simplecsv 0.1.0 → 0.1.2

data/CHANGELOG.txt

@@ -1,3 +1,15 @@
+2007-03-18 DATE Ken <itacchi@gmail.com>
+
+ * fix bug: options parameter of SimpleCSV.new do not effect SimpleCSV#parser
+ * implements SimpleCSV#options and SimpleCSV#options=
+
+2007-03-15 DATE Ken <itacchi@gmail.com>
+
+ * change SimpleCSV from module to class
+ * implements SimpleCSV.new
+ * implements SimpleCSV#parse
+ * implements SimpleCSV#write
+
 2007-03-09 DATE Ken <itacchi@gmail.com>
 
  * libcsv 1.0.0 support

data/History.txt

@@ -1,3 +1,8 @@
+2007-03-18
+
+ * 0.1.2
+ * SimpleCSV#parse
+
 2007-03-09
 
  * 0.1.0

data/README.txt

@@ -1,4 +1,4 @@
-SimpleCSV README
+SimpleCSV 0.1.2 README
 
 What is:
  SimpleCSV provides simple api to read/write csv.

data/TODO

@@ -1,2 +1,4 @@
 * stream input
-* OO I/F
+* refactoring (parse method)
+* subclass sutomizing by template method pattern
+* convenience method such as parse_file

data/ext/simplecsv/rb_simplecsv.c

@@ -1,28 +1,65 @@
 #include <ruby.h>
 #include "rb_simplecsv.h"
 
-VALUE mSimpleCSV;
+VALUE cSimpleCSV;
 VALUE eCSVError;
 VALUE eCSVParseError;
 VALUE eCSVNoMemError;
 VALUE eCSVTooBigError;
 VALUE eCSVInvalid;
 
+struct csv_parser_object {
+    struct csv_parser* _parser;
+};
 
-static void field_callback(char* s, size_t i, void* data) {
+
+#define Get_Parser(obj, parser_obj, parser) {\
+    Data_Get_Struct(obj, struct csv_parser_object, parser_obj);\
+    parser = parser_obj->_parser;\
+}
+
+
+#define Set_Options(opts, options) {\
+    switch (TYPE(opts)) {\
+        case T_NIL:\
+            options = 0;\
+            break;\
+        case T_FIXNUM:\
+            options = FIX2INT(opts);\
+            break;\
+        default:\
+            rb_raise(rb_eTypeError, "wrong argument type %s (expected Fixnum/nil)",\
+                     rb_class2name(CLASS_OF(opts)));\
+    }\
+}
+
+static void free_parser_object(struct csv_parser_object* parser_object)
+{
+    if (parser_object){
+        if (parser_object->_parser) {
+            csv_free(parser_object->_parser);
+        }
+        free(parser_object);
+    }
+}
+
+
+static void field_callback(char* s, size_t i, void* obj)
+{
     VALUE row, str;
 
-    row = rb_ivar_get(mSimpleCSV, rb_intern("'row'"));
+    row = rb_ivar_get((VALUE)obj, rb_intern("'row'"));
     str = rb_str_new(s, i);
     rb_ary_push(row, str);
 }
 
-static void row_callback(char c, void* data) {
+static void row_callback(char c, void* obj)
+{
     VALUE row;
 
-    row = rb_ivar_get(mSimpleCSV, rb_intern("'row'"));
+    row = rb_ivar_get((VALUE)obj, rb_intern("'row'"));
     rb_yield(row);
-    rb_ivar_set(mSimpleCSV, rb_intern("'row'"), rb_ary_new());
+    rb_ivar_set((VALUE)obj, rb_intern("'row'"), rb_ary_new());
 }
 
