_context.font);
523 if (parent_context.deeper_paragraph) {
524 // We are beginning a nested environment after a
525 // deeper paragraph inside the outer list environment.
526 // Therefore we don't need to output a "begin deeper".
527 context.need_end_deeper = true;
529 context.check_deeper(os);
530 context.check_layout(os);
531 unsigned int optargs = 0;
532 while (optargs < context.layout->optargs) {
533 eat_whitespace(p, os, context, false);
534 if (p.next_token().cat() == catEscape ||
535 p.next_token().character() != '[')
537 p.get_token(); // eat '['
538 begin_inset(os, "OptArg\n");
539 os << "status collapsed\n\n";
540 parse_text_in_inset(p, os, FLAG_BRACK_LAST, outer, context);
542 eat_whitespace(p, os, context, false);
545 unsigned int reqargs = 0;
546 while (LYX_FORMAT >= 392 && reqargs < context.layout->reqargs) {
547 eat_whitespace(p, os, context, false);
548 if (p.next_token().cat() != catBegin)
550 p.get_token(); // eat '{'
551 begin_inset(os, "OptArg\n");
552 os << "status collapsed\n\n";
553 parse_text_in_inset(p, os, FLAG_BRACE_LAST, outer, context);
555 eat_whitespace(p, os, context, false);
558 parse_text(p, os, FLAG_ITEM, outer, context);
559 context.check_end_layout(os);
560 if (parent_context.deeper_paragraph) {
561 // We must suppress the "end deeper" because we
562 // suppressed the "begin deeper" above.
563 context.need_end_deeper = false;
565 context.check_end_deeper(os);
566 // We don't need really a new paragraph, but
567 // we must make sure that the next item gets a \begin_layout.
568 parent_context.new_paragraph(os);
569 // Set the font size to the original value. No need to output it here
570 // (Context::begin_layout() will do that if needed)
571 parent_context.font.size = size;
576 * Output a space if necessary.
577 * This function gets called for every whitespace token.
579 * We have three cases here:
580 * 1. A space must be suppressed. Example: The lyxcode case below
581 * 2. A space may be suppressed. Example: Spaces before "\par"
582 * 3. A space must not be suppressed. Example: A space between two words
584 * We currently handle only 1. and 3 and from 2. only the case of
585 * spaces before newlines as a side effect.
587 * 2. could be used to suppress as many spaces as possible. This has two effects:
588 * - Reimporting LyX generated LaTeX files changes almost no whitespace
589 * - Superflous whitespace from non LyX generated LaTeX files is removed.
590 * The drawback is that the logic inside the function becomes
591 * complicated, and that is the reason why it is not implemented.
593 void check_space(Parser & p, ostream & os, Context & context)
595 Token const next = p.next_token();
596 Token const curr = p.curr_token();
597 // A space before a single newline and vice versa must be ignored
598 // LyX emits a newline before \end{lyxcode}.
599 // This newline must be ignored,
600 // otherwise LyX will add an additional protected space.
601 if (next.cat() == catSpace ||
602 next.cat() == catNewline ||
603 (next.cs() == "end" && context.layout->free_spacing && curr.cat() == catNewline)) {
606 context.check_layout(os);
612 * Parse all arguments of \p command
614 void parse_arguments(string const & command,
615 vector<ArgumentType> const & template_arguments,
616 Parser & p, ostream & os, bool outer, Context & context)
618 string ert = command;
619 size_t no_arguments = template_arguments.size();
620 for (size_t i = 0; i < no_arguments; ++i) {
621 switch (template_arguments[i]) {
623 // This argument contains regular LaTeX
624 handle_ert(os, ert + '{', context);
625 eat_whitespace(p, os, context, false);
626 parse_text(p, os, FLAG_ITEM, outer, context);
630 // This argument consists only of a single item.
631 // The presence of '{' or not must be preserved.
633 if (p.next_token().cat() == catBegin)
634 ert += '{' + p.verbatim_item() + '}';
636 ert += p.verbatim_item();
639 // This argument may contain special characters
640 ert += '{' + p.verbatim_item() + '}';
643 // true because we must not eat whitespace
644 ert += p.getOpt(true);
648 handle_ert(os, ert, context);
653 * Check whether \p command is a known command. If yes,
654 * handle the command with all arguments.
655 * \return true if the command was parsed, false otherwise.
657 bool parse_command(string const & command, Parser & p, ostream & os,
658 bool outer, Context & context)
660 if (known_commands.find(command) != known_commands.end()) {
661 parse_arguments(command, known_commands[command], p, os,
669 /// Parses a minipage or parbox
670 void parse_box(Parser & p, ostream & os, unsigned outer_flags,
671 unsigned inner_flags, bool outer, Context & parent_context,
672 string const & outer_type, string const & special,
673 string const & inner_type)
677 string hor_pos = "c";
678 // We need to set the height to the LaTeX default of 1\\totalheight
679 // for the case when no height argument is given
680 string height_value = "1";
681 string height_unit = "in";
682 string height_special = "totalheight";
684 if (!inner_type.empty() && p.hasOpt()) {
685 position = p.getArg('[', ']');
686 if (position != "t" && position != "c" && position != "b") {
687 cerr << "invalid position " << position << " for "
688 << inner_type << endl;
692 latex_height = p.getArg('[', ']');
693 translate_box_len(latex_height, height_value, height_unit, height_special);
696 inner_pos = p.getArg('[', ']');
697 if (inner_pos != "c" && inner_pos != "t" &&
698 inner_pos != "b" && inner_pos != "s") {
699 cerr << "invalid inner_pos "
700 << inner_pos << " for "
701 << inner_type << endl;
702 inner_pos = position;
710 if (inner_type.empty()) {
712 latex_width = "\\columnwidth";
715 latex_width = p2.getOptContent();
716 string const opt = p2.getOptContent();
719 if (hor_pos != "l" && hor_pos != "c" &&
721 cerr << "invalid hor_pos " << hor_pos
722 << " for " << outer_type << endl;
728 latex_width = p.verbatim_item();
729 translate_len(latex_width, width_value, width_unit);
730 // LyX can't handle length variables
731 bool use_ert = contains(width_unit, '\\') || contains(height_unit, '\\');
732 if (!use_ert && !outer_type.empty() && !inner_type.empty()) {
733 // Look whether there is some content after the end of the
734 // inner box, but before the end of the outer box.
735 // If yes, we need to output ERT.
737 if (inner_flags & FLAG_END)
738 p.verbatimEnvironment(inner_type);
742 if ((outer_type == "framed" && p.next_token().asInput() != "\\end") ||
743 (outer_type != "framed" && p.next_token().cat() != catEnd)) {
744 // something is between the end of the inner box and
745 // the end of the outer box, so we need to use ERT.
752 if (!outer_type.empty()) {
753 if (outer_flags & FLAG_END)
754 ss << "\\begin{" << outer_type << '}';
756 ss << '\\' << outer_type << '{';
757 if (!special.empty())
761 if (!inner_type.empty()) {
762 if (inner_flags & FLAG_END)
763 ss << "\\begin{" << inner_type << '}';
765 ss << '\\' << inner_type;
766 if (!position.empty())
767 ss << '[' << position << ']';
768 if (!latex_height.empty())
769 ss << '[' << latex_height << ']';
770 if (!inner_pos.empty())
771 ss << '[' << inner_pos << ']';
772 ss << '{' << latex_width << '}';
773 if (!(inner_flags & FLAG_END))
776 handle_ert(os, ss.str(), parent_context);
777 if (!inner_type.empty()) {
778 parse_text(p, os, inner_flags, outer, parent_context);
779 if (inner_flags & FLAG_END)
780 handle_ert(os, "\\end{" + inner_type + '}',
783 handle_ert(os, "}", parent_context);
785 if (!outer_type.empty()) {
786 parse_text(p, os, outer_flags, outer, parent_context);
787 if (outer_flags & FLAG_END)
788 handle_ert(os, "\\end{" + outer_type + '}',
791 handle_ert(os, "}", parent_context);
794 // LyX does not like empty positions, so we have
795 // to set them to the LaTeX default values here.
796 if (position.empty())
798 if (inner_pos.empty())
799 inner_pos = position;
800 parent_context.check_layout(os);
801 begin_inset(os, "Box ");
802 if (outer_type == "framed")
804 else if (outer_type == "framebox")
806 else if (outer_type == "shadowbox")
808 else if (outer_type == "shaded")
810 else if (outer_type == "doublebox")
812 else if (outer_type.empty())
815 os << outer_type << '\n';
816 os << "position \"" << position << "\"\n";
817 os << "hor_pos \"" << hor_pos << "\"\n";
818 os << "has_inner_box " << !inner_type.empty() << "\n";
819 os << "inner_pos \"" << inner_pos << "\"\n";
820 os << "use_parbox " << (inner_type == "parbox") << '\n';
821 os << "width \"" << width_value << width_unit << "\"\n";
822 os << "special \"none\"\n";
823 os << "height \"" << height_value << height_unit << "\"\n";
824 os << "height_special \"" << height_special << "\"\n";
825 os << "status open\n\n";
826 Context context(true, parent_context.textclass);
827 context.font = parent_context.font;
829 // If we have no inner box the contens will be read with the outer box
830 if (!inner_type.empty())
831 parse_text(p, os, inner_flags, outer, context);
833 // Ensure that the end of the outer box is parsed correctly:
834 // The opening brace has been eaten by parse_outer_box()
835 if (!outer_type.empty() && (outer_flags & FLAG_ITEM)) {
836 outer_flags &= ~FLAG_ITEM;
837 outer_flags |= FLAG_BRACE_LAST;
840 // Find end of outer box, output contents if inner_type is
841 // empty and output possible comments
842 if (!outer_type.empty()) {
843 // This does not output anything but comments if
844 // inner_type is not empty (see use_ert)
845 parse_text(p, os, outer_flags, outer, context);
848 context.check_end_layout(os);
850 #ifdef PRESERVE_LAYOUT
851 // LyX puts a % after the end of the minipage
852 if (p.next_token().cat() == catNewline && p.next_token().cs().size() > 1) {
854 //handle_comment(os, "%dummy", parent_context);
857 parent_context.new_paragraph(os);
859 else if (p.next_token().cat() == catSpace || p.next_token().cat() == catNewline) {
860 //handle_comment(os, "%dummy", parent_context);
863 // We add a protected space if something real follows
864 if (p.good() && p.next_token().cat() != catComment) {
865 begin_inset(os, "space ~\n");
874 void parse_outer_box(Parser & p, ostream & os, unsigned flags, bool outer,
875 Context & parent_context, string const & outer_type,
876 string const & special)
878 eat_whitespace(p, os, parent_context, false);
879 if (flags & FLAG_ITEM) {
881 if (p.next_token().cat() == catBegin)
884 cerr << "Warning: Ignoring missing '{' after \\"
885 << outer_type << '.' << endl;
886 eat_whitespace(p, os, parent_context, false);
889 unsigned int inner_flags = 0;
890 if (outer_type == "shaded") {
891 // These boxes never have an inner box
893 } else if (p.next_token().asInput() == "\\parbox") {
894 inner = p.get_token().cs();
895 inner_flags = FLAG_ITEM;
896 } else if (p.next_token().asInput() == "\\begin") {
897 // Is this a minipage?
