mmapscanner 0.3 → 0.3.1

data/README.md

@@ -53,6 +53,7 @@ Usage
 * rest はポインタ以降のデータを返します。
 * matched は正規表現に一致した部分を MmapScanner オブジェクトで返します。
 * matched(n) は正規表現の n番目の括弧に一致した部分を MmapScanner オブジェクトで返します。
+* matched_str は matched と同じですが、文字列を返します。
 * pos は現在のポインタの位置を返します。
 * pos= でポインタ位置を変更することができます。
 

data/ext/mmapscanner.c

@@ -20,6 +20,8 @@ typedef struct {
     int matched;
     size_t matched_pos;
     struct re_registers regs;
+    VALUE data;
+    VALUE dummy_str;
 } mmapscanner_t;
 
 static void mmap_free(mmap_data_t *data)
@@ -48,6 +50,12 @@ static void mmapscanner_free(mmapscanner_t *ms)
     free(ms);
 }
 
+static void mark(mmapscanner_t *ms)
+{
+    rb_gc_mark_maybe(ms->data);
+    rb_gc_mark_maybe(ms->dummy_str);
+}
+
 VALUE allocate(VALUE klass)
 {
     mmapscanner_t *ms;
@@ -55,9 +63,11 @@ VALUE allocate(VALUE klass)
     ms->offset = 0;
     ms->size = 0;
     ms->pos = 0;
+    ms->matched = 0;
     ms->matched_pos = 0;
     onig_region_init(&ms->regs);
-    return Data_Wrap_Struct(klass, 0, mmapscanner_free, ms);
+    ms->dummy_str = Qnil;
+    return Data_Wrap_Struct(klass, mark, mmapscanner_free, ms);
 }
 
 static VALUE initialize(int argc, VALUE *argv, VALUE obj)
@@ -78,7 +88,7 @@ static VALUE initialize(int argc, VALUE *argv, VALUE obj)
         Data_Get_Struct(src, mmapscanner_t, ms);
         src_offset = ms->offset;
         src_size = ms->size;
-        src_data = rb_iv_get(src, "data");
+        src_data = ms->data;
     } else if (TYPE(src) == T_FILE) {
         int fd;
         struct stat st;
@@ -109,7 +119,7 @@ static VALUE initialize(int argc, VALUE *argv, VALUE obj)
     ms->pos = 0;
     ms->matched = 0;
     ms->matched_pos = 0;
-    rb_iv_set(obj, "data", src_data);
+    ms->data = src_data;
     return Qnil;
 }
 
@@ -123,16 +133,12 @@ static VALUE size(VALUE obj)
 static VALUE to_s(VALUE obj)
 {
     mmapscanner_t *ms;
-    Data_Get_Struct(obj, mmapscanner_t, ms);
-    size_t offset = ms->offset;
-    size_t size = ms->size;
-    VALUE data = rb_iv_get(obj, "data");
     mmap_data_t *mdata;
-
-    if (TYPE(data) == T_STRING)
-        return rb_str_new(RSTRING_PTR(data)+offset, size);
-    Data_Get_Struct(data, mmap_data_t, mdata);
-    return rb_str_new(mdata->ptr+offset, size);
+    Data_Get_Struct(obj, mmapscanner_t, ms);
+    if (TYPE(ms->data) == T_STRING)
+        return rb_str_new(RSTRING_PTR(ms->data) + ms->offset, ms->size);
+    Data_Get_Struct(ms->data, mmap_data_t, mdata);
+    return rb_str_new(mdata->ptr + ms->offset, ms->size);
 }
 
 static VALUE slice(VALUE obj, VALUE pos, VALUE len)
@@ -178,38 +184,36 @@ static VALUE scan_sub(VALUE obj, VALUE re, int forward, int headonly, int sizeon
     int result;
     size_t old_pos, matched_len;
     char *ptr;
-    size_t pos, size;
-    VALUE data;
     mmap_data_t *mdata;
 
+    ms->matched = 0;
     Check_Type(re, T_REGEXP);
-    pos = ms->pos;
-    size = ms->size;
-    if (pos >= size)
+    if (ms->pos >= ms->size)
         return Qnil;
-    data = rb_iv_get(obj, "data");
-    if (TYPE(data) == T_STRING)
-        ptr = RSTRING_PTR(data);
+    if (TYPE(ms->data) == T_STRING)
+        ptr = RSTRING_PTR(ms->data);
     else {
-        Data_Get_Struct(data, mmap_data_t, mdata);
+        Data_Get_Struct(ms->data, mmap_data_t, mdata);
         ptr = mdata->ptr;
     }
     ptr += ms->offset;
 
