X-Git-Url: https://git.saurik.com/bison.git/blobdiff_plain/345532d70bdcd1e4cb4939fea038674d34fbc064..e342c3beaf9d73616e8ff6fe005b97f3a72c0476:/src/scan-gram.l diff --git a/src/scan-gram.l b/src/scan-gram.l index b9442358..7541bf9f 100644 --- a/src/scan-gram.l +++ b/src/scan-gram.l @@ -1,5 +1,6 @@ /* Bison Grammar Scanner -*- C -*- - Copyright (C) 2002 Free Software Foundation, Inc. + + Copyright (C) 2002, 2003, 2004 Free Software Foundation, Inc. This file is part of Bison, the GNU Compiler Compiler. @@ -19,140 +20,75 @@ 02111-1307 USA */ -%option debug nodefault noyywrap never-interactive +%option debug nodefault nounput noyywrap never-interactive %option prefix="gram_" outfile="lex.yy.c" %{ #include "system.h" -#include "mbswidth.h" + +#include +#include +#include + #include "complain.h" -#include "quote.h" +#include "files.h" #include "getargs.h" #include "gram.h" #include "reader.h" +#include "uniqstr.h" -/* Each time we match a string, move the end cursor to its end. */ -#define YY_USER_INIT \ -do { \ - LOCATION_RESET (*yylloc); \ - yylloc->file = infile; \ - /* This is only to avoid GCC warnings. */ \ - if (yycontrol) {;}; \ -} while (0) - -#define YY_USER_ACTION extend_location (yylloc, yytext, yyleng); -#define YY_STEP LOCATION_STEP (*yylloc) - -#define YY_INPUT(buf, result, size) ((result) = no_cr_read (yyin, buf, size)) - - -/* 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. */ - -static size_t -no_cr_read (FILE *fp, char *buf, size_t size) -{ - size_t s = fread (buf, 1, size, fp); - if (s) - { - char *w = memchr (buf, '\r', s); - if (w) - { - char const *r = ++w; - char const *lim = buf + s; - - 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; - } - } - - return s; -} - - -/* Extend *LOC to account for token TOKEN of size SIZE. */ - -static void -extend_location (location_t *loc, char const *token, int size) -{ - int line = loc->last_line; - int column = loc->last_column; - char const *p0 = token; - char const *p = token; - char const *lim = token + size; - - for (p = token; p < lim; p++) - switch (*p) - { - case '\r': - /* \r shouldn't survive no_cr_read. */ - abort (); +#define YY_USER_INIT \ + do \ + { \ + scanner_cursor.file = current_file; \ + scanner_cursor.line = 1; \ + scanner_cursor.column = 1; \ + code_start = scanner_cursor; \ + } \ + while (0) - case '\n': - line++; - column = 1; - p0 = p + 1; - break; +/* Location of scanner cursor. */ +boundary scanner_cursor; - case '\t': - column += mbsnwidth (p0, p - p0, 0); - column += 8 - ((column - 1) & 7); - p0 = p + 1; - break; - } - - loc->last_line = line; - loc->last_column = column + mbsnwidth (p0, p - p0, 0); -} +static void adjust_location (location *, char const *, size_t); +#define YY_USER_ACTION adjust_location (loc, 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)) -/* STRING_OBSTACK -- Used to store all the characters that we need to +/* OBSTACK_FOR_STRING -- 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. */ + Use STRING_GROW to append what has just been matched, and + STRING_FINISH to end the string (it puts the ending 0). + STRING_FINISH also stores this string in LAST_STRING, which can be + used, and which is used by STRING_FREE to free the last string. */ + +static struct obstack obstack_for_string; -static struct obstack string_obstack; +/* A string representing the most recently saved token. */ +static char *last_string; -#define YY_OBS_GROW \ - obstack_grow (&string_obstack, yytext, yyleng) -#define YY_OBS_FINISH \ +#define STRING_GROW \ + obstack_grow (&obstack_for_string, yytext, yyleng) + +#define STRING_FINISH \ do { \ - obstack_1grow (&string_obstack, '\0'); \ - last_string = obstack_finish (&string_obstack); \ + obstack_1grow (&obstack_for_string, '\0'); \ + last_string = obstack_finish (&obstack_for_string); \ } while (0) -#define YY_OBS_FREE \ - obstack_free (&string_obstack, last_string) +#define STRING_FREE \ + obstack_free (&obstack_for_string, last_string) +void +scanner_last_string_free (void) +{ + STRING_FREE; +} /* Within well-formed