X-Git-Url: https://git.donarmstrong.com/?a=blobdiff_plain;f=lily%2Flexer.ll;h=8bf851eea4b5e3a63cb2b240bfb9682674ff9ccf;hb=7785ac382ed511ed535adf77adc3c07649fbc9b5;hp=8a208c030f5b3802df8e74c440a59da6b24eaa32;hpb=2536405e9584f63d559653183971c0fec57ba037;p=lilypond.git

diff --git a/lily/lexer.ll b/lily/lexer.ll
index 8a208c030f..8bf851eea4 100644
--- a/lily/lexer.ll
+++ b/lily/lexer.ll
@@ -4,7 +4,7 @@
 
   source file of the LilyPond music typesetter
 
-  (c) 1996--2004 Han-Wen Nienhuys <hanwen@cs.uu.nl>
+  (c) 1996--2006 Han-Wen Nienhuys <hanwen@xs4all.nl>
            Jan Nieuwenhuizen <janneke@gnu.org>
 */
 
@@ -22,66 +22,67 @@
  */
 
 
-#include <stdio.h>
-#include <ctype.h>
-#include <errno.h>
+
+#include <cstdio>
+#include <cctype>
+#include <cerrno>
 
 /* Flex >= 2.5.29 fix; FlexLexer.h's multiple include bracing breaks
    when building the actual lexer.  */
+
 #define LEXER_CC
 
 #include <iostream>
 using namespace std;
-#include "music-function.hh"
-#include "source-file.hh"
-#include "parse-scm.hh"
-#include "lily-guile.hh"
-#include "string.hh"
-#include "string-convert.hh"
-#include "lily-lexer.hh"
+
+#include "context-def.hh"
+#include "identifier-smob.hh"
+#include "international.hh"
 #include "interval.hh"
 #include "lily-guile.hh"
-#include "parser.hh"
-#include "warn.hh"
+#include "lily-lexer.hh"
+#include "lilypond-input-version.hh"
 #include "main.hh"
+#include "music.hh"
+#include "music-function.hh"
+#include "parse-scm.hh"
+#include "parser.hh"
+#include "source-file.hh"
+#include "std-string.hh"
+#include "string-convert.hh"
 #include "version.hh"
-#include "lilypond-input-version.hh"
-#include "context-def.hh"
-#include "identifier-smob.hh"
+#include "warn.hh"
 
 /*
 RH 7 fix (?)
 */
 #define isatty HORRIBLEKLUDGE
 
-void strip_trailing_white (String&);
-void strip_leading_white (String&);
-String lyric_fudge (String s);
-int music_function_type (SCM);
-SCM lookup_markup_command (String s);
-bool is_valid_version (String s);
+void strip_trailing_white (string&);
+void strip_leading_white (string&);
+string lyric_fudge (string s);
+SCM lookup_markup_command (string s);
+bool is_valid_version (string s);
 
 
 #define start_quote()	\
 	yy_push_state (quote);\
-	yylval.string = new String
+	yylval.string = new string
 
 #define start_lyric_quote()	\
 	yy_push_state (lyric_quote);\
-	yylval.string = new String
+	yylval.string = new string
 
 #define yylval \
 	(*(YYSTYPE*)lexval)
 
-#define YY_USER_ACTION	add_lexed_char (YYLeng ());
-/*
-
-LYRICS		({AA}|{TEX})[^0-9 \t\n\f]*
+#define yylloc \
+	(*(YYLTYPE*)lexloc)
 
-*/
+#define YY_USER_ACTION	add_lexed_char (YYLeng ());
 
 
-SCM scan_fraction (String);
+SCM scan_fraction (string);
 SCM (* scm_parse_error_handler) (void *);
 
 
@@ -97,28 +98,35 @@ SCM (* scm_parse_error_handler) (void *);
 %option never-interactive 
 %option warn
 
+%x extratoken
 %x chords
-%x encoding
 %x figures
 %x incl
 %x lyrics
 %x lyric_quote
 %x longcomment
-%x markup 
+%x markup
 %x notes
 %x quote
-%x renameinput
+%x sourcefileline
+%x sourcefilename
 %x version
 
