3 * Purpose: Parser for mathed
4 * Author: Alejandro Aguilar Sierra <asierra@servidor.unam.mx>
5 * Created: January 1996
6 * Description: Parse LaTeX2e math mode code.
8 * Dependencies: Xlib, XForms
10 * Copyright: 1996, Alejandro Aguilar Sierra
14 * You are free to use and modify this code under the terms of
15 * the GNU General Public Licence version 2 or later.
23 #pragma implementation
26 #include "math_parser.h"
28 #include "math_inset.h"
29 #include "math_arrayinset.h"
30 #include "math_charinset.h"
31 #include "math_deliminset.h"
32 #include "math_factory.h"
33 #include "math_funcinset.h"
34 #include "math_macro.h"
35 #include "math_macrotable.h"
36 #include "math_macrotemplate.h"
37 #include "math_matrixinset.h"
38 #include "math_rootinset.h"
39 #include "math_scopeinset.h"
40 #include "math_sqrtinset.h"
41 #include "math_scriptinset.h"
42 #include "math_sqrtinset.h"
46 #include "support/lstrings.h"
54 bool stared(string const & s)
56 unsigned n = s.size();
57 return n && s[n - 1] == '*';
60 MathScriptInset * prevScriptInset(MathArray const & array)
62 MathInset * p = array.back();
63 return (p && p->isScriptInset()) ? static_cast<MathScriptInset *>(p) : 0;
67 MathInset * lastScriptInset(MathArray & array, bool up, int limits)
69 MathScriptInset * p = prevScriptInset(array);
71 MathInset * b = array.back();
72 if (b && b->isScriptable()) {
73 p = new MathScriptInset(up, !up, b->clone());
76 p = new MathScriptInset(up, !up);
90 // These are TeX's catcodes
92 catEscape, // 0 backslash
102 catSpace, // 10 space
103 catLetter, // 11 a-zA-Z
104 catOther, // 12 none of the above
107 catInvalid // 15 <delete>
110 CatCode theCatcode[256];
113 inline CatCode catcode(unsigned char c)
115 return theCatcode[c];
119 const unsigned char LM_TK_OPEN = '{';
120 const unsigned char LM_TK_CLOSE = '}';
123 FLAG_BRACE = 1 << 0, // an opening brace needed
124 FLAG_BRACE_LAST = 1 << 1, // last closing brace ends the parsing process
125 FLAG_RIGHT = 1 << 2, // next \\right ends the parsing process
126 FLAG_END = 1 << 3, // next \\end ends the parsing process
127 FLAG_BRACK_END = 1 << 4, // next closing bracket ends the parsing process
128 FLAG_NEWLINE = 1 << 6, // next \\\\ ends the parsing process
129 FLAG_ITEM = 1 << 7, // read a (possibly braced token)
130 FLAG_BLOCK = 1 << 8, // next block ends the parsing process
131 FLAG_LEAVE = 1 << 9, // leave the loop at the end
137 for (int i = 0; i <= 255; ++i)
138 theCatcode[i] = catOther;
139 for (int i = 'a'; i <= 'z'; ++i)
140 theCatcode[i] = catLetter;
141 for (int i = 'A'; i <= 'Z'; ++i)
142 theCatcode[i] = catLetter;
144 theCatcode['\\'] = catEscape;
145 theCatcode['{'] = catBegin;
146 theCatcode['}'] = catEnd;
147 theCatcode['$'] = catMath;
148 theCatcode['&'] = catAlign;
149 theCatcode['\n'] = catNewline;
150 theCatcode['#'] = catParameter;
151 theCatcode['^'] = catSuper;
152 theCatcode['_'] = catSub;
153 theCatcode['
\7f'] = catIgnore;
154 theCatcode[' '] = catSpace;
155 theCatcode['\t'] = catSpace;
156 theCatcode['\r'] = catSpace;
157 theCatcode['~'] = catActive;
158 theCatcode['%'] = catComment;
164 // Helper class for parsing
