fastcsv 0.0.2 → 0.0.3

data/ext/fastcsv/fastcsv.c

@@ -7,7 +7,7 @@
 // http://w3c.github.io/csvw/syntax/#ebnf
 
 // CSV implementation.
-// https://github.com/ruby/ruby/blob/master/lib/csv.rb
+// https://github.com/ruby/ruby/blob/trunk/lib/csv.rb
 
 // Ruby C extensions help.
 // https://github.com/ruby/ruby/blob/trunk/README.EXT
@@ -21,31 +21,42 @@ if (enc2 != NULL) { \
   field = rb_str_encode(field, rb_enc_from_encoding(enc), 0, Qnil); \
 }
 
-static VALUE mModule, rb_eParseError;
-static ID s_read, s_to_str, s_internal_encoding, s_external_encoding, s_string, s_encoding;
+#define FREE \
+if (buf != NULL) { \
+  free(buf); \
+} \
+if (row_sep != NULL) { \
+  free(row_sep); \
+}
+
+static VALUE cClass, cParser, eError;
+static ID s_read, s_row;
+
+// @see https://github.com/nofxx/georuby_c/blob/b3b91fd90980d7c295ac8f6012d89878ea7cd569/ext/types.h#L22
+typedef struct {
+  char *start;
+} Data;
 
 
-#line 125 "ext/fastcsv/fastcsv.rl"
+#line 170 "ext/fastcsv/fastcsv.rl"
 
 
 
-#line 33 "ext/fastcsv/fastcsv.c"
-static const int fastcsv_start = 4;
-static const int fastcsv_first_final = 4;
-static const int fastcsv_error = 0;
+#line 46 "ext/fastcsv/fastcsv.c"
+static const int raw_parse_start = 4;
+static const int raw_parse_first_final = 4;
+static const int raw_parse_error = 0;
 
-static const int fastcsv_en_main = 4;
+static const int raw_parse_en_main = 4;
 
 
-#line 128 "ext/fastcsv/fastcsv.rl"
+#line 173 "ext/fastcsv/fastcsv.rl"
 
 // 16 kB
 #define BUFSIZE 16384
 
 // @see http://rxr.whitequark.org/mri/source/io.c#4845
-static void
-rb_io_ext_int_to_encs(rb_encoding *ext, rb_encoding *intern, rb_encoding **enc, rb_encoding **enc2, int fmode)
-{
+static void rb_io_ext_int_to_encs(rb_encoding *ext, rb_encoding *intern, rb_encoding **enc, rb_encoding **enc2, int fmode) {
   int default_ext = 0;
 
   if (ext == NULL) {
@@ -70,15 +81,17 @@ rb_io_ext_int_to_encs(rb_encoding *ext, rb_encoding *intern, rb_encoding **enc,
   }
 }
 
-VALUE fastcsv(int argc, VALUE *argv, VALUE self) {
+static VALUE raw_parse(int argc, VALUE *argv, VALUE self) {
   int cs, act, have = 0, curline = 1, io = 0;
-  char *ts = 0, *te = 0, *buf = 0, *eof = 0;
+  char *ts = 0, *te = 0, *buf = 0, *eof = 0, *mark_row_sep = 0, *row_sep = NULL;
 
-  VALUE port, opts;
+  VALUE port, opts, r_encoding;
   VALUE row = rb_ary_new(), field = Qnil, bufsize = Qnil;
-  int done = 0, unclosed_line = 0, buffer_size = 0, taint = 0;
+  int done = 0, unclosed_line = 0, len_row_sep = 0, buffer_size = 0, taint = 0;
   rb_encoding *enc = NULL, *enc2 = NULL, *encoding = NULL;
-  VALUE r_encoding;
+
+  Data *d;
+  Data_Get_Struct(self, Data, d);
 
