Another long lines failure case

content.py

@@ -643,6 +643,8 @@ def removeLongWords(content: str, maxWordLength: int,
                 if wordStr not in longWordsList:
                     longWordsList.append(wordStr)
     for wordStr in longWordsList:
+        if wordStr.startswith('<p>'):
+            wordStr = wordStr.replace('<p>', '')
         if wordStr.startswith('<'):
             continue
         if len(wordStr) == 76:
@@ -678,6 +680,8 @@ def removeLongWords(content: str, maxWordLength: int,
             continue
         if '<' in wordStr:
             replaceWord = wordStr.split('<', 1)[0]
+            # if len(replaceWord) > maxWordLength:
+            #     replaceWord = replaceWord[:maxWordLength]
             content = content.replace(wordStr, replaceWord)
             wordStr = replaceWord
         if '/' in wordStr:

tests.py

@@ -1911,6 +1911,22 @@ def testActorParsing():
 def testWebLinks():
     print('testWebLinks')
 
+    exampleText = \
+        "<p>Aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa" + \
+        "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa" + \
+        "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa" + \
+        "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa" + \
+        " <a href=\"https://domain.ugh/tags/turbot\" class=\"mention " + \
+        "hashtag\" rel=\"tag\">#<span>turbot</span></a> <a href=\"" + \
+        "https://domain.ugh/tags/haddock\" class=\"mention hashtag\"" + \
+        " rel=\"tag\">#<span>haddock</span></a></p>"
+    resultText = removeLongWords(exampleText, 40, [])
+    assert resultText == "<p>Aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa" + \
+        " <a href=\"https://domain.ugh/tags/turbot\" class=\"mention " + \
+        "hashtag\" rel=\"tag\">#<span>turbot</span></a> " + \
+        "<a href=\"https://domain.ugh/tags/haddock\" " + \
+        "class=\"mention hashtag\" rel=\"tag\">#<span>haddock</span></a></p>"
+
     exampleText = \
         '<p><span class=\"h-card\"><a href=\"https://something/@orother' + \
         '\" class=\"u-url mention\">@<span>foo</span></a></span> Some ' + \