aboutsummaryrefslogtreecommitdiff
path: root/gcc/cobol/scan.l
diff options
context:
space:
mode:
Diffstat (limited to 'gcc/cobol/scan.l')
-rw-r--r--gcc/cobol/scan.l217
1 files changed, 111 insertions, 106 deletions
diff --git a/gcc/cobol/scan.l b/gcc/cobol/scan.l
index 2fc4aea..ba4c044 100644
--- a/gcc/cobol/scan.l
+++ b/gcc/cobol/scan.l
@@ -83,10 +83,13 @@ NONWORD [^[:alnum:]$_-]+
SPC [[:space:]]+
OSPC [[:space:]]*
+BLANK [[:blank:]]+
+OBLANK [[:blank:]]*
EOL \r?\n
BLANK_EOL [[:blank:]]*{EOL}
BLANK_OEOL [[:blank:]]*{EOL}?
+PICTURE [^[:space:]]+
DOTSEP [.]+[[:space:]]
DOTEOL [[:blank:]]*[.]{BLANK_EOL}
@@ -160,7 +163,7 @@ COMMA [,;][[:blank:]]*
ISNT (IS{SPC})?NOT
-COMMENTARY DATE-COMPILED|DATE-WRITTEN|INSTALLATION|SECURITY
+COMMENTARY AUTHOR|DATE-COMPILED|DATE-WRITTEN|INSTALLATION|SECURITY
SORT_MERGE SORT(-MERGE)?
@@ -174,7 +177,7 @@ SIZE_ERROR (ON[[[:space:]]+)?SIZE[[:space:]]+ERROR
VARTYPE NUMERIC|ALPHABETIC|ALPHABETIC_LOWER|ALPHABETIC_UPPER|DBCS|KANJI
NAMTYP {NAME}|{VARTYPE}
-NL [[:blank:]]*\r?\n[[:blank:]]*
+NL [[:blank:]]*{EOL}[[:blank:]]*
PUSH_FILE \f?[#]FILE{SPC}PUSH{SPC}[^\f]+\f
POP_FILE \f?[#]FILE{SPC}POP\f
@@ -182,7 +185,7 @@ LINE_DIRECTIVE ^[#]line{SPC}[[:alnum:]]+{SPC}[""''].+\n
%x procedure_div ident_state addr_of function classify
%x program_id_state comment_entries
-%x author_state date_state field_level field_state dot_state
+%x date_state field_level field_state dot_state
%x numeric_state name_state
%x quoted1 quoted2 quoteq
%x picture picture_count integer_count
@@ -238,30 +241,23 @@ WORKING-STORAGE{SPC}SECTION {
yy_push_state(field_state);
return WORKING_STORAGE_SECT; }
LOCAL-STORAGE{SPC}SECTION {
- yy_push_state(field_state);
- return LOCAL_STORAGE_SECT; }
-WORKING-STORAGE {
- return WORKING_STORAGE; }
-LOCAL-STORAGE {
- return LOCAL_STORAGE; }
-SCREEN {
- return SCREEN; }
+ yy_push_state(field_state);
+ return LOCAL_STORAGE_SECT; }
+WORKING-STORAGE { return WORKING_STORAGE; }
+LOCAL-STORAGE { return LOCAL_STORAGE; }
+SCREEN { return SCREEN; }
LINKAGE{SPC}SECTION {
yy_push_state(field_state);
return LINKAGE_SECT; }
-FUNCTION-ID { yy_push_state(ident_state);
- yy_push_state(program_id_state);
- yy_push_state(name_state); return FUNCTION; }
-
-PROGRAM-ID { yy_push_state(ident_state);
- yy_push_state(program_id_state);
- yy_push_state(name_state); return PROGRAM_ID; }
+FUNCTION-ID{OSPC}{DOTSEP}? { yy_push_state(ident_state);
+ yy_push_state(program_id_state);
+ yy_push_state(name_state); return FUNCTION; }
-PROGRAM-ID/{DOTEOL} { yy_push_state(ident_state);
- yy_push_state(name_state);
- yy_push_state(dot_state); return PROGRAM_ID; }
+PROGRAM-ID{OSPC}{DOTSEP}? { yy_push_state(ident_state);
+ yy_push_state(program_id_state);
+ yy_push_state(name_state); return PROGRAM_ID; }
PROCEDURE{SPC}DIVISION { yy_push_state(procedure_div);
return PROCEDURE_DIV; }
@@ -272,30 +268,18 @@ PROCEDURE{SPC}DIVISION { yy_push_state(procedure_div);
}
<ident_state>{
+ {BLANK_OEOL}
ID(ENTIFICATION)?{SPC}DIVISION { myless(0); yy_pop_state(); }
+ (ENVIRONMENT|DATA|PROCEDURE){SPC}DIVISION {
+ myless(0); yy_pop_state(); }
+ OPTIONS { myless(0); yy_pop_state(); }
+
AS{SPC}[""] { yy_push_state(quoted2); return AS; }
AS{SPC}[''] { yy_push_state(quoted1); return AS; }
IS { pop_return IS; }
- OPTIONS { yy_pop_state(); myless(0); }
- [[:blank:]]*(ENVIRONMENT|DATA|PROCEDURE){SPC}DIVISION/[[:space:].] {
- yy_pop_state(); myless(0); }
- [[:blank:]]*AUTHOR[[:blank:].]+{EOL}? {
- // Might not have an EOL, but stop on one.