   VALUE option;
   char quote_char = '"';
@@ -87,8 +100,8 @@ VALUE fastcsv(int argc, VALUE *argv, VALUE self) {
   taint = OBJ_TAINTED(port);
   io = rb_respond_to(port, s_read);
   if (!io) {
-    if (rb_respond_to(port, s_to_str)) {
-      port = rb_funcall(port, s_to_str, 0);
+    if (rb_respond_to(port, rb_intern("to_str"))) {
+      port = rb_funcall(port, rb_intern("to_str"), 0);
       StringValue(port);
     }
     else {
@@ -112,7 +125,7 @@ VALUE fastcsv(int argc, VALUE *argv, VALUE self) {
   // @see http://ruby-doc.org/core-2.1.1/IO.html#method-c-new-label-Open+Mode
   option = rb_hash_aref(opts, ID2SYM(rb_intern("encoding")));
   if (TYPE(option) == T_STRING) {
-    // parse_mode_enc is not in header file.
+    // `parse_mode_enc` is not in header file.
     const char *estr = StringValueCStr(option), *ptr;
     char encname[ENCODING_MAXNAMELEN+1];
     int idx, idx2;
@@ -123,17 +136,17 @@ VALUE fastcsv(int argc, VALUE *argv, VALUE self) {
     ptr = strrchr(estr, ':');
     if (ptr) {
       long len = (ptr++) - estr;
-      if (len == 0 || len > ENCODING_MAXNAMELEN) {
+      if (len == 0 || len > ENCODING_MAXNAMELEN) { // ":enc"
         idx = -1;
       }
-      else {
+      else { // "enc2:enc" or "enc:-"
         memcpy(encname, estr, len);
         encname[len] = '\0';
         estr = encname;
         idx = rb_enc_find_index(encname);
       }
     }
-    else {
+    else { // "enc"
       idx = rb_enc_find_index(estr);
     }
 
@@ -141,7 +154,7 @@ VALUE fastcsv(int argc, VALUE *argv, VALUE self) {
       ext_enc = rb_enc_from_index(idx);
     }
     else {
-      if (idx != -2) {
+      if (idx != -2) { // ":enc"
         // `unsupported_encoding` is not in header file.
         rb_warn("Unsupported encoding %s ignored", estr);
       }
@@ -150,11 +163,11 @@ VALUE fastcsv(int argc, VALUE *argv, VALUE self) {
 
     int_enc = NULL;
     if (ptr) {
-      if (*ptr == '-' && *(ptr+1) == '\0') {
+      if (*ptr == '-' && *(ptr+1) == '\0') { // "enc:-"
         /* Special case - "-" => no transcoding */
         int_enc = (rb_encoding *)Qnil;
       }
-      else {
+      else { // "enc2:enc"
         idx2 = rb_enc_find_index(ptr);
         if (idx2 < 0) {
           // `unsupported_encoding` is not in header file.
@@ -175,29 +188,33 @@ VALUE fastcsv(int argc, VALUE *argv, VALUE self) {
     rb_raise(rb_eArgError, ":encoding has to be a String");
   }
 
-  // @see https://github.com/ruby/ruby/blob/70510d026f8d86693dccaba07417488eed09b41d/lib/csv.rb#L1567
-  // @see https://github.com/ruby/ruby/blob/70510d026f8d86693dccaba07417488eed09b41d/lib/csv.rb#L2300
-  if (rb_respond_to(port, s_internal_encoding)) {
-    r_encoding = rb_funcall(port, s_internal_encoding, 0);
+  // @see CSV#raw_encoding
+  // @see https://github.com/ruby/ruby/blob/ab337e61ecb5f42384ba7d710c36faf96a454e5c/lib/csv.rb#L2290
+  if (rb_respond_to(port, rb_intern("internal_encoding"))) {
+    r_encoding = rb_funcall(port, rb_intern("internal_encoding"), 0);
     if (NIL_P(r_encoding)) {
-      r_encoding = rb_funcall(port, s_external_encoding, 0);
+      r_encoding = rb_funcall(port, rb_intern("external_encoding"), 0);
     }
   }
-  else if (rb_respond_to(port, s_string)) {
-    r_encoding = rb_funcall(rb_funcall(port, s_string, 0), s_encoding, 0);
+  else if (rb_respond_to(port, rb_intern("string"))) {
+    r_encoding = rb_funcall(rb_funcall(port, rb_intern("string"), 0), rb_intern("encoding"), 0);
   }
-  else if (rb_respond_to(port, s_encoding)) {
-    r_encoding = rb_funcall(port, s_encoding, 0);
+  else if (rb_respond_to(port, rb_intern("encoding"))) {
+    r_encoding = rb_funcall(port, rb_intern("encoding"), 0);
   }
   else {
     r_encoding = rb_enc_from_encoding(rb_ascii8bit_encoding());
   }
+
+  // @see CSV#initialize
+  // @see https://github.com/ruby/ruby/blob/ab337e61ecb5f42384ba7d710c36faf96a454e5c/lib/csv.rb#L1510
   if (NIL_P(r_encoding)) {
     r_encoding = rb_enc_from_encoding(rb_default_internal_encoding());
   }
   if (NIL_P(r_encoding)) {
     r_encoding = rb_enc_from_encoding(rb_default_external_encoding());
   }
+
   if (enc2 != NULL) {
     encoding = enc2;
   }
@@ -208,11 +225,17 @@ VALUE fastcsv(int argc, VALUE *argv, VALUE self) {
     encoding = rb_enc_get(r_encoding);
   }
 