 
@@ -79,37 +116,26 @@ static void row_callback(char c, void* data) {
  *
  * ==== Parameters
  * [str] <code>String</code> : csv string.
- * [options] <code>Interger</code> : parser options. You can specify SimpleCSV::STRICT and/or SimpleCSVREPALL_NL.
+ * [options] <code>Interger</code> : parser options. You can specify SimpleCSV::STRICT and/or SimpleCSV::REPALL_NL.
  * [row] <code>Array</code> of <code>String</code>s : parsed data.
  *
  * ==== Exceptions
- * [TypeError] <i>options</i> isn't <code>Fixnum</code>/<code>nil</code>.
+ * [TypeError] <i>options</i> isn't <code>Fixnum</code> / <code>nil</code>.
  * [SimpleCSV::Error] failed to initialize csv parser.
  * [SimpleCSV::ParseError] failed to parse.
  * [LocalJumpError] You must call this method with block.
  */
-static VALUE simplecsv_parse(int argc, VALUE* argv, VALUE klass)
+static VALUE simplecsv_s_parse(int argc, VALUE* argv, VALUE klass)
 {
     VALUE str, opts;
-    struct csv_parser *parser;
+    struct csv_parser* parser;
     int options;
     int len;
 
     rb_scan_args(argc, argv, "11", &str, &opts);
 
     Check_Type(str, T_STRING);
-
-    switch (TYPE(opts)) {
-        case T_NIL:
-            options = 0;
-            break;
-        case T_FIXNUM:
-            options = FIX2INT(opts);
-            break;
-        default:
-            rb_raise(rb_eTypeError, "wrong argument type %s (expected Fixnum/nil)",
-                     rb_class2name(CLASS_OF(opts)));
-    }
+    Set_Options(opts, options);
 
     if (csv_init(&parser, options) != 0) {
         rb_raise(eCSVError, "failed to initialize csv parser");
@@ -120,20 +146,20 @@ static VALUE simplecsv_parse(int argc, VALUE* argv, VALUE klass)
     len = RSTRING(str)->len;
     if (csv_parse(parser, StringValuePtr(str), len, field_callback, row_callback, (void*)klass) != len) {
         switch (csv_error(parser)) {
-          case CSV_EPARSE:
-              rb_raise(eCSVParseError, "error while parsing by malformed data");
-              break;
-          case CSV_ENOMEM:
-              rb_raise(eCSVNoMemError, "no memory");
-              break;
-          case CSV_ETOOBIG:
-              rb_raise(eCSVTooBigError, "too large field data");
-              break;
-          case CSV_EINVALID:
-              rb_raise(eCSVInvalid, csv_strerror(csv_error(parser)));
-              break;
-          default:
-            rb_raise(eCSVError, "failed to parse by unknown reason");
+            case CSV_EPARSE:
+                rb_raise(eCSVParseError, "error while parsing by malformed data");
+                break;
+            case CSV_ENOMEM:
+                rb_raise(eCSVNoMemError, "no memory");
+                break;
+            case CSV_ETOOBIG:
+                rb_raise(eCSVTooBigError, "too large field data");
+                break;
+            case CSV_EINVALID:
+                rb_raise(eCSVInvalid, csv_strerror(csv_error(parser)));
+                break;
+            default:
+                rb_raise(eCSVError, "failed to parse by unknown reason");
         }
     }
 
@@ -168,7 +194,7 @@ static VALUE simplecsv_parse(int argc, VALUE* argv, VALUE klass)
  * ==== Exceptions
  * [TypeError] <i>array</i> isn't <code>Array</code> or each content of array isn't <code>String</code>.
  */
-static VALUE simplecsv_write(VALUE klass, VALUE array)
+static VALUE simplecsv_s_write(VALUE klass, VALUE array)
 {
     int i; 
     VALUE ret, str;
@@ -204,18 +230,175 @@ static VALUE simplecsv_write(VALUE klass, VALUE array)
 
 
 