- yy_push_state(author_state); }
-
- {DOTEOL}
-
{COMMENTARY} { BEGIN(comment_entries); }
}
-<author_state>{
- [[:blank:]]+
- ^{BLANK_EOL}
- [^\r\n]+ { yy_pop_state();
- yylval.string = xstrdup(yytext);
- }
-}
-
<INITIAL>{
COBOL { return COBOL; }
@@ -307,6 +291,15 @@ PROCEDURE{SPC}DIVISION { yy_push_state(procedure_div);
yy_push_state(field_state);
yy_set_bol(1);
myless(0); }
+
+ END{SPC}PROGRAM { yy_push_state(name_state);
+ return program_level() > 1?
+ END_SUBPROGRAM : END_PROGRAM; }
+
+ END{SPC}FUNCTION { yy_push_state(name_state);
+ return program_level() > 1?
+ END_SUBPROGRAM /*invalid*/ :
+ END_FUNCTION; }
}
<INITIAL,procedure_div,cdf_state>{
@@ -444,6 +437,11 @@ STDOUT { return STDOUT; }
STDERR { return STDERR; }
SYSERR { return STDERR; }
+ARGUMENT-NUMBER { return ARGUMENT_NUMBER; }
+ARGUMENT-VALUE { return ARGUMENT_VALUE; }
+ENVIRONMENT-NAME { return ENVIRONMENT_NAME; }
+ENVIRONMENT-VALUE { return ENVIRONMENT_VALUE; }
+
CANCEL { return CANCEL; }
COMMIT { return COMMIT; }
COMMON { return COMMON; }
@@ -541,7 +539,7 @@ SECTION{SPC}[+-]?{INTEGERZ}/{OSPC}{DOTSEP} {
auto eotext = yytext + yyleng;
auto p = std::find_if(yytext, eotext, fisspace);
p = std::find_if(p, eotext, nonspace);
- yylval.string = p;
+ yylval.string = xstrdup(p);
return SECTION;
}
@@ -968,7 +966,9 @@ USE({SPC}FOR)? { return USE; }
return NUMSTR;
}
- PIC(TURE)?({SPC}IS)?[[:space:]]{BLANK_OEOL} {
+ PIC(TURE)?({SPC}IS)?{SPC}{PICTURE} {
+ auto pos = validate_picture();
+ myless(pos);
yy_push_state(picture); return PIC; }
ANY { return ANY; }
@@ -1150,7 +1150,7 @@ USE({SPC}FOR)? { return USE; }
yy_push_state(hex_state); }
N?X{nonseq} { dbgmsg("invalid hexadecimal value: %s", yytext);
return NO_CONDITION; }
- [[:blank:]]*\r?\n {}
+ [[:blank:]]*{EOL} {}
WORKING-STORAGE{SPC}SECTION { return WORKING_STORAGE_SECT; }
LOCAL-STORAGE{SPC}SECTION { return LOCAL_STORAGE_SECT; }
@@ -1220,7 +1220,7 @@ USE({SPC}FOR)? { return USE; }
{NP}V?/[,.]? { yylval.number = ndigit(yyleng); return picset(PIC_P); }
{N9}*V/{N9}* { yylval.number = ndigit(yyleng - 1); return picset(NINEV); }
{N9}/{N9}*[,.]? { yylval.number = ndigit(yyleng); return picset(NINES); }
- P+/[,.]?\r?\n { yylval.number = yyleng; return picset(PIC_P); }
+ P+/[,.]?{EOL} { yylval.number = yyleng; return picset(PIC_P); }
1{1,31}/({COUNT}|[(]{NAME}[)]) {
yy_push_state(picture_count);
@@ -1319,7 +1319,7 @@ USE({SPC}FOR)? { return USE; }
[""]{SPC}[&]{SPC}[""''] {