+  rb_ivar_set(self, s_row, Qnil);
+
   buffer_size = BUFSIZE;
   if (rb_ivar_defined(self, rb_intern("@buffer_size")) == Qtrue) {
     bufsize = rb_ivar_get(self, rb_intern("@buffer_size"));
     if (!NIL_P(bufsize)) {
       buffer_size = NUM2INT(bufsize);
+      // buffer_size = 0 can cause segmentation faults.
+      if (buffer_size == 0) {
+        buffer_size = BUFSIZE;
+      }
     }
   }
 
@@ -221,39 +244,47 @@ VALUE fastcsv(int argc, VALUE *argv, VALUE self) {
   }
 
   
-#line 225 "ext/fastcsv/fastcsv.c"
+#line 248 "ext/fastcsv/fastcsv.c"
 	{
-	cs = fastcsv_start;
+	cs = raw_parse_start;
 	ts = 0;
 	te = 0;
 	act = 0;
 	}
 
-#line 311 "ext/fastcsv/fastcsv.rl"
+#line 366 "ext/fastcsv/fastcsv.rl"
 
   while (!done) {
     VALUE str;
     char *p, *pe;
-    int len, space = buffer_size - have, tokstart_diff, tokend_diff;
+    int len, space = buffer_size - have, tokstart_diff, tokend_diff, start_diff, mark_row_sep_diff;
 
     if (io) {
       if (space == 0) {
-         tokstart_diff = ts - buf;
-         tokend_diff = te - buf;
+        // Not moving d->start will cause intermittent segmentation faults.
+        tokstart_diff = ts - buf;
+        tokend_diff = te - buf;
+        start_diff = d->start - buf;
+        mark_row_sep_diff = mark_row_sep - buf;
 
-         buffer_size += BUFSIZE;
-         REALLOC_N(buf, char, buffer_size);
+        buffer_size += BUFSIZE;
+        REALLOC_N(buf, char, buffer_size);
 
-         space = buffer_size - have;
+        space = buffer_size - have;
 
-         ts = buf + tokstart_diff;
-         te = buf + tokend_diff;
+        ts = buf + tokstart_diff;
+        te = buf + tokend_diff;
+        d->start = buf + start_diff;
+        mark_row_sep = buf + mark_row_sep_diff;
       }
       p = buf + have;
 
+      // Reads "`length` bytes without any conversion (binary mode)."
+      // "The resulted string is always ASCII-8BIT encoding."
+      // @see http://www.ruby-doc.org/core-2.1.4/IO.html#method-i-read
       str = rb_funcall(port, s_read, 1, INT2FIX(space));
       if (NIL_P(str)) {
-        // StringIO#read returns nil for empty string.
+        // "`nil` means it met EOF at beginning," e.g. for `StringIO.new("")`.
         len = 0;
       }
       else {
@@ -261,6 +292,7 @@ VALUE fastcsv(int argc, VALUE *argv, VALUE self) {
         memcpy(p, StringValuePtr(str), len);
       }
 
+      // "The 1 to `length`-1 bytes string means it met EOF after reading the result."
       if (len < space) {
         // EOF actions don't work in scanners, so we add a sentinel value.
         // @see http://www.complang.org/pipermail/ragel-users/2007-May/001516.html
@@ -276,9 +308,13 @@ VALUE fastcsv(int argc, VALUE *argv, VALUE self) {
       done = 1;
     }
 