900 inner = p.getArg('{', '}');
902 if (inner == "minipage") {
905 eat_whitespace(p, os, parent_context, false);
906 inner_flags = FLAG_END;
910 if (inner_flags == FLAG_END) {
911 active_environments.push_back(inner);
912 parse_box(p, os, flags, FLAG_END, outer, parent_context,
913 outer_type, special, inner);
914 active_environments.pop_back();
916 parse_box(p, os, flags, inner_flags, outer, parent_context,
917 outer_type, special, inner);
922 /// parse an unknown environment
923 void parse_unknown_environment(Parser & p, string const & name, ostream & os,
924 unsigned flags, bool outer,
925 Context & parent_context)
927 if (name == "tabbing")
928 // We need to remember that we have to handle '\=' specially
929 flags |= FLAG_TABBING;
931 // We need to translate font changes and paragraphs inside the
932 // environment to ERT if we have a non standard font.
933 // Otherwise things like
934 // \large\begin{foo}\huge bar\end{foo}
936 bool const specialfont =
937 (parent_context.font != parent_context.normalfont);
938 bool const new_layout_allowed = parent_context.new_layout_allowed;
940 parent_context.new_layout_allowed = false;
941 handle_ert(os, "\\begin{" + name + "}", parent_context);
942 parse_text_snippet(p, os, flags, outer, parent_context);
943 handle_ert(os, "\\end{" + name + "}", parent_context);
945 parent_context.new_layout_allowed = new_layout_allowed;
949 void parse_environment(Parser & p, ostream & os, bool outer,
950 string & last_env, Context & parent_context)
952 Layout const * newlayout;
953 InsetLayout const * newinsetlayout = 0;
954 string const name = p.getArg('{', '}');
955 const bool is_starred = suffixIs(name, '*');
956 string const unstarred_name = rtrim(name, "*");
957 active_environments.push_back(name);
959 if (is_math_env(name)) {
960 parent_context.check_layout(os);
961 begin_inset(os, "Formula ");
962 os << "\\begin{" << name << "}";
963 parse_math(p, os, FLAG_END, MATH_MODE);
964 os << "\\end{" << name << "}";
968 else if (name == "tabular" || name == "longtable") {
969 eat_whitespace(p, os, parent_context, false);
970 parent_context.check_layout(os);
971 begin_inset(os, "Tabular ");
972 handle_tabular(p, os, name == "longtable", parent_context);
977 else if (parent_context.textclass.floats().typeExist(unstarred_name)) {
978 eat_whitespace(p, os, parent_context, false);
979 parent_context.check_layout(os);
980 begin_inset(os, "Float " + unstarred_name + "\n");
982 os << "placement " << p.getArg('[', ']') << '\n';
983 os << "wide " << convert<string>(is_starred)
984 << "\nsideways false"
985 << "\nstatus open\n\n";
986 parse_text_in_inset(p, os, FLAG_END, outer, parent_context);
988 // We don't need really a new paragraph, but
989 // we must make sure that the next item gets a \begin_layout.
990 parent_context.new_paragraph(os);
994 else if (name == "minipage") {
995 eat_whitespace(p, os, parent_context, false);
996 parse_box(p, os, 0, FLAG_END, outer, parent_context, "", "", name);
1000 else if (name == "comment") {
1001 eat_whitespace(p, os, parent_context, false);
1002 parent_context.check_layout(os);
1003 begin_inset(os, "Note Comment\n");
1004 os << "status open\n";
1005 parse_text_in_inset(p, os, FLAG_END, outer, parent_context);
1008 skip_braces(p); // eat {} that might by set by LyX behind comments
1011 else if (name == "lyxgreyedout") {
1012 eat_whitespace(p, os, parent_context, false);
1013 parent_context.check_layout(os);
1014 begin_inset(os, "Note Greyedout\n");
1015 os << "status open\n";
1016 parse_text_in_inset(p, os, FLAG_END, outer, parent_context);
1021 else if (name == "framed" || name == "shaded") {
1022 eat_whitespace(p, os, parent_context, false);
1023 parse_outer_box(p, os, FLAG_END, outer, parent_context, name, "");
1027 else if (!parent_context.new_layout_allowed)
1028 parse_unknown_environment(p, name, os, FLAG_END, outer,
1031 // Alignment and spacing settings
1032 // FIXME (bug xxxx): These settings can span multiple paragraphs and
1033 // therefore are totally broken!
1034 // Note that \centering, raggedright, and raggedleft cannot be handled, as
1035 // they are commands not environments. They are furthermore switches that
1036 // can be ended by another switches, but also by commands like \footnote or
1037 // \parbox. So the only safe way is to leave them untouched.
1038 else if (name == "center" || name == "centering" ||
1039 name == "flushleft" || name == "flushright" ||
1040 name == "singlespace" || name == "onehalfspace" ||
1041 name == "doublespace" || name == "spacing") {
1042 eat_whitespace(p, os, parent_context, false);
1043 // We must begin a new paragraph if not already done
1044 if (! parent_context.atParagraphStart()) {
1045 parent_context.check_end_layout(os);
1046 parent_context.new_paragraph(os);
1048 if (name == "flushleft")
1049 parent_context.add_extra_stuff("\\align left\n");
1050 else if (name == "flushright")
1051 parent_context.add_extra_stuff("\\align right\n");
1052 else if (name == "center" || name == "centering")
1053 parent_context.add_extra_stuff("\\align center\n");
1054 else if (name == "singlespace")
1055 parent_context.add_extra_stuff("\\paragraph_spacing single\n");
1056 else if (name == "onehalfspace")
1057 parent_context.add_extra_stuff("\\paragraph_spacing onehalf\n");
1058 else if (name == "doublespace")
1059 parent_context.add_extra_stuff("\\paragraph_spacing double\n");
1060 else if (name == "spacing")
1061 parent_context.add_extra_stuff("\\paragraph_spacing other " + p.verbatim_item() + "\n");
1062 parse_text(p, os, FLAG_END, outer, parent_context);
1063 // Just in case the environment is empty
1064 parent_context.extra_stuff.erase();
1065 // We must begin a new paragraph to reset the alignment
1066 parent_context.new_paragraph(os);
1070 // The single '=' is meant here.
1071 else if ((newlayout = findLayout(parent_context.textclass, name, false))) {
1072 eat_whitespace(p, os, parent_context, false);
1073 Context context(true, parent_context.textclass, newlayout,
1074 parent_context.layout, parent_context.font);
1075 if (parent_context.deeper_paragraph) {
1076 // We are beginning a nested environment after a
1077 // deeper paragraph inside the outer list environment.
1078 // Therefore we don't need to output a "begin deeper".
1079 context.need_end_deeper = true;
1081 parent_context.check_end_layout(os);
1082 if (last_env == name) {
1083 // we need to output a separator since LyX would export
1084 // the two environments as one otherwise (bug 5716)
1085 docstring const sep = from_ascii("--Separator--");
1086 TeX2LyXDocClass const & textclass(parent_context.textclass);
1087 if (textclass.hasLayout(sep)) {
1088 Context newcontext(parent_context);
1089 newcontext.layout = &(textclass[sep]);
1090 newcontext.check_layout(os);
1091 newcontext.check_end_layout(os);
1093 parent_context.check_layout(os);
1094 begin_inset(os, "Note Note\n");
1095 os << "status closed\n";
1096 Context newcontext(true, textclass,
1097 &(textclass.defaultLayout()));
1098 newcontext.check_layout(os);
1099 newcontext.check_end_layout(os);
1101 parent_context.check_end_layout(os);
1104 switch (context.layout->latextype) {
1105 case LATEX_LIST_ENVIRONMENT:
1106 context.add_par_extra_stuff("\\labelwidthstring "
1107 + p.verbatim_item() + '\n');
1110 case LATEX_BIB_ENVIRONMENT:
1111 p.verbatim_item(); // swallow next arg
1117 context.check_deeper(os);
1118 parse_text(p, os, FLAG_END, outer, context);
1119 context.check_end_layout(os);
1120 if (parent_context.deeper_paragraph) {
1121 // We must suppress the "end deeper" because we
1122 // suppressed the "begin deeper" above.
1123 context.need_end_deeper = false;
1125 context.check_end_deeper(os);
1126 parent_context.new_paragraph(os);
1130 // The single '=' is meant here.
1131 else if ((newinsetlayout = findInsetLayout(parent_context.textclass, name, false))) {
1132 eat_whitespace(p, os, parent_context, false);
1133 parent_context.check_layout(os);
1134 begin_inset(os, "Flex ");
1135 os << to_utf8(newinsetlayout->name()) << '\n'
1136 << "status collapsed\n";
1137 parse_text_in_inset(p, os, FLAG_END, false, parent_context, newinsetlayout);
1141 else if (name == "appendix") {
1142 // This is no good latex style, but it works and is used in some documents...
1143 eat_whitespace(p, os, parent_context, false);
1144 parent_context.check_end_layout(os);
1145 Context context(true, parent_context.textclass, parent_context.layout,
1146 parent_context.layout, parent_context.font);
1147 context.check_layout(os);
1148 os << "\\start_of_appendix\n";
1149 parse_text(p, os, FLAG_END, outer, context);
1150 context.check_end_layout(os);
1154 else if (known_environments.find(name) != known_environments.end()) {
1155 vector<ArgumentType> arguments = known_environments[name];
1156 // The last "argument" denotes wether we may translate the
1157 // environment contents to LyX
1158 // The default required if no argument is given makes us
1159 // compatible with the reLyXre environment.