if( yytext[yyleng - 1] == '\'' ) BEGIN(quoted1);
}
- [""]-{OSPC}(\r?\n{OSPC})+[""] /* continue ... */
+ [""]-{OSPC}({EOL}{OSPC})+[""] /* continue ... */
[""] {
char *s = xstrdup(tmpstring? tmpstring : "\0");
yylval.literal.set_data(strlen(s), s);
@@ -1336,7 +1336,7 @@ USE({SPC}FOR)? { return USE; }
['']{SPC}[&]{SPC}[""''] {
if( yytext[yyleng - 1] == '"' ) BEGIN(quoted2);
}
- ['']-{OSPC}(\r?\n{OSPC})+[''] /* continue ... */
+ ['']-{OSPC}({EOL}{OSPC})+[''] /* continue ... */
[''] {
char *s = xstrdup(tmpstring? tmpstring : "\0");
yylval.literal.set_data(strlen(s), s);
@@ -1384,45 +1384,36 @@ USE({SPC}FOR)? { return USE; }
}
<program_id_state>{
- ^[[:blank:]]+
- ^{BLANK_EOL}
+ {BLANK_OEOL}
(IS)?[[:space:]]
+ AS/{SPC} { myless(0); yy_pop_state(); } /* => ident_state */
- COMMON/[.]|{SPC}[[:alnum:].] { return COMMON; }
- INITIAL/[.]|{SPC}[[:alnum:].] { return INITIAL_kw; }
- RECURSIVE { return RECURSIVE; }
- PROGRAM/[.]|{SPC}[[:alnum:].] { return PROGRAM_kw; }
-
- INITIAL { pop_return INITIAL_kw; }
- COMMON { pop_return COMMON; }
- PROGRAM { pop_return PROGRAM; }
+ INITIAL { return INITIAL_kw; }
+ COMMON { return COMMON; }
+ RECURSIVE { return RECURSIVE; }
+ PROGRAM { return PROGRAM_kw; }
- AS/{SPC} { myless(0); yy_pop_state(); } /* => ident_state */
- [[:blank:]]*{DOTSEP}[[:blank:].]+{EOL} { pop_return '.'; }
- {DOTEOL} { pop_return '.'; }
+ {DOTSEP} { pop_return '.'; }
}
-<name_state>{
- ^[[:blank:]]+
- ^{BLANK_EOL}
- {NAME} |
- {NAME}/{OSPC}[.] { yy_pop_state();
- yylval.string = xstrdup(yytext); return NAME; }
+<name_state>{ /* Either pop from here, or let the quoted state pop */
+ {BLANK_OEOL}
+ {NAME} { yy_pop_state();
+ yylval.string = xstrdup(yytext);
+ return NAME;
+ }
Z?[''] { yylval.literal.set_prefix(yytext, yyleng-1);
- yy_push_state(quoted1); }
+ BEGIN(quoted1); }
Z?[""] { yylval.literal.set_prefix(yytext, yyleng-1);
- yy_push_state(quoted2); }
-
- [.]/[[:blank:]]+. { return *yytext; }
+ BEGIN(quoted2); }
- [[:blank:]]*{DOTSEP}[[:blank:].]+{EOL} {
- yy_pop_state(); myless(0); }
- {DOTEOL} { yy_pop_state(); myless(0); }
+ . { myless(0); yy_pop_state();
+ /* Should not happen for valid inputs. */ }
}
<dot_state>{
[[:blank:]]*[.][[:blank:].]+{EOL} { pop_return '.'; }
- [[:blank:]]*[.] { pop_return '.'; }
+ [[:blank:]]*[.]+ { pop_return '.'; }
}
<date_state>{
@@ -1645,9 +1636,9 @@ B-SHIFT-RC
FUNCTION { yy_push_state(function); return FUNCTION; }
- SECTION{OSPC}[.]{SPC}/USE[[:space:]] { yylval.string = NULL; return SECTION; }
