static BOOLEAN forgetindent = FALSE; /* Don't bother applying indentation rules */
static int nested_comments; /* For counting comment nesting depth */
+static int comment_start;
/* OLD: Hacky definition of yywrap: see flex doc.
static int Return PROTO((int));
static void hsentercontext PROTO((int));
+static BOOLEAN is_commment PROTO((char*, int));
+
/* Special file handling for IMPORTS */
/* Note: imports only ever go *one deep* (hence no need for a stack) WDP 94/09 */
S [!#$%&*+./<=>?@\\^|\-~:\xa1-\xbf\xd7\xf7]
SId {S}{S}*
L [A-Z\xc0-\xd6\xd8-\xde]
-l [a-z\xdf-\xf6\xf8-\xff]
+l [a-z_\xdf-\xf6\xf8-\xff]
I {L}|{l}
i {L}|{l}|[0-9'_]
Id {I}{i}*
*/
%}
-<Code,GlaExt,StringEsc>"--"[^\n\r]*{NL}?{WS}* |
<Code,GlaExt,UserPragma,StringEsc>{WS}+ { noGap = FALSE; }
%{
* Special GHC pragma rules. Do we need a start state for interface files,
* so these won't be matched in source files? --JSM
*/
+
%}
+%{
+/* I believe the next rule is not ever matched.
+
+ The '#line ' rule is un-cool, recognising a cpp directive inside hs source.
+ Driver has now been modified to output `standard' {-# LINE ..-} pragmas
+ where possible, so the lexer should now never see cpp directives
+ like '# ' and '#line'.
+
+ -- SOF
+
<Code,GlaExt>^"# ".*{NL} {
char tempf[FILENAME_SIZE];
sscanf(yytext+1, "%d \"%[^\"]", &hslineno, tempf);
new_filename(tempf);
hsplineno = hslineno; hscolno = 0; hspcolno = 0;
}
+*/
+%}
<Code,GlaExt>"{-# LINE ".*"-}"{NL} {
/* partain: pragma-style line directive */
PUSH_STATE(UserPragma);
RETURN(SPECIALISE_UPRAGMA);
}
+<Code,GlaExt>"{-#"{WS}*"speciali"[sz]e {
+ PUSH_STATE(UserPragma);
+ RETURN(SPECIALISE_UPRAGMA);
+ }
<Code,GlaExt>"{-#"{WS}*"INLINE" {
PUSH_STATE(UserPragma);
RETURN(INLINE_UPRAGMA);
}
+<Code,GlaExt>"{-#"{WS}*"RULES" {
+ PUSH_STATE(Code); /* I'm not sure about this */
+ RETURN(RULES_UPRAGMA);
+ }
+<Code,GlaExt>"{-#"{WS}*"inline" {
+ PUSH_STATE(UserPragma);
+ RETURN(INLINE_UPRAGMA);
+ }
+<Code,GlaExt>"{-#"{WS}*"NOINLINE" {
+ PUSH_STATE(UserPragma);
+ RETURN(NOINLINE_UPRAGMA);
+ }
+<Code,GlaExt>"{-#"{WS}*"notInline" {
+ PUSH_STATE(UserPragma);
+ RETURN(NOINLINE_UPRAGMA);
+ }
<Code,GlaExt>"{-#"{WS}*"MAGIC_UNFOLDING" {
PUSH_STATE(UserPragma);
RETURN(MAGIC_UNFOLDING_UPRAGMA);
}
-<Code,GlaExt>"{-#"{WS}*"DEFOREST" {
- PUSH_STATE(UserPragma);
- RETURN(DEFOREST_UPRAGMA);
- }
<Code,GlaExt>"{-#"{WS}*"GENERATE_SPECS" {
/* these are handled by hscpp */
- nested_comments =1;
+ nested_comments =1; comment_start = hsplineno;
PUSH_STATE(Comment);
}
<Code,GlaExt>"{-#"{WS}*"OPTIONS" {
/* these are for the driver! */
- nested_comments =1;
+ nested_comments =1; comment_start = hsplineno;
