-
Notifications
You must be signed in to change notification settings - Fork 0
/
data_load.py
283 lines (250 loc) · 11.2 KB
/
data_load.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
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
import base64
import hashlib
import json
import os
from json import JSONDecodeError
from pathlib import Path
from time import sleep
from jinja2 import Environment, select_autoescape, FileSystemLoader
from slugify import slugify
import requests
from unstructured.documents.html import HTMLDocument
from constants import (
WEBSITE, DATA_DIR, COLLECTION_NAME, RESET_VECTOR_STORE, RESET_LINKEDIN_DATA, PROXYCURL_API_KEY,
SPEAKER_IMAGE_CACHE_FILE_NAME, PULL_LINKEDIN, TEMPLATED_DATA_STORE_PATH, NUBELA_REQUESTS_PER_MINUTE
)
from bs4 import BeautifulSoup
from utils import initialize_vector_store
env = Environment(
loader=FileSystemLoader("templates"),
autoescape=select_autoescape()
)
if not os.path.exists(SPEAKER_IMAGE_CACHE_FILE_NAME):
json.dump({}, open(SPEAKER_IMAGE_CACHE_FILE_NAME, "w"))
def get_image_speakers(image_uri):
"""
This is caching for the name extraction from the image data so we don't hit the
openai for expensive requests when the signature of the data hasn't changed
"""
speaker_image_cache = json.load(open(SPEAKER_IMAGE_CACHE_FILE_NAME, "r"))
image_data, base64_image = get_image_data(image_uri)
image_data_sig = hashlib.md5(image_data).hexdigest()
if image_data_sig in speaker_image_cache:
return speaker_image_cache[image_data_sig]
speaker_names = extract_names_from_image(base64_image=base64_image)
speaker_image_cache[image_data_sig] = speaker_names
json.dump(speaker_image_cache, open(SPEAKER_IMAGE_CACHE_FILE_NAME, "w"))
return speaker_names
def get_image_data(image_url):
response = requests.get(image_url)
base64_image = (
"data:" + response.headers['Content-Type'] + ";" +
"base64," + base64.b64encode(response.content).decode("utf-8")
)
return response.content, base64_image
def extract_names_from_image(base64_image):
headers = {
"Content-Type": "application/json",
"Authorization": f"Bearer {os.getenv('OPENAI_API_KEY')}"
}
payload = {
"model": "gpt-4-vision-preview",
"messages": [
{
"role": "user",
"content": [
{
"type": "text",
"text": "What are the names in this image? Names are usually given the largest fonts. "
"Return only the names in comma separated format without explanation. "
"There's no need to warn about not being able to access other information, you are only"
"going to be asked for the name."
},
{
"type": "image_url",
"image_url": {
"url": f"{base64_image}"
}
}
]
}
],
"max_tokens": 300
}
response = requests.post(
"https://api.openai.com/v1/chat/completions",
headers=headers,
json=payload
)
response_json = response.json()
speakers = response_json["choices"][0]["message"]["content"]
return speakers
def populate_data_from_website(collection):
for root, dirs, files in os.walk(os.sep.join([os.getcwd(), DATA_DIR, WEBSITE])):
for file_name in files:
full_file_name = os.path.join(root, file_name)
# here we want to handle special cases to ensure high quality data
if file_name in SPESHUL_CASES:
SPESHUL_CASES[file_name](collection, full_file_name)
# we also keep the less than high quality data
doc = HTMLDocument.from_file(full_file_name)
for i in range(len(doc.pages)):
print(f"Adding page number {i} from {file_name}")
collection.upsert(
documents=[str(doc.pages[i])],
metadatas=[{"file": file_name, "page": i, "Type": "Generic"}],
ids=[f"{file_name}_page_{i}"],
)
def handle_tracks(collection, full_file_name):
soup = BeautifulSoup(open(full_file_name), "html.parser")
track_items_list = soup.find_all("article")
for item in track_items_list:
track_event_data = structure_track_event_for_template(item)
# render data to text
template = env.get_template("track_event.jinja2")
track_event_content = template.render(track_event=track_event_data)
print(f"Adding track event speakers {track_event_data['event_speakers_names']}")
# save template output file for review
track_event_file = Path(TEMPLATED_DATA_STORE_PATH, "track_event", f"{slugify(track_event_data['event_title'])}_{track_event_data['event_date']}.txt")
track_event_file.parent.mkdir(exist_ok=True, parents=True)
with open(track_event_file, "w") as f:
f.write(track_event_content)
collection.upsert(
documents=[track_event_content],
metadatas=[{
"Track event name": track_event_data["event_title"],
"Track event speakers": track_event_data["event_speakers_names"],
"Type": "Track Event",
"source": track_event_file.as_posix().replace(os.getcwd(), ""),
}],
ids=[f"{slugify(track_event_data['event_title'])}_{track_event_data['event_date']}_track_event"],
)
def structure_track_event_for_template(track_item):
track_event_data = {}
# get the speaker image for parsing
speakers_img_wrapper = track_item.find('a', attrs={"class": "eventlist-column-thumbnail"})
speakers_img_uri = speakers_img_wrapper.find("img")
