From 67e06f65c9ef03950b327e7229a12a9b992f4537 Mon Sep 17 00:00:00 2001
From: Bob Mottram <bob@freedombone.net>
Date: Sat, 5 Dec 2020 11:11:32 +0000
Subject: [PATCH] Guess hashtag categorisations

---
 inbox.py | 37 +++++++++++++++++++++++++++++++++++++
 tests.py | 15 +++++++++++++++
 2 files changed, 52 insertions(+)

diff --git a/inbox.py b/inbox.py
index cf681eb4e..d0c2a21dc 100644
--- a/inbox.py
+++ b/inbox.py
@@ -30,6 +30,8 @@ from utils import loadJson
 from utils import saveJson
 from utils import updateLikesCollection
 from utils import undoLikesCollectionEntry
+from utils import getHashtagCategories
+from utils import setHashtagCategory
 from httpsig import verifyPostHeaders
 from session import createSession
 from session import getJson
@@ -68,6 +70,31 @@ from happening import saveEventPost
 from delete import removeOldHashtags
 
 
+def guessHashtagCategory(tagName: str, hashtagCategories: {}) -> str:
+    """Tries to guess a category for the given hashtag.
+    This works by trying to find the longest similar hashtag
+    """
+    categoryMatched = ''
+    tagMatched = ''
+    tagMatchedLen = 0
+
+    for categoryStr, hashtagList in hashtagCategories.items():
+        for hashtag in hashtagList:
+            if hashtag in tagName:
+                if not tagMatched:
+                    tagMatched = hashtag
+                    tagMatchedLen = len(tagMatched)
+                    categoryMatched = categoryStr
+                else:
+                    # match the longest tag
+                    if len(hashtag) > tagMatchedLen:
+                        tagMatched = hashtag
+                        categoryMatched = categoryStr
+    if not categoryMatched:
+        return
+    return categoryMatched
+
+
 def storeHashTags(baseDir: str, nickname: str, postJsonObject: {}) -> None:
     """Extracts hashtags from an incoming post and updates the
     relevant tags files.
@@ -91,6 +118,8 @@ def storeHashTags(baseDir: str, nickname: str, postJsonObject: {}) -> None:
         print('Creating tags directory')
         os.mkdir(tagsDir)
 
+    hashtagCategories = getHashtagCategories(baseDir)
+
     for tag in postJsonObject['object']['tag']:
         if not tag.get('type'):
             continue
@@ -122,6 +151,14 @@ def storeHashTags(baseDir: str, nickname: str, postJsonObject: {}) -> None:
                           tagsFilename + ' ' + str(e))
                 removeOldHashtags(baseDir, 3)
 
+        # automatically assign a category to the tag if possible
+        categoryFilename = tagsDir + '/' + tagName + '.category'
+        if not os.path.isfile(categoryFilename):
+            categoryStr = \
+                guessHashtagCategory(tagName, hashtagCategories)
+            if categoryStr:
+                setHashtagCategory(baseDir, tagName, categoryStr)
+
 
 def inboxStorePostToHtmlCache(recentPostsCache: {}, maxRecentPosts: int,
                               translate: {},
diff --git a/tests.py b/tests.py
index aae2345b3..be85000b3 100644
--- a/tests.py
+++ b/tests.py
@@ -71,6 +71,7 @@ from delete import sendDeleteViaServer
 from inbox import jsonPostAllowsComments
 from inbox import validInbox
 from inbox import validInboxFilenames
+from inbox import guessHashtagCategory
 from content import htmlReplaceEmailQuote
 from content import htmlReplaceQuoteMarks
 from content import dangerousMarkup
@@ -2421,8 +2422,22 @@ def testValidNickname():
     assert not validNickname(domain, nickname)
 
 
+def testGuessHashtagCategory() -> None:
+    print('testGuessHashtagCategory')
+    hashtagCategories = {
+        "foo": ["swan", "goose"],
+        "bar": ["cat", "mouse"]
+    }
+    guess = guessHashtagCategory("unspecifiedgoose", hashtagCategories)
+    assert guess == "foo"
+
+    guess = guessHashtagCategory("catpic", hashtagCategories)
+    assert guess == "bar"
+
+
 def runAllTests():
     print('Running tests...')
+    testGuessHashtagCategory()
     testValidNickname()
     testParseFeedDate()
     testFirstParagraphFromString()