PUSH_STATE(Comment);
}
<Code,GlaExt>"{-#"{WS}*"SOURCE"{WS}*"#"?"-}" {
/* these are used by `make depend' and the
compiler to indicate that a module should
be imported from source */
- nested_comments =1;
+ nested_comments =1; comment_start = hsplineno;
RETURN(SOURCE_UPRAGMA);
}
-<Code,GlaExt>"{-#"{WS}*[A-Z_]+ {
+<Code,GlaExt>"{-#"{WS}*[a-zA-Z_]+ {
fprintf(stderr, "%s:%d: Warning: Unrecognised pragma '",
input_filename, hsplineno);
format_string(stderr, (unsigned char *) yytext, yyleng);
fputs("'\n", stderr);
- nested_comments = 1;
+ nested_comments = 1; comment_start = hsplineno;
PUSH_STATE(Comment);
}
-<UserPragma>"#-}" { POP_STATE; RETURN(END_UPRAGMA); }
+<Code,GlaExt,UserPragma>"#-}" { POP_STATE; RETURN(END_UPRAGMA); }
%{
/*
<Code,GlaExt>"hiding" { RETURN(HIDING); }
<Code,GlaExt>"qualified" { RETURN(QUALIFIED); }
+<Code,GlaExt>"forall" { RETURN(FORALL); }
+
<Code,GlaExt>"_scc_" { RETURN(SCC); }
<GlaExt>"_ccall_" { RETURN(CCALL); }
<GlaExt>"_ccall_GC_" { RETURN(CCALL_GC); }
<GlaExt>"_casm_" { RETURN(CASM); }
<GlaExt>"_casm_GC_" { RETURN(CASM_GC); }
+<GlaExt>"(#" { RETURN(OUNBOXPAREN); }
+<GlaExt>"#)" { RETURN(CUNBOXPAREN); }
+<GlaExt>"foreign" { RETURN(FOREIGN); }
+<GlaExt>"export" { RETURN(EXPORT); }
+<GlaExt>"label" { RETURN(LABEL); }
+<GlaExt>"unsafe" { RETURN(UNSAFE); }
+<GlaExt>"_stdcall" { RETURN(STDCALL); }
+<GlaExt>"_ccall" { RETURN(C_CALL); }
+<GlaExt>"_pascal" { RETURN(PASCAL); }
+<GlaExt>"stdcall" { RETURN(STDCALL); }
+<GlaExt>"ccall" { RETURN(C_CALL); }
+<GlaExt>"pascal" { RETURN(PASCAL); }
+<GlaExt>"dynamic" { RETURN(DYNAMIC); }
%{
/*
<Code,GlaExt,UserPragma>"," { RETURN(COMMA); }
<Code,GlaExt>";" { RETURN(SEMI); }
<Code,GlaExt>"`" { RETURN(BQUOTE); }
-<Code,GlaExt>"_" { RETURN(WILDCARD); }
+<Code,GlaExt>"." { RETURN(DOT); }
<Code,GlaExt>".." { RETURN(DOTDOT); }
<Code,GlaExt,UserPragma>"::" { RETURN(DCOLON); }
<Code,GlaExt,UserPragma>"=" { RETURN(EQUAL); }
RETURN(isconstr(yytext) ? CONID : VARID);
}
<Code,GlaExt,UserPragma>{SId} {
- hsnewid(yytext, yyleng);
- RETURN(isconstr(yytext) ? CONSYM : VARSYM);
+ if (is_commment(yytext,yyleng)) {
+ int c;
+ while ((c = input()) != '\n' && c != '\r' && c!= EOF )
+ ;
+ if (c != EOF)
+ unput(c);
+ } else {
+ hsnewid(yytext, yyleng);
+ RETURN(isconstr(yytext) ? CONSYM : VARSYM);
+ }
}
<Code,GlaExt,UserPragma>{Mod}"."{Id}"#" {
BOOLEAN is_constr;
<CharEsc>\\ { addchar(*yytext); POP_STATE; }
<StringEsc>\\ { if (noGap) { addchar(*yytext); } POP_STATE; }
+%{
+/*
+ Not 100% correct, tokenizes "foo \ --<>--
+ \ bar"
+
+ as "foo bar", but this is not correct as per Haskell 98 report and its
+ maximal munch rule for "--"-style comments.
