From 7c6dad605fec72bdc8051fb7afabd0a3edeabc5c Mon Sep 17 00:00:00 2001 From: Anthony Wang Date: Sun, 28 Jul 2024 02:58:02 +0000 Subject: Don't index text since it interferes with other results --- server.py | 22 ++++++++++------------ 1 file changed, 10 insertions(+), 12 deletions(-) diff --git a/server.py b/server.py index d1c41d5..43d3316 100644 --- a/server.py +++ b/server.py @@ -98,18 +98,16 @@ def index(path, parent): # Modified or not in emb emb = None type = mimetypes.guess_type(path)[0] - try: - if type is None and os.path.getsize(path) < 2**16: - with open(path) as f: - emb = model.embed_text(f.read()) - elif type.startswith("audio"): - emb = model.embed_audio(path) - elif type.startswith("image"): - emb = model.embed_image(path) - elif type.startswith("video") and os.path.getsize(path) < 2**25: - emb = model.embed_video(path) - except: - print(traceback.format_exc()) + if isinstance(type, str): + try: + if type.startswith("audio"): + emb = model.embed_audio(path) + elif type.startswith("image"): + emb = model.embed_image(path) + elif type.startswith("video") and os.path.getsize(path) < 2**25: + emb = model.embed_video(path) + except: + print(traceback.format_exc()) if emb is None: # Might be in index but no longer valid -- cgit v1.2.3-70-g09d2