mmapscanner 0.3.7 → 0.3.9

data/README.md

@@ -46,8 +46,21 @@ Usage
 * scan はポインタ位置で正規表現との一致を試みます。一致した部分を返し、ポインタを進めます。一致しない場合は nil を返します。
 * scan_until は scan と同じですが、現在のポインタの位置以降で一致を試みます。
 * check は scan と同じですが、ポインタを進めません。
+* check_until は check と同じですが、現在のポインタの位置以降で一致を試みます。
 * skip は scan と同じですが、一致したバイト数を返します。
+* skip_until は skip と同じですが、現在のポインタの位置以降で一致を試みます。
 * match? は check と同じですが、一致したバイト数を返します。
+* exist? は match? と同じですが、現在のポインタの位置以降で一致を試みます。
+* scan_full(re, s, f) はポインタの位置でスキャンします。
+  * scan_full(re, true, true) は scan(re) と同じです。
+  * scan_full(re, true, false) は skip(re) と同じです。
+  * scan_full(re, false, true) は check(re) と同じです。
+  * scan_full(re, false, false) は match?(re) と同じです。
+* search_full(re, s, f) はポインタの位置以降でスキャンします。
+  * search_full(re, true, true) は scan_until(re) と同じです。
+  * search_full(re, true, false) は skip_until(re) と同じです。
+  * search_full(re, false, true) は check_until(re) と同じです。
+  * search_full(re, false, false) は exist?(re) と同じです。
 * peek は指定したバイト数分のデータを返します。ポインタは進みません。
 * eos? はポインタが末尾に達していると true を返します。
 * rest はポインタ以降のデータを返します。
@@ -56,6 +69,7 @@ Usage
 * 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。
+* terminate はポインタを末尾に移動します。
 
 Copyright
 ---------

data/ext/mmapscanner.c

@@ -53,12 +53,16 @@ static VALUE mmap_allocate(VALUE klass)
 static VALUE mmap_initialize(int argc, VALUE *argv, VALUE obj)
 {
     mmap_data_t *data;
-    Data_Get_Struct(obj, mmap_data_t, data);
-    if (data->ptr)
-        rb_raise(rb_eRuntimeError, "already mapped");
     VALUE file, voffset, vlength;
     off_t offset = 0;
     size_t length = 0;
+    int fd;
+    struct stat st;
+    void *ptr;
+
+    Data_Get_Struct(obj, mmap_data_t, data);
+    if (data->ptr)
+        rb_raise(rb_eRuntimeError, "already mapped");
     rb_scan_args(argc, argv, "12", &file, &voffset, &vlength);
     if (TYPE(file) != T_FILE)
         rb_raise(rb_eTypeError, "File object required");
@@ -66,15 +70,13 @@ static VALUE mmap_initialize(int argc, VALUE *argv, VALUE obj)
         rb_raise(rb_eRangeError, "offset out of range: %lld", NUM2LL(voffset));
     if (vlength != Qnil && NUM2LL(vlength) < 0)
         rb_raise(rb_eRangeError, "length out of range: %lld", NUM2LL(vlength));
-    int fd = FIX2INT(rb_funcall(file, rb_intern("fileno"), 0));
-    struct stat st;
+    fd = FIX2INT(rb_funcall(file, rb_intern("fileno"), 0));
     if (fstat(fd, &st) < 0)
         rb_sys_fail("fstat");
     offset = voffset == Qnil ? 0 : NUM2SIZET(voffset);
     length = vlength == Qnil ? st.st_size : NUM2SIZET(vlength);
     if (offset + length > st.st_size)
         length = st.st_size - offset;
-    void *ptr;
     if ((ptr = mmap(NULL, length, PROT_READ, MAP_SHARED, fd, offset)) == MAP_FAILED)
         rb_sys_fail("mmap");
 