+/*
+ * call-seq:
+ *   new(options = nil)
+ *
+ * Create a SimpleCSV instance object.
+ *
+ * ==== Parameters
+ * [options] <code>Interger</code> : parser options. You can specify SimpleCSV::STRICT and/or SimpleCSV::REPALL_NL.
+ *
+ * ==== Exceptions
+ * [TypeError] <i>options</i> isn't <code>Fixnum</code> / <code>nil</code>.
+ * [SimpleCSV::Error] failed to initialize csv parser.
+ */
+static VALUE simplecsv_s_new(int argc, VALUE* argv, VALUE klass)
+{
+    VALUE new_obj;
+    VALUE opts;
+    struct csv_parser_object* parser_obj;
+    struct csv_parser* parser;
+    int options;
+
+    rb_scan_args(argc, argv, "01", &opts);
+
+    Set_Options(opts, options);
+
+    new_obj = Data_Make_Struct(klass, struct csv_parser_object, NULL, free_parser_object, parser_obj);
+
+    if (csv_init(&parser, options) != 0) {
+        rb_raise(eCSVError, "failed to initialize csv parser");
+    }
+
+    parser_obj->_parser = parser;
+
+    rb_obj_call_init(new_obj, argc, argv);
+
+    rb_ivar_set(new_obj, rb_intern("'row'"), rb_ary_new());
+
+    return new_obj;
+}
+
+
+
+/* :nodoc: */
+static VALUE simplecsv_initialize(VALUE self)
+{
+    return Qnil;
+}
+
+
+
+/*
+ * options -> options
+ *
+ * Get option values.
+ *
+ * === Returns
+ * [options] <code>Interger</code> : parser options. SimpleCSV::STRICT and/or SimpleCSVREPALL_NL.
+ */
+static VALUE simplecsv_get_options(VALUE self)
+{
+    struct csv_parser_object* parser_object;
+    struct csv_parser* parser;
+
+    Get_Parser(self, parser_object, parser);
+
+    return INT2FIX(parser->options);
+}
+
+
+
+/*
+ * call-seq:
+ *   options = new_options
+ *
+ * Set option values.
+ *
+ * === Parameters
+ * [options] <code>Interger</code> : parser options. SimpleCSV::STRICT and/or SimpleCSVREPALL_NL. You can specify nil to clear options.
+ */
+static VALUE simplecsv_set_options(VALUE self, VALUE opts)
+{
+    struct csv_parser_object* parser_object;
+    struct csv_parser* parser;
+    int options;
+
+    Get_Parser(self, parser_object, parser);
+    Set_Options(opts, options);
+
+    csv_opts(parser, options);
+
+    return Qnil;
+}
+
+
+
+/*
+ * call-seq:
+ *   parse(str, options = nil){|row| ...}
+ *
+ * Parse csv string.
+ * Same as SimpleCSV.parse.
+ *
+ * ==== Parameters
+ * [str] <code>String</code> : csv string.
+ * [options] <code>Interger</code> : parser options. You can specify SimpleCSV::STRICT and/or SimpleCSVREPALL_NL.
+ * [row] <code>Array</code> of <code>String</code>s : parsed data.
+ *
+ * ==== Exceptions
+ * [TypeError] <i>options</i> isn't <code>Fixnum</code>/<code>nil</code>.
+ * [SimpleCSV::Error] failed to initialize csv parser.
+ * [SimpleCSV::ParseError] failed to parse.
+ * [LocalJumpError] You must call this method with block.
+ */
+static VALUE simplecsv_parse(VALUE self, VALUE str)
+{
+    struct csv_parser_object* parser_object;
+    struct csv_parser* parser;
+    int len;
+
+    Check_Type(str, T_STRING);
+
+    Get_Parser(self, parser_object, parser);
+
+    len = RSTRING(str)->len;
+    if (csv_parse(parser, StringValuePtr(str), len, field_callback, row_callback, (void*)self) != len) {
+        switch (csv_error(parser)) {
+            case CSV_EPARSE:
+                rb_raise(eCSVParseError, "error while parsing by malformed data");
+                break;
+            case CSV_ENOMEM:
+                rb_raise(eCSVNoMemError, "no memory");
+                break;
+            case CSV_ETOOBIG:
+                rb_raise(eCSVTooBigError, "too large field data");
+                break;
+            case CSV_EINVALID:
+                rb_raise(eCSVInvalid, csv_strerror(csv_error(parser)));
+                break;
+            default:
+                rb_raise(eCSVError, "failed to parse by unknown reason");
+        }
+    }
+
+    csv_fini(parser, field_callback, row_callback, (void*)self);
+
+    return Qnil;
+}
+
+
+
 void Init_simplecsv(void)
 {
-    mSimpleCSV = rb_define_module("SimpleCSV");
-    eCSVError = rb_define_class_under(mSimpleCSV, "Error", rb_eStandardError);
-    eCSVParseError = rb_define_class_under(mSimpleCSV, "ParseError", rb_eStandardError);
-    eCSVNoMemError = rb_define_class_under(mSimpleCSV, "NoMemoryError", rb_eStandardError);
-    eCSVTooBigError = rb_define_class_under(mSimpleCSV, "TooBigError", rb_eStandardError);
-    eCSVInvalid = rb_define_class_under(mSimpleCSV, "InvalidError", rb_eStandardError);
-
-    rb_define_const(mSimpleCSV, "STRICT", INT2FIX(CSV_STRICT));
-    rb_define_const(mSimpleCSV, "REPALL_NL", INT2FIX(CSV_REPALL_NL));
-
-    rb_define_singleton_method(mSimpleCSV, "parse", simplecsv_parse, -1);
-    rb_define_singleton_method(mSimpleCSV, "write", simplecsv_write, 1);
+    cSimpleCSV = rb_define_class("SimpleCSV", rb_cObject);
+    eCSVError = rb_define_class_under(cSimpleCSV, "Error", rb_eStandardError);
+    eCSVParseError = rb_define_class_under(cSimpleCSV, "ParseError", rb_eStandardError);
+    eCSVNoMemError = rb_define_class_under(cSimpleCSV, "NoMemoryError", rb_eStandardError);
+    eCSVTooBigError = rb_define_class_under(cSimpleCSV, "TooBigError", rb_eStandardError);
+    eCSVInvalid = rb_define_class_under(cSimpleCSV, "InvalidError", rb_eStandardError);
+
+    rb_define_const(cSimpleCSV, "STRICT", INT2FIX(CSV_STRICT));
+    rb_define_const(cSimpleCSV, "REPALL_NL", INT2FIX(CSV_REPALL_NL));
+
+    rb_define_singleton_method(cSimpleCSV, "parse", simplecsv_s_parse, -1);
+    rb_define_singleton_method(cSimpleCSV, "write", simplecsv_s_write, 1);
+    rb_define_singleton_method(cSimpleCSV, "new", simplecsv_s_new, -1);
+
+    rb_define_method(cSimpleCSV, "initialize", simplecsv_initialize, -1);
+    rb_define_method(cSimpleCSV, "options", simplecsv_get_options, 0);
+    rb_define_method(cSimpleCSV, "options=", simplecsv_set_options, 1);
+    rb_define_method(cSimpleCSV, "parse", simplecsv_parse, 1);
+    rb_define_method(cSimpleCSV, "write", simplecsv_s_write, 1);
 }

