1berry/ruby
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

* ext/date/date_parse.c (date__parse): revised the tight parser.

Browse Source 
git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@37356 b2dd03c8-39d4-4d8f-98ff-823fe69b080e
This commit is contained in:
tadf 2012-10-28 09:53:08 +00:00
parent af83fcfe4d
commit b4c6d6a3a5
2 changed files with 207 additions and 87 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
ChangeLog
View File

@ -1,3 +1,7 @@
Sun Oct 28 18:51:33 2012 Tadayoshi Funaba <tadf@dotrb.org>
* ext/date/date_parse.c (date__parse): revised the tight parser.
Sun Oct 28 15:41:50 2012 Kouhei Sutou <kou@cozmixng.org> Sun Oct 28 15:41:50 2012 Kouhei Sutou <kou@cozmixng.org>
* lib/rexml/document.rb (REXML::Document#write): Add :encoding option * lib/rexml/document.rb (REXML::Document#write): Add :encoding option

288
ext/date/date_parse.c
View File

@ -57,7 +57,8 @@ static const char *abbr_months[] = {
#define issign(c) ((c) == '-' || (c) == '+') #define issign(c) ((c) == '-' || (c) == '+')
#define asp_string() rb_str_new(" ", 1) #define asp_string() rb_str_new(" ", 1)
#ifdef TIGHT_PARSER #ifdef TIGHT_PARSER
#define asub_string() rb_str_new("\032", 1) #define asubw_string() rb_str_new("\027", 1)
#define asubt_string() rb_str_new("\024", 1)
#endif #endif
static void static void
@ -223,13 +224,17 @@ s3e(VALUE hash, VALUE y, VALUE m, VALUE d, int bc)
#define ABBR_MONTHS "jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec" #define ABBR_MONTHS "jan|feb|mar|apr|may|jun|jul|aug|sep|oct|nov|dec"
#ifdef TIGHT_PARSER #ifdef TIGHT_PARSER
#define VALID_DAYS DAYS "|tues|wednes|thur|thurs|" ABBR_DAYS #define VALID_DAYS "(?:" DAYS ")\\b" "|(?:tues|wednes|thur|thurs|" ABBR_DAYS ")\\b\\.?"
#define VALID_MONTHS MONTHS "|sept|" ABBR_MONTHS #define VALID_MONTHS "(?:" MONTHS ")\\b" "|(?:sept|" ABBR_MONTHS ")\\b\\.?"
#define DOTLESS_VALID_MONTHS "(?:" MONTHS ")\\b" "|(?:sept|" ABBR_MONTHS ")\\b"
#define BOS "\\A\\s*" #define BOS "\\A\\s*"
#define FP_CODE "\\032" #define FPW "\\027"
#define FP_COM "\\s*(?:" FP_CODE "\\s*,?)?\\s*" #define FPT "\\024"
#define COM_FP "\\s*(?:,?\\s*" FP_CODE ")?\\s*" #define FPW_COM "\\s*(?:" FPW "\\s*,?)?\\s*"
#define FP_ISO "\\s*(?:[tT]?" FP_CODE ")?" #define FPT_COM "\\s*(?:" FPT "\\s*,?)?\\s*"
#define COM_FPW "\\s*(?:,?\\s*" FPW ")?\\s*"
#define COM_FPT "\\s*(?:,?\\s*(?:@|\\b[aA][tT]\\b)?\\s*" FPT ")?\\s*"
#define TEE_FPT "\\s*(?:[tT]?" FPT ")?"
