X-Git-Url: https://git.saurik.com/bison.git/blobdiff_plain/4f25ebb043fdb762ce4c1a2564cd7056d957ed87..026816664ff8283a55f91915843a8ff0ac5cf86c:/src/scan-gram.l diff --git a/src/scan-gram.l b/src/scan-gram.l index 462be239..555e6952 100644 --- a/src/scan-gram.l +++ b/src/scan-gram.l @@ -1,11 +1,12 @@ /* Bison Grammar Scanner -*- C -*- - Copyright (C) 2002 Free Software Foundation, Inc. + + Copyright (C) 2002-2013 Free Software Foundation, Inc. This file is part of Bison, the GNU Compiler Compiler. - This program is free software; you can redistribute it and/or modify + This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by - the Free Software Foundation; either version 2 of the License, or + the Free Software Foundation, either version 3 of the License, or (at your option) any later version. This program is distributed in the hope that it will be useful, @@ -14,304 +15,568 @@ GNU General Public License for more details. You should have received a copy of the GNU General Public License - along with this program; if not, write to the Free Software - Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA - 02111-1307 USA -*/ + along with this program. If not, see . */ -%option debug nodefault noyywrap nounput never-interactive stack +%option debug nodefault noinput noyywrap never-interactive %option prefix="gram_" outfile="lex.yy.c" %{ -#include "system.h" -#include "complain.h" -#include "quote.h" -#include "getargs.h" -#include "gram.h" -#include "reader.h" - -/* Each time we match a string, move the end cursor to its end. */ -#define YY_USER_INIT \ -do { \ - LOCATION_RESET (*yylloc); \ - /* This is only to avoid GCC warnings. */ \ - if (yycontrol) {;}; \ -} while (0) - -#define YY_USER_ACTION LOCATION_COLUMNS (*yylloc, yyleng) -#define YY_LINES LOCATION_LINES (*yylloc, yyleng); lineno += yyleng; -#define YY_STEP LOCATION_STEP (*yylloc) - - -/* STRING_OBSTACK -- Used to store all the characters that we need to - keep (to construct ID, STRINGS etc.). Use the following macros to - use it. - - Use YY_OBS_GROW to append what has just been matched, and - YY_OBS_FINISH to end the string (it puts the ending 0). - YY_OBS_FINISH also stores this string in LAST_STRING, which can be - used, and which is used by YY_OBS_FREE to free the last string. */ - -static struct obstack string_obstack; -char *last_string; - -#define YY_OBS_GROW \ - obstack_grow (&string_obstack, yytext, yyleng) - -#define YY_OBS_FINISH \ - do { \ - obstack_1grow (&string_obstack, '\0'); \ - last_string = obstack_finish (&string_obstack); \ +/* Work around a bug in flex 2.5.31. See Debian bug 333231 + . */ +#undef gram_wrap +#define gram_wrap() 1 + +#define FLEX_PREFIX(Id) gram_ ## Id +#include + +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#include + +#include + +#define YY_DECL GRAM_LEX_DECL + +#define YY_USER_INIT \ + code_start = scanner_cursor = loc->start; \ + +/* Location of scanner cursor. */ +static boundary scanner_cursor; + +#define YY_USER_ACTION location_compute (loc, &scanner_cursor, yytext, yyleng); + +static size_t no_cr_read (FILE *, char *, size_t); +#define YY_INPUT(buf, result, size) ((result) = no_cr_read (yyin, buf, size)) + +#define RETURN_PERCENT_PARAM(Value) \ + RETURN_VALUE(PERCENT_PARAM, param, param_ ## Value) + +#define RETURN_PERCENT_FLAG(Value) \ + RETURN_VALUE(PERCENT_FLAG, uniqstr, uniqstr_new (Value)) + +#define RETURN_VALUE(Token, Field, Value) \ + do { \ + val->Field = Value; \ + return Token; \ + } while (0) + +#define ROLLBACK_CURRENT_TOKEN \ + do { \ + scanner_cursor.column -= mbsnwidth (yytext, yyleng, 0); \ + yyless (0); \ } while (0) -#define YY_OBS_FREE \ - do { \ - obstack_free (&string_obstack, last_string); \ +#define DEPRECATED(Msg) \ + do { \ + size_t i; \ + deprecated_directive (loc, yytext, Msg); \ + scanner_cursor.column -= mbsnwidth (Msg, strlen (Msg), 0); \ + for (i = strlen (Msg); i != 0; --i) \ + unput (Msg[i - 1]); \ } while (0) +/* A string representing the most recently saved token. */ +static char *last_string; + +/* Bracketed identifier. */ +static uniqstr bracketed_id_str = 0; +static location bracketed_id_loc; +static boundary bracketed_id_start; +static int bracketed_id_context_state = 0; + void -scanner_last_string_free (void) +gram_scanner_last_string_free (void) { - YY_OBS_FREE; + STRING_FREE; } - -static int braces_level = 0; -static int percent_percent_count = 0; - -static void handle_dollar PARAMS ((braced_code_t code_kind, - char *cp, location_t location)); -static void handle_at PARAMS ((braced_code_t code_kind, - char *cp, location_t location)); +static void handle_syncline (char *, location); +static unsigned long int scan_integer (char const *p, int base, location loc); +static int convert_ucn_to_byte (char const *hex_text); +static void unexpected_eof (boundary, char const *); +static void unexpected_newline (boundary, char const *); %} -%x SC_COMMENT -%x SC_STRING SC_CHARACTER + /* A C-like comment in directives/rules. */ +%x SC_YACC_COMMENT + /* Strings and characters in directives/rules. */ %x SC_ESCAPED_STRING SC_ESCAPED_CHARACTER -%x SC_BRACED_CODE SC_PROLOGUE SC_EPILOGUE + /* A identifier was just read in directives/rules. Special state + to capture the sequence 'identifier :'. */ +%x SC_AFTER_IDENTIFIER + + /* POSIX says that a tag must be both an id and a C union member, but + historically almost any character is allowed in a tag. We + disallow NUL, as this simplifies our implementation. We match + angle brackets in nested pairs: several languages use them for + generics/template types. */ +%x SC_TAG + + /* Four types of user code: + - prologue (code between '%{' '%}' in the first section, before %%); + - actions, printers, union, etc, (between braced in the middle section); + - epilogue (everything after the second %%). + - predicate (code between '%?