1160 ArgumentType contents = arguments.empty() ?
1163 if (!arguments.empty())
1164 arguments.pop_back();
1165 // See comment in parse_unknown_environment()
1166 bool const specialfont =
1167 (parent_context.font != parent_context.normalfont);
1168 bool const new_layout_allowed =
1169 parent_context.new_layout_allowed;
1171 parent_context.new_layout_allowed = false;
1172 parse_arguments("\\begin{" + name + "}", arguments, p, os,
1173 outer, parent_context);
1174 if (contents == verbatim)
1175 handle_ert(os, p.verbatimEnvironment(name),
1178 parse_text_snippet(p, os, FLAG_END, outer,
1180 handle_ert(os, "\\end{" + name + "}", parent_context);
1182 parent_context.new_layout_allowed = new_layout_allowed;
1186 parse_unknown_environment(p, name, os, FLAG_END, outer,
1190 active_environments.pop_back();
1194 /// parses a comment and outputs it to \p os.
1195 void parse_comment(Parser & p, ostream & os, Token const & t, Context & context)
1197 LASSERT(t.cat() == catComment, return);
1198 if (!t.cs().empty()) {
1199 context.check_layout(os);
1200 handle_comment(os, '%' + t.cs(), context);
1201 if (p.next_token().cat() == catNewline) {
1202 // A newline after a comment line starts a new
1204 if (context.new_layout_allowed) {
1205 if(!context.atParagraphStart())
1206 // Only start a new paragraph if not already
1207 // done (we might get called recursively)
1208 context.new_paragraph(os);
1210 handle_ert(os, "\n", context);
1211 eat_whitespace(p, os, context, true);
1214 // "%\n" combination
1221 * Reads spaces and comments until the first non-space, non-comment token.
1222 * New paragraphs (double newlines or \\par) are handled like simple spaces
1223 * if \p eatParagraph is true.
1224 * Spaces are skipped, but comments are written to \p os.
1226 void eat_whitespace(Parser & p, ostream & os, Context & context,
1230 Token const & t = p.get_token();
1231 if (t.cat() == catComment)
1232 parse_comment(p, os, t, context);
1233 else if ((! eatParagraph && p.isParagraph()) ||
1234 (t.cat() != catSpace && t.cat() != catNewline)) {
1243 * Set a font attribute, parse text and reset the font attribute.
1244 * \param attribute Attribute name (e.g. \\family, \\shape etc.)
1245 * \param currentvalue Current value of the attribute. Is set to the new
1246 * value during parsing.
1247 * \param newvalue New value of the attribute
1249 void parse_text_attributes(Parser & p, ostream & os, unsigned flags, bool outer,
1250 Context & context, string const & attribute,
1251 string & currentvalue, string const & newvalue)
1253 context.check_layout(os);
1254 string const oldvalue = currentvalue;
1255 currentvalue = newvalue;
1256 os << '\n' << attribute << ' ' << newvalue << "\n";
1257 parse_text_snippet(p, os, flags, outer, context);
1258 context.check_layout(os);
1259 os << '\n' << attribute << ' ' << oldvalue << "\n";
1260 currentvalue = oldvalue;
1264 /// get the arguments of a natbib or jurabib citation command
1265 void get_cite_arguments(Parser & p, bool natbibOrder,
1266 string & before, string & after)
1268 // We need to distinguish "" and "[]", so we can't use p.getOpt().
1270 // text before the citation
1272 // text after the citation
1273 after = p.getFullOpt();
1275 if (!after.empty()) {
1276 before = p.getFullOpt();
1277 if (natbibOrder && !before.empty())
1278 swap(before, after);
1283 /// Convert filenames with TeX macros and/or quotes to something LyX
1285 string const normalize_filename(string const & name)
1287 Parser p(trim(name, "\""));
1290 Token const & t = p.get_token();
1291 if (t.cat() != catEscape)
1293 else if (t.cs() == "lyxdot") {
1294 // This is used by LyX for simple dots in relative
1298 } else if (t.cs() == "space") {
1308 /// Convert \p name from TeX convention (relative to master file) to LyX
1309 /// convention (relative to .lyx file) if it is relative
1310 void fix_relative_filename(string & name)
1312 if (FileName::isAbsolute(name))
1315 name = to_utf8(makeRelPath(from_utf8(makeAbsPath(name, getMasterFilePath()).absFileName()),
1316 from_utf8(getParentFilePath())));
1320 /// Parse a NoWeb Scrap section. The initial "<<" is already parsed.
1321 void parse_noweb(Parser & p, ostream & os, Context & context)
1323 // assemble the rest of the keyword
1327 Token const & t = p.get_token();
1328 if (t.asInput() == ">" && p.next_token().asInput() == ">") {
1331 scrap = (p.good() && p.next_token().asInput() == "=");
1333 name += p.get_token().asInput();
1336 name += t.asInput();
1339 if (!scrap || !context.new_layout_allowed ||
1340 !context.textclass.hasLayout(from_ascii("Scrap"))) {
1341 cerr << "Warning: Could not interpret '" << name
1342 << "'. Ignoring it." << endl;
1346 // We use new_paragraph instead of check_end_layout because the stuff
1347 // following the noweb chunk needs to start with a \begin_layout.
1348 // This may create a new paragraph even if there was none in the
1349 // noweb file, but the alternative is an invalid LyX file. Since
1350 // noweb code chunks are implemented with a layout style in LyX they
1351 // always must be in an own paragraph.
1352 context.new_paragraph(os);
1353 Context newcontext(true, context.textclass,
1354 &context.textclass[from_ascii("Scrap")]);
1355 newcontext.check_layout(os);
1358 Token const & t = p.get_token();
1359 // We abuse the parser a bit, because this is no TeX syntax
1361 if (t.cat() == catEscape)
1362 os << subst(t.asInput(), "\\", "\n\\backslash\n");
1365 begin_inset(oss, "Newline newline");
1367 os << subst(t.asInput(), "\n", oss.str());
1369 // The scrap chunk is ended by an @ at the beginning of a line.
1370 // After the @ the line may contain a comment and/or
1371 // whitespace, but nothing else.
1372 if (t.asInput() == "@" && p.prev_token().cat() == catNewline &&
1373 (p.next_token().cat() == catSpace ||
1374 p.next_token().cat() == catNewline ||
1375 p.next_token().cat() == catComment)) {
1376 while (p.good() && p.next_token().cat() == catSpace)
1377 os << p.get_token().asInput();
1378 if (p.next_token().cat() == catComment)
1379 // The comment includes a final '\n'
1380 os << p.get_token().asInput();
1382 if (p.next_token().cat() == catNewline)
1389 newcontext.check_end_layout(os);
1393 /// detects \\def, \\long\\def and \\global\\long\\def with ws and comments
1394 bool is_macro(Parser & p)
1396 Token first = p.curr_token();
1397 if (first.cat() != catEscape || !p.good())
1399 if (first.cs() == "def")
1401 if (first.cs() != "global" && first.cs() != "long")
1403 Token second = p.get_token();
1405 while (p.good() && !p.isParagraph() && (second.cat() == catSpace ||
1406 second.cat() == catNewline || second.cat() == catComment)) {
1407 second = p.get_token();
1410 bool secondvalid = second.cat() == catEscape;
1412 bool thirdvalid = false;
1413 if (p.good() && first.cs() == "global" && secondvalid &&
1414 second.cs() == "long") {
1415 third = p.get_token();
1417 while (p.good() && !p.isParagraph() &&
1418 (third.cat() == catSpace ||
1419 third.cat() == catNewline ||
1420 third.cat() == catComment)) {
1421 third = p.get_token();
1424 thirdvalid = third.cat() == catEscape;
1426 for (int i = 0; i < pos; ++i)
1431 return (first.cs() == "global" || first.cs() == "long") &&
1432 second.cs() == "def";
1433 return first.cs() == "global" && second.cs() == "long" &&
1434 third.cs() == "def";
1438 /// Parse a macro definition (assumes that is_macro() returned true)
1439 void parse_macro(Parser & p, ostream & os, Context & context)
1441 context.check_layout(os);
1442 Token first = p.curr_token();
1445 string command = first.asInput();
1446 if (first.cs() != "def") {
1448 eat_whitespace(p, os, context, false);
1449 second = p.curr_token();
1450 command += second.asInput();
1451 if (second.cs() != "def") {
1453 eat_whitespace(p, os, context, false);
1454 third = p.curr_token();
1455 command += third.asInput();
1458 eat_whitespace(p, os, context, false);
1459 string const name = p.get_token().cs();
1460 eat_whitespace(p, os, context, false);
1466 while (p.next_token().cat() != catBegin) {
1467 if (p.next_token().cat() == catParameter) {
1472 // followed by number?
1473 if (p.next_token().cat() == catOther) {
1474 char c = p.getChar();
1476 // number = current arity + 1?
1477 if (c == arity + '0' + 1)
1482 paramtext += p.get_token().cs();
1484 paramtext += p.get_token().cs();
1489 // only output simple (i.e. compatible) macro as FormulaMacros
1490 string ert = '\\' + name + ' ' + paramtext + '{' + p.verbatim_item() + '}';
1492 context.check_layout(os);
1493 begin_inset(os, "FormulaMacro");
1494 os << "\n\\def" << ert;
1497 handle_ert(os, command + ert, context);
1500 } // anonymous namespace
1503 void parse_text(Parser & p, ostream & os, unsigned flags, bool outer,
1506 Layout const * newlayout = 0;
1507 InsetLayout const * newinsetlayout = 0;
1508 // Store the latest bibliographystyle and nocite{*} option
1509 // (needed for bibtex inset)
1511 string bibliographystyle;
1512 bool const use_natbib = used_packages.find("natbib") != used_packages.end();
1513 bool const use_jurabib = used_packages.find("jurabib") != used_packages.end();
1516 Token const & t = p.get_token();
1519 debugToken(cerr, t, flags);
1522 if (flags & FLAG_ITEM) {
1523 if (t.cat() == catSpace)
1526 flags &= ~FLAG_ITEM;
1527 if (t.cat() == catBegin) {
1528 // skip the brace and collect everything to the next matching
1530 flags |= FLAG_BRACE_LAST;
1534 // handle only this single token, leave the loop if done
1535 flags |= FLAG_LEAVE;
1538 if (t.cat() != catEscape && t.character() == ']' &&
1539 (flags & FLAG_BRACK_LAST))
1541 if (t.cat() == catEnd && (flags & FLAG_BRACE_LAST))
1544 // If there is anything between \end{env} and \begin{env} we
1545 // don't need to output a separator.