#define EOS "\\s*\\z" #define EOS "\\s*\\z"
#endif #endif
@ -272,13 +277,8 @@ match(VALUE str, VALUE pat, VALUE hash, int (*cb)(VALUE, VALUE))
return 1; return 1;
} }
#define SUBS(s,p,c) \
{ \
return subs(s, p, hash, c); \
}
static int static int
subs(VALUE str, VALUE pat, VALUE hash, int (*cb)(VALUE, VALUE)) subx(VALUE str, VALUE rep, VALUE pat, VALUE hash, int (*cb)(VALUE, VALUE))
{ {
VALUE m; VALUE m;
@ -292,39 +292,27 @@ subs(VALUE str, VALUE pat, VALUE hash, int (*cb)(VALUE, VALUE))
be = f_begin(m, INT2FIX(0)); be = f_begin(m, INT2FIX(0));
en = f_end(m, INT2FIX(0)); en = f_end(m, INT2FIX(0));
f_aset2(str, be, LONG2NUM(NUM2LONG(en) - NUM2LONG(be)), asp_string()); f_aset2(str, be, LONG2NUM(NUM2LONG(en) - NUM2LONG(be)), rep);
(*cb)(m, hash); (*cb)(m, hash);
} }
return 1; return 1;
} }
#define SUBS(s,p,c) \
{ \
return subx(s, asp_string(), p, hash, c); \
}
#ifdef TIGHT_PARSER #ifdef TIGHT_PARSER
#define SUBC(s,p,c) \ #define SUBW(s,p,c) \
{ \ { \
return subc(s, p, hash, c); \ return subx(s, asubw_string(), p, hash, c); \
} }
static int #define SUBT(s,p,c) \
subc(VALUE str, VALUE pat, VALUE hash, int (*cb)(VALUE, VALUE)) { \
{ return subx(s, asubt_string(), p, hash, c); \
VALUE m;
m = f_match(pat, str);
if (NIL_P(m))
return 0;
{
VALUE be, en;
be = f_begin(m, INT2FIX(0));
en = f_end(m, INT2FIX(0));
f_aset2(str, be, LONG2NUM(NUM2LONG(en) - NUM2LONG(be)), asub_string());
(*cb)(m, hash);
}
return 1;
} }
#endif #endif
@ -642,14 +630,17 @@ parse_day(VALUE str, VALUE hash)
#ifndef TIGHT_PARSER #ifndef TIGHT_PARSER
"\\b(" ABBR_DAYS ")[^-/\\d\\s]*" "\\b(" ABBR_DAYS ")[^-/\\d\\s]*"
#else #else
BOS "(" VALID_DAYS ")\\b\\.?" "(" VALID_DAYS ")"
"(?:,(?=\\s*\\S))?" /* , ... */
#endif #endif
; ;
static VALUE pat = Qnil; static VALUE pat = Qnil;
REGCOMP_I(pat); REGCOMP_I(pat);
#ifndef TIGHT_PARSER
SUBS(str, pat, parse_day_cb); SUBS(str, pat, parse_day_cb);
#else
SUBW(str, pat, parse_day_cb);
#endif
} }
static int static int
@ -736,7 +727,11 @@ parse_time(VALUE str, VALUE hash)
"(?:" "(?:"
"\\d+\\s*:\\s*\\d+" "\\d+\\s*:\\s*\\d+"
"(?:" "(?:"
#ifndef TIGHT_PARSER
"\\s*:\\s*\\d+(?:[,.]\\d*)?" "\\s*:\\s*\\d+(?:[,.]\\d*)?"
#else
"\\s*:\\s*\\d+(?:[,.]\\d+)?"
#endif
")?" ")?"
"|" "|"
"\\d+\\s*h(?:\\s*\\d+m?(?:\\s*\\d+s?)?)?" "\\d+\\s*h(?:\\s*\\d+m?(?:\\s*\\d+s?)?)?"