+    if (d->start == 0) {
+      d->start = p;
+    }
+
     pe = p + len;
     
-#line 282 "ext/fastcsv/fastcsv.c"
+#line 318 "ext/fastcsv/fastcsv.c"
 	{
 	if ( p == pe )
 		goto _test_eof;
@@ -296,56 +332,80 @@ tr0:
 	}
 	}
 	goto st4;
-tr10:
-#line 101 "ext/fastcsv/fastcsv.rl"
+tr5:
+#line 49 "ext/fastcsv/fastcsv.rl"
 	{
-    if (!NIL_P(field) || RARRAY_LEN(row)) {
-      rb_ary_push(row, field);
+    if (p == ts) {
+      // Unquoted empty fields are nil, not "", in Ruby.
+      field = Qnil;
     }
-    if (RARRAY_LEN(row)) {
-      rb_yield(row);
+    else if (p > ts) {
+      field = rb_enc_str_new(ts, p - ts, encoding);
+      ENCODE;
     }
   }
-#line 123 "ext/fastcsv/fastcsv.rl"
+#line 95 "ext/fastcsv/fastcsv.rl"
+	{
+    rb_ary_push(row, field);
+    field = Qnil;
+  }
+#line 166 "ext/fastcsv/fastcsv.rl"
 	{te = p+1;}
 	goto st4;
-tr16:
-#line 123 "ext/fastcsv/fastcsv.rl"
-	{te = p;p--;}
-	goto st4;
-tr17:
-#line 122 "ext/fastcsv/fastcsv.rl"
-	{te = p;p--;}
-	goto st4;
-tr18:
-#line 101 "ext/fastcsv/fastcsv.rl"
+tr9:
+#line 138 "ext/fastcsv/fastcsv.rl"
 	{
+    if (d->start == 0 || p == d->start) {
+      rb_ivar_set(self, s_row, rb_str_new2(""));
+    }
+    else if (p > d->start) {
+      rb_ivar_set(self, s_row, rb_str_new(d->start, p - d->start));
+    }
+
     if (!NIL_P(field) || RARRAY_LEN(row)) {
       rb_ary_push(row, field);
     }
+
     if (RARRAY_LEN(row)) {
       rb_yield(row);
     }
   }
-#line 122 "ext/fastcsv/fastcsv.rl"
+#line 168 "ext/fastcsv/fastcsv.rl"
+	{te = p+1;}
+	goto st4;
+tr12:
+#line 95 "ext/fastcsv/fastcsv.rl"
+	{
+    rb_ary_push(row, field);
+    field = Qnil;
+  }
+#line 166 "ext/fastcsv/fastcsv.rl"
 	{te = p+1;}
 	goto st4;
-tr20:
-#line 121 "ext/fastcsv/fastcsv.rl"
+tr15:
+#line 168 "ext/fastcsv/fastcsv.rl"
 	{te = p;p--;}
 	goto st4;
-tr21:
-#line 101 "ext/fastcsv/fastcsv.rl"
+tr16:
+#line 100 "ext/fastcsv/fastcsv.rl"
 	{
-    if (!NIL_P(field) || RARRAY_LEN(row)) {
-      rb_ary_push(row, field);
+    d->start = p;
+
+    if (len_row_sep) {
+      if (p - mark_row_sep != len_row_sep || row_sep[0] != *mark_row_sep || len_row_sep == 2 && row_sep[1] != *(mark_row_sep + 1)) {
+        FREE;
+
+        rb_raise(eError, "Unquoted fields do not allow \\r or \\n (line %d).", curline - 1);
+      }
     }
-    if (RARRAY_LEN(row)) {
-      rb_yield(row);
+    else {
+      len_row_sep = p - mark_row_sep;