-A		[a-zA-Z]
+A		[a-zA-Z\200-\377]
 AA		{A}|_
 N		[0-9]
 AN		{AA}|{N}
+ANY_CHAR	(.|\n)
 PUNCT		[?!:'`]
 ACCENT		\\[`'"^]
-NATIONAL	[\001-\006\021-\027\031\036\200-\377]
+NATIONAL	[\001-\006\021-\027\031\036]
 TEX		{AA}|-|{PUNCT}|{ACCENT}|{NATIONAL}
 WORD		{A}{AN}*
+DASHED_WORD		{A}({AN}|-)*
+DASHED_KEY_WORD		\\{DASHED_WORD}
+
+
+
 ALPHAWORD	{A}+
 DIGIT		{N}
 UNSIGNED	{N}+
@@ -137,6 +145,8 @@ LYRICS		({AA}|{TEX})[^0-9 \t\n\r\f]*
 ESCAPED		[nt\\'"]
 EXTENDER	__
 HYPHEN		--
+BOM_UTF8	\357\273\277
+
 %%
 
 
@@ -144,7 +154,33 @@ HYPHEN		--
 	// windows-suck-suck-suck
 }
 
-<INITIAL,chords,incl,markup,lyrics,notes,figures>{
+<extratoken>{ANY_CHAR}	{
+  /* Generate a token without swallowing anything */
+
+  /* First unswallow the eaten character */
+  add_lexed_char (-YYLeng ());
+  yyless (0);
+
+  /* produce requested token */
+  int type = extra_token_types_.back ();
+  extra_token_types_.pop_back ();
+  if (extra_token_types_.empty ())
+    yy_pop_state ();
+
+  return type;
+}
+
+<INITIAL,chords,lyrics,figures,notes>{BOM_UTF8} {
+  if (this->lexloc->line_number () != 1 || this->lexloc->column_number () != 0)
+    {
+      LexerError (_ ("stray UTF-8 BOM encountered").c_str ());
+      exit (1);
+    }
+  if (be_verbose_global)
+     message (_ ("Skipping UTF-8 BOM"));
+}
+
+<INITIAL,chords,figures,incl,lyrics,markup,notes>{
   "%{"	{
 	yy_push_state (longcomment);
   }
@@ -161,53 +197,67 @@ HYPHEN		--
   }
 }
 
-<INITIAL,notes>\\encoding{WHITE}* {
-	yy_push_state (encoding);
+<INITIAL,notes,figures,chords,markup>{
+	\"		{
+		start_quote ();
+	}
 }
+
 <INITIAL,chords,lyrics,notes,figures>\\version{WHITE}*	{
 	yy_push_state (version);
 }
-<INITIAL,chords,lyrics,notes,figures>\\renameinput{WHITE}*	{
-	yy_push_state (renameinput);
+<INITIAL,chords,lyrics,notes,figures>\\sourcefilename{WHITE}*	{
+	yy_push_state (sourcefilename);
 }
-<encoding>\"[^"]*\"     {
-	String s (YYText () + 1);
-	s = s.left_string (s.index_last ('\"'));
-	set_encoding (s);
-	yy_pop_state ();
+<INITIAL,chords,lyrics,notes,figures>\\sourcefileline{WHITE}*	{
+	yy_push_state (sourcefileline);
 }
 <version>\"[^"]*\"     { /* got the version number */
-	String s (YYText () + 1);
-	s = s.left_string (s.index_last ('\"'));
+	string s (YYText () + 1);
+	s = s.substr (0, s.rfind ('\"'));
 
 	yy_pop_state ();
+
+	SCM top_scope = scm_car (scm_last_pair (scopes_));
+	scm_module_define (top_scope, ly_symbol2scm ("version-seen"), SCM_BOOL_T);
+
 	if (!is_valid_version (s))
 		return INVALID;
+
+
 }
-<renameinput>\"[^"]*\"     {
-	String s (YYText ()+1);
-	s = s.left_string (s.index_last ('\"'));
+<sourcefilename>\"[^"]*\"     {
+	string s (YYText () + 1);
+	s = s.substr (0, s.rfind ('\"'));
 
-	yy_pop_state();
-	this->here_input().source_file_->name_ = s;
-	progress_indication ("\n");
-	progress_indication (_f ("input renamed to: `%s'", s.to_str0 ()));
+	yy_pop_state ();
+	this->here_input().get_source_file ()->name_ = s;
+	message (_f ("Renaming input to: `%s'", s.c_str ()));
 	progress_indication ("\n");
-	scm_module_define (ly_car (scopes_),
+	scm_module_define (scm_car (scopes_),
 		     ly_symbol2scm ("input-file-name"),
-		     scm_makfrom0str (s.to_str0()));
+		     scm_makfrom0str (s.c_str ()));
 
