[ruby-dev:16219] Re: Defining (Python) Source Code Encoding

なかだです。

At Mon, 4 Mar 2002 22:34:20 +0900,
Nobuyoshi-Nakada wrote:
> > なるほど。とりあえず合わせてみますか。ってかなり手抜き。

ONIGURUMA対応。

# 一条直也? 龍虎の拳?


Index: intern.h
===================================================================
RCS file: /cvs/ruby/src/ruby/intern.h,v
retrieving revision 1.81
diff -u -2 -p -r1.81 intern.h
--- intern.h	2002/02/20 06:33:24	1.81
+++ intern.h	2002/03/03 10:13:34
@@ -332,4 +332,5 @@ void ruby_load_script _((void));
 void ruby_init_loadpath _((void));
 void ruby_incpush _((const char*));
+void ruby_modeset _((const char*, int));
 /* signal.c */
 VALUE rb_f_kill _((int, VALUE*));
Index: parse.y
===================================================================
RCS file: /cvs/ruby/src/ruby/parse.y,v
retrieving revision 1.156
diff -u -2 -p -r1.156 parse.y
--- parse.y	2002/02/20 04:31:50	1.156
+++ parse.y	2002/03/04 12:59:09
@@ -2098,4 +2098,5 @@ yycompile(f, line)
     int n;
     NODE *node = 0;
+    const char *kcode_save;
 
     if (!compile_for_eval && rb_safe_level() == 0 &&
@@ -2121,4 +2122,5 @@ yycompile(f, line)
     }
 
+    kcode_save = rb_get_kcode();
     ruby__end__seen = 0;
     ruby_eval_tree = 0;
@@ -2137,4 +2139,5 @@ yycompile(f, line)
     in_def = 0;
     cur_mid = 0;
+    rb_set_kcode(kcode_save);
 
     if (n == 0) node = ruby_eval_tree;
@@ -2985,7 +2988,7 @@ yylex()
 
       case '#':		/* it's a comment */
-	while ((c = nextc()) != '\n') {
-	    if (c == -1)
-		return 0;
+	while (lex_p < lex_pend) {
+	    ruby_modeset(lex_p, lex_pend - lex_p);
+	    lex_p = lex_pend;
 	}
 	/* fall through */
Index: ruby.c
===================================================================
RCS file: /cvs/ruby/src/ruby/ruby.c,v
retrieving revision 1.56
diff -u -2 -p -r1.56 ruby.c
--- ruby.c	2002/02/10 07:55:40	1.56
+++ ruby.c	2002/03/05 09:57:44
@@ -19,4 +19,5 @@
 #include "dln.h"
 #include "node.h"
+#include "regex.h"
 #include <stdio.h>
 #include <sys/types.h>
@@ -335,5 +336,105 @@ require_libraries()
 }
 
+/* emacsen -*- hack */
+typedef void (*rb_modeset_setter_t) _((const char *val, const char *name));
+
+struct modeset {
+    const char *name;
+    rb_modeset_setter_t func;
+};
+
+static const struct modeset modesets[] = {
+    {"coding", (rb_modeset_setter_t)rb_set_kcode},
+};
+
+static const char modeset_pattern[] = ".*-\\*-\\s*(.*?)\\s*-\\*-";
+static const char modeset_scan_pattern[] =
+"([^\\s\'\":;]+)\\s*:\\s*(\"(?:\\\\.|[^\"])*\"|[^\"\\s;]+)[\\s;]*";
+
+#ifdef ONIGURUMA_VERSION
+static regex_t *modeset_re, *modeset_scan_re;
+#else
+static struct re_pattern_buffer modeset_re, modeset_scan_re;
+#endif
+
 static void
+Init_modeset _((void))
+{
+#if ONIGURUMA_VERSION == 110
+    RegCharEncoding enc;
+# ifdef M17N_H
+    m17n_init();
+    enc = *m17n_encoding_table;	/* default encoding */
+# else
+    enc = REG_MBLEN_TABLE[MBCTYPE_ASCII];
+# endif
+#define endof(array) ((array)+sizeof(array))
+    RegexNew(&modeset_re, (UChar *)modeset_pattern,
+	     (UChar *)endof(modeset_pattern) - 1, 0, enc, 0);
+    RegexNew(&modeset_scan_re, (UChar *)modeset_scan_pattern,
+	     (UChar *)endof(modeset_scan_pattern) - 1, 0, enc, 0);
+#else
+    re_compile_pattern(modeset_pattern,
+		       sizeof(modeset_pattern) - 1,
+		       &modeset_re);
+    re_compile_pattern(modeset_scan_pattern,
+		       sizeof(modeset_scan_pattern) - 1,
+		       &modeset_scan_re);
+#endif
+}
+
+void
+ruby_modeset(str, len)
+    const char *str;
+    int len;
+{
+#define str_copy(_s, _p, _n) ((_s) \
+	? (rb_str_resize((_s), (_n)), \
+	   MEMCPY(RSTRING(_s)->ptr, (_p), char, (_n)), (_s)) \
+	: ((_s) = rb_str_new((_p), (_n))))
+#ifdef ONIGURUMA_VERSION
+#define match(_re, _p, _n) (MEMZERO(&regs, struct re_registers, 1), \
+			    re_match((_re), (_p), (_n), 0, &regs))
+#else
+#define match(_re, _p, _n) (MEMZERO(&regs, struct re_registers, 1), \
+			    re_match(&(_re), (_p), (_n), 0, &regs))
+#endif
+    struct re_registers regs;
+    VALUE name = 0, val = 0;
+
+    if (match(modeset_re, str, len) < 0) return;
+
+    len = regs.end[1] - regs.beg[1];
+    str += regs.beg[1];
+    re_free_registers(&regs);
+
+    while (match(modeset_scan_re, str, len) >= 0) {
+	const struct modeset *p = modesets;
+	int e0 = regs.end[0];
+	int n1 = regs.end[1] - regs.beg[1];
+	int s2 = regs.beg[2];
+	int n2 = regs.end[2] - s2;
+	re_free_registers(&regs);
+
+	str_copy(name, str, n1);
+	rb_funcall(name, rb_intern("downcase!"), 0);
+
+	do {
+	    if (strncmp(p->name, RSTRING(name)->ptr, n1) == 0) {
+		str_copy(val, str + s2, n2);
+		(*p->func)(RSTRING(val)->ptr, RSTRING(name)->ptr);
+		break;
+	    }
+	} while (++p < modesets + sizeof(modesets) / sizeof(*p));
+
+	len -= e0;
+	str += e0;
+    }
+
+    if (name) rb_gc_force_recycle(name);
+    if (val) rb_gc_force_recycle(val);
+}
+
+static void
 process_sflag()
 {
@@ -843,4 +944,7 @@ load_file(fname, script)
 		}
 	    }
+	    if (RSTRING(line)->len > 1) {
+		ruby_modeset(RSTRING(line)->ptr, RSTRING(line)->len);
+	    }
 	}
 	else if (!NIL_P(c)) {
@@ -1007,4 +1111,6 @@ ruby_prog_init()
     rb_define_readonly_variable("$-a", &do_split);
     rb_global_variable(&rb_argv0);
+
+    Init_modeset();
 
 #ifdef MSDOS


-- 
--- 僕の前にBugはない。
--- 僕の後ろにBugはできる。
    中田 伸悦