+ SECTION{OSPC}[.]+{SPC}/USE[[:space:]] { yylval.string = NULL; return SECTION; }
- [.]({SPC}(EJECT|SKIP[123]))*{SPC}EXIT{OSPC}/{DOTSEP} {
+ [.]+({SPC}(EJECT|SKIP[123]))*{SPC}EXIT{OSPC}/{DOTSEP} {
// EXIT format-1 is a "continue" statement
}
{NAME}/{OSPC}{DOTSEP} {
@@ -1682,16 +1673,17 @@ B-SHIFT-RC
p += 2;
while( ISSPACE(*p) ) p++;
cbl_name_t name2;
- std::transform( p, p + sizeof(name2), name2,
- []( char ch ) {
- switch(ch) {
- case '-':
- case '_': return ch;
- default:
- if( ISALNUM(ch) ) return ch;
- }
- return '\0';
- } );
+ const char *pend = p + sizeof(name2);
+ char *pout = name2;
+ while( p < pend ) {
+ char ch = *p++;
+ if( ISALNUM(ch) || ch == '-' || ch == '_' ) {
+ *pout++ = ch;
+ } else {
+ *pout++ = '\0';
+ break;
+ }
+ }
symbol_elem_t *e = symbol_file(PROGRAM, name2);
/*
* For NAME IN FILENAME, we want the parser to handle it.
@@ -2052,7 +2044,7 @@ BASIS { yy_push_state(basis); return BASIS; }
return symbol_file(PROGRAM, yytext)? FILENAME : NAME;
}
[[:blank:]]+
- \r?\n { yy_pop_state(); }
+ {EOL} { yy_pop_state(); }
}
<raising>{
@@ -2073,49 +2065,62 @@ BASIS { yy_push_state(basis); return BASIS; }
if( include_debug() ) myless(7);
}
}
- ^[ ]*>>{OSPC}IF { yy_push_state(cdf_state); return CDF_IF; }
- ^[ ]*>>{OSPC}ELSE { return CDF_ELSE; }
- ^[ ]*>>{OSPC}END-IF { return CDF_END_IF; }
+ ^[ ]*>>{OBLANK}IF { yy_push_state(cdf_state); return CDF_IF; }
+ ^[ ]*>>{OBLANK}ELSE { return CDF_ELSE; }
+ ^[ ]*>>{OBLANK}END-IF { return CDF_END_IF; }
- ^[ ]*[$]{OSPC}IF { if( ! dialect_mf() ) {
+ ^[ ]*[$]{OBLANK}IF { if( ! dialect_mf() ) {
dialect_error(yylloc, yytext, "mf");
}
yy_push_state(cdf_state); return CDF_IF; }
- ^[ ]*[$]{OSPC}ELSE { if( ! dialect_mf() ) {
+ ^[ ]*[$]{OBLANK}ELSE { if( ! dialect_mf() ) {
dialect_error(yylloc, yytext, "mf");
}
return CDF_ELSE; }
- ^[ ]*[$]{OSPC}END { if( ! dialect_mf() ) {
+ ^[ ]*[$]{OBLANK}END { if( ! dialect_mf() ) {
dialect_error(yylloc, yytext, "mf");
}
return CDF_END_IF; }
- ^[ ]*[$]{OSPC}SET({SPC}CONSTANT)? {
+ ^[ ]*[$]{OBLANK}SET({SPC}CONSTANT)? {
if( ! dialect_mf() ) dialect_error(yylloc, yytext, "mf");
yy_push_state(cdf_state); return CDF_DEFINE; }
- ^[ ]*>>{OSPC}EVALUATE { return CDF_EVALUATE; }
- ^[ ]*>>{OSPC}WHEN { return CDF_WHEN; }
- ^[ ]*>>{OSPC}END-EVALUATE { return CDF_END_EVALUATE; }
+ ^[ ]*>>{OBLANK}EVALUATE { return CDF_EVALUATE; }
+ ^[ ]*>>{OBLANK}WHEN { return CDF_WHEN; }
+ ^[ ]*>>{OBLANK}END-EVALUATE { return CDF_END_EVALUATE; }