{' and '{' in middle section); */ +%x SC_PROLOGUE SC_BRACED_CODE SC_EPILOGUE SC_PREDICATE + /* C and C++ comments in code. */ +%x SC_COMMENT SC_LINE_COMMENT + /* Strings and characters in code. */ +%x SC_STRING SC_CHARACTER + /* Bracketed identifiers support. */ +%x SC_BRACKETED_ID SC_RETURN_BRACKETED_ID + +letter [.abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ_] +notletter [^.abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ_]{-}[%\{] +id {letter}({letter}|[-0-9])* +int [0-9]+ -id [.a-zA-Z][.a-zA-Z_0-9]* -int [0-9]+ -eols (\n|\r|\n\r|\r\n)+ -blanks [ \t\f]+ +/* Zero or more instances of backslash-newline. Following GCC, allow + white space between the backslash and the newline. */ +splice (\\[ \f\t\v]*\n)* + +/* An equal sign, with optional leading whitespaces. This is used in some + deprecated constructs. */ +eqopt ([[:space:]]*=)? %% %{ - /* At each yylex invocation, mark the current position as the - start of the next token. */ -#define TR_POS 0 -#if TR_POS - fprintf (stderr, "FOO1: %p: ", yylloc); - LOCATION_PRINT (stderr, *yylloc); - fprintf (stderr, "\n"); -#endif - YY_STEP; -#if TR_POS - fprintf (stderr, "BAR1: "); - LOCATION_PRINT (stderr, *yylloc); - fprintf (stderr, "\n"); -#endif + /* Nesting level. Either for nested braces, or nested angle brackets + (but not mixed). */ + int nesting PACIFY_CC (= 0); + + /* Parent context state, when applicable. */ + int context_state PACIFY_CC (= 0); + + /* Location of most recent identifier, when applicable. */ + location id_loc PACIFY_CC (= empty_location); + + /* Where containing code started, when applicable. Its initial + value is relevant only when yylex is invoked in the SC_EPILOGUE + start condition. */ + boundary code_start = scanner_cursor; + + /* Where containing comment or string or character literal started, + when applicable. */ + boundary token_start PACIFY_CC (= scanner_cursor); %} + /*-----------------------. + | Scanning white space. | + `-----------------------*/ + + +{ + /* Comments and white space. */ + "," { + complain (loc, Wother, _("stray ',' treated as white space")); + } + [ \f\n\t\v] | + "//".* continue; + "/*" { + token_start = loc->start; + context_state = YY_START; + BEGIN SC_YACC_COMMENT; + } + + /* #line directives are not documented, and may be withdrawn or + modified in future versions of Bison. */ + ^"#line "{int}(" \"".*"\"")?"\n" { + handle_syncline (yytext + sizeof "#line " - 1, *loc); + } +} + + /*----------------------------. | Scanning Bison directives. | `----------------------------*/ + + /* For directives that are also command line options, the regex must be + "%..." + after "[-_]"s are removed, and the directive must match the --long + option name, with a single string argument. Otherwise, add exceptions + to ../build-aux/cross-options.pl. */ + { - "%binary" return PERCENT_NONASSOC; - "%debug" return PERCENT_DEBUG; - "%define" return PERCENT_DEFINE; - "%defines" return PERCENT_DEFINES; - "%destructor" return PERCENT_DESTRUCTOR; - "%error"[-_]"verbose" return PERCENT_ERROR_VERBOSE; - "%expect" return PERCENT_EXPECT; - "%file-prefix" return PERCENT_FILE_PREFIX; - "%fixed"[-_]"output"[-_]"files" return PERCENT_YACC; - "%left" return PERCENT_LEFT; - "%locations" return PERCENT_LOCATIONS; - "%name"[-_]"prefix" return PERCENT_NAME_PREFIX; - "%no"[-_]"lines" return PERCENT_NO_LINES; - "%nonassoc" return PERCENT_NONASSOC; - "%nterm" return PERCENT_NTERM; - "%output" return PERCENT_OUTPUT; - "%prec" return PERCENT_PREC; - "%pure"[-_]"parser" return PERCENT_PURE_PARSER; - "%right" return PERCENT_RIGHT; - "%skeleton" return PERCENT_SKELETON; - "%start" return PERCENT_START; - "%term" return PERCENT_TOKEN; - "%token" return PERCENT_TOKEN; - "%token"[-_]"table" return PERCENT_TOKEN_TABLE; - "%type" return PERCENT_TYPE; - "%union" return PERCENT_UNION; - "%verbose" return PERCENT_VERBOSE; - "%yacc" return PERCENT_YACC; + "%binary" return PERCENT_NONASSOC; + "%code" return PERCENT_CODE; + "%debug" RETURN_PERCENT_FLAG("parse.trace"); + "%default-prec" return PERCENT_DEFAULT_PREC; + "%define" return PERCENT_DEFINE; + "%defines" return PERCENT_DEFINES; + "%destructor" return PERCENT_DESTRUCTOR; + "%dprec" return PERCENT_DPREC; + "%empty" return PERCENT_EMPTY; + "%error-verbose" return PERCENT_ERROR_VERBOSE; + "%expect" return PERCENT_EXPECT; + "%expect-rr" return PERCENT_EXPECT_RR; + "%file-prefix" return PERCENT_FILE_PREFIX; + "%fixed-output-files" return PERCENT_YACC; + "%initial-action" return PERCENT_INITIAL_ACTION; + "%glr-parser" return PERCENT_GLR_PARSER; + "%language" return PERCENT_LANGUAGE; + "%left" return