 }
-<encoding>. 	{
-	LexerError (_ ("No quoted string found after \\encoding").to_str0 ());
+
+<sourcefileline>{INT}	{
+	int i;
+	sscanf (YYText (), "%d", &i);
+
 	yy_pop_state ();
+	this->here_input ().get_source_file ()->set_line (here_input ().start (), i);
 }
+
 <version>. 	{
-	LexerError (_ ("No quoted string found after \\version").to_str0 ());
+	LexerError (_ ("quoted string expected after \\version").c_str ());
+	yy_pop_state ();
+}
+<sourcefilename>. 	{
+	LexerError (_ ("quoted string expected after \\sourcefilename").c_str ());
 	yy_pop_state ();
 }
-<renameinput>. 	{
-	LexerError (_ ("No quoted string found after \\renameinput").to_str0 ());
+<sourcefileline>. 	{
+	LexerError (_ ("integer expected after \\sourcefileline").c_str ());
 	yy_pop_state ();
 }
 <longcomment>{
@@ -220,8 +270,8 @@ HYPHEN		--
 		yy_pop_state ();
 	}
 	<<EOF>> 	{
-		LexerError (_ ("EOF found inside a comment").to_str0 ());
-		main_input_b_ = false;
+		LexerError (_ ("EOF found inside a comment").c_str ());
+		is_main_input_ = false;
 		if (! close_input ()) 
 		  yyterminate (); // can't move this, since it actually rets a YY_NULL
 	}
@@ -229,10 +279,10 @@ HYPHEN		--
 
 
 <INITIAL,chords,lyrics,notes,figures>\\maininput           {
-	if (!main_input_b_)
+	if (!is_main_input_)
 	{
 		start_main_input ();
-		main_input_b_ = true;
+		is_main_input_ = true;
 	}
 	else
 		error (_ ("\\maininput not allowed outside init files"));
@@ -241,52 +291,50 @@ HYPHEN		--
 <INITIAL,chords,lyrics,figures,notes>\\include           {
 	yy_push_state (incl);
 }
-<incl>\"[^"]*\";?   { /* got the include file name */
-/* FIXME: semicolon? */
-	String s (YYText ()+1);
-	s = s.left_string (s.index_last ('"'));
+<incl>\"[^"]*\"   { /* got the include file name */
+	string s (YYText ()+1);
+	s = s.substr (0, s.rfind ('"'));
 
 	new_input (s, sources_);
 	yy_pop_state ();
 }
-<incl>\\{BLACK}*;?{WHITE} { /* got the include identifier */
-/* FIXME: semicolon? */
-	String s = YYText () + 1;
+<incl>\\{BLACK}*{WHITE} { /* got the include identifier */
+	string s = YYText () + 1;
 	strip_trailing_white (s);
 	if (s.length () && (s[s.length () - 1] == ';'))
-	  s = s.left_string (s.length () - 1);
+	  s = s.substr (0, s.length () - 1);
 
 	SCM sid = lookup_identifier (s);
 	if (scm_is_string (sid)) {
 		new_input (ly_scm2string (sid), sources_);
 		yy_pop_state ();
 	} else { 
-	    String msg (_f ("wrong or undefined identifier: `%s'", s ));
+	    string msg (_f ("wrong or undefined identifier: `%s'", s ));
 
-	    LexerError (msg.to_str0 ());
+	    LexerError (msg.c_str ());
 	    SCM err = scm_current_error_port ();
 	    scm_puts ("This value was found in the table: ", err);
 	    scm_display (sid, err);
 	  }
 }
 <incl>\"[^"]*   { // backup rule
-	error (_ ("Missing end quote"));
+	error (_ ("end quote missing"));
 	exit (1);
 }
 <chords,notes,figures>{RESTNAME} 	{
-	const char *s = YYText ();
+	char const *s = YYText ();
 	yylval.scm = scm_makfrom0str (s);
 	return RESTNAME;
 }
 <chords,notes,figures>R		{
 	return MULTI_MEASURE_REST;
 }
-<INITIAL,markup,chords,lyrics,notes,figures>#	{ //embedded scm
-	//char const* s = YYText () + 1;
-	char const* s = here_str0 ();
+<INITIAL,chords,figures,lyrics,markup,notes>#	{ //embedded scm
 	int n = 0;
-	SCM sval = ly_parse_scm (s, &n, here_input (),
-		safe_global_b && main_input_b_);
+	Input hi = here_input();
+	hi.step_forward ();
+	SCM sval = ly_parse_scm (hi.start (), &n, hi,
+		be_safe_global && is_main_input_);
 
 	if (sval == SCM_UNDEFINED)
 	{
@@ -294,11 +342,11 @@ HYPHEN		--
 		error_level_ = 1;
  	}
 
-	for (int i=0; i < n; i++)
+	for (int i = 0; i < n; i++)
 	{
 		yyinput ();
 	}
-	char_count_stack_.top () += n;
+	char_count_stack_.back () += n;
 
 	if (unpack_identifier (sval) != SCM_UNDEFINED)
 	{
@@ -307,16 +355,26 @@ HYPHEN		--
 	}
 		
 	yylval.scm = sval;
-	return SCM_T;
+	return SCM_TOKEN;
 }
 <INITIAL,notes,lyrics>{ 
-	\<\<   {
+	\<\<	{
 		return DOUBLE_ANGLE_OPEN;
 	}
-	\>\>   {
+	\>\>	{
 		return DOUBLE_ANGLE_CLOSE;
 	}
 }
+
+<INITIAL,notes>{
+	\<	{
+		return ANGLE_OPEN;
+	}
+	\>	{
+		return ANGLE_CLOSE;
+	}
+}
+
 <figures>{
 	_	{
 		return FIGURE_SPACE;
@@ -343,48 +401,20 @@ HYPHEN		--
 	}
 
 	{DIGIT}		{
-		yylval.i = String_convert::dec2int (String (YYText ()));
+		yylval.i = String_convert::dec2int (string (YYText ()));
 		return DIGIT;
 	}
 	{UNSIGNED}		{
-		yylval.i = String_convert::dec2int (String (YYText ()));
+		yylval.i = String_convert::dec2int (string (YYText ()));
 		return UNSIGNED;
 	}
 	{E_UNSIGNED}	{
-		yylval.i = String_convert::dec2int (String (YYText () +1));
+		yylval.i = String_convert::dec2int (string (YYText () +1));
 		return E_UNSIGNED;
 	}
-
-	\" {
-		start_quote ();
-	}
 }
 
-\"		{
-	start_quote ();
-}
-<quote>{
-	\\{ESCAPED}	{
-		*yylval.string += to_string (escaped_char (YYText ()[1]));
-	}
-	[^\\"]+	{
-		*yylval.string += YYText ();
-	}
-	\"	{
-
-		yy_pop_state ();
-
-		/* yylval is union. Must remember STRING before setting SCM*/
-		String *sp = yylval.string;
-		yylval.scm = scm_makfrom0str (sp->to_str0 ());
-		delete sp;
-		return STRING;
-	}
-	.	{
-		*yylval.string += YYText ();
-	}
-}
-<lyric_quote>{
+<quote,lyric_quote>{
 	\\{ESCAPED}	{
 		*yylval.string += to_string (escaped_char (YYText ()[1]));
 	}
@@ -396,10 +426,10 @@ HYPHEN		--
 		yy_pop_state ();
 
 		/* yylval is union. Must remember STRING before setting SCM*/
-		String *sp = yylval.string;
-		yylval.scm = scm_makfrom0str (sp->to_str0 ());
+		string *sp = yylval.string;
+		yylval.scm = scm_makfrom0str (sp->c_str ());
 		delete sp;
-		return LYRICS_STRING;
+		return is_lyric_state () ? LYRICS_STRING : STRING;
 	}
 	.	{
 		*yylval.string += YYText ();
@@ -415,7 +445,7 @@ HYPHEN		--
 		return FRACTION;
 	}
 	{UNSIGNED}		{
-		yylval.i = String_convert::dec2int (String (YYText ()));
+		yylval.i = String_convert::dec2int (string (YYText ()));
 		return UNSIGNED;
 	}
 	{NOTECOMMAND}	{
@@ -423,7 +453,7 @@ HYPHEN		--
 	}
 	{LYRICS} {
 		/* ugr. This sux. */
-		String s (YYText ()); 
+		string s (YYText ()); 
 		if (s == "__")
 			return yylval.i = EXTENDER;
 		if (s == "--")
@@ -434,7 +464,7 @@ HYPHEN		--
 		if (c == '{' ||  c == '}') // brace open is for not confusing dumb tools.
 			here_input ().warning (
 				_ ("Brace found at end of lyric.  Did you forget a space?"));
-		yylval.scm = scm_makfrom0str (s.to_str0 ());
+		yylval.scm = scm_makfrom0str (s.c_str ());
 