1546 if (t.cat() != catSpace && t.cat() != catNewline &&
1547 t.asInput() != "\\begin")
1553 if (t.cat() == catMath) {
1554 // we are inside some text mode thingy, so opening new math is allowed
1555 context.check_layout(os);
1556 begin_inset(os, "Formula ");
1557 Token const & n = p.get_token();
1558 if (n.cat() == catMath && outer) {
1559 // TeX's $$...$$ syntax for displayed math
1561 parse_math(p, os, FLAG_SIMPLE, MATH_MODE);
1563 p.get_token(); // skip the second '$' token
1565 // simple $...$ stuff
1568 parse_math(p, os, FLAG_SIMPLE, MATH_MODE);
1574 else if (t.cat() == catSuper || t.cat() == catSub)
1575 cerr << "catcode " << t << " illegal in text mode\n";
1577 // Basic support for english quotes. This should be
1578 // extended to other quotes, but is not so easy (a
1579 // left english quote is the same as a right german
1581 else if (t.asInput() == "`" && p.next_token().asInput() == "`") {
1582 context.check_layout(os);
1583 begin_inset(os, "Quotes ");
1589 else if (t.asInput() == "'" && p.next_token().asInput() == "'") {
1590 context.check_layout(os);
1591 begin_inset(os, "Quotes ");
1598 else if (t.asInput() == ">" && p.next_token().asInput() == ">") {
1599 context.check_layout(os);
1600 begin_inset(os, "Quotes ");
1607 else if (t.asInput() == "<" && p.next_token().asInput() == "<") {
1608 context.check_layout(os);
1609 begin_inset(os, "Quotes ");
1616 else if (t.asInput() == "<"
1617 && p.next_token().asInput() == "<" && noweb_mode) {
1619 parse_noweb(p, os, context);
1622 else if (t.cat() == catSpace || (t.cat() == catNewline && ! p.isParagraph()))
1623 check_space(p, os, context);
1625 else if (t.character() == '[' && noweb_mode &&
1626 p.next_token().character() == '[') {
1627 // These can contain underscores
1629 string const s = p.getFullOpt() + ']';
1630 if (p.next_token().character() == ']')
1633 cerr << "Warning: Inserting missing ']' in '"
1634 << s << "'." << endl;
1635 handle_ert(os, s, context);
1638 else if (t.cat() == catLetter) {
1639 context.check_layout(os);
1640 // Workaround for bug 4752.
1641 // FIXME: This whole code block needs to be removed
1642 // when the bug is fixed and tex2lyx produces
1643 // the updated file format.
1644 // The replacement algorithm in LyX is so stupid that
1645 // it even translates a phrase if it is part of a word.
1646 bool handled = false;
1647 for (int const * l = known_phrase_lengths; *l; ++l) {
1648 string phrase = t.cs();
1649 for (int i = 1; i < *l && p.next_token().isAlnumASCII(); ++i)
1650 phrase += p.get_token().cs();
1651 if (is_known(phrase, known_coded_phrases)) {
1652 handle_ert(os, phrase, context);
1656 for (size_t i = 1; i < phrase.length(); ++i)
1664 else if (t.cat() == catOther ||
1665 t.cat() == catAlign ||
1666 t.cat() == catParameter) {
1667 // This translates "&" to "\\&" which may be wrong...
1668 context.check_layout(os);
1672 else if (p.isParagraph()) {
1673 if (context.new_layout_allowed)
1674 context.new_paragraph(os);
1676 handle_ert(os, "\\par ", context);
1677 eat_whitespace(p, os, context, true);
1680 else if (t.cat() == catActive) {
1681 context.check_layout(os);
1682 if (t.character() == '~') {
1683 if (context.layout->free_spacing)
1686 begin_inset(os, "space ~\n");
1693 else if (t.cat() == catBegin &&
1694 p.next_token().cat() == catEnd) {
1696 Token const prev = p.prev_token();
1698 if (p.next_token().character() == '`' ||
1699 (prev.character() == '-' &&
1700 p.next_token().character() == '-'))
1701 ; // ignore it in {}`` or -{}-
1703 handle_ert(os, "{}", context);
1707 else if (t.cat() == catBegin) {
1708 context.check_layout(os);
1709 // special handling of font attribute changes
1710 Token const prev = p.prev_token();
1711 Token const next = p.next_token();
1712 TeXFont const oldFont = context.font;
1713 if (next.character() == '[' ||
1714 next.character() == ']' ||
1715 next.character() == '*') {
1717 if (p.next_token().cat() == catEnd) {
1722 handle_ert(os, "{", context);
1723 parse_text_snippet(p, os,
1726 handle_ert(os, "}", context);
1728 } else if (! context.new_layout_allowed) {
1729 handle_ert(os, "{", context);
1730 parse_text_snippet(p, os, FLAG_BRACE_LAST,
1732 handle_ert(os, "}", context);
1733 } else if (is_known(next.cs(), known_sizes)) {
1734 // next will change the size, so we must
1736 parse_text_snippet(p, os, FLAG_BRACE_LAST,
1738 if (!context.atParagraphStart())
1740 << context.font.size << "\n";
1741 } else if (is_known(next.cs(), known_font_families)) {
1742 // next will change the font family, so we
1743 // must reset it here
1744 parse_text_snippet(p, os, FLAG_BRACE_LAST,
1746 if (!context.atParagraphStart())
1748 << context.font.family << "\n";
1749 } else if (is_known(next.cs(), known_font_series)) {
1750 // next will change the font series, so we
1751 // must reset it here
1752 parse_text_snippet(p, os, FLAG_BRACE_LAST,
1754 if (!context.atParagraphStart())
1756 << context.font.series << "\n";
1757 } else if (is_known(next.cs(), known_font_shapes)) {
1758 // next will change the font shape, so we
1759 // must reset it here
1760 parse_text_snippet(p, os, FLAG_BRACE_LAST,
1762 if (!context.atParagraphStart())
1764 << context.font.shape << "\n";
1765 } else if (is_known(next.cs(), known_old_font_families) ||
1766 is_known(next.cs(), known_old_font_series) ||
1767 is_known(next.cs(), known_old_font_shapes)) {
1768 // next will change the font family, series
1769 // and shape, so we must reset it here
1770 parse_text_snippet(p, os, FLAG_BRACE_LAST,
1772 if (!context.atParagraphStart())
1774 << context.font.family
1776 << context.font.series
1778 << context.font.shape << "\n";
1780 handle_ert(os, "{", context);
1781 parse_text_snippet(p, os, FLAG_BRACE_LAST,
1783 handle_ert(os, "}", context);
1787 else if (t.cat() == catEnd) {
1788 if (flags & FLAG_BRACE_LAST) {
1791 cerr << "stray '}' in text\n";
1792 handle_ert(os, "}", context);
1795 else if (t.cat() == catComment)
1796 parse_comment(p, os, t, context);
1799 // control sequences
1802 else if (t.cs() == "(") {
1803 context.check_layout(os);
1804 begin_inset(os, "Formula");
1806 parse_math(p, os, FLAG_SIMPLE2, MATH_MODE);
1811 else if (t.cs() == "[") {
1812 context.check_layout(os);
1813 begin_inset(os, "Formula");
1815 parse_math(p, os, FLAG_EQUATION, MATH_MODE);
1820 else if (t.cs() == "begin")
1821 parse_environment(p, os, outer, last_env, context);
1823 else if (t.cs() == "end") {
1824 if (flags & FLAG_END) {
1825 // eat environment name
1826 string const name = p.getArg('{', '}');
1827 if (name != active_environment())
1828 cerr << "\\end{" + name + "} does not match \\begin{"
1829 + active_environment() + "}\n";
1832 p.error("found 'end' unexpectedly");
1835 else if (t.cs() == "item") {
1838 bool optarg = false;
1839 if (p.next_token().cat() != catEscape &&
1840 p.next_token().character() == '[') {
1841 p.get_token(); // eat '['
1842 s = parse_text_snippet(p, FLAG_BRACK_LAST,
1847 context.check_layout(os);
1848 if (context.has_item) {
1849 // An item in an unknown list-like environment
1850 // FIXME: Do this in check_layout()!
1851 context.has_item = false;
1853 handle_ert(os, "\\item", context);
1855 handle_ert(os, "\\item ", context);
1858 if (context.layout->labeltype != LABEL_MANUAL) {
1859 // LyX does not support \item[\mybullet]
1860 // in itemize environments
1861 handle_ert(os, "[", context);
1863 handle_ert(os, "]", context);
1864 } else if (!s.empty()) {
1865 // The space is needed to separate the
1866 // item from the rest of the sentence.
1868 eat_whitespace(p, os, context, false);
1873 else if (t.cs() == "bibitem") {
1875 context.check_layout(os);
1876 begin_command_inset(os, "bibitem", "bibitem");
1877 os << "label \"" << p.getOptContent() << "\"\n";
1878 os << "key \"" << p.verbatim_item() << "\"\n";
1882 else if (is_macro(p))
1883 parse_macro(p, os, context);
1885 else if (t.cs() == "noindent") {
1887 context.add_par_extra_stuff("\\noindent\n");
1890 else if (t.cs() == "appendix") {
1891 context.add_par_extra_stuff("\\start_of_appendix\n");
1892 // We need to start a new paragraph. Otherwise the
1893 // appendix in 'bla\appendix\chapter{' would start
1895 context.new_paragraph(os);
1896 // We need to make sure that the paragraph is
1897 // generated even if it is empty. Otherwise the
1898 // appendix in '\par\appendix\par\chapter{' would
1900 context.check_layout(os);
1901 // FIXME: This is a hack to prevent paragraph
1902 // deletion if it is empty. Handle this better!
1904 "%dummy comment inserted by tex2lyx to "
1905 "ensure that this paragraph is not empty",
1907 // Both measures above may generate an additional
1908 // empty paragraph, but that does not hurt, because
1909 // whitespace does not matter here.
1910 eat_whitespace(p, os, context, true);
1913 // Starred section headings
1914 // Must attempt to parse "Section*" before "Section".
