diff options
Diffstat (limited to 'scripts')
| -rw-r--r-- | scripts/fill.image_urls.py | 11 |
1 files changed, 8 insertions, 3 deletions
diff --git a/scripts/fill.image_urls.py b/scripts/fill.image_urls.py index 0b1c2e3..952b7ea 100644 --- a/scripts/fill.image_urls.py +++ b/scripts/fill.image_urls.py @@ -5,20 +5,25 @@ import postgresql from urllib.parse import urlparse db = postgresql.open("pq://postgres:root@localhost/dumpfm") +db.execute("SET CLIENT_ENCODING to 'UNICODE'") def get_num_images_in_db(): ps = db.prepare("SELECT COUNT(*) FROM image_urls") return int(ps()[0][0]) def add_url(url): - print(url) + try: + print(url) + except UnicodeEncodeError: + print("i thought python 3 fixed the unicode shit. yet i still get unicode errors everywhere. GOOD JOB FUCKHEADS") ps = db.prepare("DELETE FROM image_urls WHERE url=$1") ps(url) ps2 = db.prepare("INSERT INTO image_urls(url) VALUES($1)") ps2(url) +# NOTE. hardcoded room numbers to index here... only indexing DUMPFM (1) and GIF (8) currently. def get_messages(lower, upper): - ps = db.prepare("SELECT content FROM messages WHERE message_id >= $1 AND message_id <= $2 ORDER BY message_id ASC") + ps = db.prepare("SELECT content FROM messages WHERE message_id >= $1 AND message_id <= $2 AND room_id IN (1,8) ORDER BY message_id ASC") rows = ps(lower, upper) return rows @@ -60,4 +65,4 @@ if __name__ == "__main__": print("added ", len(urls), " images to db") num_new_images = get_num_images_in_db() - num_existing_images percent_new_images = num_new_images / len(urls) * 100 - print(num_new_images, " were new (", percent_new_images ,"%)")
\ No newline at end of file + print(num_new_images, " were new (", percent_new_images ,"%)") |