summaryrefslogtreecommitdiff
path: root/lib/rss/xmlscanner.rb
diff options
context:
space:
mode:
Diffstat (limited to 'lib/rss/xmlscanner.rb')
-rw-r--r--lib/rss/xmlscanner.rb97
1 files changed, 97 insertions, 0 deletions
diff --git a/lib/rss/xmlscanner.rb b/lib/rss/xmlscanner.rb
new file mode 100644
index 0000000000..81dae96df5
--- /dev/null
+++ b/lib/rss/xmlscanner.rb
@@ -0,0 +1,97 @@
+require 'xmlscan/scanner'
+
+module RSS
+
+ class Parser < BaseParser
+
+ private
+ def _parse
+ begin
+ XMLScan::XMLScanner.new(@listener).parse(@rss)
+ rescue XMLScan::Error => e
+ raise NotWellFormedError.new(e.lineno){e.message}
+ end
+ end
+
+ end
+
+ class Listener < BaseListener
+
+ include XMLScan::Visitor
+ include ListenerMixin
+
+ ENTITIES = {
+ 'lt' => '<',
+ 'gt' => '>',
+ 'amp' => '&',
+ 'quot' => '"',
+ 'apos' => '\''
+ }
+
+ def on_xmldecl_version(str)
+ @version = str
+ end
+
+ def on_xmldecl_encoding(str)
+ @encoding = str
+ end
+
+ def on_xmldecl_standalone(str)
+ @standalone = str
+ end
+
+ def on_xmldecl_end
+ xmldecl(@version, @encoding, @standalone)
+ end
+
+ alias_method(:on_chardata, :text)
+ alias_method(:on_cdata, :text)
+
+ def on_etag(name)
+ tag_end(name)
+ end
+
+ def on_entityref(ref)
+ text(ENTITIES[ref])
+ end
+
+ def on_charref(code)
+ text([code].pack('U'))
+ end
+
+ alias_method(:on_charref_hex, :on_charref)
+
+ def on_stag(name)
+ @attrs = {}
+ end
+
+ def on_attribute(name)
+ @attrs[name] = @current_attr = ''
+ end
+
+ def on_attr_value(str)
+ @current_attr << str
+ end
+
+ def on_attr_entityref(ref)
+ @current_attr << ENTITIES[ref]
+ end
+
+ def on_attr_charref(code)
+ @current_attr << [code].pack('U')
+ end
+
+ alias_method(:on_attr_charref_hex, :on_attr_charref)
+
+ def on_stag_end(name)
+ tag_start(name, @attrs)
+ end
+
+ def on_stag_end_empty(name)
+ tag_start(name, @attrs)
+ tag_end(name)
+ end
+
+ end
+
+end