new feature: ignore images in html2text

2013-03-20 23:28:48 +01:00
parent 7cadfb8bd1
commit f53338c3b1
1 changed files with 3 additions and 0 deletions
--- a/atomstrom.py
+++ b/atomstrom.py
@@ -31,6 +31,7 @@ class Feed(Base):
    readability = Column(Boolean)
    fullpage = Column(Boolean)
    html2textsummary = Column(Boolean)
+    html2textignoreimages = Column(Boolean)
    enabled = Column(Boolean)

    def __init__(self, url, daily, readability, fullpage, enabled, html2textsummary):
@@ -270,6 +271,8 @@ def process_feed_entry(session, feed, entry):
            print '    converting summary'
            h2t = html2text.HTML2Text()
            h2t.body_width = 0
+            if feed.html2textignoreimages:
+                h2t.ignore_images = True
            summary = thisentry.summary.decode('latin-1')
            summary = h2t.handle(summary)
            thisentry.summary = summary.encode('latin-1', 'replace')