Remove html from searched profile descriptions

content.py

@@ -400,6 +400,24 @@ def removeTextFormatting(content: str) -> str:
     return content
 
 
+def removeHtml(content: str) -> str:
+    """Removes html links from the given content.
+    Used to ensure that profile descriptions don't contain dubious content
+    """
+    if '<' not in content:
+        return content
+    removing = False
+    result = ''
+    for ch in content:
+        if ch == '<':
+            removing = True
+        elif ch == '>':
+            removing = False
+        elif not removing:
+            result += ch
+    return result
+
+
 def removeLongWords(content: str, maxWordLength: int,
                     longWordsList: []) -> str:
     """Breaks up long words so that on mobile screens this doesn't

tests.py

@@ -64,6 +64,7 @@ from media import getAttachmentMediaType
 from delete import sendDeleteViaServer
 from inbox import validInbox
 from inbox import validInboxFilenames
+from content import removeHtml
 from content import addWebLinks
 from content import replaceEmojiFromTags
 from content import addHtmlTags
@@ -1873,8 +1874,18 @@ def testSiteIsActive():
     assert(not siteIsActive('https://notarealwebsite.a.b.c'))
 
 
+def testRemoveHtml():
+    print('testRemoveHtml')
+    testStr = 'This string has no html.'
+    assert(removeHtml(testStr) == testStr)
+    testStr = 'This string <a href="1234.567">has html</a>.'
+    print(removeHtml(testStr))
+    assert(removeHtml(testStr) == 'This string has html.')
+
+
 def runAllTests():
     print('Running tests...')
+    testRemoveHtml()
     testSiteIsActive()
     testJsonld()
     testRemoveTextFormatting()

webinterface.py

@@ -63,6 +63,7 @@ from content import getMentionsFromHtml
 from content import addHtmlTags
 from content import replaceEmojiFromTags
 from content import removeLongWords
+from content import removeHtml
 from config import getConfigParam
 from skills import getSkills
 from cache import getPersonFromCache
@@ -6215,6 +6216,8 @@ def htmlProfileAfterSearch(recentPostsCache: {}, maxRecentPosts: int,
                     profileJson['summary'].replace('<br>', '\n')
                 avatarDescription = avatarDescription.replace('<p>', '')
                 avatarDescription = avatarDescription.replace('</p>', '')
+                if '<' in avatarDescription:
+                    avatarDescription = removeHtml(avatarDescription)
         profileStr = ' <div class="hero-image">'
         profileStr += '  <div class="hero-text">'
         if avatarUrl: