Update rag_llamaindex.py
Browse files- rag_llamaindex.py +7 -7
rag_llamaindex.py
CHANGED
@@ -40,7 +40,7 @@ class LlamaIndexRAG(BaseRAG):
|
|
40 |
out_path = out_dir / "gpt-4.pdf"
|
41 |
|
42 |
if not out_path.exists():
|
43 |
-
r = requests.get(PDF_URL)
|
44 |
with open(out_path, "wb") as f:
|
45 |
f.write(r.content)
|
46 |
|
@@ -50,13 +50,13 @@ class LlamaIndexRAG(BaseRAG):
|
|
50 |
# Web
|
51 |
SimpleWebPageReader = download_loader("SimpleWebPageReader")
|
52 |
loader = SimpleWebPageReader()
|
53 |
-
docs.extend(loader.load_data(urls = [WEB_URL]))
|
54 |
#print("docs = " + str(len(docs)))
|
55 |
|
56 |
# YouTube
|
57 |
loader = YoutubeTranscriptReader()
|
58 |
-
docs.extend(loader.load_data(ytlinks = [YOUTUBE_URL_1,
|
59 |
-
YOUTUBE_URL_2]))
|
60 |
#print("docs = " + str(len(docs)))
|
61 |
|
62 |
return docs
|
@@ -72,10 +72,10 @@ class LlamaIndexRAG(BaseRAG):
|
|
72 |
|
73 |
def get_vector_store(self):
|
74 |
return MongoDBAtlasVectorSearch(
|
75 |
-
MongoClient(MONGODB_ATLAS_CLUSTER_URI),
|
76 |
db_name = MONGODB_DB_NAME,
|
77 |
-
collection_name = MONGODB_COLLECTION_NAME,
|
78 |
-
index_name = MONGODB_INDEX_NAME
|
79 |
)
|
80 |
|
81 |
def ingestion(self, config):
|
|
|
40 |
out_path = out_dir / "gpt-4.pdf"
|
41 |
|
42 |
if not out_path.exists():
|
43 |
+
r = requests.get(self.PDF_URL)
|
44 |
with open(out_path, "wb") as f:
|
45 |
f.write(r.content)
|
46 |
|
|
|
50 |
# Web
|
51 |
SimpleWebPageReader = download_loader("SimpleWebPageReader")
|
52 |
loader = SimpleWebPageReader()
|
53 |
+
docs.extend(loader.load_data(urls = [self.WEB_URL]))
|
54 |
#print("docs = " + str(len(docs)))
|
55 |
|
56 |
# YouTube
|
57 |
loader = YoutubeTranscriptReader()
|
58 |
+
docs.extend(loader.load_data(ytlinks = [self.YOUTUBE_URL_1,
|
59 |
+
self.YOUTUBE_URL_2]))
|
60 |
#print("docs = " + str(len(docs)))
|
61 |
|
62 |
return docs
|
|
|
72 |
|
73 |
def get_vector_store(self):
|
74 |
return MongoDBAtlasVectorSearch(
|
75 |
+
MongoClient(self.MONGODB_ATLAS_CLUSTER_URI),
|
76 |
db_name = MONGODB_DB_NAME,
|
77 |
+
collection_name = self.MONGODB_COLLECTION_NAME,
|
78 |
+
index_name = self.MONGODB_INDEX_NAME
|
79 |
)
|
80 |
|
81 |
def ingestion(self, config):
|