 
 		return LYRICS_STRING;
@@ -455,12 +485,9 @@ HYPHEN		--
 		return FRACTION;
 	}
 	{UNSIGNED}		{
-		yylval.i = String_convert::dec2int (String (YYText ()));
+		yylval.i = String_convert::dec2int (string (YYText ()));
 		return UNSIGNED;
 	}
-	\" {
-		start_quote ();
-	}
 	-  {
 		return CHORD_MINUS;
 	}
@@ -483,25 +510,16 @@ HYPHEN		--
 
 
 <markup>{
-	\" {
-		start_quote ();
-	}
-	\< {
-		return '<';
-	}
-	\> {
-		return '>';
-	}
 	\\score {
 		return SCORE;
 	}
 	{MARKUPCOMMAND} {
-		String str (YYText () + 1);
+		string str (YYText () + 1);
 		SCM s = lookup_markup_command (str);
 
-		if (scm_is_pair (s) && scm_is_symbol (ly_cdr (s)) ) {
-			yylval.scm = ly_car(s);
-			SCM tag = ly_cdr(s);
+		if (scm_is_pair (s) && scm_is_symbol (scm_cdr (s)) ) {
+			yylval.scm = scm_car(s);
+			SCM tag = scm_cdr(s);
 			if (tag == ly_symbol2scm("markup0"))
 				return MARKUP_HEAD_MARKUP0;
 			if (tag == ly_symbol2scm("empty"))
@@ -521,7 +539,7 @@ HYPHEN		--
 			else if (tag == ly_symbol2scm ("scheme0-scheme1-scheme2"))
 				return MARKUP_HEAD_SCM0_SCM1_SCM2;
 			else {
-				programming_error ("No parser tag defined for this signature. Abort"); 
+				programming_error ("no parser tag defined for this markup signature"); 
 				ly_display_scm (s);
 				assert(false);
 			}
@@ -532,14 +550,14 @@ HYPHEN		--
 		return YYText ()[0];
 	}
 	[^#{}"\\ \t\n\r\f]+ {
-		String s (YYText ()); 
+		string s (YYText ()); 
 
 		char c = s[s.length () - 1];
 		/* brace open is for not confusing dumb tools.  */
 		if (c == '{' ||  c == '}')
 			here_input ().warning (
 				_ ("Brace found at end of markup.  Did you forget a space?"));
-		yylval.scm = scm_makfrom0str (s.to_str0 ());
+		yylval.scm = scm_makfrom0str (s.c_str ());
 
 
 		return STRING;
@@ -549,10 +567,10 @@ HYPHEN		--
 	}
 }
 
-<<EOF>> {
-	if (main_input_b_)
+<*><<EOF>> {
+	if (is_main_input_)
 	{
-		main_input_b_ = false;
+		is_main_input_ = false;
 		if (!close_input ())
  	        /* Returns YY_NULL */
 			yyterminate ();
@@ -562,6 +580,14 @@ HYPHEN		--
  	  	yyterminate ();
 }
 
+<INITIAL>{
+	{DASHED_WORD}	{
+		return scan_bare_word (YYText ());
+	}
+	{DASHED_KEY_WORD}	{
+		return scan_escaped_word (YYText () + 1);
+	}
+}
 
 {WORD}	{
 	return scan_bare_word (YYText ());
@@ -571,15 +597,16 @@ HYPHEN		--
 }
 {REAL}		{
 	Real r;
-	int cnv=sscanf (YYText (), "%lf", &r);
+	int cnv = sscanf (YYText (), "%lf", &r);
 	assert (cnv == 1);
+	(void) cnv;
 
-	yylval.scm = scm_make_real (r);
+	yylval.scm = scm_from_double (r);
 	return REAL;
 }
 
 {UNSIGNED}	{
-	yylval.i = String_convert::dec2int (String (YYText ()));
+	yylval.i = String_convert::dec2int (string (YYText ()));
 	return UNSIGNED;
 }
 