-    reg = rb_reg_prepare_re(re, rb_str_new("", 0));
+    if (ms->dummy_str == Qnil)
+        ms->dummy_str = rb_str_new("", 0);
+    reg = rb_reg_prepare_re(re, ms->dummy_str);
     tmpreg = reg != RREGEXP(re)->ptr;
     if (!tmpreg) RREGEXP(re)->usecnt++;
 
     if (headonly) {
-        result = onig_match(reg, (UChar*)(ptr+pos),
-                            (UChar*)(ptr+size),
-                            (UChar*)(ptr+pos),
+        result = onig_match(reg, (UChar*)(ptr + ms->pos),
+                            (UChar*)(ptr + ms->size),
+                            (UChar*)(ptr + ms->pos),
                             &ms->regs, ONIG_OPTION_NONE);
     } else {
-        result = onig_search(reg, (UChar*)(ptr+pos),
-                             (UChar*)(ptr+size),
-                             (UChar*)(ptr+pos),
-                             (UChar*)(ptr+size),
+        result = onig_search(reg, (UChar*)(ptr + ms->pos),
+                             (UChar*)(ptr + ms->size),
+                             (UChar*)(ptr + ms->pos),
+                             (UChar*)(ptr + ms->size),
                              &ms->regs, ONIG_OPTION_NONE);
     }
     if (!tmpreg) RREGEXP(re)->usecnt--;
@@ -223,12 +227,10 @@ static VALUE scan_sub(VALUE obj, VALUE re, int forward, int headonly, int sizeon
     }
     if (result < 0)
         return Qnil;
-    old_pos = pos;
+    old_pos = ms->pos;
     matched_len = ms->regs.end[0];
-    if (forward) {
-        pos += matched_len;
-        ms->pos = pos;
-    }
+    if (forward)
+        ms->pos += matched_len;
     ms->matched = 1;
     ms->matched_pos = old_pos;
 
@@ -286,27 +288,50 @@ static VALUE rest(VALUE obj)
     return rb_funcall(cMmapScanner, rb_intern("new"), 2, obj, SIZET2NUM(ms->pos));
 }
 
-static VALUE matched(int argc, VALUE *argv, VALUE obj)
+static int matched_sub(int argc, VALUE *argv, mmapscanner_t *ms, size_t *pos, size_t *len)
 {
-    mmapscanner_t *ms;
-    Data_Get_Struct(obj, mmapscanner_t, ms);
-    VALUE nth;
     int i = 0;
-    size_t pos, len;
-
-    if (rb_scan_args(argc, argv, "01", &nth) == 1)
-        i = NUM2LONG(nth);
     if (ms->matched == 0)
-        return Qnil;
+        return 0;
+    if (argc == 0)
+        i = 0;
+    else if (argc == 1)
+        i = NUM2LONG(argv[0]);
+    else
+        rb_raise(rb_eArgError, "wrong number of arguments (%d for 0..1)", argc);
     if (i < 0)
-        return Qnil;
+        return 0;
     if (i >= ms->regs.num_regs)
+        return 0;
+    *pos = ms->matched_pos + ms->regs.beg[i];
+    *len = ms->regs.end[i] - ms->regs.beg[i];
+    return 1;
+}
+
+static VALUE matched(int argc, VALUE *argv, VALUE obj)
+{
+    mmapscanner_t *ms;
+    Data_Get_Struct(obj, mmapscanner_t, ms);
+    size_t pos, len;
+    if (matched_sub(argc, argv, ms, &pos, &len) == 0)
         return Qnil;
-    pos = ms->matched_pos + ms->regs.beg[i];
-    len = ms->regs.end[i] - ms->regs.beg[i];
     return rb_funcall(cMmapScanner, rb_intern("new"), 3, obj, SIZET2NUM(pos), SIZET2NUM(len));
 }
 
+static VALUE matched_str(int argc, VALUE *argv, VALUE obj)
+{
+    mmapscanner_t *ms;
+    Data_Get_Struct(obj, mmapscanner_t, ms);
+    mmap_data_t *mdata;
+    size_t pos, len;
+    if (matched_sub(argc, argv, ms, &pos, &len) == 0)
+        return Qnil;
+    if (TYPE(ms->data) == T_STRING)
+        return rb_str_new(RSTRING_PTR(ms->data)+ms->offset+pos, len);
+    Data_Get_Struct(ms->data, mmap_data_t, mdata);
+    return rb_str_new(mdata->ptr+ms->offset+pos, len);
+}
+
 void Init_mmapscanner(void)
 {
     cMmapScanner = rb_define_class("MmapScanner", rb_cObject);
@@ -329,6 +354,7 @@ void Init_mmapscanner(void)
     rb_define_method(cMmapScanner, "eos?", eos_p, 0);
     rb_define_method(cMmapScanner, "rest", rest, 0);
     rb_define_method(cMmapScanner, "matched", matched, -1);
+    rb_define_method(cMmapScanner, "matched_str", matched_str, -1);
 
     cMmap = rb_define_class_under(cMmapScanner, "Mmap", rb_cObject);
 }

metadata

@@ -2,7 +2,7 @@
 name: mmapscanner
 version: !ruby/object:Gem::Version 
   prerelease: 
-  version: "0.3"
+  version: 0.3.1
 platform: ruby
 authors: 
 - TOMITA Masahiro
@@ -10,7 +10,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2011-03-21 00:00:00 +09:00
+date: 2011-03-23 00:00:00 +09:00
 default_executable: 
 dependencies: []