-
Notifications
You must be signed in to change notification settings - Fork 0
/
index_arxiv.py
166 lines (141 loc) · 5.28 KB
/
index_arxiv.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
import pandas as pd
import weaviate
import logging
import os
from dotenv import load_dotenv
ARXIV_JSON = "data/arXiv.cs.CL.embedv3.jsonl" # "data/arxiv.cs.CL.json"
SCHEMA_NAME = "ArxivDocument_CS_CL"
def index_data(cohere_api_key: str, weaviate_url: str, weaviate_api_key: str):
"""Index Data into Weaviate"""
logging.info(f"Loading data from '{ARXIV_JSON}'")
df = pd.read_json(ARXIV_JSON, lines=True)
logging.info(f"Initializing Weaviate Client: '{weaviate_url}'")
client = weaviate.Client(
url=weaviate_url,
auth_client_secret=weaviate.AuthApiKey(api_key=weaviate_api_key),
additional_headers={"X-Cohere-Api-Key": cohere_api_key})
logging.info(
f"Deleting '{SCHEMA_NAME}' schema in Weaviate: '{weaviate_url}'")
client.schema.delete_class(SCHEMA_NAME)
"""
Weaviate generates vector embeddings at the object level (rather than for individual properties).
text2vec-* modules generate vectors from text objects.
It vectorizes only properties that use the text data type (unless skipped)
See: https://weaviate.io/developers/weaviate/config-refs/schema#vectorizer
"""
logging.info(
f"Creating '{SCHEMA_NAME}' schema in Weaviate: '{weaviate_url}'")
class_obj = {
"class": SCHEMA_NAME,
"description": "This class contains Arxiv Documents in the CS.CL category",
"vectorIndexType": "hnsw",
"vectorizer": "text2vec-cohere",
"vectorIndexConfig": {
"distance": "cosine" # Set to "cosine" for English models; "dot" for multilingual models
},
"moduleConfig": {
"text2vec-cohere": {
"model": "embed-english-v3.0",
"truncate": "RIGHT",
"vectorizeClassName": False
}
},
"properties": [
{
"name": "url",
"dataType": ["text"],
"indexFilterable": False,
"indexSearchable": False,
"vectorizePropertyName": False
},
{
"name": "url_pdf",
"dataType": ["text"],
"indexFilterable": False,
"indexSearchable": False,
"vectorizePropertyName": False
},
{
"name": "title",
"dataType": ["text"]
},
{
"name": "authors",
"dataType": ["text"]
},
{
"name": "categories",
"dataType": ["text"]
},
{
"name": "abstract",
"dataType": ["text"]
},
{
"name": "update_date",
"dataType": ["date"],
},
{
"name": "publication_date",
"dataType": ["date"],
},
]
}
client.schema.create_class(class_obj)
logging.info(f"Importing data to Weaviate: '{weaviate_url}'")
try:
with client.batch as batch:
batch.batch_size = 100
for item in df.itertuples():
properties = {
"url": item.id,
"url_pdf": item.link_pdf,
"title": item.title,
"authors": item.authors,
"categories": item.categories,
"abstract": item.summary,
"update_date": item.updated,
"publication_date": item.published,
}
if (item.embeddings):
batch.add_data_object(
data_object=properties,
class_name=SCHEMA_NAME,
vector=item.embeddings["summary"])
else:
batch.add_data_object(
data_object=properties,
class_name=SCHEMA_NAME)
except Exception as ex:
logging.error(f"Unexpected Error: {ex}")
raise
def load_environment_vars() -> dict:
"""Load required environment variables. Raise an exception if any are missing."""
load_dotenv()
cohere_api_key = os.getenv("COHERE_API_KEY")
weaviate_url = os.getenv("WEAVIATE_URL")
weaviate_api_key = os.getenv("WEAVIATE_API_KEY")
if not cohere_api_key:
raise EnvironmentError("COHERE_API_KEY environment variable not set.")
if not weaviate_url:
raise EnvironmentError("WEAVIATE_URL environment variable not set.")
if not weaviate_api_key:
raise EnvironmentError(
"WEAVIATE_API_KEY environment variable not set.")
logging.info("Environment variables loaded.")
return {"COHERE_API_KEY": cohere_api_key, "WEAVIATE_URL": weaviate_url, "WEAVIATE_API_KEY": weaviate_api_key}
def main():
logging.basicConfig(level=logging.INFO,
format="%(asctime)s [%(levelname)s] %(message)s")
try:
env_vars = load_environment_vars()
index_data(env_vars["COHERE_API_KEY"],
env_vars["WEAVIATE_URL"], env_vars["WEAVIATE_API_KEY"])
except EnvironmentError as ee:
logging.error(f"Environment Error: {ee}")
raise
except Exception as ex:
logging.error(f"Unexpected Error: {ex}")
raise
if __name__ == "__main__":
main()