1915 else if ((p.next_token().asInput() == "*") &&
1916 context.new_layout_allowed &&
1917 (newlayout = findLayout(context.textclass, t.cs() + '*', true))) {
1920 output_command_layout(os, p, outer, context, newlayout);
1924 // Section headings and the like
1925 else if (context.new_layout_allowed &&
1926 (newlayout = findLayout(context.textclass, t.cs(), true))) {
1928 output_command_layout(os, p, outer, context, newlayout);
1932 else if (t.cs() == "caption") {
1934 context.check_layout(os);
1936 begin_inset(os, "Caption\n\n");
1937 Context newcontext(true, context.textclass);
1938 newcontext.font = context.font;
1939 newcontext.check_layout(os);
1940 if (p.next_token().cat() != catEscape &&
1941 p.next_token().character() == '[') {
1942 p.get_token(); // eat '['
1943 begin_inset(os, "OptArg\n");
1944 os << "status collapsed\n";
1945 parse_text_in_inset(p, os, FLAG_BRACK_LAST, outer, context);
1947 eat_whitespace(p, os, context, false);
1949 parse_text(p, os, FLAG_ITEM, outer, context);
1950 context.check_end_layout(os);
1951 // We don't need really a new paragraph, but
1952 // we must make sure that the next item gets a \begin_layout.
1953 context.new_paragraph(os);
1956 newcontext.check_end_layout(os);
1959 else if (t.cs() == "includegraphics") {
1960 bool const clip = p.next_token().asInput() == "*";
1963 string const arg = p.getArg('[', ']');
1964 map<string, string> opts;
1965 vector<string> keys;
1966 split_map(arg, opts, keys);
1968 opts["clip"] = string();
1969 string name = normalize_filename(p.verbatim_item());
1971 string const path = getMasterFilePath();
1972 // We want to preserve relative / absolute filenames,
1973 // therefore path is only used for testing
1974 if (!makeAbsPath(name, path).exists()) {
1975 // The file extension is probably missing.
1976 // Now try to find it out.
1977 string const dvips_name =
1978 find_file(name, path,
1979 known_dvips_graphics_formats);
1980 string const pdftex_name =
1981 find_file(name, path,
1982 known_pdftex_graphics_formats);
1983 if (!dvips_name.empty()) {
1984 if (!pdftex_name.empty()) {
1985 cerr << "This file contains the "
1987 "\"\\includegraphics{"
1989 "However, files\n\""
1990 << dvips_name << "\" and\n\""
1991 << pdftex_name << "\"\n"
1992 "both exist, so I had to make a "
1993 "choice and took the first one.\n"
1994 "Please move the unwanted one "
1995 "someplace else and try again\n"
1996 "if my choice was wrong."
2000 } else if (!pdftex_name.empty()) {
2006 if (makeAbsPath(name, path).exists())
2007 fix_relative_filename(name);
2009 cerr << "Warning: Could not find graphics file '"
2010 << name << "'." << endl;
2012 context.check_layout(os);
2013 begin_inset(os, "Graphics ");
2014 os << "\n\tfilename " << name << '\n';
2015 if (opts.find("width") != opts.end())
2017 << translate_len(opts["width"]) << '\n';
2018 if (opts.find("height") != opts.end())
2020 << translate_len(opts["height"]) << '\n';
2021 if (opts.find("scale") != opts.end()) {
2022 istringstream iss(opts["scale"]);
2026 os << "\tscale " << val << '\n';
2028 if (opts.find("angle") != opts.end()) {
2029 os << "\trotateAngle "
2030 << opts["angle"] << '\n';
2031 vector<string>::const_iterator a =
2032 find(keys.begin(), keys.end(), "angle");
2033 vector<string>::const_iterator s =
2034 find(keys.begin(), keys.end(), "width");
2035 if (s == keys.end())
2036 s = find(keys.begin(), keys.end(), "height");
2037 if (s == keys.end())
2038 s = find(keys.begin(), keys.end(), "scale");
2039 if (s != keys.end() && distance(s, a) > 0)
2040 os << "\tscaleBeforeRotation\n";
2042 if (opts.find("origin") != opts.end()) {
2044 string const opt = opts["origin"];
2045 if (opt.find('l') != string::npos) ss << "left";
2046 if (opt.find('r') != string::npos) ss << "right";
2047 if (opt.find('c') != string::npos) ss << "center";
2048 if (opt.find('t') != string::npos) ss << "Top";
2049 if (opt.find('b') != string::npos) ss << "Bottom";
2050 if (opt.find('B') != string::npos) ss << "Baseline";
2051 if (!ss.str().empty())
2052 os << "\trotateOrigin " << ss.str() << '\n';
2054 cerr << "Warning: Ignoring unknown includegraphics origin argument '" << opt << "'\n";
2056 if (opts.find("keepaspectratio") != opts.end())
2057 os << "\tkeepAspectRatio\n";
2058 if (opts.find("clip") != opts.end())
2060 if (opts.find("draft") != opts.end())
2062 if (opts.find("bb") != opts.end())
2063 os << "\tBoundingBox "
2064 << opts["bb"] << '\n';
2065 int numberOfbbOptions = 0;
2066 if (opts.find("bbllx") != opts.end())
2067 numberOfbbOptions++;
2068 if (opts.find("bblly") != opts.end())
2069 numberOfbbOptions++;
2070 if (opts.find("bburx") != opts.end())
2071 numberOfbbOptions++;
2072 if (opts.find("bbury") != opts.end())
2073 numberOfbbOptions++;
2074 if (numberOfbbOptions == 4)
2075 os << "\tBoundingBox "
2076 << opts["bbllx"] << " " << opts["bblly"] << " "
2077 << opts["bburx"] << " " << opts["bbury"] << '\n';
2078 else if (numberOfbbOptions > 0)
2079 cerr << "Warning: Ignoring incomplete includegraphics boundingbox arguments.\n";
2080 numberOfbbOptions = 0;
2081 if (opts.find("natwidth") != opts.end())
2082 numberOfbbOptions++;
2083 if (opts.find("natheight") != opts.end())
2084 numberOfbbOptions++;
2085 if (numberOfbbOptions == 2)
2086 os << "\tBoundingBox 0bp 0bp "
2087 << opts["natwidth"] << " " << opts["natheight"] << '\n';
2088 else if (numberOfbbOptions > 0)
2089 cerr << "Warning: Ignoring incomplete includegraphics boundingbox arguments.\n";
2090 ostringstream special;
2091 if (opts.find("hiresbb") != opts.end())
2092 special << "hiresbb,";
2093 if (opts.find("trim") != opts.end())
2095 if (opts.find("viewport") != opts.end())
2096 special << "viewport=" << opts["viewport"] << ',';
2097 if (opts.find("totalheight") != opts.end())
2098 special << "totalheight=" << opts["totalheight"] << ',';
2099 if (opts.find("type") != opts.end())
2100 special << "type=" << opts["type"] << ',';
2101 if (opts.find("ext") != opts.end())
2102 special << "ext=" << opts["ext"] << ',';
2103 if (opts.find("read") != opts.end())
2104 special << "read=" << opts["read"] << ',';
2105 if (opts.find("command") != opts.end())
2106 special << "command=" << opts["command"] << ',';
2107 string s_special = special.str();
2108 if (!s_special.empty()) {
2109 // We had special arguments. Remove the trailing ','.
2110 os << "\tspecial " << s_special.substr(0, s_special.size() - 1) << '\n';
2112 // TODO: Handle the unknown settings better.
2113 // Warn about invalid options.
2114 // Check whether some option was given twice.
2118 else if (t.cs() == "footnote" ||
2119 (t.cs() == "thanks" && context.layout->intitle)) {
2121 context.check_layout(os);
2122 begin_inset(os, "Foot\n");
2123 os << "status collapsed\n\n";
2124 parse_text_in_inset(p, os, FLAG_ITEM, false, context);
2128 else if (t.cs() == "marginpar") {
2130 context.check_layout(os);
2131 begin_inset(os, "Marginal\n");
2132 os << "status collapsed\n\n";
2133 parse_text_in_inset(p, os, FLAG_ITEM, false, context);
2137 else if (t.cs() == "ensuremath") {
2139 context.check_layout(os);
2140 string const s = p.verbatim_item();
2141 //FIXME: this never triggers in UTF8
2142 if (s == "\xb1" || s == "\xb3" || s == "\xb2" || s == "\xb5")
2145 handle_ert(os, "\\ensuremath{" + s + "}",
2149 else if (t.cs() == "makeindex" || t.cs() == "maketitle") {
2150 // FIXME: Somehow prevent title layouts if
2151 // "maketitle" was not found
2153 skip_spaces_braces(p);
2156 else if (t.cs() == "tableofcontents") {
2157 context.check_layout(os);
2158 begin_command_inset(os, "toc", "tableofcontents");
2160 skip_spaces_braces(p);
2163 else if (t.cs() == "listoffigures") {
2164 context.check_layout(os);
2165 begin_inset(os, "FloatList figure\n");
2167 skip_spaces_braces(p);
2170 else if (t.cs() == "listoftables") {
2171 context.check_layout(os);
2172 begin_inset(os, "FloatList table\n");
2174 skip_spaces_braces(p);
2177 else if (t.cs() == "listof") {
2178 p.skip_spaces(true);
2179 string const name = p.get_token().cs();
2180 if (context.textclass.floats().typeExist(name)) {
2181 context.check_layout(os);
2182 begin_inset(os, "FloatList ");
2185 p.get_token(); // swallow second arg
2187 handle_ert(os, "\\listof{" + name + "}", context);
2190 else if (t.cs() == "textrm")
2191 parse_text_attributes(p, os, FLAG_ITEM, outer,
2192 context, "\\family",
2193 context.font.family, "roman");
2195 else if (t.cs() == "textsf")
2196 parse_text_attributes(p, os, FLAG_ITEM, outer,
2197 context, "\\family",
2198 context.font.family, "sans");
2200 else if (t.cs() == "texttt")
2201 parse_text_attributes(p, os, FLAG_ITEM, outer,
2202 context, "\\family",
2203 context.font.family, "typewriter");
2205 else if (t.cs() == "textmd")
2206 parse_text_attributes(p, os, FLAG_ITEM, outer,
2207 context, "\\series",
2208 context.font.series, "medium");
2210 else if (t.cs() == "textbf")
2211 parse_text_attributes(p, os, FLAG_ITEM, outer,
2212 context, "\\series",
2213 context.font.series, "bold");
2215 else if (t.cs() == "textup")
2216 parse_text_attributes(p, os, FLAG_ITEM, outer,
2218 context.font.shape, "up");
2220 else if (t.cs() == "textit")
2221 parse_text_attributes(p, os, FLAG_ITEM, outer,
2223 context.font.shape, "italic");
2225 else if (t.cs() == "textsl")
2226 parse_text_attributes(p, os, FLAG_ITEM, outer,
2228 context.font.shape, "slanted");
2230 else if (t.cs() == "textsc")
2231 parse_text_attributes(p, os, FLAG_ITEM, outer,
2233 context.font.shape, "smallcaps");
2235 else if (t.cs() == "textnormal" || t.cs() == "normalfont") {
2236 context.check_layout(os);
2237 TeXFont oldFont = context.font;
2238 context.font.init();
2239 context.font.size = oldFont.size;
2240 os << "\n\\family " << context.font.family << "\n";
2241 os << "\n\\series " << context.font.series << "\n";
2242 os << "\n\\shape " << context.font.shape << "\n";
2243 if (t.cs() == "textnormal") {
2244 parse_text_snippet(p, os, FLAG_ITEM, outer, context);
2245 output_font_change(os, context.font, oldFont);
2246 context.font = oldFont;
2248 eat_whitespace(p, os, context, false);
2251 else if (t.cs() == "textcolor") {
2252 // scheme is \textcolor{color name}{text}
2253 string const color = p.verbatim_item();
2254 // we only support the predefined colors of the color package
2255 if (color == "black" || color == "blue" || color == "cyan"
2256 || color == "green" || color == "magenta" || color == "red"
2257 || color == "white" || color == "yellow") {
2258 context.check_layout(os);
2259 os << "\n\\color " << color << "\n";
2260 parse_text_snippet(p, os, FLAG_ITEM, outer, context);
2261 context.check_layout(os);
2262 os << "\n\\color inherit\n";
2264 // for custom defined colors
2265 handle_ert(os, t.asInput() + "{" + color + "}", context);
2268 else if (t.cs() == "underbar") {
2269 // Do NOT handle \underline.
2270 // \underbar cuts through y, g, q, p etc.,
2271 // \underline does not.
2272 context.check_layout(os);
2273 os << "\n\\bar under\n";
2274 parse_text_snippet(p, os, FLAG_ITEM, outer, context);
2275 context.check_layout(os);
2276 os << "\n\\bar default\n";
2279 else if (t.cs() == "emph" || t.cs() == "noun") {
2280 context.check_layout(os);
2281 os << "\n\\" << t.cs() << " on\n";
2282 parse_text_snippet(p, os, FLAG_ITEM, outer, context);
2283 context.check_layout(os);
2284 os << "\n\\" << t.cs() << " default\n";
2287 else if (t.cs() == "lyxline") {
2288 context.check_layout(os);
2292 else if (is_known(t.cs(), known_phrases) ||
2293 (t.cs() == "protect" &&
2294 p.next_token().cat() == catEscape &&
2295 is_known(p.next_token().cs(), known_phrases))) {
2296 // LyX sometimes puts a \protect in front, so we have to ignore it
2297 // FIXME: This needs to be changed when bug 4752 is fixed.
2298 char const * const * where = is_known(
2299 t.cs() == "protect" ? p.get_token().cs() : t.cs(),
2301 context.check_layout(os);
2302 os << known_coded_phrases[where - known_phrases];
2303 skip_spaces_braces(p);
2306 else if (is_known(t.cs(), known_ref_commands)) {
2307 context.check_layout(os);
2308 begin_command_inset(os, "ref", t.cs());
2309 // LyX cannot handle newlines in a latex command
2310 // FIXME: Move the substitution into parser::getOpt()?
2311 os << subst(p.getOpt(), "\n", " ");
2312 os << "reference " << '"' << subst(p.verbatim_item(), "\n", " ") << '"' << "\n";
2316 else if (use_natbib &&
2317 is_known(t.cs(), known_natbib_commands) &&
2318 ((t.cs() != "citefullauthor" &&
2319 t.cs() != "citeyear" &&
2320 t.cs() != "citeyearpar") ||
2321 p.next_token().asInput() != "*")) {
2322 context.check_layout(os);
2323 string command = t.cs();
2324 if (p.next_token().asInput() == "*") {
2328 if (command == "citefullauthor")
2329 // alternative name for "\\citeauthor*"
2330 command = "citeauthor*";
2332 // text before the citation
2334 // text after the citation
2336 get_cite_arguments(p, true, before, after);
2338 if (command == "cite") {
2339 // \cite without optional argument means
2340 // \citet, \cite with at least one optional
2341 // argument means \citep.
2342 if (before.empty() && after.empty())
2347 if (before.empty() && after == "[]")
2348 // avoid \citet[]{a}
2350 else if (before == "[]" && after == "[]") {
2351 // avoid \citet[][]{a}
2355 // remove the brackets around after and before
2356 if (!after.empty()) {
2358 after.erase(after.length() - 1, 1);
2359 // LyX cannot handle newlines in the parameter
2360 after = subst(after, "\n", " ");
2362 if (!before.empty()) {
2364 before.erase(before.length() - 1, 1);
2365 // LyX cannot handle newlines in the parameter
2366 before = subst(before, "\n", " ");
2368 begin_command_inset(os, "citation", command);
2369 os << "after " << '"' << after << '"' << "\n";
2370 os << "before " << '"' << before << '"' << "\n";
2371 os << "key " << '"' << p.verbatim_item() << '"' << "\n";
2375 else if (use_jurabib &&
2376 is_known(t.cs(), known_jurabib_commands) &&
2377 (t.cs() == "cite" || p.next_token().asInput() != "*")) {
2378 context.check_layout(os);
2379 string command = t.cs();
2380 if (p.next_token().asInput() == "*") {
2384 char argumentOrder = '\0';
2385 vector<string> const & options = used_packages["jurabib"];
2386 if (find(options.begin(), options.end(),
2387 "natbiborder") != options.end())
2388 argumentOrder = 'n';
2389 else if (find(options.begin(), options.end(),
2390 "jurabiborder") != options.end())
2391 argumentOrder = 'j';
2393 // text before the citation
2395 // text after the citation
2397 get_cite_arguments(p, argumentOrder != 'j', before, after);
2399 string const citation = p.verbatim_item();
2400 if (!before.empty() && argumentOrder == '\0') {
2401 cerr << "Warning: Assuming argument order "
2402 "of jurabib version 0.6 for\n'"
2403 << command << before << after << '{'
2404 << citation << "}'.\n"
2405 "Add 'jurabiborder' to the jurabib "
2406 "package options if you used an\n"
2407 "earlier jurabib version." << endl;
2409 if (!after.empty()) {
2411 after.erase(after.length() - 1, 1);
2413 if (!before.empty()) {
2415 before.erase(before.length() - 1, 1);
2417 begin_command_inset(os, "citation", command);
2418 os << "after " << '"' << after << '"' << "\n";
2419 os << "before " << '"' << before << '"' << "\n";
2420 os << "key " << '"' << citation << '"' << "\n";
2424 else if (t.cs() == "cite"
2425 || t.cs() == "nocite") {
2426 context.check_layout(os);
2427 // LyX cannot handle newlines in a latex command
2428 string after = subst(p.getOptContent(), "\n", " ");
2429 string key = subst(p.verbatim_item(), "\n", " ");
2430 // store the case that it is "\nocite{*}" to use it later for
2433 begin_command_inset(os, "citation", t.cs());
2434 os << "after " << '"' << after << '"' << "\n";
2435 os << "key " << '"' << key << '"' << "\n";
2437 } else if (t.cs() == "nocite")
2441 else if (t.cs() == "index") {
2442 context.check_layout(os);
2443 begin_inset(os, "Index\n");
2444 os << "status collapsed\n";
2445 parse_text_in_inset(p, os, FLAG_ITEM, false, context, "Index");
2449 else if (t.cs() == "nomenclature") {
2450 context.check_layout(os);
2451 begin_command_inset(os, "nomenclature", "nomenclature");
2452 // LyX cannot handle newlines in a latex command
2453 string prefix = subst(p.getOptContent(), "\n", " ");
2454 if (!prefix.empty())
2455 os << "prefix " << '"' << prefix << '"' << "\n";
2456 os << "symbol " << '"' << subst(p.verbatim_item(), "\n", " ") << '"' << "\n";
2457 os << "description " << '"' << subst(p.verbatim_item(), "\n", " ") << '"' << "\n";
2461 else if (t.cs() == "label") {
2462 context.check_layout(os);
2463 begin_command_inset(os, "label", "label");
2464 // LyX cannot handle newlines in a latex command
2465 os << "name " << '"' << subst(p.verbatim_item(), "\n", " ") << '"' << "\n";
2469 else if (t.cs() == "printindex") {
2470 context.check_layout(os);
2471 begin_command_inset(os, "index_print", "printindex");
2473 skip_spaces_braces(p);
2476 else if (t.cs() == "printnomenclature") {
2477 context.check_layout(os);
2478 begin_command_inset(os, "nomencl_print", "printnomenclature");
2480 skip_spaces_braces(p);
2483 else if (LYX_FORMAT >= 408 &&
2484 (t.cs() == "textsuperscript" || t.cs() == "textsubscript")) {
2485 context.check_layout(os);
2486 begin_inset(os, "script ");
2487 os << t.cs().substr(4) << '\n';
2488 parse_text_in_inset(p, os, FLAG_ITEM, false, context);
2492 else if (is_known(t.cs(), known_quotes)) {
2493 char const * const * where = is_known(t.cs(), known_quotes);
2494 context.check_layout(os);
2495 begin_inset(os, "Quotes ");
2496 os << known_coded_quotes[where - known_quotes];
2498 // LyX adds {} after the quote, so we have to eat
2499 // spaces here if there are any before a possible
2501 eat_whitespace(p, os, context, false);
2505 else if (is_known(t.cs(), known_sizes) &&
2506 context.new_layout_allowed) {
2507 char const * const * where = is_known(t.cs(), known_sizes);
2508 context.check_layout(os);
2509 TeXFont const oldFont = context.font;
2510 context.font.size = known_coded_sizes[where - known_sizes];
2511 output_font_change(os, oldFont, context.font);
2512 eat_whitespace(p, os, context, false);
2515 else if (is_known(t.cs(), known_font_families) &&
2516 context.new_layout_allowed) {
2517 char const * const * where =
2518 is_known(t.cs(), known_font_families);
2519 context.check_layout(os);
2520 TeXFont const oldFont = context.font;
2521 context.font.family =
2522 known_coded_font_families[where - known_font_families];
2523 output_font_change(os, oldFont, context.font);
2524 eat_whitespace(p, os, context, false);
2527 else if (is_known(t.cs(), known_font_series) &&
2528 context.new_layout_allowed) {
2529 char const * const * where =
2530 is_known(t.cs(), known_font_series);
2531 context.check_layout(os);
2532 TeXFont const oldFont = context.font;
2533 context.font.series =
2534 known_coded_font_series[where - known_font_series];
2535 output_font_change(os, oldFont, context.font);
2536 eat_whitespace(p, os, context, false);
2539 else if (is_known(t.cs(), known_font_shapes) &&
2540 context.new_layout_allowed) {
2541 char const * const * where =
2542 is_known(t.cs(), known_font_shapes);
2543 context.check_layout(os);
2544 TeXFont const oldFont = context.font;
