Generate cleaner plaintext

diff --git a/karmaworld/apps/notes/management/commands/strip_html_to_text.py b/karmaworld/apps/notes/management/commands/strip_html_to_text.py
index bf32b3198a0a78f7a82ddb2eddfd7e4c4b0ed5c0..8d42981678d80b1ed40ea627dec578109c1b64fb 100644 (file)
--- a/karmaworld/apps/notes/management/commands/strip_html_to_text.py
+++ b/karmaworld/apps/notes/management/commands/strip_html_to_text.py
@@ -4,7 +4,7 @@
 
 import html2text
 from django.core.management.base import BaseCommand
-from apps.notes.models import Note
+from karmaworld.apps.notes.models import Note
 
 class Command(BaseCommand):
     """ Command to process notes with html, and without text
@@ -13,17 +13,21 @@ class Command(BaseCommand):
     help = "Take all notes with the .html property and use that to fill Note.text by stripping html"
 
     def handle(self, *args, **kwargs):
-        """ On all calls, clean all notes with html and not text using lxml """
+        """ On all calls, clean all notes with html and not text using html2text """
         notes = Note.objects.filter(html__isnull=False).filter(text__isnull=True)
         cleaned_notes = 0
         for note in notes:
             try:
                 h = html2text.HTML2Text()
+                h.escape_snob = True
+                h.unicode_snob = True
                 h.ignore_links = True
                 h.ignore_images = True
+                h.ignore_emphasis = True
                 note.text = h.handle(note.html)
                 note.save()
                 cleaned_notes += 1
+                print 'Processed {n}'.format(n=note)
             except Exception, e:
                 print note
                 print e