@@ -599,13 +626,13 @@ HYPHEN		--
 }
 
 <INITIAL,lyrics,notes,figures>\\. {
-    char c= YYText ()[1];
+    char c = YYText ()[1];
 
     switch (c) {
     case '>':
-	return E_BIGGER;
+	return E_ANGLE_CLOSE;
     case '<':
-	return E_SMALLER;
+	return E_ANGLE_OPEN;
     case '!':
 	return E_EXCLAMATION;
     case '(':
@@ -613,9 +640,11 @@ HYPHEN		--
     case ')':
 	return E_CLOSE;
     case '[':
-	return E_LEFTSQUARE;
+	return E_BRACKET_OPEN;
+    case '+':
+	return E_PLUS;
     case ']':
-	return E_RIGHTSQUARE;
+	return E_BRACKET_CLOSE;
     case '~':
 	return E_TILDE;
     case '\\':
@@ -627,13 +656,27 @@ HYPHEN		--
 }
 
 <*>.		{
-	String msg = _f ("invalid character: `%c'", YYText ()[0]);
-	LexerError (msg.to_str0 ());
+	string msg = _f ("invalid character: `%c'", YYText ()[0]);
+	LexerError (msg.c_str ());
 	return YYText ()[0];
 }
 
 %%
 
+/* Make the lexer generate a token of the given type as the next token. 
+ TODO: make it possible to define a value for the token as well */
+void
+Lily_lexer::push_extra_token (int token_type)
+{
+	if (extra_token_types_.empty ())
+	{
+		if (YY_START != extratoken)
+			hidden_state_ = YY_START;
+		yy_push_state (extratoken);
+	}
+	extra_token_types_.push_back (token_type);
+}
+
 void
 Lily_lexer::push_chord_state (SCM tab)
 {
@@ -676,7 +719,8 @@ void
 Lily_lexer::pop_state ()
 {
 	if (YYSTATE == notes || YYSTATE == chords)
-		pitchname_tab_stack_ = ly_cdr (pitchname_tab_stack_);
+		pitchname_tab_stack_ = scm_cdr (pitchname_tab_stack_);
+
 	yy_pop_state ();
 }
 
@@ -689,22 +733,36 @@ Lily_lexer::identifier_type (SCM sid)
 
 
 int
-Lily_lexer::scan_escaped_word (String str)
+Lily_lexer::scan_escaped_word (string str)
 {
 	// use more SCM for this.
 
-//	SCM sym = ly_symbol2scm (str.to_str0 ());
+//	SCM sym = ly_symbol2scm (str.c_str ());
+
+	int i = lookup_keyword (str);
+ 	if (i == MARKUP && is_lyric_state ())
+ 		return LYRIC_MARKUP;
+	if (i != -1)
+		return i;
 
-	int l = lookup_keyword (str);
-	if (l != -1) {
-		return l;
-	}
 	SCM sid = lookup_identifier (str);
 	if (is_music_function (sid))
 	{
 		yylval.scm = get_music_function_transform (sid);
-		
-		return music_function_type (yylval.scm);
+
+		SCM s = scm_object_property (yylval.scm, ly_symbol2scm ("music-function-signature"));
+		push_extra_token (EXPECT_NO_MORE_ARGS);
+		for (; scm_is_pair (s); s = scm_cdr (s))
+		{
+			if (scm_car (s) == ly_music_p_proc)
+				push_extra_token (EXPECT_MUSIC);
+			else if (scm_car (s) == ly_lily_module_constant ("markup?"))
+				push_extra_token (EXPECT_MARKUP);
+			else if (ly_is_procedure (scm_car (s)))
+				push_extra_token (EXPECT_SCM);
+			else programming_error ("Function parameter without type-checking predicate");
+		}
+		return MUSIC_FUNCTION;
 	}
 
 	if (sid != SCM_UNDEFINED)
@@ -713,25 +771,25 @@ Lily_lexer::scan_escaped_word (String str)
 		return identifier_type (sid);
 	}
 
-	String msg (_f ("unknown escaped string: `\\%s'", str));	
-	LexerError (msg.to_str0 ());
+	string msg (_f ("unknown escaped string: `\\%s'", str));	
+	LexerError (msg.c_str ());
 