@ -764,7 +759,7 @@ parse_time(VALUE str, VALUE hash)
#ifndef TIGHT_PARSER #ifndef TIGHT_PARSER
SUBS(str, pat, parse_time_cb); SUBS(str, pat, parse_time_cb);
#else #else
SUBC(str, pat, parse_time_cb); SUBT(str, pat, parse_time_cb);
#endif #endif
} }
@ -804,18 +799,18 @@ parse_eu(VALUE str, VALUE hash)
static const char pat_source[] = static const char pat_source[] =
#ifdef TIGHT_PARSER #ifdef TIGHT_PARSER
BOS BOS
FP_COM FPW_COM FPT_COM
#endif #endif
#ifndef TIGHT_PARSER #ifndef TIGHT_PARSER
"'?(\\d+)[^-\\d\\s]*" "'?(\\d+)[^-\\d\\s]*"
#else #else
"'?(\\d+)(?:st|nd|rd|th)?\\b" "(\\d+)(?:st|nd|rd|th)?\\b"
#endif #endif
"\\s*" "\\s*"
#ifndef TIGHT_PARSER #ifndef TIGHT_PARSER
"(" ABBR_MONTHS ")[^-\\d\\s']*" "(" ABBR_MONTHS ")[^-\\d\\s']*"
#else #else
"(" VALID_MONTHS ")\\b\\.?" "(" VALID_MONTHS ")"
#endif #endif
"(?:" "(?:"
"\\s*" "\\s*"
@ -828,7 +823,7 @@ parse_eu(VALUE str, VALUE hash)
#endif #endif
")?" ")?"
#ifdef TIGHT_PARSER #ifdef TIGHT_PARSER
COM_FP COM_FPT COM_FPW
EOS EOS
#endif #endif
; ;
@ -875,19 +870,19 @@ parse_us(VALUE str, VALUE hash)
static const char pat_source[] = static const char pat_source[] =
#ifdef TIGHT_PARSER #ifdef TIGHT_PARSER
BOS BOS
FP_COM FPW_COM FPT_COM
#endif #endif
#ifndef TIGHT_PARSER #ifndef TIGHT_PARSER
"\\b(" ABBR_MONTHS ")[^-\\d\\s']*" "\\b(" ABBR_MONTHS ")[^-\\d\\s']*"
#else #else
"\\b(" VALID_MONTHS ")\\b\\.?" "\\b(" VALID_MONTHS ")"
#endif #endif
"\\s*" "\\s*"
#ifndef TIGHT_PARSER #ifndef TIGHT_PARSER
"('?\\d+)[^-\\d\\s']*" "('?\\d+)[^-\\d\\s']*"
#else #else
"('?\\d+)(?:st|nd|rd|th)?\\b" "(\\d+)(?:st|nd|rd|th)?\\b"
COM_FP COM_FPT
#endif #endif
"(?:" "(?:"
"\\s*,?" "\\s*,?"
@ -899,7 +894,7 @@ parse_us(VALUE str, VALUE hash)
"('?-?\\d+)" "('?-?\\d+)"
")?" ")?"
#ifdef TIGHT_PARSER #ifdef TIGHT_PARSER
COM_FP COM_FPT COM_FPW
EOS EOS
#endif #endif
; ;
@ -918,6 +913,11 @@ parse_iso_cb(VALUE m, VALUE hash)
mon = rb_reg_nth_match(2, m); mon = rb_reg_nth_match(2, m);
d = rb_reg_nth_match(3, m); d = rb_reg_nth_match(3, m);
#ifdef TIGHT_PARSER
if (*RSTRING_PTR(y) == '\'' && *RSTRING_PTR(d) == '\'')
return 0;
#endif
s3e(hash, y, mon, d, 0); s3e(hash, y, mon, d, 0);
return 1; return 1;
} }
@ -930,8 +930,9 @@ parse_iso(VALUE str, VALUE hash)
"('?[-+]?\\d+)-(\\d+)-('?-?\\d+)" "('?[-+]?\\d+)-(\\d+)-('?-?\\d+)"
#else #else
BOS BOS
FPW_COM FPT_COM
"('?[-+]?\\d+)-(\\d+)-('?-?\\d+)" "('?[-+]?\\d+)-(\\d+)-('?-?\\d+)"
FP_ISO TEE_FPT COM_FPW
EOS EOS
#endif #endif
; ;
@ -967,8 +968,9 @@ parse_iso21(VALUE str, VALUE hash)
"\\b(\\d{2}|\\d{4})?-?w(\\d{2})(?:-?(\\d))?\\b" "\\b(\\d{2}|\\d{4})?-?w(\\d{2})(?:-?(\\d))?\\b"
#else #else
BOS BOS
FPW_COM FPT_COM
"(\\d{2}|\\d{4})?-?w(\\d{2})(?:-?(\\d))?" "(\\d{2}|\\d{4})?-?w(\\d{2})(?:-?(\\d))?"