@@ -159,7 +161,16 @@ static VALUE initialize(int argc, VALUE *argv, VALUE obj)
         src = ms->data;
         src_size_defined = 1;
     } else if (TYPE(src) == T_FILE) {
-        src = rb_funcall(cMmap, rb_intern("new"), 1, src);
+        int fd = FIX2INT(rb_funcall(src, rb_intern("fileno"), 0));
+        struct stat st;
+        if (fstat(fd, &st) < 0)
+            rb_sys_fail("fstat");
+        if (st.st_size == 0) {
+            src = rb_str_new(NULL, 0);
+            src_size = 0;
+        } else {
+            src = rb_funcall(cMmap, rb_intern("new"), 1, src);
+        }
     }
     if (rb_obj_class(src) == cMmap) {
         if (!src_size_defined) {
@@ -260,8 +271,8 @@ static VALUE pos(VALUE obj)
 static VALUE set_pos(VALUE obj, VALUE pos)
 {
     mmapscanner_t *ms;
-    Data_Get_Struct(obj, mmapscanner_t, ms);
     size_t p, size;
+    Data_Get_Struct(obj, mmapscanner_t, ms);
 
     if (NUM2LL(pos) < 0)
         rb_raise(rb_eRangeError, "out of range: %lld", NUM2LL(pos));
@@ -276,7 +287,6 @@ static VALUE set_pos(VALUE obj, VALUE pos)
 static VALUE scan_sub(VALUE obj, VALUE re, int forward, int headonly, int sizeonly)
 {
     mmapscanner_t *ms;
-    Data_Get_Struct(obj, mmapscanner_t, ms);
     regex_t *rb_reg_prepare_re(VALUE re, VALUE str);
     regex_t *reg;
     int tmpreg;
@@ -285,6 +295,7 @@ static VALUE scan_sub(VALUE obj, VALUE re, int forward, int headonly, int sizeon
     char *ptr;
     mmap_data_t *mdata;
 
+    Data_Get_Struct(obj, mmapscanner_t, ms);
     ms->matched = 0;
     Check_Type(re, T_REGEXP);
    if (ms->pos > ms->size)
@@ -355,6 +366,18 @@ static VALUE scan_sub(VALUE obj, VALUE re, int forward, int headonly, int sizeon
     return create_from_mmapscanner(obj, old_pos, matched_len);
 }
 
+static VALUE scan_full(VALUE obj, VALUE re, VALUE forward, VALUE ret_ms)
+{
+    return scan_sub(obj, re, (forward != Qnil && forward != Qfalse), 1,
+                    (ret_ms == Qnil || ret_ms == Qfalse));
+}
+
+static VALUE search_full(VALUE obj, VALUE re, VALUE forward, VALUE ret_ms)
+{
+    return scan_sub(obj, re, (forward != Qnil && forward != Qfalse), 0,
+                    (ret_ms == Qnil || ret_ms == Qfalse));
+}
+
 static VALUE scan(VALUE obj, VALUE re)
 {
     return scan_sub(obj, re, 1, 1, 0);
@@ -370,21 +393,36 @@ static VALUE check(VALUE obj, VALUE re)
     return scan_sub(obj, re, 0, 1, 0);
 }
 
+static VALUE check_until(VALUE obj, VALUE re)
+{
+    return scan_sub(obj, re, 0, 0, 0);
+}
+
 static VALUE skip(VALUE obj, VALUE re)
 {
     return scan_sub(obj, re, 1, 1, 1);
 }
 
+static VALUE skip_until(VALUE obj, VALUE re)
+{
+    return scan_sub(obj, re, 1, 0, 1);
+}
+
 static VALUE match_p(VALUE obj, VALUE re)
 {
     return scan_sub(obj, re, 0, 1, 1);
 }
 
+static VALUE exist_p(VALUE obj, VALUE re)
+{
+    return scan_sub(obj, re, 0, 0, 1);
+}
+
 static VALUE peek(VALUE obj, VALUE size)
 {
     mmapscanner_t *ms;
-    Data_Get_Struct(obj, mmapscanner_t, ms);
     size_t sz = NUM2SIZET(size);
+    Data_Get_Struct(obj, mmapscanner_t, ms);
     if (sz > ms->size - ms->pos)
         sz = ms->size - ms->pos;
     return create_from_mmapscanner(obj, ms->pos, sz);
@@ -404,7 +442,7 @@ static VALUE rest(VALUE obj)
     return create_from_mmapscanner(obj, ms->pos, ms->size - ms->pos);
 }
 