rules, RULE_LENGTH is the number of values in the current rule so far, which says where to find `$0' with respect @@ -162,24 +98,25 @@ static struct obstack string_obstack; Outside of well-formed rules, RULE_LENGTH has an undefined value. */ static int rule_length; -static void handle_dollar (braced_code_t code_kind, - char *cp, location_t location); -static void handle_at (braced_code_t code_kind, - char *cp, location_t location); -static void handle_syncline (char *args, location_t *location); +static void handle_dollar (int token_type, char *cp, location loc); +static void handle_at (int token_type, char *cp, location loc); +static void handle_syncline (char *args); +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_end_of_file (location_t *, char const *); +static void unexpected_eof (boundary, char const *); +static void unexpected_newline (boundary, char const *); %} %x SC_COMMENT SC_LINE_COMMENT SC_YACC_COMMENT %x SC_STRING SC_CHARACTER +%x SC_AFTER_IDENTIFIER %x SC_ESCAPED_STRING SC_ESCAPED_CHARACTER -%x SC_BRACED_CODE SC_PROLOGUE SC_EPILOGUE +%x SC_PRE_CODE SC_BRACED_CODE SC_PROLOGUE SC_EPILOGUE letter [.abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ_] id {letter}({letter}|[0-9])* directive %{letter}({letter}|[0-9]|-)* -int [0-9]+ +int [0-9]+ /* 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 @@ -195,18 +132,50 @@ splice (\\[ \f\t\v]*\n)* /* Nesting level of the current code in braces. */ int braces_level IF_LINT (= 0); - /* Scanner context when scanning C code. */ - int c_context IF_LINT (= 0); + /* Parent context state, when applicable. */ + int context_state IF_LINT (= 0); + + /* Token type to return, when applicable. */ + int token_type IF_LINT (= 0); + + /* Location of most recent identifier, when applicable. */ + location id_loc IF_LINT (= empty_location); - /* A string representing the most recently saved token. */ - char *last_string; + /* 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; - /* At each yylex invocation, mark the current position as the - start of the next token. */ - YY_STEP; + /* Where containing comment or string or character literal started, + when applicable. */ + boundary token_start IF_LINT (= scanner_cursor); %} + /*-----------------------. + | Scanning white space. | + `-----------------------*/ + + +{ + /* Comments and white space. */ + "," warn_at (*loc, _("stray `,' treated as white space")); + [ \f\n\t\v] | + "//".* ; + "/*" { + 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); + } +} + + /*----------------------------. | Scanning Bison directives. | `----------------------------*/ @@ -214,94 +183,92 @@ splice (\\[ \f\t\v]*\n)* { "%binary" return PERCENT_NONASSOC; "%debug" return PERCENT_DEBUG; + "%default"[-_]"prec" return PERCENT_DEFAULT_PREC; "%define" return PERCENT_DEFINE; "%defines" return PERCENT_DEFINES; - "%destructor" return PERCENT_DESTRUCTOR; + "%destructor" token_type = PERCENT_DESTRUCTOR; BEGIN SC_PRE_CODE; "%dprec" return PERCENT_DPREC; "%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" token_type = PERCENT_INITIAL_ACTION; BEGIN SC_PRE_CODE; "%glr-parser" return PERCENT_GLR_PARSER; "%left" return PERCENT_LEFT; + "%lex-param" token_type = PERCENT_LEX_PARAM; BEGIN SC_PRE_CODE; "%locations" return PERCENT_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; - "%parse-param" return PERCENT_PARSE_PARAM; + "%parse-param" token_type = PERCENT_PARSE_PARAM; BEGIN SC_PRE_CODE; "%prec" rule_length--; return PERCENT_PREC; - "%printer" return PERCENT_PRINTER; + "%printer" token_type = PERCENT_PRINTER; BEGIN SC_PRE_CODE; "%pure"[-_]"parser" return PERCENT_PURE_PARSER; "%right" return PERCENT_RIGHT; - "%lex-param" return PERCENT_LEX_PARAM; "%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; + "%union" token_type = PERCENT_UNION; BEGIN SC_PRE_CODE; "%verbose" return PERCENT_VERBOSE; "%yacc" return PERCENT_YACC; - {directive} { - complain_at (*yylloc, _("invalid directive: %s"), quote (yytext)); - YY_STEP; + {directive} { + complain_at (*loc, _("invalid directive: %s"), quote (yytext)); } - ^"#line "{int}" \""[^\"]*"\"\n" handle_syncline (yytext + strlen ("#line "), yylloc); YY_STEP; - "=" return EQUAL; - ":" rule_length = 0; return COLON; "|" rule_length = 0; return PIPE; - "," return COMMA; ";" return SEMICOLON; - [ \f\n\t\v] YY_STEP; - - {id} { - yylval->symbol = symbol_get (yytext, *yylloc); + {id} { + val->symbol = symbol_get (yytext, *loc); + id_loc = *loc; rule_length++; - return ID; + BEGIN SC_AFTER_IDENTIFIER; } {int} { - unsigned long num; - errno = 0; - num = strtoul (yytext, 0, 10); - if (INT_MAX < num || errno) - { - complain_at (*yylloc, _("integer out of range: %s"), quote (yytext)); - num = INT_MAX; - } - yylval->integer = num; + 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; BEGIN SC_ESCAPED_CHARACTER; + "'" STRING_GROW; token_start = loc->start; BEGIN SC_ESCAPED_CHARACTER; /* Strings. */ - "\"" YY_OBS_GROW; BEGIN SC_ESCAPED_STRING; - - /* Comments. */ - "/*" BEGIN SC_YACC_COMMENT; - "//".* YY_STEP; + "\"" STRING_GROW; token_start = loc->start; BEGIN SC_ESCAPED_STRING; /* Prologue. */ - "%{" BEGIN SC_PROLOGUE; + "%{" code_start = loc->start; BEGIN SC_PROLOGUE; /* Code in between braces. */ - "{" YY_OBS_GROW; braces_level = 0; BEGIN SC_BRACED_CODE; + "{" { + STRING_GROW; + token_type = BRACED_CODE; + braces_level = 0; + code_start = loc->start; + BEGIN SC_BRACED_CODE; + } /* A type. */ "<"{tag}">" { - obstack_grow (&string_obstack, yytext + 1, yyleng - 2); - YY_OBS_FINISH; - yylval->string = last_string; + obstack_grow (&obstack_for_string, yytext + 1, yyleng - 2); + STRING_FINISH; + val->uniqstr = uniqstr_new (last_string); + STRING_FREE; return TYPE; } @@ -313,8 +280,39 @@ splice (\\[ \f\t\v]*\n)* } . { - complain_at (*yylloc, _("invalid character: %s"), quote (yytext)); - YY_STEP; + complain_at (*loc, _("invalid character: %s"), quote (yytext)); + } + + <> { + loc->start = loc->end = scanner_cursor; + yyterminate (); + } +} + + + /*-----------------------------------------------------------------. + | Scanning after an identifier, checking whether a colon is next. | + `-----------------------------------------------------------------*/ + + +{ + ":" { + rule_length = 0; + *loc = id_loc; + BEGIN INITIAL; + return ID_COLON; + } + . { + scanner_cursor.column -= mbsnwidth (yytext, yyleng, 0); + yyless (0); + *loc = id_loc; + BEGIN INITIAL; + return ID; + } + <> { + *loc = id_loc; + BEGIN INITIAL; + return ID; } } @@ -325,13 +323,9 @@ splice (\\[ \f\t\v]*\n)* { - "*/" { - YY_STEP; - BEGIN INITIAL; - } - + "*/" BEGIN context_state; .|\n ; - <> unexpected_end_of_file (yylloc, "*/"); + <> unexpected_eof (token_start, "*/"); BEGIN context_state; } @@ -341,8 +335,8 @@ splice (\\[ \f\t\v]*\n)* { - "*"{splice}"/" YY_OBS_GROW; BEGIN c_context; - <> unexpected_end_of_file (yylloc, "*/"); + "*"{splice}"/" STRING_GROW; BEGIN context_state; + <> unexpected_eof (token_start, "*/"); BEGIN context_state; } @@ -352,54 +346,60 @@ splice (\\[ \f\t\v]*\n)* { - "\n" YY_OBS_GROW; BEGIN c_context; - {splice} YY_OBS_GROW; - <> BEGIN c_context; + "\n" STRING_GROW; BEGIN context_state; + {splice} STRING_GROW; + <> BEGIN context_state; } - /*----------------------------------------------------------------. - | Scanning a C string, including its escapes. The initial `"' is | - | already eaten. | - `----------------------------------------------------------------*/ + /*------------------------------------------------. + | Scanning a Bison string, including its escapes. | + | The initial quote is already eaten. | + `------------------------------------------------*/ { "\"" { - YY_OBS_GROW; - YY_OBS_FINISH; - yylval->string = last_string; + STRING_GROW; + STRING_FINISH; + loc->start = token_start; + val->chars = last_string; rule_length++; BEGIN INITIAL; return STRING; } - - .