FP_ISO TEE_FPT COM_FPW
EOS EOS
#endif #endif
; ;
@ -996,8 +998,9 @@ parse_iso22(VALUE str, VALUE hash)
"-w-(\\d)\\b" "-w-(\\d)\\b"
#else #else
BOS BOS
FPW_COM FPT_COM
"-w-(\\d)" "-w-(\\d)"
FP_ISO TEE_FPT COM_FPW
EOS EOS
#endif #endif
; ;
@ -1030,8 +1033,9 @@ parse_iso23(VALUE str, VALUE hash)
"--(\\d{2})?-(\\d{2})\\b" "--(\\d{2})?-(\\d{2})\\b"
#else #else
BOS BOS
FPW_COM FPT_COM
"--(\\d{2})?-(\\d{2})" "--(\\d{2})?-(\\d{2})"
FP_ISO TEE_FPT COM_FPW
EOS EOS
#endif #endif
; ;
@ -1064,8 +1068,9 @@ parse_iso24(VALUE str, VALUE hash)
"--(\\d{2})(\\d{2})?\\b" "--(\\d{2})(\\d{2})?\\b"
#else #else
BOS BOS
FPW_COM FPT_COM
"--(\\d{2})(\\d{2})?" "--(\\d{2})(\\d{2})?"
FP_ISO TEE_FPT COM_FPW
EOS EOS
#endif #endif
; ;
@ -1097,8 +1102,9 @@ parse_iso25(VALUE str, VALUE hash)
"[,.](\\d{2}|\\d{4})-\\d{3}\\b" "[,.](\\d{2}|\\d{4})-\\d{3}\\b"
#else #else
BOS BOS
FPW_COM FPT_COM
"[,.](\\d{2}|\\d{4})-\\d{3}" "[,.](\\d{2}|\\d{4})-\\d{3}"
FP_ISO TEE_FPT COM_FPW
EOS EOS
#endif #endif
; ;
@ -1108,8 +1114,9 @@ parse_iso25(VALUE str, VALUE hash)
"\\b(\\d{2}|\\d{4})-(\\d{3})\\b" "\\b(\\d{2}|\\d{4})-(\\d{3})\\b"
#else #else
BOS BOS
FPW_COM FPT_COM
"(\\d{2}|\\d{4})-(\\d{3})" "(\\d{2}|\\d{4})-(\\d{3})"
FP_ISO TEE_FPT COM_FPW
EOS EOS
#endif #endif
; ;
@ -1141,8 +1148,9 @@ parse_iso26(VALUE str, VALUE hash)
"\\d-\\d{3}\\b" "\\d-\\d{3}\\b"
#else #else
BOS BOS
FPW_COM FPT_COM
"\\d-\\d{3}" "\\d-\\d{3}"
FP_ISO TEE_FPT COM_FPW
EOS EOS
#endif #endif
; ;
@ -1152,8 +1160,9 @@ parse_iso26(VALUE str, VALUE hash)
"\\b-(\\d{3})\\b" "\\b-(\\d{3})\\b"
#else #else
BOS BOS
FPW_COM FPT_COM
"-(\\d{3})" "-(\\d{3})"
FP_ISO TEE_FPT COM_FPW
EOS EOS
#endif #endif
; ;
@ -1231,8 +1240,9 @@ parse_jis(VALUE str, VALUE hash)
"\\b([mtsh])(\\d+)\\.(\\d+)\\.(\\d+)" "\\b([mtsh])(\\d+)\\.(\\d+)\\.(\\d+)"
#else #else
BOS BOS
FPW_COM FPT_COM
"([mtsh])(\\d+)\\.(\\d+)\\.(\\d+)" "([mtsh])(\\d+)\\.(\\d+)\\.(\\d+)"
FP_ISO TEE_FPT COM_FPW
EOS EOS
#endif #endif
