From 09d63088f06da6173b7a469c333d807b10e25e5a Mon Sep 17 00:00:00 2001
From: Bob Mottram <bob@freedombone.net>
Date: Sun, 31 May 2020 17:31:33 +0100
Subject: [PATCH] Remove old hashtags when new ones are created

---
 inbox.py        |  2 ++
 webinterface.py | 46 ++++++++++++++++++++++++++++++++--------------
 2 files changed, 34 insertions(+), 14 deletions(-)

diff --git a/inbox.py b/inbox.py
index fa269206..51130f97 100644
--- a/inbox.py
+++ b/inbox.py
@@ -56,6 +56,7 @@ from posts import sendSignedJson
 from posts import sendToFollowersThread
 from webinterface import individualPostAsHtml
 from webinterface import getIconsDir
+from webinterface import removeOldHashtags
 from question import questionUpdateVotes
 from media import replaceYouTube
 from git import isGitPatch
@@ -108,6 +109,7 @@ def storeHashTags(baseDir: str, nickname: str, postJsonObject: {}) -> None:
                 except Exception as e:
                     print('WARN: Failed to write entry to tags file ' +
                           tagsFilename + ' ' + str(e))
+                removeOldHashtags(baseDir, 3)
 
 
 def inboxStorePostToHtmlCache(recentPostsCache: {}, maxRecentPosts: int,
diff --git a/webinterface.py b/webinterface.py
index c3de636b..4e66d5ac 100644
--- a/webinterface.py
+++ b/webinterface.py
@@ -5725,15 +5725,15 @@ def htmlCalendar(translate: {},
     return calendarStr
 
 
-def htmlHashTagSwarm(baseDir: str, actor: str) -> str:
-    """Returns a tag swarm of today's hashtags
+def removeOldHashtags(baseDir: str, maxMonths: int) -> str:
+    """Remove old hashtags
     """
-    currTime = datetime.utcnow()
-    daysSinceEpoch = (currTime - datetime(1970, 1, 1)).days
-    maxDaysSinceEpoch = (currTime - datetime(1970, 3, 1)).days
-    daysSinceEpochStr = str(daysSinceEpoch) + ' '
-    tagSwarm = []
+    if maxMonths > 11:
+        maxMonths = 11
+    maxDaysSinceEpoch = \
+        (datetime.utcnow() - datetime(1970, 1 + maxMonths, 1)).days
     removeHashtags = []
+
     for subdir, dirs, files in os.walk(baseDir + '/tags'):
         for f in files:
             tagsFilename = os.path.join(baseDir + '/tags', f)
@@ -5743,10 +5743,35 @@ def htmlHashTagSwarm(baseDir: str, actor: str) -> str:
             modTimesinceEpoc = os.path.getmtime(tagsFilename)
             lastModifiedDate = datetime.fromtimestamp(modTimesinceEpoc)
             fileDaysSinceEpoch = (lastModifiedDate - datetime(1970, 1, 1)).days
+
             # check of the file is too old
             if fileDaysSinceEpoch < maxDaysSinceEpoch:
                 removeHashtags.append(tagsFilename)
+
+    for removeFilename in removeHashtags:
+        try:
+            os.remove(removeFilename)
+        except BaseException:
+            pass
+
+
+def htmlHashTagSwarm(baseDir: str, actor: str) -> str:
+    """Returns a tag swarm of today's hashtags
+    """
+    currTime = datetime.utcnow()
+    daysSinceEpoch = (currTime - datetime(1970, 1, 1)).days
+    daysSinceEpochStr = str(daysSinceEpoch) + ' '
+    tagSwarm = []
+
+    for subdir, dirs, files in os.walk(baseDir + '/tags'):
+        for f in files:
+            tagsFilename = os.path.join(baseDir + '/tags', f)
+            if not os.path.isfile(tagsFilename):
                 continue
+            # get last modified datetime
+            modTimesinceEpoc = os.path.getmtime(tagsFilename)
+            lastModifiedDate = datetime.fromtimestamp(modTimesinceEpoc)
+            fileDaysSinceEpoch = (lastModifiedDate - datetime(1970, 1, 1)).days
             # check if the file was last modified today
             if fileDaysSinceEpoch != daysSinceEpoch:
                 continue
@@ -5791,13 +5816,6 @@ def htmlHashTagSwarm(baseDir: str, actor: str) -> str:
                     if lineCtr >= maxLineCtr:
                         break
 
-    # remove old hashtags
-    for removeFilename in removeHashtags:
-        try:
-            os.remove(removeFilename)
-        except BaseException:
-            pass
-
     if not tagSwarm:
         return ''
     tagSwarm.sort()