+      row_sep = ALLOC_N(char, p - mark_row_sep);
+      memcpy(row_sep, mark_row_sep, p - mark_row_sep);
     }
   }
-#line 121 "ext/fastcsv/fastcsv.rl"
-	{te = p+1;}
+#line 167 "ext/fastcsv/fastcsv.rl"
+	{te = p;p--;}
 	goto st4;
 st4:
 #line 1 "NONE"
@@ -357,12 +417,12 @@ st4:
 case 4:
 #line 1 "NONE"
 	{ts = p;}
-#line 361 "ext/fastcsv/fastcsv.c"
+#line 421 "ext/fastcsv/fastcsv.c"
 	switch( (*p) ) {
-		case 0: goto tr14;
+		case 0: goto tr13;
 		case 10: goto tr3;
 		case 13: goto tr4;
-		case 34: goto tr15;
+		case 34: goto tr14;
 		case 44: goto tr5;
 	}
 	goto st1;
@@ -381,7 +441,7 @@ case 1:
 tr2:
 #line 1 "NONE"
 	{te = p+1;}
-#line 40 "ext/fastcsv/fastcsv.rl"
+#line 49 "ext/fastcsv/fastcsv.rl"
 	{
     if (p == ts) {
       // Unquoted empty fields are nil, not "", in Ruby.
@@ -392,33 +452,41 @@ tr2:
       ENCODE;
     }
   }
-#line 101 "ext/fastcsv/fastcsv.rl"
+#line 138 "ext/fastcsv/fastcsv.rl"
 	{
+    if (d->start == 0 || p == d->start) {
+      rb_ivar_set(self, s_row, rb_str_new2(""));
+    }
+    else if (p > d->start) {
+      rb_ivar_set(self, s_row, rb_str_new(d->start, p - d->start));
+    }
+
     if (!NIL_P(field) || RARRAY_LEN(row)) {
       rb_ary_push(row, field);
     }
+
     if (RARRAY_LEN(row)) {
       rb_yield(row);
     }
   }
-#line 123 "ext/fastcsv/fastcsv.rl"
+#line 168 "ext/fastcsv/fastcsv.rl"
 	{act = 3;}
 	goto st5;
 st5:
 	if ( ++p == pe )
 		goto _test_eof5;
 case 5:
-#line 412 "ext/fastcsv/fastcsv.c"
+#line 480 "ext/fastcsv/fastcsv.c"
 	switch( (*p) ) {
 		case 0: goto tr2;
 		case 10: goto tr3;
 		case 13: goto tr4;
-		case 34: goto tr16;
+		case 34: goto tr15;
 		case 44: goto tr5;
 	}
 	goto st1;
 tr3:
-#line 40 "ext/fastcsv/fastcsv.rl"
+#line 49 "ext/fastcsv/fastcsv.rl"
 	{
     if (p == ts) {
       // Unquoted empty fields are nil, not "", in Ruby.
@@ -429,8 +497,19 @@ tr3:
       ENCODE;
     }
   }
-#line 91 "ext/fastcsv/fastcsv.rl"
+#line 117 "ext/fastcsv/fastcsv.rl"
 	{
+    mark_row_sep = p;
+
+    curline++;
+
+    if (d->start == 0 || p == d->start) {
+      rb_ivar_set(self, s_row, rb_str_new2(""));
+    }
+    else if (p > d->start) {
+      rb_ivar_set(self, s_row, rb_str_new(d->start, p - d->start));
+    }
+
     if (!NIL_P(field) || RARRAY_LEN(row)) { // same as new_field
       rb_ary_push(row, field);
       field = Qnil;
@@ -438,21 +517,22 @@ tr3:
 