; ;
@ -1251,6 +1261,11 @@ parse_vms11_cb(VALUE m, VALUE hash)
mon = rb_reg_nth_match(2, m); mon = rb_reg_nth_match(2, m);
y = rb_reg_nth_match(3, m); y = rb_reg_nth_match(3, m);
#ifdef TIGHT_PARSER
if (*RSTRING_PTR(d) == '\'' && !NIL_P(y) && *RSTRING_PTR(y) == '\'')
return 0;
#endif
mon = INT2FIX(mon_num(mon)); mon = INT2FIX(mon_num(mon));
s3e(hash, y, mon, d, 0); s3e(hash, y, mon, d, 0);
@ -1266,9 +1281,10 @@ parse_vms11(VALUE str, VALUE hash)
"-('?-?\\d+)" "-('?-?\\d+)"
#else #else
BOS BOS
"('?-?\\d+)-(" VALID_MONTHS ")" FPW_COM FPT_COM
"('?-?\\d+)-(" DOTLESS_VALID_MONTHS ")"
"-('?-?\\d+)" "-('?-?\\d+)"
COM_FP COM_FPT COM_FPW
EOS EOS
#endif #endif
; ;
@ -1287,6 +1303,11 @@ parse_vms12_cb(VALUE m, VALUE hash)
d = rb_reg_nth_match(2, m); d = rb_reg_nth_match(2, m);
y = rb_reg_nth_match(3, m); y = rb_reg_nth_match(3, m);
#ifdef TIGHT_PARSER
if (*RSTRING_PTR(d) == '\'' && !NIL_P(y))
return 0;
#endif
mon = INT2FIX(mon_num(mon)); mon = INT2FIX(mon_num(mon));
s3e(hash, y, mon, d, 0); s3e(hash, y, mon, d, 0);
@ -1302,9 +1323,10 @@ parse_vms12(VALUE str, VALUE hash)
"-('?-?\\d+)(?:-('?-?\\d+))?" "-('?-?\\d+)(?:-('?-?\\d+))?"
#else #else
BOS BOS
"(" VALID_MONTHS ")" FPW_COM FPT_COM
"(" DOTLESS_VALID_MONTHS ")"
"-('?-?\\d+)(?:-('?-?\\d+))?" "-('?-?\\d+)(?:-('?-?\\d+))?"
COM_FP COM_FPT COM_FPW
EOS EOS
#endif #endif
; ;
@ -1336,6 +1358,15 @@ parse_sla_cb(VALUE m, VALUE hash)
mon = rb_reg_nth_match(2, m); mon = rb_reg_nth_match(2, m);
d = rb_reg_nth_match(3, m); d = rb_reg_nth_match(3, m);
#ifdef TIGHT_PARSER
if (*RSTRING_PTR(y) == '\'' &&
(*RSTRING_PTR(mon) == '\'' ||
!NIL_P(d) && *RSTRING_PTR(d) == '\''))
return 0;
if (*RSTRING_PTR(mon) == '\'' && !NIL_P(d))
return 0;
#endif
s3e(hash, y, mon, d, 0); s3e(hash, y, mon, d, 0);
return 1; return 1;
} }
@ -1348,8 +1379,9 @@ parse_sla(VALUE str, VALUE hash)
"('?-?\\d+)/\\s*('?\\d+)(?:\\D\\s*('?-?\\d+))?" "('?-?\\d+)/\\s*('?\\d+)(?:\\D\\s*('?-?\\d+))?"
