diff options
Diffstat (limited to 'src/scan.l')
| -rw-r--r-- | src/scan.l | 281 |
1 files changed, 159 insertions, 122 deletions
@@ -38,6 +38,12 @@ extern bool tablesverify, tablesext; extern int trlcontxt; /* Set in parse.y for each rule. */ extern const char *escaped_qstart, *escaped_qend; +#define M4QSTART "[""[" +#define M4QEND "]""]" + +#define ESCAPED_QSTART "[" M4QEND M4QSTART "[" M4QEND M4QSTART +#define ESCAPED_QEND M4QEND "]" M4QSTART M4QEND "]" M4QSTART + #define ACTION_ECHO add_action( yytext ) #define ACTION_IFDEF(def, should_define) \ { \ @@ -45,8 +51,8 @@ extern const char *escaped_qstart, *escaped_qend; action_define( def, 1 ); \ } -#define ACTION_ECHO_QSTART add_action (escaped_qstart) -#define ACTION_ECHO_QEND add_action (escaped_qend) +#define ACTION_ECHO_QSTART add_action (ESCAPED_QSTART) +#define ACTION_ECHO_QEND add_action (ESCAPED_QEND) #define ACTION_M4_IFDEF(def, should_define) \ do{ \ @@ -68,18 +74,20 @@ extern const char *escaped_qstart, *escaped_qend; #define RETURNNAME \ if(yyleng < MAXLINE) \ { \ - strcpy( nmstr, yytext ); \ + strncpy( nmstr, yytext, sizeof(nmstr) ); \ + return NAME; \ } \ else \ - { \ + do { \ synerr(_("Input line too long\n")); \ FLEX_EXIT(EXIT_FAILURE); \ - } \ - return NAME; + } while (0) #define PUT_BACK_STRING(str, start) \ - for ( i = strlen( str ) - 1; i >= start; --i ) \ - unput((str)[i]) + { size_t i = strlen( str ); \ + while ( i > start ) \ + unput((str)[--i]); \ + } #define CHECK_REJECT(str) \ if ( all_upper( str ) ) \ @@ -93,6 +101,23 @@ extern const char *escaped_qstart, *escaped_qend; if ( getenv("POSIXLY_CORRECT") ) \ posix_compat = true; +#define START_CODEBLOCK(x) do { \ + /* Emit the needed line directive... */\ + if (indented_code == false) { \ + linenum++; \ + line_directive_out(NULL, 1); \ + } \ + add_action(M4QSTART); \ + yy_push_state(CODEBLOCK); \ + if ((indented_code = x)) ACTION_ECHO; \ +} while(0) + +#define END_CODEBLOCK do { \ + yy_pop_state();\ + add_action(M4QEND); \ + if (!indented_code) line_directive_out(NULL, 0);\ +} while (0) + %} %option caseless nodefault noreject stack noyy_top_state @@ -104,7 +129,9 @@ extern const char *escaped_qstart, *escaped_qend; %x GROUP_WITH_PARAMS %x GROUP_MINUS_PARAMS %x EXTENDED_COMMENT -%x COMMENT_DISCARD +%x COMMENT_DISCARD CODE_COMMENT +%x SECT3_NOESCAPE +%x CHARACTER_CONSTANT WS [[:blank:]]+ OPTWS [[:blank:]]* @@ -125,8 +152,8 @@ CCL_EXPR ("[:"^?[[:alpha:]]+":]") LEXOPT [aceknopr] -M4QSTART "[[" -M4QEND "]]" +M4QSTART "[""[" +M4QEND "]""]" %% static int bracelevel, didadef, indented_code; @@ -134,22 +161,17 @@ M4QEND "]]" static int option_sense; int doing_codeblock = false; - int i, brace_depth=0, brace_start_line=0; + int brace_depth=0, brace_start_line=0; char nmdef[MAXLINE]; <INITIAL>{ - ^{WS} indented_code = true; BEGIN(CODEBLOCK); - ^"/*" ACTION_ECHO; yy_push_state( COMMENT ); + ^{WS} START_CODEBLOCK(true); + ^"/*" add_action("/*[""["); yy_push_state( COMMENT ); ^#{OPTWS}line{WS} yy_push_state( LINEDIR ); ^"%s"{NAME}? return SCDECL; ^"%x"{NAME}? return XSCDECL; - ^"%{".