-	yylval.scm = scm_makfrom0str (str.to_str0 ());
+	yylval.scm = scm_makfrom0str (str.c_str ());
 
 	return STRING;
 }
 
 int
-Lily_lexer::scan_bare_word (String str)
+Lily_lexer::scan_bare_word (string str)
 {
-	SCM sym = ly_symbol2scm (str.to_str0 ());
+	SCM sym = ly_symbol2scm (str.c_str ());
 	if ((YYSTATE == notes) || (YYSTATE == chords)) {
 		SCM handle = SCM_BOOL_F;
 		if (scm_is_pair (pitchname_tab_stack_))
-			handle = scm_hashq_get_handle (ly_car (pitchname_tab_stack_), sym);
+			handle = scm_hashq_get_handle (scm_car (pitchname_tab_stack_), sym);
 		
 		if (scm_is_pair (handle)) {
-			yylval.scm = ly_cdr (handle);
+			yylval.scm = scm_cdr (handle);
 			if (unsmob_pitch (yylval.scm)) 
 	                    return (YYSTATE == notes) ? NOTENAME_PITCH : TONICNAME_PITCH;
 			else if (scm_is_symbol (yylval.scm))
@@ -739,73 +797,81 @@ Lily_lexer::scan_bare_word (String str)
 		}
 		else if ((handle = scm_hashq_get_handle (chordmodifier_tab_, sym))!= SCM_BOOL_F)
 		{
-		    yylval.scm = ly_cdr (handle);
+		    yylval.scm = scm_cdr (handle);
 		    return CHORD_MODIFIER;
 		}
 	}
 
-	yylval.scm = scm_makfrom0str (str.to_str0 ());
+	yylval.scm = scm_makfrom0str (str.c_str ());
 	return STRING;
 }
 
+int
+Lily_lexer::get_state () const
+{
+	if (YY_START == extratoken)
+		return hidden_state_;
+	else
+		return YY_START;
+}
+
 bool
 Lily_lexer::is_note_state () const
 {
-	return YY_START == notes;
+	return get_state () == notes;
 }
 
 bool
 Lily_lexer::is_chord_state () const
 {
-	return YY_START == chords;
+	return get_state () == chords;
 }
 
 bool
 Lily_lexer::is_lyric_state () const
 {
-	return YY_START == lyrics;
+	return get_state () == lyrics;
 }
 
 bool
 Lily_lexer::is_figure_state () const
 {
-	return YY_START == figures;
+	return get_state () == figures;
 }
 
 /*
- urg, belong to String (_convert)
+ urg, belong to string (_convert)
  and should be generalised 
  */
 void
-strip_leading_white (String&s)
+strip_leading_white (string&s)
 {
-	int i=0;
-	for (;  i < s.length (); i++) 
+	ssize i = 0;
+	for (;  i < s.length (); i++)
 		if (!isspace (s[i]))
 			break;
 
-	s = s.nomid_string (0, i);
+	s = s.substr (i);
 }
 
 void
-strip_trailing_white (String&s)
+strip_trailing_white (string&s)
 {
-	int i=s.length ();	
+	ssize i = s.length ();	
 	while (i--) 
 		if (!isspace (s[i]))
 			break;
 
-	s = s.left_string (i+1);
+	s = s.substr (0, i + 1);
 }
 
 
 
-/* 2.1.2x something -> \property -> \set. */ 
-Lilypond_version oldest_version ("2.2.0");
+Lilypond_version oldest_version ("2.7.38");
 
 
 bool
-is_valid_version (String s)
+is_valid_version (string s)
 {
   Lilypond_version current ( MAJOR_VERSION "." MINOR_VERSION "." PATCH_LEVEL );
   Lilypond_version ver (s);
@@ -822,25 +888,25 @@ is_valid_version (String s)
 /*
   substitute _ and \,
 */
-String
-lyric_fudge (String s)
+string
+lyric_fudge (string s)
 {
-  char  * chars  =s.get_copy_str0 ();
+  char *chars = string_copy (s);
 
-  for (char * p = chars; *p ; p++)
+  for (char *p = chars; *p ; p++)
     {
       if (*p == '_' && (p == chars || *(p-1) != '\\'))
 	*p = ' ';
     }
   
-  s = String (chars);
+  s = string (chars);
   delete[] chars;
 