#else #else
BOS BOS
FPW_COM FPT_COM
"('?-?\\d+)/\\s*('?\\d+)(?:(?:[-/]|\\s+)\\s*('?-?\\d+))?" "('?-?\\d+)/\\s*('?\\d+)(?:(?:[-/]|\\s+)\\s*('?-?\\d+))?"
COM_FP COM_FPT COM_FPW
EOS EOS
#endif #endif
; ;
@ -1369,6 +1401,11 @@ parse_sla2_cb(VALUE m, VALUE hash)
mon = rb_reg_nth_match(2, m); mon = rb_reg_nth_match(2, m);
y = rb_reg_nth_match(3, m); y = rb_reg_nth_match(3, m);
#ifdef TIGHT_PARSER
if (*RSTRING_PTR(d) == '\'' && !NIL_P(y) && *RSTRING_PTR(y) == '\'')
return 0;
#endif
mon = INT2FIX(mon_num(mon)); mon = INT2FIX(mon_num(mon));
s3e(hash, y, mon, d, 0); s3e(hash, y, mon, d, 0);
@ -1380,8 +1417,9 @@ parse_sla2(VALUE str, VALUE hash)
{ {
static const char pat_source[] = static const char pat_source[] =
BOS BOS
"('?-?\\d+)/\\s*(" VALID_MONTHS ")(?:(?:[-/]|\\s+)\\s*('?-?\\d+))?" FPW_COM FPT_COM
COM_FP "('?-?\\d+)/\\s*(" DOTLESS_VALID_MONTHS ")(?:(?:[-/]|\\s+)\\s*('?-?\\d+))?"
COM_FPT COM_FPW
EOS EOS
; ;
static VALUE pat = Qnil; static VALUE pat = Qnil;
@ -1399,6 +1437,11 @@ parse_sla3_cb(VALUE m, VALUE hash)
d = rb_reg_nth_match(2, m); d = rb_reg_nth_match(2, m);
y = rb_reg_nth_match(3, m); y = rb_reg_nth_match(3, m);
#ifdef TIGHT_PARSER
if (*RSTRING_PTR(d) == '\'' && !NIL_P(y))
return 0;
#endif
mon = INT2FIX(mon_num(mon)); mon = INT2FIX(mon_num(mon));
s3e(hash, y, mon, d, 0); s3e(hash, y, mon, d, 0);
@ -1410,8 +1453,9 @@ parse_sla3(VALUE str, VALUE hash)
{ {
static const char pat_source[] = static const char pat_source[] =
BOS BOS
"(" VALID_MONTHS ")/\\s*('?\\d+)(?:(?:[-/]|\\s+)\\s*('?-?\\d+))?" FPW_COM FPT_COM
COM_FP "(" DOTLESS_VALID_MONTHS ")/\\s*('?\\d+)(?:(?:[-/]|\\s+)\\s*('?-?\\d+))?"