     rb_yield(row);
     row = rb_ary_new();
-  }
-#line 28 "ext/fastcsv/fastcsv.rl"
-	{
-    curline++;
   }
 	goto st6;
-tr19:
-#line 28 "ext/fastcsv/fastcsv.rl"
+tr10:
+#line 117 "ext/fastcsv/fastcsv.rl"
 	{
+    mark_row_sep = p;
+
     curline++;
-  }
-	goto st6;
-tr11:
-#line 91 "ext/fastcsv/fastcsv.rl"
-	{
+
+    if (d->start == 0 || p == d->start) {
+      rb_ivar_set(self, s_row, rb_str_new2(""));
+    }
+    else if (p > d->start) {
+      rb_ivar_set(self, s_row, rb_str_new(d->start, p - d->start));
+    }
+
     if (!NIL_P(field) || RARRAY_LEN(row)) { // same as new_field
       rb_ary_push(row, field);
       field = Qnil;
@@ -460,22 +540,16 @@ tr11:
 
     rb_yield(row);
     row = rb_ary_new();
-  }
-#line 28 "ext/fastcsv/fastcsv.rl"
-	{
-    curline++;
   }
 	goto st6;
 st6:
 	if ( ++p == pe )
 		goto _test_eof6;
 case 6:
-#line 474 "ext/fastcsv/fastcsv.c"
-	if ( (*p) == 0 )
-		goto tr18;
-	goto tr17;
+#line 550 "ext/fastcsv/fastcsv.c"
+	goto tr16;
 tr4:
-#line 40 "ext/fastcsv/fastcsv.rl"
+#line 49 "ext/fastcsv/fastcsv.rl"
 	{
     if (p == ts) {
       // Unquoted empty fields are nil, not "", in Ruby.
@@ -486,8 +560,19 @@ tr4:
       ENCODE;
     }
   }
-#line 91 "ext/fastcsv/fastcsv.rl"
+#line 117 "ext/fastcsv/fastcsv.rl"
 	{
+    mark_row_sep = p;
+
+    curline++;
+
+    if (d->start == 0 || p == d->start) {
+      rb_ivar_set(self, s_row, rb_str_new2(""));
+    }
+    else if (p > d->start) {
+      rb_ivar_set(self, s_row, rb_str_new(d->start, p - d->start));
+    }
+
     if (!NIL_P(field) || RARRAY_LEN(row)) { // same as new_field
       rb_ary_push(row, field);
       field = Qnil;
@@ -495,15 +580,22 @@ tr4:
 
     rb_yield(row);
     row = rb_ary_new();
-  }
-#line 28 "ext/fastcsv/fastcsv.rl"
-	{
-    curline++;
   }
 	goto st7;
-tr12:
-#line 91 "ext/fastcsv/fastcsv.rl"
+tr11:
+#line 117 "ext/fastcsv/fastcsv.rl"
 	{
+    mark_row_sep = p;
+
+    curline++;
+
+    if (d->start == 0 || p == d->start) {
+      rb_ivar_set(self, s_row, rb_str_new2(""));
+    }
+    else if (p > d->start) {
+      rb_ivar_set(self, s_row, rb_str_new(d->start, p - d->start));
+    }
+
     if (!NIL_P(field) || RARRAY_LEN(row)) { // same as new_field
       rb_ary_push(row, field);
       field = Qnil;
@@ -511,24 +603,20 @@ tr12:
 