-  int i =0;	
-  if ((i=s.index ("\\,")) != -1)   // change "\," to TeX's "\c "
+  ssize i = 0;	
+  if ((i = s.find ("\\,")) != NPOS)   // change "\," to TeX's "\c "
     {
-      * (s.get_str0 () + i + 1) = 'c';
-      s = s.left_string (i+2) + " " + s.right_string (s.length ()-i-2);
+      * (((char*)s.c_str ()) + i + 1) = 'c';
+      s = s.substr (0, i + 2) + " " + s.substr (i - 2);
     }
 
   return s;
@@ -850,87 +916,52 @@ lyric_fudge (String s)
 Convert "NUM/DEN" into a '(NUM . DEN) cons.
 */
 SCM
-scan_fraction (String frac)
+scan_fraction (string frac)
 {
-	int i = frac.index ('/');
-	int l = frac.length ();
-	String left = frac.left_string (i);
-	String right = frac.right_string (l - i - 1);
+	ssize i = frac.find ('/');
+	string left = frac.substr (0, i);
+	string right = frac.substr (i + 1, (frac.length () - i + 1));
 
 	int n = String_convert::dec2int (left);
 	int d = String_convert::dec2int (right);
-	return scm_cons (scm_int2num (n), scm_int2num (d));
+	return scm_cons (scm_from_int (n), scm_from_int (d));
+}
+
+SCM
+lookup_markup_command (string s)
+{
+	SCM proc = ly_lily_module_constant ("lookup-markup-command");
+	return scm_call_1 (proc, scm_makfrom0str (s.c_str ()));
 }
 
-// Breaks for flex 2.5.31
-#if 0
-/* avoid silly flex induced gcc warnings */
-static void yy_push_state (int) {;}
-static void yy_pop_state () {;}
-static int yy_top_state () { return 0; }
+/* Shut up lexer warnings.  */
+#if YY_STACK_USED
 
 static void
-avoid_silly_flex_induced_gcc_warnings ()
+yy_push_state (int)
 {
-	(void)yy_start_stack_ptr;
-	(void)yy_start_stack_depth;
-	(void)yy_start_stack;
-	yy_push_state (0);
-	yy_pop_state ();
-	yy_top_state ();
-	avoid_silly_flex_induced_gcc_warnings ();
 }
-#endif
 
-SCM
-lookup_markup_command (String s)
+static void
+yy_pop_state ()
 {
-	SCM proc = ly_scheme_function ("lookup-markup-command");
-	return scm_call_1 (proc, scm_makfrom0str (s.to_str0 ()));
 }
 
-
-int
-music_function_type (SCM func)
+static int
+yy_top_state ()
 {
-	SCM type= scm_object_property (func, ly_symbol2scm ("music-function-signature-keyword"));
-	if (type == ly_symbol2scm ("scm"))
-	{
-		return MUSIC_FUNCTION_SCM;
-	}
-	else if (type == ly_symbol2scm ("music"))
-	{
-		return MUSIC_FUNCTION_MUSIC;
-	}
-	else if (type == ly_symbol2scm ("scm-music"))
-	{
-		return MUSIC_FUNCTION_SCM_MUSIC;
-	}
-	else if (type == ly_symbol2scm ("scm-scm"))
-	{
-		return MUSIC_FUNCTION_SCM_SCM;
-	}
-	else if (type == ly_symbol2scm ("music-music"))
-	{
-		return MUSIC_FUNCTION_MUSIC_MUSIC;
-	}
-	else if (type == ly_symbol2scm ("scm-music-music"))
-	{
-		return MUSIC_FUNCTION_SCM_MUSIC_MUSIC;
-	}
-	else if (type == ly_symbol2scm ("scm-scm-music"))
-	{
-		return MUSIC_FUNCTION_SCM_SCM_MUSIC;
-	}
-	else if (type == ly_symbol2scm ("noarg"))
-	{
-		return MUSIC_FUNCTION;
-	}
-	else
-		{
-		/* TODO: print location */
-		error ("Can not find signature for music function.");
-		}
+  return 0;
+}
 
-	return MUSIC_FUNCTION_SCM;
+static void
+silence_lexer_warnings ()
+{
+   (void) yy_start_stack_ptr;
+   (void) yy_start_stack_depth;
+   (void) yy_start_stack;
+   (void) yy_push_state;
+   (void) yy_pop_state;
+   (void) yy_top_state;
+   (void) silence_lexer_warnings;
 }
+#endif