COM_FPT COM_FPW
EOS EOS
; ;
static VALUE pat = Qnil; static VALUE pat = Qnil;
@ -1430,6 +1474,11 @@ parse_dot_cb(VALUE m, VALUE hash)
mon = rb_reg_nth_match(2, m); mon = rb_reg_nth_match(2, m);
d = rb_reg_nth_match(3, m); d = rb_reg_nth_match(3, m);
#ifdef TIGHT_PARSER
if (*RSTRING_PTR(y) == '\'' && *RSTRING_PTR(d) == '\'')
return 0;
#endif
s3e(hash, y, mon, d, 0); s3e(hash, y, mon, d, 0);
return 1; return 1;
} }
@ -1442,8 +1491,9 @@ parse_dot(VALUE str, VALUE hash)
"('?-?\\d+)\\.\\s*('?\\d+)\\.\\s*('?-?\\d+)" "('?-?\\d+)\\.\\s*('?\\d+)\\.\\s*('?-?\\d+)"
#else #else
BOS BOS
"('?-?\\d+)\\.\\s*('?\\d+)\\.\\s*('?-?\\d+)" FPW_COM FPT_COM
COM_FP "('?-?\\d+)\\.\\s*(\\d+)\\.\\s*('?-?\\d+)"
COM_FPT COM_FPW
EOS EOS
#endif #endif
; ;
@ -1463,6 +1513,11 @@ parse_dot2_cb(VALUE m, VALUE hash)
mon = rb_reg_nth_match(2, m); mon = rb_reg_nth_match(2, m);
y = rb_reg_nth_match(3, m); y = rb_reg_nth_match(3, m);
#ifdef TIGHT_PARSER
if (*RSTRING_PTR(d) == '\'' && !NIL_P(y) && *RSTRING_PTR(y) == '\'')
return 0;
#endif
mon = INT2FIX(mon_num(mon)); mon = INT2FIX(mon_num(mon));
s3e(hash, y, mon, d, 0); s3e(hash, y, mon, d, 0);
@ -1474,8 +1529,9 @@ parse_dot2(VALUE str, VALUE hash)
{ {
static const char pat_source[] = static const char pat_source[] =
BOS BOS
"('?-?\\d+)\\.\\s*(" VALID_MONTHS ")(?:(?:[./]|\\s+)\\s*('?-?\\d+))?" FPW_COM FPT_COM
COM_FP "('?-?\\d+)\\.\\s*(" DOTLESS_VALID_MONTHS ")(?:(?:[./])\\s*('?-?\\d+))?"
COM_FPT COM_FPW
EOS EOS
; ;
static VALUE pat = Qnil; static VALUE pat = Qnil;
@ -1493,6 +1549,11 @@ parse_dot3_cb(VALUE m, VALUE hash)
d = rb_reg_nth_match(2, m); d = rb_reg_nth_match(2, m);
y = rb_reg_nth_match(3, m); y = rb_reg_nth_match(3, m);
#ifdef TIGHT_PARSER
if (*RSTRING_PTR(d) == '\'' && !NIL_P(y))
return 0;
#endif
mon = INT2FIX(mon_num(mon)); mon = INT2FIX(mon_num(mon));
s3e(hash, y, mon, d, 0); s3e(hash, y, mon, d, 0);
@ -1504,8 +1565,9 @@ parse_dot3(VALUE str, VALUE hash)
{ {
static const char pat_source[] = static const char pat_source[] =
BOS BOS
"(" VALID_MONTHS ")\\.\\s*('?\\d+)(?:(?:[./]|\\s+)\\s*('?-?\\d+))?" FPW_COM FPT_COM
COM_FP "(" DOTLESS_VALID_MONTHS ")\\.\\s*('?\\d+)(?:(?:[./])\\s*('?-?\\d+))?"