     rb_yield(row);
     row = rb_ary_new();
-  }
-#line 28 "ext/fastcsv/fastcsv.rl"
-	{
-    curline++;
   }
 	goto st7;
 st7:
 	if ( ++p == pe )
 		goto _test_eof7;
 case 7:
-#line 525 "ext/fastcsv/fastcsv.c"
-	switch( (*p) ) {
-		case 0: goto tr18;
-		case 10: goto tr19;
-	}
-	goto tr17;
-tr5:
-#line 40 "ext/fastcsv/fastcsv.rl"
+#line 613 "ext/fastcsv/fastcsv.c"
+	if ( (*p) == 10 )
+		goto st6;
+	goto tr16;
+tr13:
+#line 1 "NONE"
+	{te = p+1;}
+#line 49 "ext/fastcsv/fastcsv.rl"
 	{
     if (p == ts) {
       // Unquoted empty fields are nil, not "", in Ruby.
@@ -539,73 +627,40 @@ tr5:
       ENCODE;
     }
   }
-#line 86 "ext/fastcsv/fastcsv.rl"
-	{
-    rb_ary_push(row, field);
-    field = Qnil;
-  }
-	goto st8;
-tr13:
-#line 86 "ext/fastcsv/fastcsv.rl"
-	{
-    rb_ary_push(row, field);
-    field = Qnil;
-  }
-	goto st8;
-st8:
-	if ( ++p == pe )
-		goto _test_eof8;
-case 8:
-#line 560 "ext/fastcsv/fastcsv.c"
-	if ( (*p) == 0 )
-		goto tr21;
-	goto tr20;
-tr14:
-#line 1 "NONE"
-	{te = p+1;}
-#line 101 "ext/fastcsv/fastcsv.rl"
+#line 138 "ext/fastcsv/fastcsv.rl"
 	{
+    if (d->start == 0 || p == d->start) {
+      rb_ivar_set(self, s_row, rb_str_new2(""));
+    }
+    else if (p > d->start) {
+      rb_ivar_set(self, s_row, rb_str_new(d->start, p - d->start));
+    }
+
     if (!NIL_P(field) || RARRAY_LEN(row)) {
       rb_ary_push(row, field);
     }
+
     if (RARRAY_LEN(row)) {
       rb_yield(row);
     }
   }
-#line 40 "ext/fastcsv/fastcsv.rl"
-	{
-    if (p == ts) {
-      // Unquoted empty fields are nil, not "", in Ruby.
-      field = Qnil;
-    }
-    else if (p > ts) {
-      field = rb_enc_str_new(ts, p - ts, encoding);
-      ENCODE;
-    }
-  }
-#line 123 "ext/fastcsv/fastcsv.rl"
+#line 168 "ext/fastcsv/fastcsv.rl"
 	{act = 3;}
-	goto st9;
-st9:
+	goto st8;
+st8:
 	if ( ++p == pe )
-		goto _test_eof9;
-case 9:
-#line 594 "ext/fastcsv/fastcsv.c"
+		goto _test_eof8;
+case 8:
+#line 655 "ext/fastcsv/fastcsv.c"
 	switch( (*p) ) {
-		case 10: goto tr16;
-		case 13: goto tr16;
-		case 34: goto tr16;
-		case 44: goto tr16;
+		case 10: goto tr15;
+		case 13: goto tr15;
+		case 34: goto tr15;
+		case 44: goto tr15;
 	}
 	goto st1;
-tr8:
-#line 28 "ext/fastcsv/fastcsv.rl"
-	{
-    curline++;
-  }
-	goto st2;
-tr15:
-#line 32 "ext/fastcsv/fastcsv.rl"
+tr14:
+#line 41 "ext/fastcsv/fastcsv.rl"
 	{
     unclosed_line = curline;
   }
@@ -614,19 +669,17 @@ st2:
 	if ( ++p == pe )
 		goto _test_eof2;
 case 2:
-#line 618 "ext/fastcsv/fastcsv.c"
+#line 673 "ext/fastcsv/fastcsv.c"
 	switch( (*p) ) {
 		case 0: goto st0;
-		case 10: goto tr8;
-		case 13: goto tr8;
-		case 34: goto tr9;
+		case 34: goto tr8;
 	}
 	goto st2;
 st0:
 cs = 0;
 	goto _out;
-tr9:
-#line 51 "ext/fastcsv/fastcsv.rl"
+tr8:
+#line 60 "ext/fastcsv/fastcsv.rl"
 	{
     if (p == ts) {
       field = rb_enc_str_new("", 0, encoding);
@@ -653,7 +706,7 @@ tr9:
         reader++;
       }
 
-      field = rb_enc_str_new(copy, writer - copy, enc);
+      field = rb_enc_str_new(copy, writer - copy, encoding);
       ENCODE;
 
       if (copy != NULL) {
@@ -661,7 +714,7 @@ tr9:
       }
     }
   }
-#line 36 "ext/fastcsv/fastcsv.rl"
+#line 45 "ext/fastcsv/fastcsv.rl"
 	{
     unclosed_line = 0;
   }
@@ -670,13 +723,13 @@ st3:
 	if ( ++p == pe )
 		goto _test_eof3;
 case 3:
-#line 674 "ext/fastcsv/fastcsv.c"
+#line 727 "ext/fastcsv/fastcsv.c"
 	switch( (*p) ) {
-		case 0: goto tr10;
-		case 10: goto tr11;
-		case 13: goto tr12;
+		case 0: goto tr9;
+		case 10: goto tr10;
+		case 13: goto tr11;
 		case 34: goto st2;
-		case 44: goto tr13;
+		case 44: goto tr12;
 	}
 	goto st0;
 	}
@@ -686,7 +739,6 @@ case 3:
 	_test_eof6: cs = 6; goto _test_eof; 
 	_test_eof7: cs = 7; goto _test_eof; 
 	_test_eof8: cs = 8; goto _test_eof; 
-	_test_eof9: cs = 9; goto _test_eof; 
 	_test_eof2: cs = 2; goto _test_eof; 
 	_test_eof3: cs = 3; goto _test_eof; 
 
