Changeset 503 for trunk

Timestamp:

01/07/11 15:44:00 (13 years ago)

Author:

bas

Message:

changed some code an made an function html_2_text that uses an external program to convert the html to text so we can use it for further processing.

Location:

trunk

Files:

• ChangeLog (modified) (4 diffs)
• email2trac.py.in (modified) (2 diffs)

trunk/ChangeLog

@@ -10 +10 @@
    * Support for parentdir. Can be used to avoid sections for each project
      or if all projects have a common parent directory, you can leave out
-     the project drectory setting for each project. 
+     the project drectory setting for each project, eg:
+      * parentdir: /data/trac/oss/projects
 
      If an section defines a project directory then this can NOT be overriden 
@@ -44 +45 @@
    * Fixed some bugs in function update_ticket_fields:
        - Can not clear values, repoterd by thomas dot moschny add gmx dot de
-             closes #228
+     closes #228
        - Values were updated regardless if there are not allowed for the ticket
-             field.
+     field.
      Author: Bas van der Vlies
 
@@ -60 +61 @@
    * Added html conversion via external commmand: eg email2trac.conf
       - html2text_cmd: /usr/bin/html2text -nobs
+      - html2text_cmd: /usr/bin/w3m -dump
+      - html2text_cmd: /usr/bin/lynx -dump
+      - html2text_cmd: /usr/bin/links -dump
      closes #218
      Author: Bas van der Vlies
@@ -75 +79 @@
      also honour the permission model if set, closes #198
      an example how to use it:
-        email2trac.conf:
+    email2trac.conf:
                 [project]
-                workflow_closed: reopen
-                workflow_infoneeded: provideinfo
-
-        trac.ini:
-                [ticket-workflow]
-                reopen = closed -> reopened 
-                reopen.permissions = TICKET_CREATE 
-                reopen.operations = del_resolution 
-                
-                provideinfo = infoneeded -> moreinfo
-                provideinfo.permissions = TICKET_CREATE
-                provideinfo.name = info provided
+            workflow_closed: reopen
+            xworkflow_infoneeded: provideinfo
+
+    trac.ini:
+    [ticket-workflow]
+    reopen = closed -> reopened 
+    reopen.permissions = TICKET_CREATE 
+    reopen.operations = del_resolution 
+
+    provideinfo = infoneeded -> moreinfo
+    provideinfo.permissions = TICKET_CREATE
+    provideinfo.name = info provided
 
      Authors: kroseneg add schmidham dot net and Bas van der Vlies

trunk/email2trac.py.in

@@ -585 +585 @@
                         self.logger.debug(util.text.unicode_quote(s))
 
+
+        def html_2_txt(self, data):
+                """
+                Various routines to convert html syntax to valid trac wiki syntax
+                """ 
+                self.logger.debug('function html_2_txt')
+
+                ## This routine make an safe html that can be include
+                #  in trac, but no further text processing can be done
+                #
+#               try:
+#                       from lxml.html.clean import Cleaner
+#                       tags_rm = list()
+#                       tags_rm.append('body')
+#
+#                       cleaner = Cleaner(remove_tags=tags_rm )
+#                       parsed_data = cleaner.clean_html(data)
+#                       parsed_data = '\n{{{\n#!html\n' + parsed_data + '\n}}}\n'
+#
+#                       return parsed_data
+#                       
+#               except ImportError::
+#                       pass
+
+                parsed_data = None
+                if self.parameters.html2text_cmd: 
+                        tmp_file = tempfile.mktemp('email2trac.html')
+                        cmd = '%s %s' %(self.parameters.html2text_cmd, tmp_file) 
+                        self.logger.debug('\t html2text conversion %s'%(cmd)) 
+        
+                        if self.parameters.dry_run:
+                                print 'DRY_RUN: html2text conversion command: %s\n' %(cmd)
+
+                        else: 
+                                f = open(tmp_file, "w+")
+                                f.write(data)
+                                f.close()
+
+                                lines = os.popen(cmd).readlines()
+                                parsed_data =  ''.join(lines)
+
+                                os.unlink(tmp_file)
+
+                else:
+                        self.logger.debug('\t No html2text conversion tool specified in email2trac.conf') 
+
+                return parsed_data
+
 ########## TRAC ticket functions  ###########################################################
 
@@ -1833 +1881 @@
                         if content_type == 'text/html':
 
-                                if self.parameters.html2text_cmd:
-
-                                        tmp_file = tempfile.mktemp('.email2trac')
-                                        cmd = '%s %s' %(self.parameters.html2text_cmd, tmp_file) 
-                                        self.logger.debug('\t html2text conversion %s'%(cmd)) 
-                                        
-                                        if self.parameters.dry_run:
-                                                print 'DRY_RUN: html2text conversion command: %s\n' %(cmd)
-                                        else:
-                                                f = open(tmp_file, "w+")
-                                                f.write(body_text)
-                                                f.close()
-
-                                                lines = os.popen(cmd).readlines()
-                                                body_text = ''.join(lines)
-
-                                                os.unlink(tmp_file)
-
-                                else:
-                                        self.logger.debug('\t No html2text conversion tool specified in email2trac.conf') 
+                                body_text = self.html_2_txt(body_text)
+                                if not body_text:
                                         continue