log_formatter.py
changeset 69 1f182913b1f2
parent 65 8b50694f841e
child 72 5ade0288f2ec
--- a/log_formatter.py	Mon Feb 09 12:59:03 2009 +0200
+++ b/log_formatter.py	Mon Feb 09 13:19:00 2009 +0200
@@ -2,8 +2,7 @@
     Format LogLines into some other representation
 """
 
-# for escape
-import cgi
+import re, xml.sax.saxutils
 
 from log_line import LogTypes
 
@@ -72,6 +71,29 @@
 
         abstract
 
+class BaseHTMLFormatter (object) :
+    """
+        Implements some HTML-formatting utils
+    """
+
+    URL_REGEXP = re.compile(r"http://\S+")
+
+    def _process_links (self, line) :
+        """
+            Processed the rendered line, adding in <a href>'s for things that look like URLs, returning the new line.
+
+            The line should already be escaped
+        """
+
+        def _encode_url (match) :
+            # encode URL
+            url_html = match.group(0)
+            url_link = xml.sax.saxutils.unescape(url_html)
+
+            return '<a href="%(url_link)s">%(url_html)s</a>' % dict(url_link=url_link, url_html=url_html)
+
+        return self.URL_REGEXP.sub(_encode_url, line)
+
 class IrssiTextFormatter (LogFormatter) :
     """
         Implements format_txt for irssi-style output
@@ -88,7 +110,7 @@
             # using __TYPES
             yield self._format_line_text(line, self.__FMT, full_timestamps)
 
-class IrssiFormatter (IrssiTextFormatter) :
+class IrssiFormatter (IrssiTextFormatter, BaseHTMLFormatter) :
     """
         Implements plain black-and-white irssi-style formatting
     """
@@ -107,7 +129,13 @@
         # format using IrssiTextFormatter
         for line in self.format_txt(lines, full_timestamps) :
             # escape HTML
-            yield cgi.escape(line)
+            line = xml.sax.saxutils.escape(line)
+
+            # process links
+            line = self._process_links(line)
+
+            # yield
+            yield line
 
         # close pre
         yield "</pre>"