PERCENT_LEFT; + "%lex-param" RETURN_PERCENT_PARAM(lex); + "%locations" RETURN_PERCENT_FLAG("locations"); + "%merge" return PERCENT_MERGE; + "%name-prefix" return PERCENT_NAME_PREFIX; + "%no-default-prec" return PERCENT_NO_DEFAULT_PREC; + "%no-lines" return PERCENT_NO_LINES; + "%nonassoc" return PERCENT_NONASSOC; + "%nondeterministic-parser" return PERCENT_NONDETERMINISTIC_PARSER; + "%nterm" return PERCENT_NTERM; + "%output" return PERCENT_OUTPUT; + "%param" RETURN_PERCENT_PARAM(both); + "%parse-param" RETURN_PERCENT_PARAM(parse); + "%prec" return PERCENT_PREC; + "%precedence" return PERCENT_PRECEDENCE; + "%printer" return PERCENT_PRINTER; + "%pure-parser" RETURN_PERCENT_FLAG("api.pure"); + "%require" return PERCENT_REQUIRE; + "%right" return PERCENT_RIGHT; + "%skeleton" return PERCENT_SKELETON; + "%start" return PERCENT_START; + "%term" return PERCENT_TOKEN; + "%token" return PERCENT_TOKEN; + "%token-table" return PERCENT_TOKEN_TABLE; + "%type" return PERCENT_TYPE; + "%union" return PERCENT_UNION; + "%verbose" return PERCENT_VERBOSE; + "%yacc" return PERCENT_YACC; + + /* deprecated */ + "%default"[-_]"prec" DEPRECATED("%default-prec"); + "%error"[-_]"verbose" DEPRECATED("%define parse.error verbose"); + "%expect"[-_]"rr" DEPRECATED("%expect-rr"); + "%file-prefix"{eqopt} DEPRECATED("%file-prefix"); + "%fixed"[-_]"output"[-_]"files" DEPRECATED("%fixed-output-files"); + "%name"[-_]"prefix"{eqopt} DEPRECATED("%name-prefix"); + "%no"[-_]"default"[-_]"prec" DEPRECATED("%no-default-prec"); + "%no"[-_]"lines" DEPRECATED("%no-lines"); + "%output"{eqopt} DEPRECATED("%output"); + "%pure"[-_]"parser" DEPRECATED("%pure-parser"); + "%token"[-_]"table" DEPRECATED("%token-table"); + + "%"{id} { + complain (loc, complaint, _("invalid directive: %s"), quote (yytext)); + } "=" return EQUAL; - ":" return COLON; "|" return PIPE; ";" return SEMICOLON; - {eols} YY_LINES; YY_STEP; - {blanks} YY_STEP; - {id} { - yylval->symbol = getsym (yytext, *yylloc); - return ID; + {id} { + val->uniqstr = uniqstr_new (yytext); + id_loc = *loc; + bracketed_id_str = NULL; + BEGIN SC_AFTER_IDENTIFIER; } - {int} yylval->integer = strtol (yytext, 0, 10); return INT; + {int} { + val->integer = scan_integer (yytext, 10, *loc); + return INT; + } + 0[xX][0-9abcdefABCDEF]+ { + val->integer = scan_integer (yytext, 16, *loc); + return INT; + } - /* Characters. We don't check there is only one. */ - \' YY_OBS_GROW; yy_push_state (SC_ESCAPED_CHARACTER); + /* Identifiers may not start with a digit. Yet, don't silently + accept "1FOO" as "1 FOO". */ + {int}{id} { + complain (loc, complaint, _("invalid identifier: %s"), quote (yytext)); + } - /* Strings. */ - \" YY_OBS_GROW; yy_push_state (SC_ESCAPED_STRING); + /* Characters. */ + "'" token_start = loc->start; BEGIN SC_ESCAPED_CHARACTER; - /* Comments. */ - "/*" yy_push_state (SC_COMMENT); - "//".* YY_STEP; + /* Strings. */ + "\"" token_start = loc->start; BEGIN SC_ESCAPED_STRING; /* Prologue. */ - "%{" yy_push_state (SC_PROLOGUE); + "%{" code_start = loc->start; BEGIN SC_PROLOGUE; /* Code in between braces. */ - "{" YY_OBS_GROW; ++braces_level; yy_push_state (SC_BRACED_CODE); + "{" { + STRING_GROW; + nesting = 0; + code_start = loc->start; + BEGIN SC_BRACED_CODE; + } - /* A type. */ - "<"[^>]+">" { - obstack_grow (&string_obstack, yytext + 1, yyleng - 2); - YY_OBS_FINISH; - yylval->string = last_string; - return TYPE; + /* Semantic predicate. */ + "%?"[ \f\n\t\v]*"{" { + nesting = 0; + code_start = loc->start; + BEGIN SC_PREDICATE; } + /* A type. */ + "<*>" return TAG_ANY; + "<>" return TAG_NONE; + "<" { + nesting = 0; + token_start = loc->start; + BEGIN SC_TAG; + } - "%%" { + "%%" { + static int percent_percent_count; if (++percent_percent_count == 2) - yy_push_state (SC_EPILOGUE); + BEGIN SC_EPILOGUE; return PERCENT_PERCENT; } - . { - LOCATION_PRINT (stderr, *yylloc); - fprintf (stderr, ": invalid character: `%c'\n", *yytext); - YY_STEP; + "[" { + bracketed_id_str = NULL; + bracketed_id_start = loc->start; + bracketed_id_context_state = YY_START; + BEGIN SC_BRACKETED_ID; + } + + [^\[%A-Za-z0-9_<>{}\"\'*;|=/, \f\n\t\v]+|. { + complain (loc, complaint, "%s: %s", + ngettext ("invalid character", "invalid characters", yyleng), + quote_mem (yytext, yyleng)); + } + + <> { + loc->start = loc->end = scanner_cursor; + yyterminate (); } } - /*------------------------------------------------------------. - | Whatever the start condition (but those which correspond to | - | entity `swallowed' by Bison: SC_ESCAPED_STRING and | - | SC_ESCAPED_CHARACTER), no M4 character must escape as is. | - `------------------------------------------------------------*/ + /*--------------------------------------------------------------. + | Supporting \0 complexifies our implementation for no expected | + | added value. | + `--------------------------------------------------------------*/ - + { - \[ if (YY_START != SC_COMMENT) obstack_sgrow (&string_obstack, "@<:@"); - \] if (YY_START != SC_COMMENT) obstack_sgrow (&string_obstack, "@:>@"); + \0 complain (loc, complaint, _("invalid null character")); } + /*-----------------------------------------------------------------. + | Scanning after an identifier, checking whether a colon is next. | + `-----------------------------------------------------------------*/ - /*-----------------------------------------------------------. - | Scanning a C comment. The initial `/ *' is already eaten. | - `-----------------------------------------------------------*/ + +{ + "[" { + if (bracketed_id_str) + { + ROLLBACK_CURRENT_TOKEN; + BEGIN SC_RETURN_BRACKETED_ID; + *loc = id_loc; + return ID; + } + else + { + bracketed_id_start = loc->start; + bracketed_id_context_state = YY_START; + BEGIN SC_BRACKETED_ID; + } + } + ":" { + BEGIN (bracketed_id_str ? SC_RETURN_BRACKETED_ID : INITIAL); + *loc = id_loc; + return ID_COLON; + } + . { + ROLLBACK_CURRENT_TOKEN; + BEGIN (bracketed_id_str ? SC_RETURN_BRACKETED_ID : INITIAL); + *loc = id_loc; + return ID; + } + <> { + BEGIN (bracketed_id_str ? SC_RETURN_BRACKETED_ID : INITIAL); + *loc = id_loc; + return ID; + } +} - + /*--------------------------------. + | Scanning bracketed identifiers. | + `--------------------------------*/ + + { - "*/" { /* End of the comment. */ - if (yy_top_state () == INITIAL) + {id} { + if (bracketed_id_str) { - YY_STEP; + complain (loc, complaint, + _("unexpected identifier in bracketed name: %s"), + quote (yytext)); } else { - YY_OBS_GROW; + bracketed_id_str = uniqstr_new (yytext); + bracketed_id_loc = *loc; + } + } + "]" { + BEGIN bracketed_id_context_state; + if (bracketed_id_str) + { + if (INITIAL == bracketed_id_context_state) + { + val->uniqstr = bracketed_id_str; + bracketed_id_str = 0; + *loc = bracketed_id_loc; + return BRACKETED_ID; + } } - yy_pop_state (); + else + complain (loc, complaint, _("an identifier expected")); } - [^\[\]*\n\r]+ if (yy_top_state () != INITIAL) YY_OBS_GROW; - {eols} if (yy_top_state () != INITIAL) YY_OBS_GROW; YY_LINES; - . /* Stray `*'. */if (yy_top_state () != INITIAL) YY_OBS_GROW; + [^\].A-Za-z0-9_/ \f\n\t\v]+|. { + complain (loc, complaint, "%s: %s", + ngettext ("invalid character in bracketed name", + "invalid characters in bracketed name", yyleng), + quote_mem (yytext, yyleng)); + } <> { - LOCATION_PRINT (stderr, *yylloc); - fprintf (stderr, ": unexpected end of file in a comment\n"); - yy_pop_state (); + BEGIN bracketed_id_context_state; + unexpected_eof (bracketed_id_start, "]"); } } + +{ + . { + ROLLBACK_CURRENT_TOKEN; + val->uniqstr = bracketed_id_str; + bracketed_id_str = 0; + *loc = bracketed_id_loc; + BEGIN INITIAL; + return BRACKETED_ID; + } +} - /*----------------------------------------------------------------. - | Scanning a C string, including its escapes. The initial `"' is | - | already eaten. | - `----------------------------------------------------------------*/ - + /*---------------------------------------------------------------. + | Scanning a Yacc comment. The initial '/ *' is already eaten. | + `---------------------------------------------------------------*/ + + { - \" { - assert (yy_top_state () == INITIAL); - YY_OBS_GROW; - YY_OBS_FINISH; - yylval->string = last_string; - yy_pop_state (); - return STRING; - } + "*/" BEGIN context_state; + .|\n continue; + <> unexpected_eof (token_start, "*/"); BEGIN context_state; +} + - [^\"\n\r\\]+ YY_OBS_GROW; + /*------------------------------------------------------------. + | Scanning a C comment. The initial '/ *' is already eaten. | + `------------------------------------------------------------*/ - {eols} obstack_1grow (&string_obstack, '\n'); YY_LINES; + +{ + "*"{splice}"/" STRING_GROW; BEGIN context_state; + <> unexpected_eof (token_start, "*/"); BEGIN context_state; +} - <> { - LOCATION_PRINT (stderr, *yylloc); - fprintf (stderr, ": unexpected end of file in a string\n"); - assert (yy_top_state () == INITIAL); - YY_OBS_FINISH; - yylval->string = last_string; - yy_pop_state (); + + /*--------------------------------------------------------------. + | Scanning a line comment. The initial '//' is already eaten. | + `--------------------------------------------------------------*/ + + +{ + "\n" STRING_GROW; BEGIN context_state; + {splice} STRING_GROW; + <> BEGIN context_state; +} + + + /*------------------------------------------------. + | Scanning a Bison string, including its escapes. | + | The initial quote is already eaten. | + `------------------------------------------------*/ + + +{ + "\"" { + STRING_FINISH; + loc->start = token_start; + val->code = last_string; + BEGIN INITIAL; return STRING; } + <> unexpected_eof (token_start, "\""); + "\n" unexpected_newline (token_start, "\""); } - /*---------------------------------------------------------------. - | Scanning a C character, decoding its escapes. The initial "'" | - | is already eaten. | - `---------------------------------------------------------------*/ + /*----------------------------------------------------------. + | Scanning a Bison character literal, decoding its escapes. | + | The initial quote is already eaten. | + `----------------------------------------------------------*/ { - \' { - YY_OBS_GROW; - assert (yy_top_state () == INITIAL); + "'" { + STRING_FINISH; + loc->start = token_start; + val->character = last_string[0]; + + /* FIXME: Eventually, make these errors. */ + if (last_string[0] == '\0') { - YY_OBS_FINISH; - yylval->symbol = getsym (last_string, *yylloc); - symbol_class_set (yylval->symbol, token_sym); - symbol_user_token_number_set (yylval->symbol, last_string[1]); - YY_OBS_FREE; - yy_pop_state (); - return ID; + complain (loc, Wother, _("empty character literal")); + /* '\0' seems dangerous even if we are about to complain. */ + val->character = '\''; } + else if (last_string[1] != '\0') + complain (loc, Wother, + _("extra characters in character literal")); + STRING_FREE; + BEGIN INITIAL; + return CHAR; } + "\n" unexpected_newline (token_start, "'"); + <> unexpected_eof (token_start, "'"); +} - [^\'\n\r\\] YY_OBS_GROW; - {eols} obstack_1grow (&string_obstack, '\n'); YY_LINES; - <> { - LOCATION_PRINT (stderr, *yylloc); - fprintf (stderr, ": unexpected end of file in a character\n"); - assert (yy_top_state () == INITIAL); - YY_OBS_FINISH; - yylval->string = last_string; - yy_pop_state (); - return CHARACTER; + /*--------------------------------------------------------------. + | Scanning a tag. The initial angle bracket is already eaten. | + `--------------------------------------------------------------*/ + + +{ + ">" { + --nesting; + if (nesting < 0) + { + STRING_FINISH; + loc->start = token_start; + val->uniqstr = uniqstr_new (last_string); + STRING_FREE; + BEGIN INITIAL; + return TAG; + } + STRING_GROW; } -} + ([^<>]|->)+ STRING_GROW; + "<"+ STRING_GROW; nesting += yyleng; + + <> unexpected_eof (token_start, ">"); +} /*----------------------------. | Decode escaped characters. | @@ -319,95 +584,77 @@ blanks [ \t\f]+ { - \\[0-7]{3} { - long c = strtol (yytext + 1, 0, 8); - if (c > 255) - { - LOCATION_PRINT (stderr, *yylloc); - fprintf (stderr, ": invalid escape: %s\n", quote (yytext)); - YY_STEP; - } + \\[0-7]{1,3} { + unsigned long int c = strtoul (yytext + 1, NULL, 8); + if (!c || UCHAR_MAX < c) + complain (loc, complaint, _("invalid number after \\-escape: %s"), + yytext+1); else - obstack_1grow (&string_obstack, c); + obstack_1grow (&obstack_for_string, c); } - \\x[0-9a-fA-F]{2} { - obstack_1grow (&string_obstack, strtol (yytext + 2, 0, 16)); + \\x[0-9abcdefABCDEF]+ { + verify (UCHAR_MAX < ULONG_MAX); + unsigned long int c = strtoul (yytext + 2, NULL, 16); + if (!c || UCHAR_MAX < c) + complain (loc, complaint, _("invalid number after \\-escape: %s"), + yytext+1); + else + obstack_1grow (&obstack_for_string, c); } - \\a obstack_1grow (&string_obstack, '\a'); - \\b obstack_1grow (&string_obstack, '\b'); - \\f obstack_1grow (&string_obstack, '\f'); - \\n obstack_1grow (&string_obstack, '\n'); - \\r obstack_1grow (&string_obstack, '\r'); - \\t obstack_1grow (&string_obstack, '\t'); - \\v obstack_1grow (&string_obstack, '\v'); - \\[\\""] obstack_1grow (&string_obstack, yytext[1]); - \\(.|\n) { - LOCATION_PRINT (stderr, *yylloc); - fprintf (stderr, ": unrecognized escape: %s\n", quote (yytext)); - YY_OBS_GROW; + \\a obstack_1grow (&obstack_for_string, '\a'); + \\b obstack_1grow (&obstack_for_string, '\b'); + \\f obstack_1grow (&obstack_for_string, '\f'); + \\n obstack_1grow (&obstack_for_string, '\n'); + \\r obstack_1grow (&obstack_for_string, '\r'); + \\t obstack_1grow (&obstack_for_string, '\t'); + \\v obstack_1grow (&obstack_for_string, '\v'); + + /* \\[\"\'?\\] would be shorter, but it confuses xgettext. */ + \\("\""|"'"|"?"|"\\") obstack_1grow (&obstack_for_string, yytext[1]); + + \\(u|U[0-9abcdefABCDEF]{4})[0-9abcdefABCDEF]{4} { + int c = convert_ucn_to_byte (yytext); + if (c <= 0) + complain (loc, complaint, _("invalid number after \\-escape: %s"), + yytext+1); + else + obstack_1grow (&obstack_for_string, c); + } + \\(.|\n) { + char const *p = yytext + 1; + /* Quote only if escaping won't make the character visible. */ + if (c_isspace ((unsigned char) *p) && c_isprint ((unsigned char) *p)) + p = quote (p); + else + p = quotearg_style_mem (escape_quoting_style, p, 1); + complain (loc, complaint, _("invalid character after \\-escape: %s"), + p); } - /* FLex wants this rule, in case of a `\<>'. */ - \\ YY_OBS_GROW; } + /*--------------------------------------------. + | Scanning user-code characters and strings. | + `--------------------------------------------*/ - /*----------------------------------------------------------. - | Scanning a C character without decoding its escapes. The | - | initial "'" is already eaten. | - `----------------------------------------------------------*/ + +{ + {splice}|\\{splice}[^\n\[\]] STRING_GROW; +} { - \' { - YY_OBS_GROW; - assert (yy_top_state () != INITIAL); - yy_pop_state (); - } - - [^\[\]\'\n\r\\]+ YY_OBS_GROW; - \\(.|\n) YY_OBS_GROW; - /* FLex wants this rule, in case of a `\<>'. */ - \\ YY_OBS_GROW; - - {eols} YY_OBS_GROW; YY_LINES; - - <> { - LOCATION_PRINT (stderr, *yylloc); - fprintf (stderr, ": unexpected end of file in a character\n"); - assert (yy_top_state () != INITIAL); - yy_pop_state (); - } + "'" STRING_GROW; BEGIN context_state; + \n unexpected_newline (token_start, "'"); + <> unexpected_eof (token_start, "'"); } - - /*----------------------------------------------------------------. - | Scanning a C string, without decoding its escapes. The initial | - | `"' is already eaten. | - `----------------------------------------------------------------*/ - { - \" { - assert (yy_top_state () != INITIAL); - YY_OBS_GROW; - yy_pop_state (); - } - - [^\[\]\"\n\r\\]+ YY_OBS_GROW; - \\(.|\n) YY_OBS_GROW; - /* FLex wants this rule, in case of a `\<>'. */ - \\ YY_OBS_GROW; - - {eols} YY_OBS_GROW; YY_LINES; - - <> { - LOCATION_PRINT (stderr, *yylloc); - fprintf (stderr, ": unexpected end of file in a string\n"); - assert (yy_top_state () != INITIAL); - yy_pop_state (); - } + "\"" STRING_GROW; BEGIN context_state; + \n unexpected_newline (token_start, "\""); + <> unexpected_eof (token_start, "\""); } @@ -415,66 +662,86 @@ blanks [ \t\f]+ | Strings, comments etc. can be found in user code. | `---------------------------------------------------*/ - + { - /* Characters. We don't check there is only one. */ - \' YY_OBS_GROW; yy_push_state (SC_CHARACTER); + "'" { + STRING_GROW; + context_state = YY_START; + token_start = loc->start; + BEGIN SC_CHARACTER; + } + "\"" { + STRING_GROW; + context_state = YY_START; + token_start = loc->start; + BEGIN SC_STRING; + } + "/"{splice}"*" { + STRING_GROW; + context_state = YY_START; + token_start = loc->start; + BEGIN SC_COMMENT; + } + "/"{splice}"/" { + STRING_GROW; + context_state = YY_START; + BEGIN SC_LINE_COMMENT; + } +} - /* Strings. */ - \" YY_OBS_GROW; yy_push_state (SC_STRING); - /* Comments. */ - "/*" YY_OBS_GROW; yy_push_state (SC_COMMENT); - "//".* YY_OBS_GROW; - /* Not comments. */ - "/" YY_OBS_GROW; -} + /*-----------------------------------------------------------. + | Scanning some code in braces (actions, predicates). The | + | initial "{" is already eaten. | + `-----------------------------------------------------------*/ + +{ + "{"|"<"{splice}"%" STRING_GROW; nesting++; + "%"{splice}">" STRING_GROW; nesting--; - /*---------------------------------------------------------------. - | Scanning some code in braces (%union and actions). The initial | - | "{" is already eaten. | - `---------------------------------------------------------------*/ + /* Tokenize '<<%' correctly (as '<<' '%') rather than incorrrectly + (as '<' '<%'). */ + "<"{splice}"<" STRING_GROW; + + <> unexpected_eof (code_start, "}"); +} { "}" { - YY_OBS_GROW; - if (--braces_level == 0) + obstack_1grow (&obstack_for_string, '}'); + + --nesting; + if (nesting < 0) { - yy_pop_state (); - YY_OBS_FINISH; - yylval->string = last_string; - return BRACED_CODE; + STRING_FINISH; + loc->start = code_start; + val->code = last_string; + BEGIN INITIAL; + return BRACED_CODE; } } +} - "{" YY_OBS_GROW; braces_level++; - - "$"("<"[^>]+">")?(-?[0-9]+|"$") { handle_dollar (current_braced_code, - yytext, *yylloc); } - "@"(-?[0-9]+|"$") { handle_at (current_braced_code, - yytext, *yylloc); } - - [^$@\[\]/\'\"\{\}\n\r]+ YY_OBS_GROW; - {eols} YY_OBS_GROW; YY_LINES; - - /* A lose $, or /, or etc. */ - . YY_OBS_GROW; - - <> { - LOCATION_PRINT (stderr, *yylloc); - fprintf (stderr, ": unexpected end of file in a braced code\n"); - yy_pop_state (); - YY_OBS_FINISH; - yylval->string = last_string; - return BRACED_CODE; + +{ + "}" { + --nesting; + if (nesting < 0) + { + STRING_FINISH; + loc->start = code_start; + val->code = last_string; + BEGIN INITIAL; + return BRACED_PREDICATE; + } + else + obstack_1grow (&obstack_for_string, '}'); } - } - /*--------------------------------------------------------------. | Scanning some prologue: from "%{" (already scanned) to "%}". | `--------------------------------------------------------------*/ @@ -482,241 +749,244 @@ blanks [ \t\f]+ { "%}" { - yy_pop_state (); - YY_OBS_FINISH; - yylval->string = last_string; + STRING_FINISH; + loc->start = code_start; + val->code = last_string; + BEGIN INITIAL; return PROLOGUE; } - [^%\[\]/\'\"\n\r]+ YY_OBS_GROW; - "%" YY_OBS_GROW; - {eols} YY_OBS_GROW; YY_LINES; - - <> { - LOCATION_PRINT (stderr, *yylloc); - fprintf (stderr, ": unexpected end of file in a prologue\n"); - yy_pop_state (); - YY_OBS_FINISH; - yylval->string = last_string; - return PROLOGUE; - } + <> unexpected_eof (code_start, "%}"); } /*---------------------------------------------------------------. | Scanning the epilogue (everything after the second "%%", which | - | has already been eaten. | + | has already been eaten). | `---------------------------------------------------------------*/ { - ([^\[\]]|{eols})+ YY_OBS_GROW; - <> { - yy_pop_state (); - YY_OBS_FINISH; - yylval->string = last_string; + STRING_FINISH; + loc->start = code_start; + val->code = last_string; + BEGIN INITIAL; return EPILOGUE; } } -%% + /*-----------------------------------------------------. + | By default, grow the string obstack with the input. | + `-----------------------------------------------------*/ -/*------------------------------------------------------------------. -| CP is pointing to a wannabee semantic value (i.e., a `$'). | -| | -| Possible inputs: $[]($|integer) | -| | -| Output to the STRING_OBSTACK a reference to this semantic value. | -`------------------------------------------------------------------*/ +. | + \n STRING_GROW; -static inline void -handle_action_dollar (char *cp, location_t location) -{ - const char *type_name = NULL; +%% - ++cp; +/* Read bytes from FP into buffer BUF of size SIZE. Return the + number of bytes read. Remove '\r' from input, treating \r\n + and isolated \r as \n. */ - /* Get the type name if explicit. */ - if (*cp == '<') +static size_t +no_cr_read (FILE *fp, char *buf, size_t size) +{ + size_t bytes_read = fread (buf, 1, size, fp); + if (bytes_read) { - type_name = ++cp; - while (*cp != '>') - ++cp; - *cp = '\0'; - ++cp; + char *w = memchr (buf, '\r', bytes_read); + if (w) + { + char const *r = ++w; + char const *lim = buf + bytes_read; + + for (;;) + { + /* Found an '\r'. Treat it like '\n', but ignore any + '\n' that immediately follows. */ + w[-1] = '\n'; + if (r == lim) + { + int ch = getc (fp); + if (ch != '\n' && ungetc (ch, fp) != ch) + break; + } + else if (*r == '\n') + r++; + + /* Copy until the next '\r'. */ + do + { + if (r == lim) + return w - buf; + } + while ((*w++ = *r++) != '\r'); + } + + return w - buf; + } } - if (*cp == '$') - { - if (!type_name) - type_name = symbol_list_n_type_name_get (current_rule, location, 0); - if (!type_name && typed) - complain_at (location, _("$$ of `%s' has no declared type"), - symbol_tag_get (current_rule->sym)); - if (!type_name) - type_name = ""; - obstack_fgrow1 (&string_obstack, - "]b4_lhs_value([%s])[", type_name); - } - else if (isdigit (*cp) || *cp == '-') - { - /* RULE_LENGTH is the number of values in the current rule so - far, which says where to find `$0' with respect to the top of - the stack. It is not the same as the rule->length in the - case of mid rule actions. */ - int rule_length = symbol_list_length (current_rule->next); - int n = strtol (cp, &cp, 10); - - if (n > rule_length) - complain_at (location, _("invalid value: %s%d"), "$", n); - else - { - if (!type_name && n > 0) - type_name = symbol_list_n_type_name_get (current_rule, location, - n); - if (!type_name && typed) - complain_at (location, _("$%d of `%s' has no declared type"), - n, symbol_tag_get (current_rule->sym)); - if (!type_name) - type_name = ""; - obstack_fgrow3 (&string_obstack, - "]b4_rhs_value([%d], [%d], [%s])[", - rule_length, n, type_name); - } - } - else - { - char buf[] = "$c"; - buf[1] = *cp; - complain_at (location, _("%s is invalid"), quote (buf)); - } + return bytes_read; } -/*---------------------------------------. -| CP is pointing to $$ in a destructor. | -`---------------------------------------*/ -static inline void -handle_destructor_dollar (char *cp, location_t location) +/*------------------------------------------------------. +| Scan NUMBER for a base-BASE integer at location LOC. | +`------------------------------------------------------*/ + +static unsigned long int +scan_integer (char const *number, int base, location loc) { - ++cp; - if (*cp == '$') - { - obstack_sgrow (&string_obstack, "]b4_dollar_dollar["); - } - else + verify (INT_MAX < ULONG_MAX); + unsigned long int num = strtoul (number, NULL, base); + + if (INT_MAX < num) { - char buf[] = "$c"; - buf[1] = *cp; - complain_at (location, _("%s is invalid"), quote (buf)); + complain (&loc, complaint, _("integer out of range: %s"), + quote (number)); + num = INT_MAX; } + + return num; } -/*-----------------------------------------------------------------. -| Dispatch onto handle_action_dollar, or handle_destructor_dollar, | -| depending upon CODE_KIND. | -`-----------------------------------------------------------------*/ +/*------------------------------------------------------------------. +| Convert universal character name UCN to a single-byte character, | +| and return that character. Return -1 if UCN does not correspond | +| to a single-byte character. | +`------------------------------------------------------------------*/ -static void -handle_dollar (braced_code_t braced_code_kind, - char *text, location_t location) +static int +convert_ucn_to_byte (char const *ucn) { - switch (braced_code_kind) - { - case action_braced_code: - handle_action_dollar (text, location); - break; + verify (UCHAR_MAX <= INT_MAX); + unsigned long int code = strtoul (ucn + 2, NULL, 16); + + /* FIXME: Currently we assume Unicode-compatible unibyte characters + on ASCII hosts (i.e., Latin-1 on hosts with 8-bit bytes). On + non-ASCII hosts we support only the portable C character set. + These limitations should be removed once we add support for + multibyte characters. */ + + if (UCHAR_MAX < code) + return -1; + +#if ! ('$' == 0x24 && '@' == 0x40 && '`' == 0x60 && '~' == 0x7e) + { + /* A non-ASCII host. Use CODE to index into a table of the C + basic execution character set, which is guaranteed to exist on + all Standard C platforms. This table also includes '$', '@', + and '`', which are not in the basic execution character set but + which are unibyte characters on all the platforms that we know + about. */ + static signed char const table[] = + { + '\0', -1, -1, -1, -1, -1, -1, '\a', + '\b', '\t', '\n', '\v', '\f', '\r', -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, + -1, -1, -1, -1, -1, -1, -1, -1, + ' ', '!', '"', '#', '$', '%', '&', '\'', + '(', ')', '*', '+', ',', '-', '.', '/', + '0', '1', '2', '3', '4', '5', '6', '7', + '8', '9', ':', ';', '<', '=', '>', '?', + '@', 'A', 'B', 'C', 'D', 'E', 'F', 'G', + 'H', 'I', 'J', 'K', 'L', 'M', 'N', 'O', + 'P', 'Q', 'R', 'S', 'T', 'U', 'V', 'W', + 'X', 'Y', 'Z', '[', '\\', ']', '^', '_', + '`', 'a', 'b', 'c', 'd', 'e', 'f', 'g', + 'h', 'i', 'j', 'k', 'l', 'm', 'n', 'o', + 'p', 'q', 'r', 's', 't', 'u', 'v', 'w', + 'x', 'y', 'z', '{', '|', '}', '~' + }; + + code = code < sizeof table ? table[code] : -1; + } +#endif - case destructor_braced_code: - handle_destructor_dollar (text, location); - break; - } + return code; } -/*------------------------------------------------------. -| TEXT is a location token (i.e., a `@...'). Output to | -| STRING_OBSTACK a reference to this location. | -`------------------------------------------------------*/ +/*---------------------------------------------------------------------. +| Handle '#line INT( "FILE")?\n'. ARGS has already skipped '#line '. | +`---------------------------------------------------------------------*/ -static inline void -handle_action_at (char *text, location_t location) +static void +handle_syncline (char *args, location loc) { - locations_flag = 1; - ++text; - - if (*text == '$') - { - obstack_sgrow (&string_obstack, "]b4_lhs_location["); - } - else if (isdigit (*text) || *text == '-') + char *file; + unsigned long int lineno = strtoul (args, &file, 10); + if (INT_MAX <= lineno) { - /* RULE_LENGTH is the number of values in the current rule so - far, which says where to find `$0' with respect to the top of - the stack. It is not the same as the rule->length in the - case of mid rule actions. */ - int rule_length = symbol_list_length (current_rule->next); - int n = strtol (text, &text, 10); - - if (n > rule_length) - complain_at (location, _("invalid value: %s%d"), "@", n); - else - obstack_fgrow2 (&string_obstack, "]b4_rhs_location([%d], [%d])[", - rule_length, n); + complain (&loc, Wother, _("line number overflow")); + lineno = INT_MAX; } - else + + file = strchr (file, '"'); + if (file) { - char buf[] = "@c"; - buf[1] = *text; - complain_at (location, _("%s is invalid"), quote (buf)); + *strchr (file + 1, '"') = '\0'; + current_file = uniqstr_new (file + 1); } + boundary_set (&scanner_cursor, current_file, lineno, 1); } -/*--------------------------------------------. -| TEXT is expexted tp be @$ in a destructor. | -`--------------------------------------------*/ +/*----------------------------------------------------------------. +| For a token or comment starting at START, report message MSGID, | +| which should say that an end marker was found before the | +| expected TOKEN_END. Then, pretend that TOKEN_END was found. | +`----------------------------------------------------------------*/ -static inline void -handle_destructor_at (char *text, location_t location) +static void +unexpected_end (boundary start, char const *msgid, char const *token_end) { - ++text; - if (*text == '$') - { - obstack_sgrow (&string_obstack, "]b4_at_dollar["); - } - else - { - char buf[] = "$c"; - buf[1] = *text; - complain_at (location, _("%s is invalid"), quote (buf)); - } + location loc; + loc.start = start; + loc.end = scanner_cursor; + size_t i = strlen (token_end); + +/* Adjust scanner cursor so that any later message does not count + the characters about to be inserted. */ + scanner_cursor.column -= i; + + while (i != 0) + unput (token_end[--i]); + + token_end = quote (token_end); + /* Instead of '\'', display "'". */ + if (STREQ (token_end, "'\\''")) + token_end = "\"'\""; + complain (&loc, complaint, _(msgid), token_end); } -/*-------------------------------------------------------------------. -| Dispatch onto handle_action_at, or handle_destructor_at, depending | -| upon CODE_KIND. | -`-------------------------------------------------------------------*/ +/*------------------------------------------------------------------------. +| Report an unexpected EOF in a token or comment starting at START. | +| An end of file was encountered and the expected TOKEN_END was missing. | +| After reporting the problem, pretend that TOKEN_END was found. | +`------------------------------------------------------------------------*/ static void -handle_at (braced_code_t braced_code_kind, - char *text, location_t location) +unexpected_eof (boundary start, char const *token_end) { - switch (braced_code_kind) - { - case action_braced_code: - handle_action_at (text, location); - break; + unexpected_end (start, N_("missing %s at end of file"), token_end); +} - case destructor_braced_code: - handle_destructor_at (text, location); - break; - } + +/*----------------------------------------. +| Likewise, but for unexpected newlines. | +`----------------------------------------*/ + +static void +unexpected_newline (boundary start, char const *token_end) +{ + unexpected_end (start, N_("missing %s at end of line"), token_end); } @@ -725,9 +995,9 @@ handle_at (braced_code_t braced_code_kind, `-------------------------*/ void -scanner_initialize (void) +gram_scanner_initialize (void) { - obstack_init (&string_obstack); + obstack_init (&obstack_for_string); } @@ -736,7 +1006,9 @@ scanner_initialize (void) `-----------------------------------------------*/ void -scanner_free (void) +gram_scanner_free (void) { - obstack_free (&string_obstack, 0); + obstack_free (&obstack_for_string, 0); + /* Reclaim Flex's buffers. */ + yylex_destroy (); }