+
+ ^[ ]*>>{OBLANK}CALL-CONVENTION{BLANK}C { return CALL_VERBATIM; }
+ ^[ ]*>>{OBLANK}CALL-CONVENTION{BLANK}COBOL { return CALL_COBOL; }
+ ^[ ]*>>{OBLANK}CALL-CONVENTION{BLANK}VERBATIM { return CALL_VERBATIM; }
+
+ ^[ ]*>>{OBLANK}DEFINE { yy_push_state(cdf_state); return CDF_DEFINE; }
+ ^[ ]*>>{OBLANK}DISPLAY { return CDF_DISPLAY; }
+ ^[ ]*>>{OBLANK}TURN { yy_push_state(exception); return TURN; }
+ ^[ ]*>>{OBLANK}COBOL-WORDS { yy_push_state(cobol_words); return COBOL_WORDS; }
- ^[ ]*>>{OSPC}CALL-CONVENTION{SPC}C { return CALL_VERBATIM; }
- ^[ ]*>>{OSPC}CALL-CONVENTION{SPC}COBOL { return CALL_COBOL; }
- ^[ ]*>>{OSPC}CALL-CONVENTION{SPC}VERBATIM { return CALL_VERBATIM; }
+ ^[ ]*>>{OBLANK}SOURCE{BLANK}FORMAT { return SOURCE_FORMAT; }
- ^[ ]*>>{OSPC}DEFINE { yy_push_state(cdf_state); return CDF_DEFINE; }
- ^[ ]*>>{OSPC}DISPLAY { return CDF_DISPLAY; }
- ^[ ]*>>{OSPC}TURN { yy_push_state(exception); return TURN; }
- ^[ ]*>>{OSPC}COBOL-WORDS { yy_push_state(cobol_words); return COBOL_WORDS; }
+ ^[ ]*>>{OBLANK}PUSH { return CDF_PUSH; }
+ ^[ ]*>>{OBLANK}POP { return CDF_POP; }
- ^[ ]*>>{OSPC}{NAME} {
+ ^[ ]*>>{OBLANK}{NAME} {
error_msg(yylloc, "unknown CDF token: %s", yytext);
}
+
OTHER { return OTHER; }
OVERRIDE { return OVERRIDE; }
PARAMETER { return PARAMETER_kw; }
THRU { return THRU; }
TRUE { return TRUE_kw; }
+
+ ALL { return ALL; }
+ CALL-CONVENTION { return CALL_CONVENTION; }
+ COBOL-WORDS { return COBOL_WORDS; }
+ DEFINE { return CDF_DEFINE; }
+ SOURCE{BLANK}FORMAT { return SOURCE_FORMAT; }
+
}
<cobol_words>{
@@ -2165,10 +2170,10 @@ BASIS { yy_push_state(basis); return BASIS; }
<*>OR { return OR; }
<*>AND { return AND; }
-<*>{DOTSEP}[[:blank:].]+$ { return '.'; }
+<*>{DOTSEP} { return '.'; }
<*>[().=*/+&-] { return *yytext; }
<*>[[:blank:]]+
-<*>\r?\n
+<*>{EOL}
<*>{
{COMMA}
@@ -2369,7 +2374,7 @@ BASIS { yy_push_state(basis); return BASIS; }
POINTER { return POINTER; }
POSITIVE { return POSITIVE; }
PROCEDURE { return PROCEDURE; }
- PROGRAM { return PROGRAM; }
+ PROGRAM { return PROGRAM_kw; }
PROGRAM-ID { return PROGRAM_ID; }
PROPERTY { return PROPERTY; }
PROTOTYPE { return PROTOTYPE; }
@@ -2411,7 +2416,7 @@ BASIS { yy_push_state(basis); return BASIS; }
SCREEN { return SCREEN; }
SD { return SD; }
SEARCH { return SEARCH; }
- SECTION { return SECTION; }
+ SECTION { yylval.string = NULL; return SECTION; }
SELECT { return SELECT; }
SENTENCE { return SENTENCE; }
SEPARATE { return SEPARATE; }