@@ -695,32 +747,26 @@ case 3:
 	{
 	switch ( cs ) {
 	case 1: goto tr0;
-	case 5: goto tr16;
-	case 6: goto tr17;
-	case 7: goto tr17;
-	case 8: goto tr20;
-	case 9: goto tr16;
+	case 5: goto tr15;
+	case 6: goto tr16;
+	case 7: goto tr16;
+	case 8: goto tr15;
 	}
 	}
 
 	_out: {}
 	}
 
-#line 359 "ext/fastcsv/fastcsv.rl"
+#line 427 "ext/fastcsv/fastcsv.rl"
+
+    if (done && cs < raw_parse_first_final) {
+      FREE;
 
-    if (done && cs < fastcsv_first_final) {
-      if (buf != NULL) {
-        free(buf);
-      }
       if (unclosed_line) {
-        rb_raise(rb_eParseError, "Unclosed quoted field on line %d.", unclosed_line);
+        rb_raise(eError, "Unclosed quoted field on line %d.", unclosed_line);
       }
-      // Ruby raises different errors for illegal quoting, depending on whether
-      // a quoted string is followed by a string ("Unclosed quoted field on line
-      // %d.") or by a string ending in a quote ("Missing or stray quote in line
-      // %d"). These precisions are kind of bogus, but we can try using $!.
       else {
-        rb_raise(rb_eParseError, "Illegal quoting in line %d.", curline);
+        rb_raise(eError, "Illegal quoting in line %d.", curline);
       }
     }
 
@@ -735,23 +781,35 @@ case 3:
     }
   }
 
-  if (buf != NULL) {
-    free(buf);
-  }
+  FREE;
 
   return Qnil;
 }
 
+// @see https://github.com/ruby/ruby/blob/trunk/README.EXT#L616
+static VALUE allocate(VALUE class) {
+  // @see https://github.com/nofxx/georuby_c/blob/b3b91fd90980d7c295ac8f6012d89878ea7cd569/ext/line.c#L66
+  Data *d = ALLOC(Data);
+  d->start = 0;
+  // @see https://github.com/nofxx/georuby_c/blob/b3b91fd90980d7c295ac8f6012d89878ea7cd569/ext/point.h#L26
+  // rb_gc_mark(d->start) or rb_gc_mark(d) cause warning "passing argument 1 of ‘rb_gc_mark’ makes integer from pointer without a cast"
+  // free(d->start) causes error "pointer being freed was not allocated"
+  return Data_Wrap_Struct(class, NULL, free, d);
+}
+
+// @see http://tenderlovemaking.com/2009/12/18/writing-ruby-c-extensions-part-1.html
+// @see http://tenderlovemaking.com/2010/12/11/writing-ruby-c-extensions-part-2.html
 void Init_fastcsv() {
   s_read = rb_intern("read");
-  s_to_str = rb_intern("to_str");
-  s_internal_encoding = rb_intern("internal_encoding");
-  s_external_encoding = rb_intern("external_encoding");
-  s_string = rb_intern("string");
-  s_encoding = rb_intern("encoding");
-
-  mModule = rb_define_module("FastCSV");
-  rb_define_attr(rb_singleton_class(mModule), "buffer_size", 1, 1);
-  rb_define_singleton_method(mModule, "raw_parse", fastcsv, -1);
-  rb_eParseError = rb_define_class_under(mModule, "ParseError", rb_eStandardError);
+  s_row = rb_intern("@row");
+
+  cClass = rb_define_class("FastCSV", rb_const_get(rb_cObject, rb_intern("CSV"))); // class FastCSV < CSV
+  cParser = rb_define_class_under(cClass, "Parser", rb_cObject);                   //   class Parser
+  rb_define_alloc_func(cParser, allocate);                                         //
+  rb_define_method(cParser, "raw_parse", raw_parse, -1);                           //     def raw_parse(port, opts = nil); end
+  rb_define_attr(cParser, "row", 1, 0);                                            //     attr_reader :row
+  rb_define_attr(cParser, "buffer_size", 1, 1);                                    //     attr_accessor :buffer_size
+                                                                                   //   end
+  eError = rb_define_class_under(cClass, "MalformedCSVError", rb_eRuntimeError);   //   class MalformedCSVError < RuntimeError
+                                                                                   // end
 }