-static int matched_sub(int argc, VALUE *argv, mmapscanner_t *ms, int *pos, int *len)
+static int matched_sub(int argc, VALUE *argv, mmapscanner_t *ms, size_t *pos, size_t *len)
 {
     int i = 0;
     if (ms->matched == 0)
@@ -429,8 +467,8 @@ static int matched_sub(int argc, VALUE *argv, mmapscanner_t *ms, int *pos, int *
 static VALUE matched(int argc, VALUE *argv, VALUE obj)
 {
     mmapscanner_t *ms;
-    Data_Get_Struct(obj, mmapscanner_t, ms);
     size_t pos, len;
+    Data_Get_Struct(obj, mmapscanner_t, ms);
     if (matched_sub(argc, argv, ms, &pos, &len) == 0)
         return Qnil;
     return create_from_mmapscanner(obj, pos, len);
@@ -439,9 +477,9 @@ static VALUE matched(int argc, VALUE *argv, VALUE obj)
 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;
+    Data_Get_Struct(obj, mmapscanner_t, ms);
     if (matched_sub(argc, argv, ms, &pos, &len) == 0)
         return Qnil;
     if (TYPE(ms->data) == T_STRING)
@@ -452,6 +490,14 @@ static VALUE matched_str(int argc, VALUE *argv, VALUE obj)
     return rb_str_new(mdata->ptr+ms->offset+pos, len);
 }
 
+static VALUE terminate(VALUE obj)
+{
+    mmapscanner_t *ms;
+    Data_Get_Struct(obj, mmapscanner_t, ms);
+    ms->pos = ms->size;
+    return obj;
+}
+
 void Init_mmapscanner(void)
 {
     cMmapScanner = rb_define_class("MmapScanner", rb_cObject);
@@ -466,16 +512,22 @@ void Init_mmapscanner(void)
     rb_define_method(cMmapScanner, "inspect", inspect, 0);
     rb_define_method(cMmapScanner, "pos", pos, 0);
     rb_define_method(cMmapScanner, "pos=", set_pos, 1);
+    rb_define_method(cMmapScanner, "scan_full", scan_full, 3);
+    rb_define_method(cMmapScanner, "search_full", search_full, 3);
     rb_define_method(cMmapScanner, "scan", scan, 1);
     rb_define_method(cMmapScanner, "scan_until", scan_until, 1);
     rb_define_method(cMmapScanner, "check", check, 1);
+    rb_define_method(cMmapScanner, "check_until", check_until, 1);
     rb_define_method(cMmapScanner, "skip", skip, 1);
+    rb_define_method(cMmapScanner, "skip_until", skip_until, 1);
     rb_define_method(cMmapScanner, "match?", match_p, 1);
+    rb_define_method(cMmapScanner, "exist?", exist_p, 1);
     rb_define_method(cMmapScanner, "peek", peek, 1);
     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);
+    rb_define_method(cMmapScanner, "terminate", terminate, 0);
 
     cMmap = rb_define_class_under(cMmapScanner, "Mmap", rb_cObject);
     rb_define_alloc_func(cMmap, mmap_allocate);