+
+ For the moment, not deemed worthy to fix.
+*/
+%}
+<StringEsc>"--"[^\n\r]*{NL}?{WS}* { noGap=FALSE; }
+
<CharEsc,StringEsc>["'] { addchar(*yytext); POP_STATE; }
<CharEsc,StringEsc>NUL { addchar('\000'); POP_STATE; }
<CharEsc,StringEsc>SOH { addchar('\001'); POP_STATE; }
%}
<Code,GlaExt,UserPragma,StringEsc>"{-" {
- noGap = FALSE; nested_comments = 1; PUSH_STATE(Comment);
+ noGap = FALSE; nested_comments = 1; comment_start = hsplineno; PUSH_STATE(Comment);
}
<Comment>[^-{]* |
<Comment>"-}" { if (--nested_comments == 0) POP_STATE; }
<Comment>(.|\n) ;
+
%{
/*
* Illegal characters. This used to be a single rule, but we might as well
hsperror("unterminated character literal");
}
<Comment><<EOF>> {
+ char errbuf[ERR_BUF_SIZE];
hsplineno = hslineno; hspcolno = hscolno;
- hsperror("unterminated comment");
+ sprintf(errbuf, "unterminated comment (which started on line %d)", comment_start);
+ hsperror(errbuf);
}
<String,StringEsc><<EOF>> {
hsplineno = hslineno; hspcolno = hscolno;
forcing insertion of ; or } as appropriate
*/
+#ifdef HSP_DEBUG
+#define LAYOUT_DEBUG
+#endif
+
+
static BOOLEAN
hsshouldindent(void)
{
void
hssetindent(void)
{
-#ifdef HSP_DEBUG
+#ifdef LAYOUT_DEBUG
fprintf(stderr, "hssetindent:hscolno=%d,hspcolno=%d,INDENTPT[%d]=%d\n", hscolno, hspcolno, icontexts, INDENTPT);
#endif
void
hsincindent(void)
{
-#ifdef HSP_DEBUG
+#ifdef LAYOUT_DEBUG
fprintf(stderr, "hsincindent:hscolno=%d,hspcolno=%d,INDENTPT[%d]=%d\n", hscolno, hspcolno, icontexts, INDENTPT);
#endif
hsentercontext(indenttab[icontexts] & ~1);
}
forgetindent = FALSE;
indenttab[icontexts] = indent;
-#ifdef HSP_DEBUG
+#ifdef LAYOUT_DEBUG
fprintf(stderr, "hsentercontext:indent=%d,hscolno=%d,hspcolno=%d,INDENTPT[%d]=%d\n", indent, hscolno, hspcolno, icontexts, INDENTPT);
#endif
}
hsendindent(void)
{
--icontexts;
-#ifdef HSP_DEBUG
+#ifdef LAYOUT_DEBUG
fprintf(stderr, "hsendindent:hscolno=%d,hspcolno=%d,INDENTPT[%d]=%d\n", hscolno, hspcolno, icontexts, INDENTPT);
#endif
}
/*
* Return checks the indentation level and returns ;, } or the specified token.
*/
-
static int
Return(int tok)
{
#ifdef HSP_DEBUG
extern int yyleng;
#endif
-
if (hsshouldindent()) {
if (hspcolno < INDENTPT) {
#ifdef HSP_DEBUG
return (SEMI);
}
}
+
hssttok = -1;
#ifdef HSP_DEBUG
fprintf(stderr, "returning %d (%d:%d)\n", tok, hspcolno, INDENTPT);
return isconstr(dot+1);
}
+
+static
+BOOLEAN
+is_commment(char* lexeme, int len)
+{
+ char* ptr;
+ int i;
+
+ if (len < 2) {
+ return FALSE;
+ }
+
+ for(i=0;i<len;i++) {
+ if (lexeme[i] != '-') return FALSE;
+ }
+ return TRUE;
+}
+
projects / ghc-hetmet.git / blobdiff