data/lib/simplecsv.rb

@@ -1,3 +1,3 @@
-Dir[File.join(File.dirname(__FILE__), 'simplecsv/**/*.rb')].sort.each { |lib| require lib }
-
 require 'simplecsv.so'
+
+Dir[File.join(File.dirname(__FILE__), 'simplecsv/**/*.rb')].sort.each { |lib| require lib }

data/lib/simplecsv/version.rb

@@ -2,7 +2,7 @@ module SimpleCSV #:nodoc:
   module VERSION #:nodoc:
     MAJOR = 0
     MINOR = 1
-    TINY  = 0
+    TINY  = 2
 
     STRING = [MAJOR, MINOR, TINY].join('.')
   end

data/test/simplecsv/simplecsv_test.rb

@@ -17,14 +17,20 @@ Try to type below.
     assert_equal('constant', defined? SimpleCSV::REPALL_NL)
   end
 
-  def test_module
-    assert_instance_of Module, SimpleCSV
-    assert_raise(NoMethodError) do
+  def test_class
+    assert_instance_of Class, SimpleCSV
+  end
+
+  def test_instantiate
+    assert_nothing_raised(NoMethodError) do
       SimpleCSV.new
     end
+    assert_raise(TypeError) do
+      SimpleCSV.new("invalid")
+    end
   end
 
-  def test_parse
+  def test_s_parse
     assert_raise(ArgumentError) do
       SimpleCSV.parse()
     end
@@ -146,7 +152,7 @@ _CSV_
     assert_equal 3, rows
   end
 
