From 01cc700706212b4a833cfc08a5ccaf2e4a09fc44 Mon Sep 17 00:00:00 2001
From: artlef <artlef@localhost>
Date: Sat, 2 Jun 2018 18:23:05 +0200
Subject: [PATCH] Improve parsing

---
 ParseHtml.hs | 22 +++++++++++++++-------
 1 file changed, 15 insertions(+), 7 deletions(-)

diff --git a/ParseHtml.hs b/ParseHtml.hs
index 8ccedee..6d21030 100644
--- a/ParseHtml.hs
+++ b/ParseHtml.hs
@@ -1,4 +1,4 @@
-module ParseHtml (printDef, printDifferentDef) where
+module ParseHtml (printDef, printDifferentDef, filterSupTags, findRelevantTags) where
 
 import Text.HTML.TagSoup
 import Prelude
@@ -54,10 +54,18 @@ parseFullDefTags :: T.Text -> [Tag String]
 parseFullDefTags = findRelevantTags . parseTags . T.unpack
 
 findRelevantTags :: [Tag String] -> [Tag String]
-findRelevantTags [] = []
-findRelevantTags (x:xs) = if x ~== "<span class=tlf_cdefinition"
-    || x ~== "<span class=tlf_cexemple"
-    then maybe (f xs) (\y -> y:(f xs)) (L.find (~== TagText "") xs)
-    else (f xs)
+findRelevantTags tags =
+    filter (~== TagText "") filteredTags
+ where
+    filteredTags =
+        takeWhile (~/= "<div id=footer>") $
+        dropWhile (~/= "<span class=tlf_cdefinition>") $
+        filterSupTags tags
+
+filterSupTags :: [Tag String] -> [Tag String]
+filterSupTags tags = filter (/= TagText "willBeDeleted") $ snd $ L.mapAccumL transformSupText (TagText "") tags
   where
-    f = findRelevantTags
+    transformSupText x y
+        | y ~== "<sup>" = (TagText "willBeDeleted", TagText "willBeDeleted")
+        | x == TagText "willBeDeleted" = (TagText "", TagText "willBeDeleted")
+        | otherwise = (x,y)