|\n YY_OBS_GROW; - <> unexpected_end_of_file (yylloc, "\""); + \n unexpected_newline (token_start, "\""); BEGIN INITIAL; + <> unexpected_eof (token_start, "\""); BEGIN INITIAL; } - /*---------------------------------------------------------------. - | 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; - YY_OBS_FINISH; - yylval->symbol = symbol_get (last_string, *yylloc); - symbol_class_set (yylval->symbol, token_sym, *yylloc); - symbol_user_token_number_set (yylval->symbol, - (unsigned char) last_string[1], *yylloc); - YY_OBS_FREE; + unsigned char last_string_1; + STRING_GROW; + STRING_FINISH; + loc->start = token_start; + val->symbol = symbol_get (last_string, *loc); + symbol_class_set (val->symbol, token_sym, *loc); + last_string_1 = last_string[1]; + symbol_user_token_number_set (val->symbol, last_string_1, *loc); + STRING_FREE; rule_length++; BEGIN INITIAL; return ID; } + \n unexpected_newline (token_start, "'"); BEGIN INITIAL; + <> unexpected_eof (token_start, "'"); BEGIN INITIAL; +} - .|\n YY_OBS_GROW; - <> unexpected_end_of_file (yylloc, "'"); + +{ + \0 complain_at (*loc, _("invalid null character")); } @@ -410,83 +410,74 @@ splice (\\[ \f\t\v]*\n)* { \\[0-7]{1,3} { - unsigned long c = strtoul (yytext + 1, 0, 8); + unsigned long int c = strtoul (yytext + 1, 0, 8); if (UCHAR_MAX < c) - { - complain_at (*yylloc, _("invalid escape sequence: %s"), - quote (yytext)); - YY_STEP; - } + complain_at (*loc, _("invalid escape sequence: %s"), quote (yytext)); + else if (! c) + complain_at (*loc, _("invalid null character: %s"), quote (yytext)); else - obstack_1grow (&string_obstack, c); + obstack_1grow (&obstack_for_string, c); } - \\x[0-9a-fA-F]+ { - unsigned long c; - errno = 0; + \\x[0-9abcdefABCDEF]+ { + unsigned long int c; + set_errno (0); c = strtoul (yytext + 2, 0, 16); - if (UCHAR_MAX < c || errno) - { - complain_at (*yylloc, _("invalid escape sequence: %s"), - quote (yytext)); - YY_STEP; - } + if (UCHAR_MAX < c || get_errno ()) + complain_at (*loc, _("invalid escape sequence: %s"), quote (yytext)); + else if (! c) + complain_at (*loc, _("invalid null character: %s"), quote (yytext)); else - obstack_1grow (&string_obstack, c); + 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]); - \\(u|U[0-9a-fA-F]{4})[0-9a-fA-F]{4} { + \\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_at (*yylloc, _("invalid escape sequence: %s"), - quote (yytext)); - YY_STEP; - } + complain_at (*loc, _("invalid escape sequence: %s"), quote (yytext)); + else if (! c) + complain_at (*loc, _("invalid null character: %s"), quote (yytext)); else - obstack_1grow (&string_obstack, c); + obstack_1grow (&obstack_for_string, c); } \\(.|\n) { - complain_at (*yylloc, _("unrecognized escape sequence: %s"), - quote (yytext)); - YY_OBS_GROW; + complain_at (*loc, _("unrecognized escape sequence: %s"), quote (yytext)); + STRING_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; BEGIN c_context; - \\{splice}[^\[\]] YY_OBS_GROW; - {splice} YY_OBS_GROW; - <> unexpected_end_of_file (yylloc, "'"); + "'" STRING_GROW; BEGIN context_state; + \n unexpected_newline (token_start, "'"); BEGIN context_state; + <> unexpected_eof (token_start, "'"); BEGIN context_state; } - - /*----------------------------------------------------------------. - | Scanning a C string, without decoding its escapes. The initial | - | `"' is already eaten. | - `----------------------------------------------------------------*/ - { - "\"" YY_OBS_GROW; BEGIN c_context; - \\{splice}[^\[\]] YY_OBS_GROW; - {splice} YY_OBS_GROW; - <> unexpected_end_of_file (yylloc, "\""); + "\"" STRING_GROW; BEGIN context_state; + \n unexpected_newline (token_start, "\""); BEGIN context_state; + <> unexpected_eof (token_start, "\""); BEGIN context_state; } @@ -496,10 +487,64 @@ splice (\\[ \f\t\v]*\n)* { - "'" YY_OBS_GROW; c_context = YY_START; BEGIN SC_CHARACTER; - "\"" YY_OBS_GROW; c_context = YY_START; BEGIN SC_STRING; - "/"{splice}"*" YY_OBS_GROW; c_context = YY_START; BEGIN SC_COMMENT; - "/"{splice}"/" YY_OBS_GROW; c_context = YY_START; BEGIN SC_LINE_COMMENT; + "'" { + 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; + } +} + + + /*---------------------------------------------------------------. + | Scanning after %union etc., possibly followed by white space. | + | For %union only, allow arbitrary C code