module LWN.XHTML (
XHTML,
XML,
+ full_story_urls,
image_srcs,
+ full_story_link,
+ full_story_paragraph,
is_image,
parse_lwn,
preprocess,
remove_byline,
+ remove_full_story_paragraphs,
remove_title,
to_xhtml,
to_xml,
import Text.HandsomeSoup (css)
import Text.XML.HXT.Core (
(>>>),
+ (/>),
ArrowXml,
IOStateArrow,
SysConfigList,
XmlTree,
changeAttrValue,
+ deep,
getAttrValue,
hasAttrValue,
hasName,
+ hasText,
+ ifA,
isElem,
mkName,
no,
processTopDown,
readString,
setElemName,
+ this,
when,
withParseHTML,
withValidate,
withWarnings,
yes)
+
import LWN.URI (URL, try_make_absolute_url)
import Misc (contains)
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
>>>
getAttrValue "src"
+
+full_story_paragraph :: (ArrowXml a) => a XmlTree XmlTree
+full_story_paragraph =
+ isElem
+ >>>
+ hasName "p"
+ >>>
+ ifA
+ (this /> full_story_link)
+ this
+ none
+
+
+-- Without regard to the parent paragraph.
+full_story_link :: (ArrowXml a) => a XmlTree XmlTree
+full_story_link =
+ isElem
+ >>>
+ hasName "a"
+ >>>
+ ifA
+ (this /> hasText (== "Full Story"))
+ this
+ none
+
+
+-- | Get the hrefs of all full story links.
+full_story_urls :: (ArrowXml a) => a XmlTree URL
+full_story_urls =
+ deep $
+ full_story_paragraph
+ />
+ full_story_link
+ >>>
+ getAttrValue "href"
+
+
+
make_image_srcs_absolute :: (ArrowXml a) => a XmlTree XmlTree
make_image_srcs_absolute =
processTopDown (make_srcs_absolute `when` is_image)