From: Michael Orlitzky <michael@orlitzky.com>
Date: Sat, 14 Jul 2012 02:43:46 +0000 (-0400)
Subject: Remove full story paragraphs if we aren't going to use them.
X-Git-Tag: v0.0.1~12
X-Git-Url: https://gitweb.michael.orlitzky.com/?a=commitdiff_plain;h=ce53d0d1ff76cc45f1f66504976a0549999677c0;p=dead%2Flwn-epub.git

Remove full story paragraphs if we aren't going to use them.
---

diff --git a/src/LWN/Page.hs b/src/LWN/Page.hs
index 3dccf46..ccc2590 100644
--- a/src/LWN/Page.hs
+++ b/src/LWN/Page.hs
@@ -58,6 +58,7 @@ import LWN.XHTML (
   is_image,
   preprocess,
   remove_byline,
+  remove_full_story_paragraphs,
   remove_title,
   to_xhtml,
   to_xml,
@@ -196,7 +197,8 @@ parse cfg xml = do
              story_map <- download_full_stories cfg xml
              return $ xml >>> insert_full_stories story_map
            else do
-             return xml
+             -- Get rid of them if we don't want them.
+             return $ xml >>> remove_full_story_paragraphs
 
   let clean_xml = fs_xml >>> preprocess
   image_map <- download_images clean_xml
diff --git a/src/LWN/XHTML.hs b/src/LWN/XHTML.hs
index 5d3bbd1..c54bada 100644
--- a/src/LWN/XHTML.hs
+++ b/src/LWN/XHTML.hs
@@ -9,6 +9,7 @@ module LWN.XHTML (
   parse_lwn,
   preprocess,
   remove_byline,
+  remove_full_story_paragraphs,
   remove_title,
   to_xhtml,
   to_xml,
@@ -90,6 +91,11 @@ is_link =
   isElem >>> hasName "a"
 
 
+remove_full_story_paragraphs  :: (ArrowXml a) => a XmlTree XmlTree
+remove_full_story_paragraphs =
+  processTopDown $ none `when` full_story_paragraph
+
+
 remove_comment_links  :: (ArrowXml a) => a XmlTree XmlTree
 remove_comment_links =
   processTopDown $ kill_comments `when` is_link