-  def test_write
+  def test_s_write
     assert_equal '', SimpleCSV.write([])
 
     assert_equal '"1","2","3"', SimpleCSV.write(["1", "2", "3"])
@@ -168,4 +174,141 @@ _CSV_
       SimpleCSV.write([1, 2, 3])
     end
   end
+
+  def test_parse
+    simplecsv = SimpleCSV.new
+
+    assert_raise(ArgumentError) do
+      simplecsv.parse()
+    end
+
+    assert_raise(TypeError) do
+      simplecsv.parse(1)
+    end
+
+    assert_raise(LocalJumpError) do
+      simplecsv.parse('no block given')
+    end
+
+    assert_nothing_raised do
+      simplecsv.parse(''){}
+    end
+
+    assert_nil simplecsv.parse('')
+
+    simplecsv.parse(<<_CSV_) do |row|
+test
+_CSV_
+      assert_instance_of Array, row
+      assert_instance_of String, row.first
+      assert_equal 1, row.size
+      assert_equal ['test'], row
+    end
+
+    simplecsv.parse(<<_CSV_) do |row|
+1,2,3
+_CSV_
+      assert_instance_of Array, row
+      assert_instance_of String, row.first
+      assert_equal 3, row.size
+      assert_equal ['1', '2', '3'], row
+    end
+
+    rows = 0
+    simplecsv.parse(<<_CSV_) do |row|
+1,2,3
+2,3,4
+_CSV_
+      assert_instance_of Array, row
+      assert_instance_of String, row.first
+      assert_equal 3, row.size
+      rows += 1
+    end
+    assert_equal 2, rows
+
+    rows = 0
+    simplecsv.parse("1,2,3\r\n2,3,4\n") do |row|
+      rows += 1
+    end
+    assert_equal 2, rows
+
+    rows = 0
+    simplecsv.parse("1,2,3\r2,3,4\n") do |row|
+      rows += 1
+    end
+    assert_equal 2, rows
+
+    rows = 0
+    simplecsv.parse("1,2,3\n2,3,4\n \t\n") do |row|
+      rows += 1
+    end
+    assert_equal 2, rows
+
+    simplecsv.parse(" 1,2 , 3 ,\t4,5\t,\t6\t, 7\t,\t8 ") do |row|
+      assert_equal ['1', '2', '3', '4', '5', '6', '7', '8'], row
+    end
+
+    simplecsv.parse(<<_CSV_) do |row|
+" 1 "
+_CSV_
+      assert_equal [' 1 '], row
+    end
+
+    simplecsv.parse(<<_CSV_) do |row|
+"1", "2", "3"
+_CSV_
+      assert_equal ['1', '2', '3'], row
+    end
+
+    simplecsv.parse(<<_CSV_) do |row|
+"1, 2, 3"
+_CSV_
+      assert_equal ['1, 2, 3'], row
+    end
+
+    simplecsv.parse(<<_CSV_) do |row|
+"""1"
+_CSV_
+      assert_equal ['"1'], row
+    end
+
+    simplecsv.parse(<<_CSV_) do |row|
+"ab"c"
+_CSV_
+      assert_equal ['ab"c'], row
+    end
+
+    simplecsv = SimpleCSV.new(SimpleCSV::STRICT)
+    assert_raise(SimpleCSV::ParseError) do
+      simplecsv.parse(<<_CSV_){}
+"ab"c"
+_CSV_
+    end
+
+  end
+
+  def test_write
+    simplecsv = SimpleCSV.new
+
+    assert_equal '', simplecsv.write([])
+
+    assert_equal '"1","2","3"', simplecsv.write(["1", "2", "3"])
+ 
+    assert_equal '""""', simplecsv.write(['"'])
+
+    assert_equal 256, simplecsv.write(['"'*127]).size
+    assert_equal '"' * 256, simplecsv.write(['"'*127])
+
+    assert_raise(ArgumentError) do
+      simplecsv.write()
+    end
+
+    assert_raise(TypeError) do
+      simplecsv.write(1)
+    end
+
+    assert_raise(TypeError) do
+      simplecsv.write([1, 2, 3])
+    end
+  end
 end

metadata

@@ -3,8 +3,8 @@ rubygems_version: 0.8.11
 specification_version: 1
 name: simplecsv
 version: !ruby/object:Gem::Version 
-  version: 0.1.0
-date: 2007-03-09 00:00:00 +09:00
+  version: 0.1.2
+date: 2007-03-18 00:00:00 +09:00
 summary: SimpleCSV privides simple csv reading/writing api.
 require_paths: 
 - lib