170 Token() : cs_(), char_(0), cat_(catIgnore) {}
172 Token(char c, CatCode cat) : cs_(), char_(c), cat_(cat) {}
174 Token(const string & cs) : cs_(cs), char_(0), cat_(catIgnore) {}
177 bool operator==(Token const & t) const;
179 bool operator!=(Token const & t) const;
181 string const & cs() const { return cs_; }
183 CatCode cat() const { return cat_; }
185 char character() const { return char_; }
187 string asString() const;
198 string Token::asString() const
200 return cs_.size() ? cs_ : string(1, char_);
203 bool Token::operator==(Token const & t) const
205 return char_ == t.char_ && cat_ == t.cat_ && cs_ == t.cs_;
208 bool Token::operator!=(Token const & t) const
210 return char_ != t.char_ && cat_ != t.cat_ && cs_ != t.cs_;
213 ostream & operator<<(ostream & os, Token const & t)
216 os << "\\" << t.cs();
218 os << "[" << t.character() << "," << t.cat() << "]";
227 Parser(LyXLex & lex);
229 Parser(istream & is);
232 MathMacroTemplate * parse_macro();
234 MathMatrixInset * parse_normal();
236 void parse_into(MathArray & array, unsigned flags);
238 int lineno() const { return lineno_; }
244 string getArg(char lf, char rf);
248 void error(string const & msg);
250 void parse_lines(MathGridInset * p, bool numbered, bool outmost);
252 latexkeys const * read_delim();
256 void tokenize(istream & is);
258 void tokenize(string const & s);
260 void push_back(Token const & t);
264 Token const & prevToken() const;
266 Token const & nextToken() const;
268 Token const & getToken();
270 void lex(string const & s);
277 std::vector<Token> tokens_;
289 Parser::Parser(LyXLex & lexer)
290 : lineno_(lexer.getLineNo()), pos_(0)
292 tokenize(lexer.getStream());
296 Parser::Parser(istream & is)
297 : lineno_(0), pos_(0)
303 void Parser::push_back(Token const & t)
305 tokens_.push_back(t);
309 void Parser::pop_back()
315 Token const & Parser::prevToken() const
317 static const Token dummy;
318 return pos_ > 0 ? tokens_[pos_ - 1] : dummy;
322 Token const & Parser::nextToken() const
324 static const Token dummy;
325 return good() ? tokens_[pos_] : dummy;
329 Token const & Parser::getToken()
331 static const Token dummy;
332 return good() ? tokens_[pos_++] : dummy;
336 void Parser::putback()
342 bool Parser::good() const
344 return pos_ < tokens_.size();
348 char Parser::getChar()
351 lyxerr << "The input stream is not well..." << endl;
352 return tokens_[pos_++].character();
356 string Parser::getArg(char lf, char rg)
364 while ((c = getChar()) != rg && good())
371 void Parser::tokenize(istream & is)
373 // eat everything up to the next \end_inset or end of stream
374 // and store it in s for further tokenization
379 if (s.size() >= 10 && s.substr(s.size() - 10) == "\\end_inset") {
380 s = s.substr(0, s.size() - 10);
390 void Parser::tokenize(string const & buffer)
392 static bool init_done = false;
399 istringstream is(buffer, ios::in || ios::binary);
404 switch (catcode(c)) {
408 if (catcode(c) == catNewline)
409 ; //push_back(Token("par"));
411 push_back(Token(' ', catSpace));
418 while (is.get(c) && catcode(c) != catNewline)
427 if (catcode(c) == catLetter) {
428 while (is.get(c) && catcode(c) == catLetter)
430 if (catcode(c) == catSpace)
431 while (is.get(c) && catcode(c) == catSpace)
440 push_back(Token(c, catcode(c)));
444 //lyxerr << "\nTokens: ";
445 //for (unsigned i = 0; i < tokens_.size(); ++i)
446 // lyxerr << tokens_[i];
451 void Parser::error(string const & msg)
453 lyxerr << "Line ~" << lineno_ << ": Math parse error: " << msg << endl;
457 void Parser::parse_lines(MathGridInset * p, bool numbered, bool outmost)
459 const int cols = p->ncols();
461 // save global variables
462 bool const saved_num = curr_num_;
463 string const saved_label = curr_label_;
465 for (int row = 0; true; ++row) {
466 // reset global variables
467 curr_num_ = numbered;
471 for (int col = 0; col < cols; ++col) {
472 //lyxerr << "reading cell " << row << " " << col << "\n";
473 parse_into(p->cell(col + row * cols), FLAG_BLOCK);
476 if (prevToken().cat() != catAlign) {
477 //lyxerr << "less cells read than normal in row/col: "
478 // << row << " " << col << "\n";
484 MathMatrixInset * m = static_cast<MathMatrixInset *>(p);
485 m->numbered(row, curr_num_);
486 m->label(row, curr_label_);
487 if (curr_skip_.size()) {
488 m->vskip(LyXLength(curr_skip_), row);
494 if (prevToken() != Token("\\")) {
495 //lyxerr << "no newline here\n";
502 // restore "global" variables
503 curr_num_ = saved_num;
504 curr_label_ = saved_label;
508 MathMacroTemplate * Parser::parse_macro()
510 while (nextToken().cat() == catSpace)
513 if (getToken().cs() != "newcommand") {
514 lyxerr << "\\newcommand expected\n";
518 if (getToken().cat() != catBegin) {
519 lyxerr << "'{' expected\n";
523 string name = getToken().cs();
525 if (getToken().cat() != catEnd) {
526 lyxerr << "'}' expected\n";
530 string arg = getArg('[', ']');
531 int narg = arg.empty() ? 0 : atoi(arg.c_str());
532 //lyxerr << "creating macro " << name << " with " << narg << "args\n";
533 MathMacroTemplate * p = new MathMacroTemplate(name, narg);
534 parse_into(p->cell(0), FLAG_BRACE | FLAG_BRACE_LAST);
539 MathMatrixInset * Parser::parse_normal()
541 Token const & t = getToken();
543 if (t.cat() == catMath || t.cs() == "(") {
544 MathMatrixInset * p = new MathMatrixInset(LM_OT_SIMPLE);
545 parse_into(p->cell(0), 0);
549 if (!t.cs().size()) {
550 lyxerr << "start of math expected, got '" << t << "'\n";
554 string const & cs = t.cs();
559 MathMatrixInset * p = new MathMatrixInset(LM_OT_EQUATION);
560 parse_into(p->cell(0), 0);
561 p->numbered(0, curr_num_);
562 p->label(0, curr_label_);
567 lyxerr << "'begin' of un-simple math expected, got '" << cs << "'\n";
571 string const name = getArg('{', '}');
573 if (name == "equation" || name == "equation*") {
574 curr_num_ = stared(name);
576 MathMatrixInset * p = new MathMatrixInset(LM_OT_EQUATION);
577 parse_into(p->cell(0), FLAG_END);
578 p->numbered(0, curr_num_);
579 p->label(0, curr_label_);
583 if (name == "eqnarray" || name == "eqnarray*") {
584 MathMatrixInset * p = new MathMatrixInset(LM_OT_EQNARRAY);
585 parse_lines(p, stared(name), true);
589 if (name == "align" || name == "align*") {
590 MathMatrixInset * p = new MathMatrixInset(LM_OT_ALIGN);
591 p->halign(getArg('{', '}'));
592 parse_lines(p, stared(name), true);
596 if (name == "alignat" || name == "alignat*") {
597 MathMatrixInset * p = new MathMatrixInset(LM_OT_ALIGNAT);
598 p->halign(getArg('{', '}'));
599 parse_lines(p, stared(name), true);
603 lyxerr[Debug::MATHED] << "1: unknown math environment: " << name << "\n";
608 latexkeys const * Parser::read_delim()
610 Token const & t = getToken();
611 latexkeys const * l = in_word_set(t.asString());
612 return l ? l : in_word_set(".");
616 void Parser::parse_into(MathArray & array, unsigned flags)
618 MathTextCodes yyvarcode = LM_TC_MIN;
624 Token const & t = getToken();
626 lyxerr << "t: " << t << " flags: " << flags << "'\n";
627 //array.dump(lyxerr);
630 if (flags & FLAG_ITEM) {
632 if (t.cat() == catBegin) {
633 // skip the brace and collect everything to the next matching
635 flags |= FLAG_BRACE_LAST;
638 // handle only this single token, leave the loop if done
643 if (flags & FLAG_BRACE) {
644 if (t.cat() != catBegin) {
645 error("Expected {. Maybe you forgot to enclose an argument in {}");
649 flags &= ~FLAG_BRACE;
654 if (flags & FLAG_BLOCK) {
655 if (t.cat() == catEnd || t.cat() == catAlign || t.cs() == "\\")
657 if (t.cs() == "end") {
666 if (t.cat() == catMath)
669 else if (t.cat() == catLetter)
670 array.push_back(new MathCharInset(t.character(), yyvarcode));
672 else if (t.cat() == catSpace && yyvarcode == LM_TC_TEXTRM)
673 array.push_back(new MathCharInset(' ', yyvarcode));
675 else if (t.cat() == catParameter) {
676 Token const & n = getToken();
677 MathMacroArgument * p = new MathMacroArgument(n.character() - '0');
681 else if (t.cat() == catBegin) {
682 //lyxerr << " creating ScopeInset\n";
683 array.push_back(new MathScopeInset);
684 parse_into(array.back()->cell(0), FLAG_BRACE_LAST);
687 else if (t.cat() == catEnd) {
688 if (!(flags & FLAG_BRACE_LAST))
689 lyxerr << " ##### unexpected end of block\n";
693 else if (t.cat() == catAlign) {
694 lyxerr << "found tab unexpectedly, array: '" << array << "'\n";
698 else if (t.cat() == catSuper)
699 parse_into(lastScriptInset(array, true, limits)->cell(0), FLAG_ITEM);
701 else if (t.cat() == catSub)
702 parse_into(lastScriptInset(array, false, limits)->cell(1), FLAG_ITEM);
704 else if (t.character() == ']' && (flags & FLAG_BRACK_END))
707 else if (t.cat() == catOther)
708 array.push_back(new MathCharInset(t.character(), yyvarcode));
713 else if (t.cs() == "protect")
716 else if (t.cs() == "end")
719 else if (t.cs() == ")")
722 else if (t.cs() == "]")
725 else if (t.cs() == "\\") {
726 curr_skip_ = getArg('[', ']');
727 if (!(flags & FLAG_NEWLINE))
728 lyxerr[Debug::MATHED]
729 << "found newline unexpectedly, array: '" << array << "'\n";
733 else if (t.cs() == "limits")
736 else if (t.cs() == "nolimits")
739 else if (t.cs() == "nonumber")
742 else if (t.cs() == "number")
745 else if (t.cs() == "sqrt") {
748 array.push_back(new MathRootInset);
749 parse_into(array.back()->cell(0), FLAG_BRACK_END);
750 parse_into(array.back()->cell(1), FLAG_ITEM);
753 array.push_back(new MathSqrtInset);
754 parse_into(array.back()->cell(0), FLAG_ITEM);
758 else if (t.cs() == "left") {
759 latexkeys const * l = read_delim();
761 parse_into(ar, FLAG_RIGHT);