speakers = get_image_speakers(speakers_img_uri["src"])
track_event_data["event_speakers_names"] = speakers
# get the event title
event_title = track_item.find('a', attrs={"class": "eventlist-title-link"}).contents[0]
track_event_data["event_title"] = event_title
# get the event date
event_date = track_item.find('time', attrs={"class": "event-date"}).contents[0]
track_event_data["event_date"] = event_date
# get the event start time
event_start_time = track_item.find('time', attrs={"class": "event-time-localized-start"}).contents[0]
track_event_data["event_start_time"] = event_start_time
# get the event end time
event_end_time = track_item.find('time', attrs={"class": "event-time-localized-end"}).contents[0]
track_event_data["event_end_time"] = event_end_time
# get the event description
event_description = track_item.find("div", attrs={"class": "eventlist-excerpt"}).text
track_event_data["description"] = event_description
return track_event_data
def handle_good_tech_fest_utah_2_1(collection, full_file_name):
soup = BeautifulSoup(open(full_file_name), "html.parser")
user_items_list = soup.find("ul", attrs={"data-controller": "UserItemsListSimple"})
speaker_data_str = user_items_list.attrs["data-current-context"]
speaker_data = json.loads(speaker_data_str)
for idx, speaker in enumerate(speaker_data["userItems"]):
speaker_info = []
description_soup = BeautifulSoup(speaker["description"], "html.parser")
speaker_name = speaker["title"]
speaker_name_slug = slugify(speaker_name)
bio_uri = speaker["button"]["buttonLink"]
speaker_info.append(f"Speaker Name: {speaker_name}")
speaker_info.append(f"LinkedIn: {bio_uri}")
speaker_info.append(f"Titles, Organizations: {description_soup.text}")
print(f"Adding Virtual Breakout Speaker: {speaker['title']}")
collection.upsert(
documents=["\n".join(speaker_info)],
metadatas=[{
"speaker": speaker["title"],
"type": "Virtual Speaker",
"source": full_file_name,
}],
ids=[f"{speaker_name_slug}_speaker_{idx}"],
)
bio_content = None
if "linkedin" in bio_uri:
print("request on linked in")
bio_content = get_linkedin_data(bio_uri, speaker_name_slug)
if bio_content:
print(f"Adding bio for {speaker['title']}")
collection.upsert(
documents=[bio_content],
metadatas=[{
"speaker": speaker["title"],
"type": "Virtual Speaker Bio",
"source": bio_uri,
}],
ids=[f"{speaker_name_slug}_speaker_bio"],
)
def get_linkedin_data(bio_uri, speaker_name_slug):
print(f"Get linked in data for {speaker_name_slug=}")
def _pull_leakedin_data(bio_uri):
if not PULL_LINKEDIN:
print(f"LinkedIn requests skipped due to {PULL_LINKEDIN=}")
return None
api_endpoint = "https://nubela.co/proxycurl/api/v2/linkedin"
headers = {"Authorization": f"Bearer {PROXYCURL_API_KEY}"}
params = {"url": bio_uri}
response = requests.get(
api_endpoint,
params=params,
headers=headers
)
# throttle for Nubela rate limits
sleep(60 / NUBELA_REQUESTS_PER_MINUTE)
return response.json()
linkedin_datadir = os.sep.join([os.getcwd(), DATA_DIR, "linkedin.com"])
linkedin_filename = os.sep.join([os.getcwd(), DATA_DIR, "linkedin.com", speaker_name_slug])
linkedin_filename = linkedin_filename + ".json"
if not os.path.exists(linkedin_datadir):
os.makedirs(linkedin_datadir)
file_exists = os.path.exists(linkedin_filename)
if not file_exists:
os.mknod(linkedin_filename)
dest_file = open(linkedin_filename, "r+")
if not file_exists or RESET_LINKEDIN_DATA:
# make call and cache for future use
bio_data = _pull_leakedin_data(bio_uri=bio_uri)
if bio_data:
print(f"writing {bio_data=} to {linkedin_filename=}")
json.dump(bio_data, dest_file)
else:
# use stored data
print(f"reading from cache for {linkedin_filename=}")
try:
bio_data = json.load(dest_file)
except JSONDecodeError:
print(f"Error reading linkedin data from {linkedin_filename=}")
bio_data = None
bio_content = None
# 'code' only exists if the user record isn't found
if bio_data and not bio_data.get("code"):
template = env.get_template("linkedin_bios.jinja2")
bio_content = template.render(bio=bio_data)
# write the template data for review
bio_file = Path(TEMPLATED_DATA_STORE_PATH, "bio", slugify(bio_data["full_name"]) + ".txt")
bio_file.parent.mkdir(exist_ok=True, parents=True)
with open(bio_file, "w") as f:
f.write(bio_content)
return bio_content
SPESHUL_CASES = {
"good-tech-fest-utah-2-1": handle_good_tech_fest_utah_2_1,
"ai-data-science-track.html": handle_tracks,
"cyber-security-track.html": handle_tracks,
"engineering-track.html": handle_tracks,
"technical-leadership-management-track.html": handle_tracks,
"product-development-management-track.html": handle_tracks,
"governance-collaboration-track.html": handle_tracks,
"ethics-responsibility-track.html": handle_tracks,
}
if __name__ == "__main__":
vector_store = initialize_vector_store(
collection_name=COLLECTION_NAME,
reset_vector_store=RESET_VECTOR_STORE,
)
populate_data_from_website(vector_store)