COM_FPT COM_FPW
EOS EOS
; ;
static VALUE pat = Qnil; static VALUE pat = Qnil;
@ -1533,8 +1595,9 @@ parse_year(VALUE str, VALUE hash)
"'(\\d+)\\b" "'(\\d+)\\b"
#else #else
BOS BOS
FPW_COM FPT_COM
"'(\\d+)" "'(\\d+)"
COM_FP COM_FPT COM_FPW
EOS EOS
#endif #endif
; ;
@ -1562,8 +1625,9 @@ parse_mon(VALUE str, VALUE hash)
"\\b(" ABBR_MONTHS ")\\S*" "\\b(" ABBR_MONTHS ")\\S*"
#else #else
BOS BOS
"(" VALID_MONTHS ")\\.?" FPW_COM FPT_COM
COM_FP "(" VALID_MONTHS ")"
COM_FPT COM_FPW
EOS EOS
#endif #endif
; ;
@ -1591,8 +1655,9 @@ parse_mday(VALUE str, VALUE hash)
"(\\d+)(st|nd|rd|th)\\b" "(\\d+)(st|nd|rd|th)\\b"
#else #else
BOS BOS
FPW_COM FPT_COM
"(\\d+)(st|nd|rd|th)" "(\\d+)(st|nd|rd|th)"
COM_FP COM_FPT COM_FPW
EOS EOS
#endif #endif
; ;
@ -1914,15 +1979,51 @@ parse_frag(VALUE str, VALUE hash)
#ifdef TIGHT_PARSER #ifdef TIGHT_PARSER
static int static int
check_leftover(VALUE str, VALUE hash) parse_dummy_cb(VALUE m, VALUE hash)
{ {
static const char pat_source[] = "\\A\\s*" FP_CODE "?\\s*\\z"; return 1;
}
static int
parse_wday_only(VALUE str, VALUE hash)
{
static const char pat_source[] = "\\A\\s*" FPW "\\s*\\z";
static VALUE pat = Qnil; static VALUE pat = Qnil;
REGCOMP_0(pat); REGCOMP_0(pat);
if (NIL_P(f_match(pat, str))) SUBS(str, pat, parse_dummy_cb);
return 0; }
static int
parse_time_only(VALUE str, VALUE hash)
{
static const char pat_source[] = "\\A\\s*" FPT "\\s*\\z";
static VALUE pat = Qnil;
REGCOMP_0(pat);
SUBS(str, pat, parse_dummy_cb);
}
static int
parse_wday_and_time(VALUE str, VALUE hash)
{
static const char pat_source[] = "\\A\\s*(" FPW "\\s+" FPT "|" FPT "\\s+" FPW ")\\s*\\z";
static VALUE pat = Qnil;
REGCOMP_0(pat);
SUBS(str, pat, parse_dummy_cb);
}
static unsigned
have_invalid_char_p(VALUE s)
{
long i;
for (i = 0; i < RSTRING_LEN(s); i++)
if (iscntrl((unsigned char)RSTRING_PTR(s)[i]) &&
!isspace((unsigned char)RSTRING_PTR(s)[i]))
return 1; return 1;
return 0;
} }
#endif #endif
@ -1956,11 +2057,20 @@ check_class(VALUE s)
#define HAVE_ELEM_P(x) ((check_class(str) & (x)) == (x)) #define HAVE_ELEM_P(x) ((check_class(str) & (x)) == (x))
#ifdef TIGHT_PARSER
#define PARSER_ERROR return rb_hash_new()
#endif
VALUE VALUE
date__parse(VALUE str, VALUE comp) date__parse(VALUE str, VALUE comp)
{ {
VALUE backref, hash; VALUE backref, hash;
#ifdef TIGHT_PARSER
if (have_invalid_char_p(str))
PARSER_ERROR;
#endif
backref = rb_backref_get(); backref = rb_backref_get();
rb_match_busy(backref); rb_match_busy(backref);
@ -2040,6 +2150,17 @@ date__parse(VALUE str, VALUE comp)
if (parse_ddd(str, hash)) if (parse_ddd(str, hash))
goto ok; goto ok;
#ifdef TIGHT_PARSER
if (parse_wday_only(str, hash))
goto ok;
if (parse_time_only(str, hash))
goto ok;
if (parse_wday_and_time(str, hash))
goto ok;
PARSER_ERROR; /* not found */
#endif
ok: ok:
#ifndef TIGHT_PARSER #ifndef TIGHT_PARSER
if (HAVE_ELEM_P(HAVE_ALPHA)) if (HAVE_ELEM_P(HAVE_ALPHA))
@ -2048,11 +2169,6 @@ date__parse(VALUE str, VALUE comp)
parse_frag(str, hash); parse_frag(str, hash);
#endif #endif
#ifdef TIGHT_PARSER
if (!check_leftover(str, hash))
return rb_hash_new();
#endif
{ {
if (RTEST(ref_hash("_comp"))) { if (RTEST(ref_hash("_comp"))) {
VALUE y; VALUE y;