762 latexkeys const * r = read_delim();
763 MathDelimInset * dl = new MathDelimInset(l, r);
768 else if (t.cs() == "right") {
769 if (!(flags & FLAG_RIGHT))
770 error("Unmatched right delimiter");
777 lyxerr[Debug::MATHED] << "LM_TK_STY not implemented\n";
778 //MathArray tmp = array;
779 //MathSizeInset * p = new MathSizeInset(MathStyles(lval_->id));
780 //array.push_back(p);
781 //parse_into(p->cell(0), FLAG_BRACE_FONT);
786 if (MathMacroTable::hasTemplate(sval_)) {
787 MathMacro * m = MathMacroTable::cloneTemplate(sval_);
788 for (int i = 0; i < m->nargs(); ++i)
789 parse_into(m->cell(i), FLAG_ITEM);
791 m->metrics(LM_ST_TEXT);
793 array.push_back(new MathFuncInset(sval_));
797 array.push_back(new MathCharInset(ival_, LM_TC_SPECIAL));
801 else if (t.cs() == "begin") {
802 string const name = getArg('{', '}');
803 if (name == "array") {
804 string const valign = getArg('[', ']') + 'c';
805 string const halign = getArg('{', '}');
806 MathArrayInset * m = new MathArrayInset(halign.size(), 1);
807 m->valign(valign[0]);
809 parse_lines(m, false, false);
812 lyxerr[Debug::MATHED] << "unknow math inset begin '" << name << "'\n";
815 else if (t.cs() == "label") {
817 parse_into(ar, FLAG_ITEM);
820 curr_label_ = os.str();
822 //curr_label_ = getArg('{', '}');
825 else if (t.cs() == "choose" || t.cs() == "over") {
827 MathInset * p = createMathInset(t.cs());
828 p->cell(0).swap(array);
830 parse_into(p->cell(1), FLAG_BLOCK);
833 else if (t.cs().size()) {
835 latexkeys const * l = in_word_set(t.cs());
837 if (l->token == LM_TK_FONT) {
838 //lyxerr << "starting font\n";
840 parse_into(ar, FLAG_ITEM);
841 for (MathArray::iterator it = ar.begin(); it != ar.end(); ++it)
842 (*it)->handleFont(static_cast<MathTextCodes>(l->id));
844 //lyxerr << "ending font\n";
847 else if (l->token == LM_TK_OLDFONT)
848 yyvarcode = static_cast<MathTextCodes>(l->id);
851 MathInset * p = createMathInset(t.cs());
852 for (int i = 0; i < p->nargs(); ++i)
853 parse_into(p->cell(i), FLAG_ITEM);
859 MathInset * p = createMathInset(t.cs());
861 for (int i = 0; i < p->nargs(); ++i)
862 parse_into(p->cell(i), FLAG_ITEM);
865 error("Unrecognized token");
866 //lyxerr[Debug::MATHED] << "[" << t << "]\n";
873 if (flags & FLAG_LEAVE) {
874 flags &= ~FLAG_LEAVE;
880 lyxerr << " Math Panic, expect problems!\n";
881 // Search for the end command.
885 } while (good() && t.cs() != "end");
889 } // anonymous namespace
893 MathArray mathed_parse_cell(string const & str)
895 istringstream is(str.c_str());
898 parser.parse_into(ar, 0);
904 MathMacroTemplate * mathed_parse_macro(string const & str)
906 istringstream is(str.c_str());
908 return parser.parse_macro();
911 MathMacroTemplate * mathed_parse_macro(istream & is)
914 return parser.parse_macro();
917 MathMacroTemplate * mathed_parse_macro(LyXLex & lex)
920 return parser.parse_macro();
925 MathMatrixInset * mathed_parse_normal(string const & str)
927 istringstream is(str.c_str());
929 return parser.parse_normal();
932 MathMatrixInset * mathed_parse_normal(istream & is)
935 return parser.parse_normal();
938 MathMatrixInset * mathed_parse_normal(LyXLex & lex)
941 return parser.parse_normal();