*{NL} { - ++linenum; - line_directive_out(NULL, 1); - indented_code = false; - BEGIN(CODEBLOCK); - } + ^"%{".*{NL} START_CODEBLOCK(false); ^"%top"[[:blank:]]*"{"[[:blank:]]*{NL} { brace_start_line = linenum; ++linenum; @@ -185,7 +207,7 @@ M4QEND "]]" ^{NAME} { if(yyleng < MAXLINE) { - strcpy( nmstr, yytext ); + strncpy( nmstr, yytext, sizeof(nmstr) ); } else { @@ -203,14 +225,18 @@ M4QEND "]]" } -<COMMENT>{ - "*/" ACTION_ECHO; yy_pop_state(); - "*" ACTION_ECHO; - {M4QSTART} ACTION_ECHO_QSTART; - {M4QEND} ACTION_ECHO_QEND; - [^*\n] ACTION_ECHO; +<COMMENT,CODE_COMMENT>{ /* */ + [^\[\]\*\n]* ACTION_ECHO; + . ACTION_ECHO; + {NL} ++linenum; ACTION_ECHO; } +<COMMENT>{ + "*/" add_action("*/]""]"); yy_pop_state(); +} +<CODE_COMMENT>{ + "*/" ACTION_ECHO; yy_pop_state(); +} <COMMENT_DISCARD>{ /* This is the same as COMMENT, but is discarded rather than output. */ @@ -223,7 +249,7 @@ M4QEND "]]" <EXTENDED_COMMENT>{ ")" yy_pop_state(); [^\n\)]+ ; - {NL} ++linenum; + {NL} ++linenum; } <LINEDIR>{ @@ -237,19 +263,19 @@ M4QEND "]]" } . /* ignore spurious characters */ } +<ACTION,CODEBLOCK,ACTION_STRING,PERCENT_BRACE_ACTION,CHARACTER_CONSTANT,COMMENT,CODE_COMMENT>{ + {M4QSTART} ACTION_ECHO_QSTART; + {M4QEND} ACTION_ECHO_QEND; +} <CODEBLOCK>{ - ^"%}".*{NL} ++linenum; BEGIN(INITIAL); - - {M4QSTART} ACTION_ECHO_QSTART; - {M4QEND} ACTION_ECHO_QEND; - . ACTION_ECHO; - + ^"%}".*{NL} ++linenum; END_CODEBLOCK; + [^\n%\[\]]* ACTION_ECHO; + . ACTION_ECHO; {NL} { ++linenum; ACTION_ECHO; - if ( indented_code ) - BEGIN(INITIAL); + if ( indented_code ) END_CODEBLOCK; } } @@ -274,10 +300,9 @@ M4QEND "]]" {M4QSTART} buf_strnappend(&top_buf, escaped_qstart, (int) strlen(escaped_qstart)); {M4QEND} buf_strnappend(&top_buf, escaped_qend, (int) strlen(escaped_qend)); - - [^{}\r\n] { - buf_strnappend(&top_buf, yytext, yyleng); - } + ([^{}\r\n\[\]]+)|[^{}\r\n] { + buf_strnappend(&top_buf, yytext, yyleng); + } <<EOF>> { linenum = brace_start_line; @@ -293,7 +318,7 @@ M4QEND "]]" {NOT_WS}[^\r\n]* { if(yyleng < MAXLINE) { - strcpy( nmdef, yytext ); + strncpy( nmdef, yytext, sizeof(nmdef) ); } else { @@ -301,12 +326,12 @@ M4QEND "]]" FLEX_EXIT(EXIT_FAILURE); } /* Skip trailing whitespace. */ - for ( i = strlen( nmdef ) - 1; - i >= 0 && (nmdef[i] == ' ' || nmdef[i] == '\t'); - --i ) - ; - - nmdef[i + 1] = '\0'; + { + size_t i = strlen( nmdef ); + while (i > 0 && (nmdef[i-1] == ' ' || nmdef[i-1] == '\t')) + --i; + nmdef[i] = '\0'; + } ndinstal( nmstr, nmdef ); didadef = true; @@ -338,8 +363,6 @@ M4QEND "]]" interactive = option_sense; } array yytext_is_array = option_sense; - ansi-definitions ansi_func_defs = option_sense; - ansi-prototypes ansi_func_protos = option_sense; backup backing_up_report = option_sense; batch interactive = ! option_sense; bison-bridge bison_bridge_lval = option_sense; @@ -437,7 +460,7 @@ M4QEND "]]" \"[^"\n]*\" { if(yyleng-1 < MAXLINE) { - strcpy( nmstr, yytext + 1 ); + strncpy( nmstr, yytext + 1, sizeof(nmstr) ); } else { @@ -462,19 +485,20 @@ M4QEND "]]" ^"%{".* ++bracelevel; yyless( 2 ); /* eat only %{ */ ^"%}".* --bracelevel; yyless( 2 ); /* eat only %} */ - ^{WS}.* ACTION_ECHO; /* indented code in prolog */ - - ^{NOT_WS}.* { /* non-indented code */ - if ( bracelevel <= 0 ) - { /* not in %{ ... %} */ - yyless( 0 ); /* put it all back */ - yy_set_bol( 1 ); - mark_prolog(); - BEGIN(SECT2); - } - else - ACTION_ECHO; - } + ^{WS} START_CODEBLOCK(true); /* indented code in prolog */ + + ^{NOT_WS}.* { + /* non-indented code */ + if ( bracelevel <= 0 ) { + /* not in %{ ... %} */ + yyless( 0 ); /* put it all back */ + yy_set_bol( 1 ); + mark_prolog(); + BEGIN(SECT2); + } else { + START_CODEBLOCK(true); + } + } . ACTION_ECHO; {NL} ++linenum; ACTION_ECHO; @@ -532,6 +556,7 @@ M4QEND "]]" yyless(amt); } else { + add_action("]""]"); continued_action = true; ++linenum; return '\n'; @@ -601,9 +626,10 @@ M4QEND "]]" ^"%%".* { sectnum = 3; - BEGIN(SECT3); + BEGIN(no_section3_escape ? SECT3_NOESCAPE : SECT3); outn("/* Begin user sect3 */"); yyterminate(); /* to stop the parser */ + } "["({FIRST_CCL_CHAR}|{CCL_EXPR})({CCL_CHAR}|{CCL_EXPR})* { @@ -611,7 +637,7 @@ M4QEND "]]" if(yyleng < MAXLINE) { - strcpy( nmstr, yytext ); + strncpy( nmstr, yytext, sizeof(nmstr) ); } else { @@ -669,7 +695,7 @@ M4QEND "]]" if(yyleng-1 < MAXLINE) { - strcpy( nmstr, yytext + 1 ); + strncpy( nmstr, yytext + 1, sizeof(nmstr) ); } else { @@ -685,7 +711,7 @@ nmstr[yyleng - 2 - end_is_ws] = '\0'; /* chop trailing brace */ else { /* push back name surrounded by ()'s */ - int len = strlen( nmdefptr ); + size_t len = strlen( nmdefptr ); if (end_is_ws) unput(end_ch); @@ -876,35 +902,31 @@ nmstr[yyleng - 2 - end_is_ws] = '\0'; /* chop trailing brace */ <PERCENT_BRACE_ACTION>{ {OPTWS}"%}".* bracelevel = 0; - <ACTION>"/*" ACTION_ECHO; yy_push_state( COMMENT ); + <ACTION>"/*" ACTION_ECHO; yy_push_state( CODE_COMMENT ); <CODEBLOCK,ACTION>{ - "reject" { - ACTION_ECHO; - CHECK_REJECT(yytext); - } - "yymore" { - ACTION_ECHO; - CHECK_YYMORE(yytext); - } + "reject" { + ACTION_ECHO; + CHECK_REJECT(yytext); + } + "yymore" { + ACTION_ECHO; + CHECK_YYMORE(yytext); + } } - {M4QSTART} ACTION_ECHO_QSTART; - {M4QEND} ACTION_ECHO_QEND; - . ACTION_ECHO; - {NL} { - ++linenum; - ACTION_ECHO; - if ( bracelevel == 0 || - (doing_codeblock && indented_code) ) - { - if ( doing_rule_action ) - add_action( "\tYY_BREAK\n" ); - - doing_rule_action = doing_codeblock = false; - BEGIN(SECT2); - } - } + . ACTION_ECHO; + {NL} { + ++linenum; + ACTION_ECHO; + if (bracelevel <= 0 || (doing_codeblock && indented_code)) { + if ( doing_rule_action ) + add_action( "\tYY_BREAK]""]\n" ); + + doing_rule_action = doing_codeblock = false; + BEGIN(SECT2); + } + } } @@ -912,37 +934,41 @@ nmstr[yyleng - 2 - end_is_ws] = '\0'; /* chop trailing brace */ <ACTION>{ "{" ACTION_ECHO; ++bracelevel; "}" ACTION_ECHO; --bracelevel; - {M4QSTART} ACTION_ECHO_QSTART; - {M4QEND} ACTION_ECHO_QEND; - [^[:alpha:]_{}"'/\n\[\]]+ ACTION_ECHO; - [\[\]] ACTION_ECHO; - {NAME} ACTION_ECHO; - "'"([^'\\\n]|\\.)*"'" ACTION_ECHO; /* character constant */ + [^[:alpha:]_{}\"'/\n\[\]]+ ACTION_ECHO; + {NAME} ACTION_ECHO; + "'"([^\'\\\n]|\\.)"'" ACTION_ECHO; /* character constant */ + "'" ACTION_ECHO; BEGIN(CHARACTER_CONSTANT); \" ACTION_ECHO; BEGIN(ACTION_STRING); - {NL} { - ++linenum; - ACTION_ECHO; - if ( bracelevel == 0 ) - { - if ( doing_rule_action ) - add_action( "\tYY_BREAK\n" ); + {NL} { + ++linenum; + ACTION_ECHO; + if (bracelevel <= 0) { + if ( doing_rule_action ) + add_action( "\tYY_BREAK]""]\n" ); - doing_rule_action = false; - BEGIN(SECT2); - } - } - . ACTION_ECHO; + doing_rule_action = false; + BEGIN(SECT2); + } + } + . ACTION_ECHO; } <ACTION_STRING>{ - [^"\\\n]+ ACTION_ECHO; - \\. ACTION_ECHO; - {NL} ++linenum; ACTION_ECHO; BEGIN(ACTION); + [^\[\]\"\\\n]+ ACTION_ECHO; \" ACTION_ECHO; BEGIN(ACTION); - . ACTION_ECHO; +} +<CHARACTER_CONSTANT>{ + [^\[\]\'\\\n]+ ACTION_ECHO; + \' ACTION_ECHO; BEGIN(ACTION); +} +<ACTION_STRING,CHARACTER_CONSTANT>{ + (\\\n)* ACTION_ECHO; + \\(\\\n)*. ACTION_ECHO; + {NL} ++linenum; ACTION_ECHO; if (bracelevel <= 0) { BEGIN(SECT2); } else { BEGIN(ACTION); } + . ACTION_ECHO; } -<COMMENT,COMMENT_DISCARD,ACTION,ACTION_STRING><<EOF>> { +<COMMENT,CODE_COMMENT,COMMENT_DISCARD,ACTION,ACTION_STRING,CHARACTER_CONSTANT><<EOF>> { synerr( _( "EOF encountered inside an action" ) ); yyterminate(); } @@ -961,15 +987,26 @@ nmstr[yyleng - 2 - end_is_ws] = '\0'; /* chop trailing brace */ return CHAR; } - <SECT3>{ - {M4QSTART} fwrite (escaped_qstart, 1, strlen(escaped_qstart), yyout); - {M4QEND} fwrite (escaped_qend, 1, strlen(escaped_qend), yyout); - [^\[\]\n]*(\n?) ECHO; - (.|\n) ECHO; - <<EOF>> sectnum = 0; yyterminate(); + {M4QSTART} fputs(escaped_qstart, yyout); + {M4QEND} fputs(escaped_qend, yyout); + [^\[\]]* ECHO; + [][] ECHO; + <<EOF>> { + sectnum = 0; + yyterminate(); + } +} +<SECT3_NOESCAPE>{ + {M4QSTART} fprintf(yyout, "[""[%s]""]", escaped_qstart); + {M4QEND} fprintf(yyout, "[""[%s]""]", escaped_qend); + [^][]* ECHO; + [][] ECHO; + <<EOF>> { + sectnum = 0; + yyterminate(); + } } - <*>.|\n format_synerr( _( "bad character: %s" ), yytext ); %% |