data/spec/mmapscanner_spec.rb

@@ -105,6 +105,20 @@ describe MmapScanner do
         subject.pos.should == 0
       end
     end
+    describe '#check_until' do
+      it 'returns matched data as MmapScanner' do
+        ret = subject.check_until(/123/)
+        ret.class.should == MmapScanner
+        ret.to_s.should == '0123'
+      end
+      it 'returns nil if not matched' do
+        subject.check_until(/abc/).should be_nil
+      end
+      it 'do not forward current position' do
+        ret = subject.check_until(/123/)
+        subject.pos.should == 0
+      end
+    end
     describe '#skip' do
       it 'returns length of matched data' do
         subject.skip(/\d{10}/).should == 10
@@ -117,6 +131,18 @@ describe MmapScanner do
         subject.pos.should == 10
       end
     end
+    describe '#skip_until' do
+      it 'returns length of matched data' do
+        subject.skip_until(/123/).should == 4
+      end
+      it 'returns nil if not matched' do
+        subject.skip_until(/abc/).should be_nil
+      end
+      it 'forward current position' do
+          subject.skip_until(/123/)
+          subject.pos.should == 4
+        end
+    end
     describe '#match?' do
       it 'returns length of matched data' do
         subject.match?(/\d{10}/).should == 10
@@ -129,6 +155,18 @@ describe MmapScanner do
         subject.pos.should == 0
       end
     end
+    describe '#exist?' do
+      it 'returns length of matched data' do
+        subject.exist?(/123/).should == 4
+      end
+      it 'returns nil if not matched' do
+        subject.exist?(/abc/).should be_nil
+      end
+      it 'do not forward current position' do
+        subject.exist?(/123/)
+        subject.pos.should == 0
+      end
+    end
     describe '#matched' do
       it 'returns matched data after scan' do
         subject.scan(/\d{6}/)
@@ -173,6 +211,64 @@ describe MmapScanner do
         subject.eos?.should == false
       end
     end
+    describe '#scan_full(re, true, true)' do
+      it 'is same as #scan' do
+        ret = subject.scan_full(/\d{10}/, true, true)
+        ret.class.should == MmapScanner
+        ret.to_s.should == '0123456789'
+        subject.pos.should == 10
+      end
+    end
+    describe '#scan_full(re, true, false)' do
+      it 'is same as #skip' do
+        ret = subject.scan_full(/\d{10}/, true, false)
+        ret.should == 10
+        subject.pos.should == 10
+      end
+    end
+    describe '#scan_full(re, false, true)' do
+      it 'is same as #check' do
+        ret = subject.scan_full(/\d{10}/, false, true)
+        ret.to_s.should == '0123456789'
+        subject.pos.should == 0
+      end
+    end
+    describe '#scan_full(re, false, false)' do
+      it 'is same as #match?' do
+        ret = subject.scan_full(/\d{10}/, false, false)
+        ret.should == 10
+        subject.pos.should == 0
+      end
+    end
+    describe '#search_full(re, true, true)' do
+      it 'is same as #scan_until' do
+        ret = subject.search_full(/789/, true, true)
+        ret.class.should == MmapScanner
+        ret.to_s.should == '0123456789'
+        subject.pos.should == 10
+      end
+    end
+    describe '#search_full(re, true, false)' do
+      it 'is same as #skip_until' do
+        ret = subject.search_full(/789/, true, false)
+        ret.should == 10
+        subject.pos.should == 10
+      end
+    end
+    describe '#search_full(re, false, true)' do
+      it 'is same as #check_until' do
+        ret = subject.search_full(/789/, false, true)
+        ret.to_s.should == '0123456789'
+        subject.pos.should == 0
+      end
+    end
+    describe '#search_full(re, false, false)' do
+      it 'is same as #exist?' do
+        ret = subject.search_full(/789/, false, false)
+        ret.should == 10
+        subject.pos.should == 0
+      end
+    end
     describe '#rest' do
       it 'returns rest data as MmapScanner' do
         subject.pos = 9997
@@ -185,6 +281,12 @@ describe MmapScanner do
         subject.rest.to_s.should == ''
       end
     end
+    describe '#terminate' do
+      it 'set position to end of MmapScanner area' do
+        subject.terminate.should == subject
+        subject.pos.should == 10000
+      end
+    end
     describe '.new with position' do
       it '#size is length of rest data' do
         if src.respond_to? :size
@@ -220,6 +322,21 @@ describe MmapScanner do
         subject.data.should be_kind_of MmapScanner::Mmap
       end
     end
+    context 'empty file' do
+      before do
+        tmpf = Tempfile.new 'mmapscanner_empty_file'
+        @file = File.open(tmpf.path)
+      end
+      it '#size returns 0' do
+        subject.size.should == 0
+      end
+      it '#to_s returns empty String' do
+        subject.to_s.should == ''
+      end
+      it '#eos? returns true' do
+        subject.eos?.should == true
+      end
+    end
   end
 
   context 'with String' do

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: mmapscanner
 version: !ruby/object:Gem::Version
-  version: 0.3.7
+  version: 0.3.9
   prerelease: 
 platform: ruby
 authors:
@@ -9,7 +9,7 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2011-11-20 00:00:00.000000000 Z
+date: 2012-07-19 00:00:00.000000000 Z
 dependencies: []
 description: 
 email: tommy@tmtm.org
@@ -43,9 +43,10 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       version: '0'
 requirements: []
 rubyforge_project: 
-rubygems_version: 1.8.11
+rubygems_version: 1.8.23
 signing_key: 
 specification_version: 3
 summary: MmapScanner like StringScanner but it use mmap(2)-ed data
 test_files:
 - spec/mmapscanner_spec.rb
+has_rdoc: