added skip to sax parser

ext/ox/ox.c

@@ -711,7 +711,8 @@ load_file(int argc, VALUE *argv, VALUE self) {
  * @param [Hash] options parse options
  * @param [true|false] :convert_special flag indicating special characters like < are converted
  * @param [true|false] :symbolize flag indicating the parser symbolize element and attribute names
- * @param [true|false] :smart flag indicating the parser use hints if available (use with html)
+ * @param [true|false] :smart flag indicating the parser uses hints if available (use with html)
+ * @param [:skip_return|:skip_white] :skip flag indicating the parser skips \r or collpase white space into a single space. Default (skip nothing)
  */
 static VALUE
 sax_parse(int argc, VALUE *argv, VALUE self) {
@@ -720,6 +721,7 @@ sax_parse(int argc, VALUE *argv, VALUE self) {
     options.symbolize = 1;
     options.convert_special = 0;
     options.smart = 0;
+    options.skip = NoSkip;
 
     if (argc < 2) {
 	rb_raise(ox_parse_error_class, "Wrong number of arguments to sax_parse.\n");
@@ -737,6 +739,13 @@ sax_parse(int argc, VALUE *argv, VALUE self) {
 	if (Qnil != (v = rb_hash_lookup(h, symbolize_sym))) {
 	    options.symbolize = (Qtrue == v);
 	}
+	if (Qnil != (v = rb_hash_lookup(h, skip_sym))) {
+	    if (skip_return_sym == v) {
+		options.skip = CrSkip;
+	    } else if (skip_white_sym == v) {
+		options.skip = SpcSkip;
+	    }
+	}
     }
     ox_sax_parse(argv[0], argv[1], &options);
 

ext/ox/sax.c

@@ -907,6 +907,7 @@ read_element_end(SaxDrive dr) {
 
 static char
 read_text(SaxDrive dr) {
+    VALUE	args[1];
     char        c;
     int		line = dr->buf.line;
     int		col = dr->buf.col - 1;
@@ -923,8 +924,6 @@ read_text(SaxDrive dr) {
 	*(dr->buf.tail - 1) = '\0';
     }
     if (dr->has.value) {
-        VALUE   args[1];
-
 	if (dr->has.line) {
 	    rb_ivar_set(dr->handler, ox_at_line_id, LONG2NUM(line));
 	}
@@ -934,11 +933,19 @@ read_text(SaxDrive dr) {
 	*args = dr->value_obj;
         rb_funcall2(dr->handler, ox_value_id, 1, args);
     } else if (dr->has.text) {
-        VALUE   args[1];
-
         if (dr->options.convert_special) {
             ox_sax_collapse_special(dr, dr->buf.str, line, col);
         }
+	switch (dr->options.skip) {
+	case CrSkip:
+	    buf_collapse_return(dr->buf.str);
+	    break;
+	case SpcSkip:
+	    buf_collapse_white(dr->buf.str);
+	    break;
+	default:
+	    break;
+	}
         args[0] = rb_str_new2(dr->buf.str);
 #if HAS_ENCODING_SUPPORT
         if (0 != dr->encoding) {

ext/ox/sax.h

@@ -35,11 +35,13 @@
 #include "sax_has.h"
 #include "sax_stack.h"
 #include "sax_hint.h"
+#include "ox.h"
 
 typedef struct _SaxOptions {
     int			symbolize;
     int			convert_special;
     int			smart;
+    SkipMode		skip;
 } *SaxOptions;
 
 typedef struct _SaxDrive {
@@ -60,6 +62,7 @@ typedef struct _SaxDrive {
 #endif
 } *SaxDrive;
 
+extern void	ox_collapse_return(char *str);
 extern void	ox_sax_parse(VALUE handler, VALUE io, SaxOptions options);
 extern void	ox_sax_drive_cleanup(SaxDrive dr);
 extern void	ox_sax_drive_error(SaxDrive dr, const char *msg);

ext/ox/sax_as.c

@@ -146,6 +146,16 @@ sax_value_as_s(VALUE self) {
     if (dr->options.convert_special) {
 	ox_sax_collapse_special(dr, dr->buf.str, dr->buf.line, dr->buf.col);
     }
+    switch (dr->options.skip) {
+    case CrSkip:
+	buf_collapse_return(dr->buf.str);
+	break;
+    case SpcSkip:
+	buf_collapse_white(dr->buf.str);
+	break;
+    default:
+	break;
+    }
     rs = rb_str_new2(dr->buf.str);
 #if HAS_ENCODING_SUPPORT
     if (0 != dr->encoding) {

ext/ox/sax_buf.h

@@ -195,4 +195,40 @@ buf_checkback(Buf buf, CheckPt cp) {
     return cp->c;
 }
 
+static inline void
+buf_collapse_return(char *str) {
+    char	*back = str;
+
+    for (; '\0' != *str; str++) {
+	if ('\r' != *str) {
+	    *back++ = *str;
+	}
+    }
+    *back = '\0';
+}
+
+static inline void
+buf_collapse_white(char *str) {
+    char	*s = str;
+    char	*back = str;
+
+    for (; '\0' != *s; s++) {
+	switch(*s) {
+	case ' ':
+	case '\t':
+	case '\f':
+	case '\n':
+	case '\r':
+	    if (back == str || ' ' != *(back - 1)) {
+		*back++ = ' ';
+	    }
+	    break;
+	default:
+	    *back++ = *s;
+	    break;
+	}
+    }
+    *back = '\0';
+}
+
 #endif /* __OX_SAX_BUF_H__ */

lib/ox/version.rb

@@ -1,5 +1,5 @@
 
 module Ox
   # Current version of the module. 
-  VERSION = '2.1.2b1'
+  VERSION = '2.1.2b2'
 end

test/sax/helpers.rb

@@ -1,6 +1,5 @@
 require 'handlers'
 
-
 module SaxTestHelpers
 
   # A helper method to initiate a sax parsing using a specified xml

test/sax/sax_test.rb

@@ -645,6 +645,27 @@ def test_sax_value_string
     assert_equal('cheese', handler.item)
   end
 
+  def test_sax_skip_none
+    handler = TypeSax.new(:as_s)
+    xml = %{<top>  Pete\r\n  Ohler</top>}
+    Ox.sax_parse(handler, StringIO.new(xml))
+    assert_equal(%{  Pete\r\n  Ohler}, handler.item)
+  end
+
+  def test_sax_skip_return
+    handler = TypeSax.new(:as_s)
+    xml = %{<top>  Pete\r\n  Ohler</top>}
+    Ox.sax_parse(handler, StringIO.new(xml), :skip => :skip_return)
+    assert_equal(%{  Pete\n  Ohler}, handler.item)
+  end
+
+  def test_sax_skip_white
+    handler = TypeSax.new(:as_s)
+    xml = %{<top>  Pete\r\n  Ohler</top>}
+    Ox.sax_parse(handler, StringIO.new(xml), :skip => :skip_white)
+    assert_equal(%{ Pete Ohler}, handler.item)
+  end
+
   def test_sax_value_float
     handler = TypeSax.new(:as_f)
     Ox.sax_parse(handler, StringIO.new(%{<top>7</top>}))

test/tests.rb

@@ -363,7 +363,6 @@ def test_skip_none
     xml = %{<top>  Pete\r\n  Ohler</top>}
     doc = Ox.load(xml, :mode => :generic, :symbolize_keys => false, :skip => :skip_none)
     x2 = Ox.dump(doc, :indent => 0)
-    puts "*** '#{x2}'"
     assert_equal(%{\n<top>  Pete\r\n  Ohler</top>\n}, x2)
   end