2545 context.font.shape =
2546 known_coded_font_shapes[where - known_font_shapes];
2547 output_font_change(os, oldFont, context.font);
2548 eat_whitespace(p, os, context, false);
2550 else if (is_known(t.cs(), known_old_font_families) &&
2551 context.new_layout_allowed) {
2552 char const * const * where =
2553 is_known(t.cs(), known_old_font_families);
2554 context.check_layout(os);
2555 TeXFont const oldFont = context.font;
2556 context.font.init();
2557 context.font.size = oldFont.size;
2558 context.font.family =
2559 known_coded_font_families[where - known_old_font_families];
2560 output_font_change(os, oldFont, context.font);
2561 eat_whitespace(p, os, context, false);
2564 else if (is_known(t.cs(), known_old_font_series) &&
2565 context.new_layout_allowed) {
2566 char const * const * where =
2567 is_known(t.cs(), known_old_font_series);
2568 context.check_layout(os);
2569 TeXFont const oldFont = context.font;
2570 context.font.init();
2571 context.font.size = oldFont.size;
2572 context.font.series =
2573 known_coded_font_series[where - known_old_font_series];
2574 output_font_change(os, oldFont, context.font);
2575 eat_whitespace(p, os, context, false);
2578 else if (is_known(t.cs(), known_old_font_shapes) &&
2579 context.new_layout_allowed) {
2580 char const * const * where =
2581 is_known(t.cs(), known_old_font_shapes);
2582 context.check_layout(os);
2583 TeXFont const oldFont = context.font;
2584 context.font.init();
2585 context.font.size = oldFont.size;
2586 context.font.shape =
2587 known_coded_font_shapes[where - known_old_font_shapes];
2588 output_font_change(os, oldFont, context.font);
2589 eat_whitespace(p, os, context, false);
2592 else if (t.cs() == "selectlanguage") {
2593 context.check_layout(os);
2594 // save the language for the case that a
2595 // \foreignlanguage is used
2597 context.font.language = babel2lyx(p.verbatim_item());
2598 os << "\n\\lang " << context.font.language << "\n";
2601 else if (t.cs() == "foreignlanguage") {
2602 string const lang = babel2lyx(p.verbatim_item());
2603 parse_text_attributes(p, os, FLAG_ITEM, outer,
2605 context.font.language, lang);
2608 else if (t.cs() == "inputencoding") {
2609 // nothing to write here
2610 string const enc = subst(p.verbatim_item(), "\n", " ");
2614 else if (t.cs() == "ldots") {
2615 context.check_layout(os);
2616 os << "\\SpecialChar \\ldots{}\n";
2617 skip_spaces_braces(p);
2620 else if (t.cs() == "lyxarrow") {
2621 context.check_layout(os);
2622 os << "\\SpecialChar \\menuseparator\n";
2623 skip_spaces_braces(p);
2626 else if (t.cs() == "textcompwordmark") {
2627 context.check_layout(os);
2628 os << "\\SpecialChar \\textcompwordmark{}\n";
2629 skip_spaces_braces(p);
2632 else if (t.cs() == "slash") {
2633 context.check_layout(os);
2634 os << "\\SpecialChar \\slash{}\n";
2635 skip_spaces_braces(p);
2638 else if (t.cs() == "nobreakdash" && p.next_token().asInput() == "-") {
2639 context.check_layout(os);
2640 os << "\\SpecialChar \\nobreakdash-\n";
2644 else if (t.cs() == "textquotedbl") {
2645 context.check_layout(os);
2650 else if (t.cs() == "@" && p.next_token().asInput() == ".") {
2651 context.check_layout(os);
2652 os << "\\SpecialChar \\@.\n";
2656 else if (t.cs() == "-") {
2657 context.check_layout(os);
2658 os << "\\SpecialChar \\-\n";
2661 else if (t.cs() == "textasciitilde") {
2662 context.check_layout(os);
2664 skip_spaces_braces(p);
2667 else if (t.cs() == "textasciicircum") {
2668 context.check_layout(os);
2670 skip_spaces_braces(p);
2673 else if (t.cs() == "textbackslash") {
2674 context.check_layout(os);
2675 os << "\n\\backslash\n";
2676 skip_spaces_braces(p);
2679 else if (t.cs() == "_" || t.cs() == "&" || t.cs() == "#"
2680 || t.cs() == "$" || t.cs() == "{" || t.cs() == "}"
2682 context.check_layout(os);
2686 else if (t.cs() == "char") {
2687 context.check_layout(os);
2688 if (p.next_token().character() == '`') {
2690 if (p.next_token().cs() == "\"") {
2695 handle_ert(os, "\\char`", context);
2698 handle_ert(os, "\\char", context);
2702 else if (t.cs() == "verb") {
2703 context.check_layout(os);
2704 char const delimiter = p.next_token().character();
2705 string const arg = p.getArg(delimiter, delimiter);
2707 oss << "\\verb" << delimiter << arg << delimiter;
2708 handle_ert(os, oss.str(), context);
2711 // Problem: \= creates a tabstop inside the tabbing environment
2712 // and else an accent. In the latter case we really would want
2713 // \={o} instead of \= o.
2714 else if (t.cs() == "=" && (flags & FLAG_TABBING))
2715 handle_ert(os, t.asInput(), context);
2717 // accents (see Table 6 in Comprehensive LaTeX Symbol List)
2718 else if (t.cs().size() == 1
2719 && contains("\"'.=^`bcdHkrtuv~", t.cs())) {
2720 context.check_layout(os);
2721 // try to see whether the string is in unicodesymbols
2723 string command = t.asInput() + "{"
2724 + trim(p.verbatim_item())
2726 docstring s = encodings.fromLaTeXCommand(from_utf8(command), rem);
2729 cerr << "When parsing " << command
2730 << ", result is " << to_utf8(s)
2731 << "+" << to_utf8(rem) << endl;
2734 // we did not find a non-ert version
2735 handle_ert(os, command, context);
2738 else if (t.cs() == "\\") {
2739 context.check_layout(os);
2741 handle_ert(os, "\\\\" + p.getOpt(), context);
2742 else if (p.next_token().asInput() == "*") {
2744 // getOpt() eats the following space if there
2745 // is no optional argument, but that is OK
2746 // here since it has no effect in the output.
2747 handle_ert(os, "\\\\*" + p.getOpt(), context);
2750 begin_inset(os, "Newline newline");
2755 else if (t.cs() == "newline" ||
2756 (t.cs() == "linebreak" && !p.hasOpt())) {
2757 context.check_layout(os);
2758 begin_inset(os, "Newline ");
2761 skip_spaces_braces(p);
2764 else if (t.cs() == "input" || t.cs() == "include"
2765 || t.cs() == "verbatiminput") {
2766 string name = t.cs();
2767 if (t.cs() == "verbatiminput"
2768 && p.next_token().asInput() == "*")
2769 name += p.get_token().asInput();
2770 context.check_layout(os);
2771 string filename(normalize_filename(p.getArg('{', '}')));
2772 string const path = getMasterFilePath();
2773 // We want to preserve relative / absolute filenames,
2774 // therefore path is only used for testing
2775 if ((t.cs() == "include" || t.cs() == "input") &&
2776 !makeAbsPath(filename, path).exists()) {
2777 // The file extension is probably missing.
2778 // Now try to find it out.
2779 string const tex_name =
2780 find_file(filename, path,
2781 known_tex_extensions);
2782 if (!tex_name.empty())
2783 filename = tex_name;
2785 bool external = false;
2787 if (makeAbsPath(filename, path).exists()) {
2788 string const abstexname =
2789 makeAbsPath(filename, path).absFileName();
2790 string const abslyxname =
2791 changeExtension(abstexname, ".lyx");
2792 string const absfigname =
2793 changeExtension(abstexname, ".fig");
2794 fix_relative_filename(filename);
2795 string const lyxname =
2796 changeExtension(filename, ".lyx");
2798 external = FileName(absfigname).exists();
2799 if (t.cs() == "input") {
2800 string const ext = getExtension(abstexname);
2802 // Combined PS/LaTeX:
2803 // x.eps, x.pstex_t (old xfig)
2804 // x.pstex, x.pstex_t (new xfig, e.g. 3.2.5)
2805 FileName const absepsname(
2806 changeExtension(abstexname, ".eps"));
2807 FileName const abspstexname(
2808 changeExtension(abstexname, ".pstex"));
2809 bool const xfigeps =
2810 (absepsname.exists() ||
2811 abspstexname.exists()) &&
2814 // Combined PDF/LaTeX:
2815 // x.pdf, x.pdftex_t (old xfig)
2816 // x.pdf, x.pdf_t (new xfig, e.g. 3.2.5)
2817 FileName const abspdfname(
2818 changeExtension(abstexname, ".pdf"));
2819 bool const xfigpdf =
2820 abspdfname.exists() &&
2821 (ext == "pdftex_t" || ext == "pdf_t");
2825 // Combined PS/PDF/LaTeX:
2826 // x_pspdftex.eps, x_pspdftex.pdf, x.pspdftex
2827 string const absbase2(
2828 removeExtension(abstexname) + "_pspdftex");
2829 FileName const abseps2name(
2830 addExtension(absbase2, ".eps"));
2831 FileName const abspdf2name(
2832 addExtension(absbase2, ".pdf"));
2833 bool const xfigboth =
2834 abspdf2name.exists() &&
2835 abseps2name.exists() && ext == "pspdftex";
2837 xfig = xfigpdf || xfigeps || xfigboth;
2838 external = external && xfig;
2841 outname = changeExtension(filename, ".fig");
2843 // Don't try to convert, the result
2844 // would be full of ERT.
2846 } else if (t.cs() != "verbatiminput" &&
2847 tex2lyx(abstexname, FileName(abslyxname),
2854 cerr << "Warning: Could not find included file '"
2855 << filename << "'." << endl;
2859 begin_inset(os, "External\n");
2860 os << "\ttemplate XFig\n"
2861 << "\tfilename " << outname << '\n';
2863 begin_command_inset(os, "include", name);
2864 os << "preview false\n"
2865 "filename \"" << outname << "\"\n";
2870 else if (t.cs() == "bibliographystyle") {
2871 // store new bibliographystyle
2872 bibliographystyle = p.verbatim_item();
2873 // If any other command than \bibliography and
2874 // \nocite{*} follows, we need to output the style
2875 // (because it might be used by that command).
2876 // Otherwise, it will automatically be output by LyX.
2879 for (Token t2 = p.get_token(); p.good(); t2 = p.get_token()) {
2880 if (t2.cat() == catBegin)
2882 if (t2.cat() != catEscape)
2884 if (t2.cs() == "nocite") {
2885 if (p.getArg('{', '}') == "*")
2887 } else if (t2.cs() == "bibliography")
2894 "\\bibliographystyle{" + bibliographystyle + '}',
2899 else if (t.cs() == "bibliography") {
2900 context.check_layout(os);
2901 begin_command_inset(os, "bibtex", "bibtex");
2902 if (!btprint.empty()) {
2903 os << "btprint " << '"' << "btPrintAll" << '"' << "\n";
2904 // clear the string because the next BibTeX inset can be without the
2905 // \nocite{*} option
2908 os << "bibfiles " << '"' << p.verbatim_item() << '"' << "\n";
2909 // Do we have a bibliographystyle set?
2910 if (!bibliographystyle.empty())
2911 os << "options " << '"' << bibliographystyle << '"' << "\n";
2915 else if (t.cs() == "parbox")
2916 parse_box(p, os, 0, FLAG_ITEM, outer, context, "", "", t.cs());
2918 else if (t.cs() == "ovalbox" || t.cs() == "Ovalbox" ||
2919 t.cs() == "shadowbox" || t.cs() == "doublebox")
2920 parse_outer_box(p, os, FLAG_ITEM, outer, context, t.cs(), "");
2922 else if (t.cs() == "framebox") {
2923 string special = p.getOpt();
2924 special += p.getOpt();
2925 parse_outer_box(p, os, FLAG_ITEM, outer, context, t.cs(), special);
2928 //\makebox() is part of the picture environment and different from \makebox{}
2929 //\makebox{} will be parsed by parse_box when bug 2956 is fixed
2930 else if (t.cs() == "makebox") {
2931 string arg = t.asInput();
2932 if (p.next_token().character() == '(')
2933 //the syntax is: \makebox(x,y)[position]{content}
2934 arg += p.getFullParentheseArg();
2936 //the syntax is: \makebox[width][position]{content}
2937 arg += p.getFullOpt();
2938 handle_ert(os, arg + p.getFullOpt(), context);
2941 else if (t.cs() == "smallskip" ||
2942 t.cs() == "medskip" ||
2943 t.cs() == "bigskip" ||
2944 t.cs() == "vfill") {
2945 context.check_layout(os);
2946 begin_inset(os, "VSpace ");
2949 skip_spaces_braces(p);
2952 else if (is_known(t.cs(), known_spaces)) {
2953 char const * const * where = is_known(t.cs(), known_spaces);
2954 context.check_layout(os);
2955 begin_inset(os, "space ");
2956 os << '\\' << known_coded_spaces[where - known_spaces]
2959 // LaTeX swallows whitespace after all spaces except
2960 // "\\,". We have to do that here, too, because LyX
2961 // adds "{}" which would make the spaces significant.
2963 eat_whitespace(p, os, context, false);
2964 // LyX adds "{}" after all spaces except "\\ " and
2965 // "\\,", so we have to remove "{}".
2966 // "\\,{}" is equivalent to "\\," in LaTeX, so we
2967 // remove the braces after "\\,", too.
2972 else if (t.cs() == "newpage" ||
2973 (t.cs() == "pagebreak" && !p.hasOpt()) ||
2974 t.cs() == "clearpage" ||
2975 t.cs() == "cleardoublepage") {
2976 context.check_layout(os);
2977 begin_inset(os, "Newpage ");
2980 skip_spaces_braces(p);
2983 else if (t.cs() == "DeclareRobustCommand" ||
2984 t.cs() == "DeclareRobustCommandx" ||
2985 t.cs() == "newcommand" ||
2986 t.cs() == "newcommandx" ||
2987 t.cs() == "providecommand" ||
2988 t.cs() == "providecommandx" ||
2989 t.cs() == "renewcommand" ||
2990 t.cs() == "renewcommandx") {
2991 // DeclareRobustCommand, DeclareRobustCommandx,
2992 // providecommand and providecommandx could be handled
2993 // by parse_command(), but we need to call
2994 // add_known_command() here.
2995 string name = t.asInput();
2996 if (p.next_token().asInput() == "*") {
2997 // Starred form. Eat '*'
3001 string const command = p.verbatim_item();
3002 string const opt1 = p.getOpt();
3003 string const opt2 = p.getFullOpt();
3004 add_known_command(command, opt1, !opt2.empty());
3005 string const ert = name + '{' + command + '}' +
3007 '{' + p.verbatim_item() + '}';
3009 if (t.cs() == "DeclareRobustCommand" ||
3010 t.cs() == "DeclareRobustCommandx" ||
3011 t.cs() == "providecommand" ||
3012 t.cs() == "providecommandx" ||
3013 name[name.length()-1] == '*')
3014 handle_ert(os, ert, context);
3016 context.check_layout(os);
3017 begin_inset(os, "FormulaMacro");
3023 else if (t.cs() == "let" && p.next_token().asInput() != "*") {
3024 // let could be handled by parse_command(),
3025 // but we need to call add_known_command() here.
3026 string ert = t.asInput();
3029 if (p.next_token().cat() == catBegin) {
3030 name = p.verbatim_item();
3031 ert += '{' + name + '}';
3033 name = p.verbatim_item();
3038 if (p.next_token().cat() == catBegin) {
3039 command = p.verbatim_item();
3040 ert += '{' + command + '}';
3042 command = p.verbatim_item();
3045 // If command is known, make name known too, to parse
3046 // its arguments correctly. For this reason we also
3047 // have commands in syntax.default that are hardcoded.
3048 CommandMap::iterator it = known_commands.find(command);
3049 if (it != known_commands.end())
3050 known_commands[t.asInput()] = it->second;
3051 handle_ert(os, ert, context);
3054 else if (t.cs() == "hspace" || t.cs() == "vspace") {
3055 bool starred = false;
3056 if (p.next_token().asInput() == "*") {
3060 string name = t.asInput();
3061 string const length = p.verbatim_item();
3064 bool valid = splitLatexLength(length, valstring, unit);
3065 bool known_hspace = false;
3066 bool known_vspace = false;
3067 bool known_unit = false;
3070 istringstream iss(valstring);
3073 if (t.cs()[0] == 'h') {
3074 if (unit == "\\fill") {
3079 known_hspace = true;
3082 if (unit == "\\smallskipamount") {
3084 known_vspace = true;
3085 } else if (unit == "\\medskipamount") {
3087 known_vspace = true;
3088 } else if (unit == "\\bigskipamount") {
3090 known_vspace = true;
3091 } else if (unit == "\\fill") {
3093 known_vspace = true;
3097 if (!known_hspace && !known_vspace) {
3098 switch (unitFromString(unit)) {
3119 if (t.cs()[0] == 'h' && (known_unit || known_hspace)) {
3120 // Literal horizontal length or known variable
3121 context.check_layout(os);
3122 begin_inset(os, "space ");
3130 if (known_unit && !known_hspace)
3132 << translate_len(length);
3134 } else if (known_unit || known_vspace) {
3135 // Literal vertical length or known variable
3136 context.check_layout(os);
3137 begin_inset(os, "VSpace ");
3145 // LyX can't handle other length variables in Inset VSpace/space
3150 handle_ert(os, name + '{' + unit + '}', context);
3151 else if (value == -1.0)
3152 handle_ert(os, name + "{-" + unit + '}', context);
3154 handle_ert(os, name + '{' + valstring + unit + '}', context);
3156 handle_ert(os, name + '{' + length + '}', context);
3160 // The single '=' is meant here.
3161 else if ((newinsetlayout = findInsetLayout(context.textclass, t.cs(), true))) {
3163 context.check_layout(os);
3164 begin_inset(os, "Flex ");
3165 os << to_utf8(newinsetlayout->name()) << '\n'
3166 << "status collapsed\n";
3167 parse_text_in_inset(p, os, FLAG_ITEM, false, context, newinsetlayout);
3172 // try to see whether the string is in unicodesymbols
3173 // Only use text mode commands, since we are in text mode here,
3174 // and math commands may be invalid (bug 6797)
3176 docstring s = encodings.fromLaTeXCommand(from_utf8(t.asInput()),
3177 rem, Encodings::TEXT_CMD);
3180 cerr << "When parsing " << t.cs()
3181 << ", result is " << to_utf8(s)
3182 << "+" << to_utf8(rem) << endl;
3183 context.check_layout(os);
3185 skip_spaces_braces(p);
3187 //cerr << "#: " << t << " mode: " << mode << endl;
3188 // heuristic: read up to next non-nested space
3190 string s = t.asInput();
3191 string z = p.verbatim_item();
3192 while (p.good() && z != " " && z.size()) {
3193 //cerr << "read: " << z << endl;
3195 z = p.verbatim_item();
3197 cerr << "found ERT: " << s << endl;
3198 handle_ert(os, s + ' ', context);
3201 string name = t.asInput();
3202 if (p.next_token().asInput() == "*") {
3203 // Starred commands like \vspace*{}
3204 p.get_token(); // Eat '*'
3207 if (!parse_command(name, p, os, outer, context))
3208 handle_ert(os, name, context);
3212 if (flags & FLAG_LEAVE) {
3213 flags &= ~FLAG_LEAVE;