Untracked hashtags report

app/config.py

@@ -21,3 +21,5 @@ class Config(object):
     MAX_POSTS_TO_FETCH = 200
     POINTS_BOOSTED = 1
     POINTS_FAVOURITED = 1
+    TOP_HASHTAGS_TO_REPORT = 3
+    TOP_POSTS_TO_REPORT = 3

app/urma.py

@@ -1,5 +1,6 @@
 #! /usr/bin/env python
 
+import argparse
 import datetime
 import ipdb
 import os
@@ -7,6 +8,7 @@ import pickle
 import random
 import requests
 import stackprinter
+import subprocess
 import sys
 
 from config import Config
@@ -28,6 +30,10 @@ from models import (
     Posts,
     PostTags,
 )
+from sqlalchemy import (
+    func,
+    select,
+)
 
 from typing import List, Optional, Union
 
@@ -101,7 +107,7 @@ class MastodonAPI:
         _ = self.mastodon.status_unbookmark(post_id)
 
 
-def main() -> None:
+def update_database() -> None:
     """
     Main loop
     """
@@ -142,6 +148,31 @@ def get_and_process_favourited(session, mastapi):
         favourited = mastapi.mastodon.fetch_next(favourited)
 
 
+def get_database_name():
+    """Return database name as string"""
+
+    with Session() as session:
+        dbname = session.bind.engine.url.database
+
+    return dbname
+
+
+def get_version_string():
+    """Return Urma version as string"""
+
+    try:
+        return str(
+            subprocess.check_output(
+                ['git', 'describe'], stderr=subprocess.STDOUT
+            )
+        ).strip('\'b\\n')
+    except subprocess.CalledProcessError as exc_info:
+        gitproc = subprocess.Popen(['git', 'rev-parse', 'HEAD'],
+                                   stdout=subprocess.PIPE)
+        (stdout, _) = gitproc.communicate()
+        return stdout.strip()[:7].decode("utf-8")
+
+
 def process_bookmarked_posts(session: Session,
                              posts: List[Posts], me_id: int) -> bool:
     """
@@ -241,6 +272,46 @@ def _process_post(session: Session, post: Posts, me_id) -> Posts:
     return rec
 
 
+def report():
+    """Print report"""
+
+    print(f"Urma version: {get_version_string()}")
+    print(f"Database: {get_database_name()}")
+    print(f"Date: {datetime.datetime.now().strftime('%c')}")
+    print()
+
+    with Session() as session:
+        # Find the most popular hashtags that we don't follow
+        print("Hashtags you don't follow that feature in posts you like")
+        print("--------------------------------------------------------")
+        top_unfollowed_tags = (
+            session.execute(
+                select(Hashtags, func.count(Hashtags.name))
+                .join(PostTags).join(Posts)
+                .where(Posts.favourited == 1, Hashtags.followed == 0)
+                .group_by(Hashtags.name)
+                .order_by(func.count(Hashtags.name).desc())
+                .limit(Config.TOP_HASHTAGS_TO_REPORT))
+            .all()
+        )
+
+        # How many times was each hashtag in a post we didnt' like?
+        for (hashtag, like) in top_unfollowed_tags:
+            dislike = (
+                session.execute(
+                    select(func.count(Posts.id))
+                    .join(PostTags).join(Hashtags)
+                    .where(Posts.favourited == 0, Hashtags.id == hashtag.id)
+                ).scalars()
+                .all()[0]
+            )
+
+            print(
+                f"Hashtag {hashtag.name} {like=}, {dislike=} "
+                f"({like * 100 / (like + dislike):.2f}% liked)"
+            )
+
+
 def update_followed_accounts(session: Session, mastapi: MastodonAPI) -> None:
     """
     Retrieve list of followed accounts and update accounts
@@ -313,10 +384,31 @@ if __name__ == "__main__":
     If command line arguments given, carry out requested function and
     exit. Otherwise run full application.
     """
-
     try:
         Base.metadata.create_all(engine)
-        sys.exit(main())
+
+        p = argparse.ArgumentParser()
+        # Only allow at most one option to be specified
+        group = p.add_mutually_exclusive_group()
+        group.add_argument('-u', '--update',
+                           action="store_true", dest="update_database",
+                           default=False, help="Update database from Mastodon")
+        group.add_argument('-r', '--report',
+                           action="store_true", dest="report",
+                           default=False, help="Report")
+        args = p.parse_args()
+
+        # Run as required
+        if args.update_database:
+            log.debug("Updating database")
+            update_database()
+        elif args.report:
+            log.debug("Report")
+            report()
+        else:
+            # For now, default to updating database
+            update_database()
+
     except Exception as exc:
 
         if os.environ["URMA_ENV"] != "DEVELOPMENT":