to appear before the | + | following brace, as an extension to POSIX. | + `---------------------------------------------------------------*/ + + +{ + . { + bool valid = yytext[0] == '{' || token_type == PERCENT_UNION; + scanner_cursor.column -= mbsnwidth (yytext, yyleng, 0); + yyless (0); + + if (valid) + { + braces_level = -1; + code_start = loc->start; + BEGIN SC_BRACED_CODE; + } + else + { + complain_at (*loc, _("missing `{' in `%s'"), + token_name (token_type)); + obstack_sgrow (&obstack_for_string, "{}"); + STRING_FINISH; + val->chars = last_string; + BEGIN INITIAL; + return token_type; + } + } + + <> unexpected_eof (scanner_cursor, "{}"); BEGIN INITIAL; } @@ -510,31 +555,49 @@ splice (\\[ \f\t\v]*\n)* { - "{"|"<"{splice}"%" YY_OBS_GROW; braces_level++; - "%"{splice}">" YY_OBS_GROW; braces_level--; + "{"|"<"{splice}"%" STRING_GROW; braces_level++; + "%"{splice}">" STRING_GROW; braces_level--; "}" { - YY_OBS_GROW; - braces_level--; - if (braces_level < 0) + bool outer_brace = --braces_level < 0; + + /* As an undocumented Bison extension, append `;' before the last + brace in braced code, so that the user code can omit trailing + `;'. But do not append `;' if emulating Yacc, since Yacc does + not append one. + + FIXME: Bison should warn if a semicolon seems to be necessary + here, and should omit the semicolon if it seems unnecessary + (e.g., after ';', '{', or '}', each followed by comments or + white space). Such a warning shouldn't depend on --yacc; it + should depend on a new --pedantic option, which would cause + Bison to warn if it detects an extension to POSIX. --pedantic + should also diagnose other Bison extensions like %yacc. + Perhaps there should also be a GCC-style --pedantic-errors + option, so that such warnings are diagnosed as errors. */ + if (outer_brace && token_type == BRACED_CODE && ! yacc_flag) + obstack_1grow (&obstack_for_string, ';'); + + obstack_1grow (&obstack_for_string, '}'); + + if (outer_brace) { - YY_OBS_FINISH; - yylval->string = last_string; + STRING_FINISH; rule_length++; + loc->start = code_start; + val->chars = last_string; BEGIN INITIAL; - return BRACED_CODE; + return token_type; } } /* Tokenize `<<%' correctly (as `<<' `%') rather than incorrrectly (as `<' `<%'). */ - "<"{splice}"<" YY_OBS_GROW; + "<"{splice}"<" STRING_GROW; - "$"("<"{tag}">")?(-?[0-9]+|"$") { handle_dollar (current_braced_code, - yytext, *yylloc); } - "@"(-?[0-9]+|"$") { handle_at (current_braced_code, - yytext, *yylloc); } + "$"("<"{tag}">")?(-?[0-9]+|"$") handle_dollar (token_type, yytext, *loc); + "@"(-?[0-9]+|"$") handle_at (token_type, yytext, *loc); - <> unexpected_end_of_file (yylloc, "}"); + <> unexpected_eof (code_start, "}"); BEGIN INITIAL; } @@ -545,13 +608,14 @@ splice (\\[ \f\t\v]*\n)* { "%}" { - YY_OBS_FINISH; - yylval->string = last_string; + STRING_FINISH; + loc->start = code_start; + val->chars = last_string; BEGIN INITIAL; return PROLOGUE; } - <> unexpected_end_of_file (yylloc, "%}"); + <> unexpected_eof (code_start, "%}"); BEGIN INITIAL; } @@ -563,43 +627,143 @@ splice (\\[ \f\t\v]*\n)* { <> { - YY_OBS_FINISH; - yylval->string = last_string; + STRING_FINISH; + loc->start = code_start; + val->chars = last_string; BEGIN INITIAL; return EPILOGUE; } } - /*----------------------------------------------------------------. - | By default, grow the string obstack with the input, escaping M4 | - | quoting characters. | - `----------------------------------------------------------------*/ + /*-----------------------------------------. + | Escape M4 quoting characters in C code. | + `-----------------------------------------*/ { - \[ obstack_sgrow (&string_obstack, "@<:@"); - \] obstack_sgrow (&string_obstack, "@:>@"); - .|\n YY_OBS_GROW; + \$ obstack_sgrow (&obstack_for_string, "$]["); + \@ obstack_sgrow (&obstack_for_string, "@@"); + \[ obstack_sgrow (&obstack_for_string, "@{"); + \] obstack_sgrow (&obstack_for_string, "@}"); } + /*-----------------------------------------------------. + | By default, grow the string obstack with the input. | + `-----------------------------------------------------*/ + +. | +\n STRING_GROW; + %% +/* Keeps track of the maximum number of semantic values to the left of + a handle (those referenced by $0, $-1, etc.) are required by the + semantic actions of this grammar. */ +int max_left_semantic_context = 0; + +/* Set *LOC and adjust scanner cursor to account for token TOKEN of + size SIZE. */ + +static void +adjust_location (location *loc, char const *token, size_t size) +{ + int line = scanner_cursor.line; + int column = scanner_cursor.column; + char const *p0 = token; + char const *p = token; + char const *lim = token + size; + + loc->start = scanner_cursor; + + for (p = token; p < lim; p++) + switch (*p) + { + case '\n': + line++; + column = 1; + p0 = p + 1; + break; + + case '\t': + column += mbsnwidth (p0, p - p0, 0); + column += 8 - ((column - 1) & 7); + p0 = p + 1; + break; + } + + scanner_cursor.line = line; + scanner_cursor.column = column + mbsnwidth (p0, p - p0, 0); + + loc->end = scanner_cursor; +} + + +/* Read bytes from FP into buffer BUF of size SIZE. Return the + number of bytes read. Remove '\r' from input, treating \r\n + and isolated \r as \n. */ + +static size_t +no_cr_read (FILE *fp, char *buf, size_t size) +{ + size_t bytes_read = fread (buf, 1, size, fp); + if (bytes_read) + { + char *w = memchr (buf, '\r', bytes_read); + if (w) + { + char const *r = ++w; + char const *lim = buf + bytes_read; + + for (;;) + { + /* Found an '\r'. Treat it like '\n', but ignore any + '\n' that immediately follows. */ + w[-1] = '\n'; + if (r == lim) + { + int ch = getc (fp); + if (ch != '\n' && ungetc (ch, fp) != ch) + break; + } + else if (*r == '\n') + r++; + + /* Copy until the next '\r'. */ + do + { + if (r == lim) + return w - buf; + } + while ((*w++ = *r++) != '\r'); + } + + return w - buf; + } + } + + return bytes_read; +} + + /*------------------------------------------------------------------. | TEXT is pointing to a wannabee semantic value (i.e., a `$'). | | | | Possible inputs: $[]($|integer) | | | -| Output to the STRING_OBSTACK a reference to this semantic value. | +| Output to OBSTACK_FOR_STRING a reference to this semantic value. | `------------------------------------------------------------------*/ -static inline void -handle_action_dollar (char *text, location_t location) +static inline bool +handle_action_dollar (char *text, location loc) { const char *type_name = NULL; char *cp = text + 1; + if (! current_rule) + return false; + /* Get the type name if explicit. */ if (*cp == '<') { @@ -613,150 +777,162 @@ handle_action_dollar (char *text, location_t location) if (*cp == '$') { if (!type_name) - type_name = symbol_list_n_type_name_get (current_rule, location, 0); + type_name = symbol_list_n_type_name_get (current_rule, loc, 0); if (!type_name && typed) - complain_at (location, _("$$ of `%s' has no declared type"), + complain_at (loc, _("$$ of `%s' has no declared type"), current_rule->sym->tag); if (!type_name) type_name = ""; - obstack_fgrow1 (&string_obstack, + obstack_fgrow1 (&obstack_for_string, "]b4_lhs_value([%s])[", type_name); } else { - long num; - errno = 0; + long int num; + set_errno (0); num = strtol (cp, 0, 10); - if (INT_MIN <= num && num <= rule_length && ! errno) + if (INT_MIN <= num && num <= rule_length && ! get_errno ()) { int n = num; + if (1-n > max_left_semantic_context) + max_left_semantic_context = 1-n; if (!type_name && n > 0) - type_name = symbol_list_n_type_name_get (current_rule, location, - n); + type_name = symbol_list_n_type_name_get (current_rule, loc, n); if (!type_name && typed) - complain_at (location, _("$%d of `%s' has no declared type"), - n, current_rule->sym->tag); + complain_at (loc, _("$%d of `%s' has no declared type"), + n, current_rule->sym->tag); if (!type_name) type_name = ""; - obstack_fgrow3 (&string_obstack, + obstack_fgrow3 (&obstack_for_string, "]b4_rhs_value([%d], [%d], [%s])[", rule_length, n, type_name); } else - complain_at (location, _("integer out of range: %s"), quote (text)); + complain_at (loc, _("integer out of range: %s"), quote (text)); } -} - -/*---------------------------------------------------------------. -| TEXT is expected to be $$ in some code associated to a symbol: | -| destructor or printer. | -`---------------------------------------------------------------*/ - -static inline void -handle_symbol_code_dollar (char *text, location_t location) -{ - char *cp = text + 1; - if (*cp == '$') - obstack_sgrow (&string_obstack, "]b4_dollar_dollar["); - else - complain_at (location, _("invalid value: %s"), quote (text)); + return true; } -/*-----------------------------------------------------------------. -| Dispatch onto handle_action_dollar, or handle_destructor_dollar, | -| depending upon CODE_KIND. | -`-----------------------------------------------------------------*/ +/*----------------------------------------------------------------. +| Map `$?' onto the proper M4 symbol, depending on its TOKEN_TYPE | +| (are we in an action?). | +`----------------------------------------------------------------*/ static void -handle_dollar (braced_code_t braced_code_kind, - char *text, location_t location) +handle_dollar (int token_type, char *text, location loc) { - switch (braced_code_kind) + switch (token_type) { - case action_braced_code: - handle_action_dollar (text, location); + case BRACED_CODE: + if (handle_action_dollar (text, loc)) + return; break; - case destructor_braced_code: - case printer_braced_code: - handle_symbol_code_dollar (text, location); + case PERCENT_DESTRUCTOR: + case PERCENT_INITIAL_ACTION: + case PERCENT_PRINTER: + if (text[1] == '$') + { + obstack_sgrow (&obstack_for_string, "]b4_dollar_dollar["); + return; + } + break; + + default: break; } + + complain_at (loc, _("invalid value: %s"), quote (text)); } /*------------------------------------------------------. | TEXT is a location token (i.e., a `@...'). Output to | -| STRING_OBSTACK a reference to this location. | +| OBSTACK_FOR_STRING a reference to this location. | `------------------------------------------------------*/ -static inline void -handle_action_at (char *text, location_t location) +static inline bool +handle_action_at (char *text, location loc) { char *cp = text + 1; - locations_flag = 1; + locations_flag = true; + + if (! current_rule) + return false; if (*cp == '$') - { - obstack_sgrow (&string_obstack, "]b4_lhs_location["); - } + obstack_sgrow (&obstack_for_string, "]b4_lhs_location["); else { - long num; - errno = 0; + long int num; + set_errno (0); num = strtol (cp, 0, 10); - if (INT_MIN <= num && num <= rule_length && ! errno) + if (INT_MIN <= num && num <= rule_length && ! get_errno ()) { int n = num; - obstack_fgrow2 (&string_obstack, "]b4_rhs_location([%d], [%d])[", + obstack_fgrow2 (&obstack_for_string, "]b4_rhs_location([%d], [%d])[", rule_length, n); } else - complain_at (location, _("integer out of range: %s"), quote (text)); + complain_at (loc, _("integer out of range: %s"), quote (text)); } -} - -/*---------------------------------------------------------------. -| TEXT is expected to be @$ in some code associated to a symbol: | -| destructor or printer. | -`---------------------------------------------------------------*/ - -static inline void -handle_symbol_code_at (char *text, location_t location) -{ - char *cp = text + 1; - if (*cp == '$') - obstack_sgrow (&string_obstack, "]b4_at_dollar["); - else - complain_at (location, _("invalid value: %s"), quote (text)); + return true; } -/*-------------------------------------------------------------------. -| Dispatch onto handle_action_at, or handle_destructor_at, depending | -| upon CODE_KIND. | -`-------------------------------------------------------------------*/ +/*----------------------------------------------------------------. +| Map `@?' onto the proper M4 symbol, depending on its TOKEN_TYPE | +| (are we in an action?). | +`----------------------------------------------------------------*/ static void -handle_at (braced_code_t braced_code_kind, - char *text, location_t location) +handle_at (int token_type, char *text, location loc) { - switch (braced_code_kind) + switch (token_type) { - case action_braced_code: - handle_action_at (text, location); + case BRACED_CODE: + handle_action_at (text, loc); + return; + + case PERCENT_INITIAL_ACTION: + case PERCENT_DESTRUCTOR: + case PERCENT_PRINTER: + if (text[1] == '$') + { + obstack_sgrow (&obstack_for_string, "]b4_at_dollar["); + return; + } break; - case destructor_braced_code: - case printer_braced_code: - handle_symbol_code_at (text, location); + default: break; } + + complain_at (loc, _("invalid value: %s"), quote (text)); +} + + +/*------------------------------------------------------. +| Scan NUMBER for a base-BASE integer at location LOC. | +`------------------------------------------------------*/ + +static unsigned long int +scan_integer (char const *number, int base, location loc) +{ + unsigned long int num; + set_errno (0); + num = strtoul (number, 0, base); + if (INT_MAX < num || get_errno ()) + { + complain_at (loc, _("integer out of range: %s"), quote (number)); + num = INT_MAX; + } + return num; } @@ -769,7 +945,7 @@ handle_at (braced_code_t braced_code_kind, static int convert_ucn_to_byte (char const *ucn) { - unsigned long code = strtoul (ucn + 2, 0, 16); + unsigned long int code = strtoul (ucn + 2, 0, 16); /* FIXME: Currently we assume Unicode-compatible unibyte characters on ASCII hosts (i.e., Latin-1 on hosts with 8-bit bytes). On @@ -821,39 +997,54 @@ convert_ucn_to_byte (char const *ucn) `----------------------------------------------------------------*/ static void -handle_syncline (char *args, location_t *location) +handle_syncline (char *args) { int lineno = strtol (args, &args, 10); const char *file = NULL; file = strchr (args, '"') + 1; *strchr (file, '"') = 0; - /* FIXME: Leaking... Can't free, as some locations are still - pointing to the old file name. */ - infile = xstrdup (file); - location->file = infile; - location->last_line = lineno; + scanner_cursor.file = current_file = uniqstr_new (file); + scanner_cursor.line = lineno; + scanner_cursor.column = 1; } -/*-------------------------------------------------------------. -| Report an unexpected end of file at LOC. An end of file was | -| encountered and the expected TOKEN_END was missing. After | -| reporting the problem, pretend that TOKEN_END was found. | -`-------------------------------------------------------------*/ +/*----------------------------------------------------------------. +| 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. | +`----------------------------------------------------------------*/ static void -unexpected_end_of_file (location_t *loc, char const *token_end) +unexpected_end (boundary start, char const *msgid, char const *token_end) { - size_t i = strlen (token_end); + location loc; + loc.start = start; + loc.end = scanner_cursor; + complain_at (loc, _(msgid), token_end); +} - complain_at (*loc, _("missing `%s' at end of file"), token_end); - /* Adjust location's last column so that any later message does not - mention the characters just inserted. */ - loc->last_column -= i; +/*------------------------------------------------------------------------. +| 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. | +`------------------------------------------------------------------------*/ - while (i != 0) - unput (token_end[--i]); +static void +unexpected_eof (boundary start, char const *token_end) +{ + unexpected_end (start, N_("missing `%s' at end of file"), token_end); +} + + +/*----------------------------------------. +| 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); } @@ -864,7 +1055,7 @@ unexpected_end_of_file (location_t *loc, char const *token_end) void scanner_initialize (void) { - obstack_init (&string_obstack); + obstack_init (&obstack_for_string); } @@ -875,7 +1066,7 @@ scanner_initialize (void) void scanner_free (void) { - obstack_free (&string_obstack, 0); + obstack_free (&obstack_for_string, 0); /* Reclaim Flex's buffers. */ yy_delete_buffer (YY_CURRENT_BUFFER); }