Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -3,26 +3,34 @@ import requests
|
|
3 |
import time
|
4 |
from duckduckgo_search import DDGS
|
5 |
|
6 |
-
#
|
|
|
7 |
def search_articles(name: str) -> str:
|
|
|
8 |
keywords = ['founders', 'partners', 'funders', 'owners']
|
9 |
search_query = f'"{name}" ({" OR ".join(keywords)}) site:news'
|
10 |
try:
|
|
|
11 |
with DDGS() as ddgs:
|
12 |
results = list(ddgs.text(search_query, max_results=3))
|
|
|
13 |
if not results:
|
14 |
return f"No articles found for {name}"
|
|
|
15 |
articles = []
|
16 |
for i, result in enumerate(results, 1):
|
17 |
-
article = f"**{i}. {result
|
18 |
-
article += f"Source: {result
|
19 |
-
article += f"{result
|
20 |
articles.append(article)
|
|
|
21 |
return "\n\n".join(articles)
|
22 |
except Exception as e:
|
23 |
-
return f"Search failed: {str(e)}"
|
|
|
24 |
|
25 |
def extract_entities(search_results: str) -> str:
|
|
|
26 |
modal_endpoint = "https://msoaresdiego--mistral-llm-endpoint-fastapi-app.modal.run/generate"
|
27 |
prompt = f"""Extract all person names and organization names from the following text.
|
28 |
Format as:
|
@@ -32,17 +40,18 @@ Text: {search_results}"""
|
|
32 |
try:
|
33 |
response = requests.post(
|
34 |
modal_endpoint,
|
35 |
-
json={"prompt": prompt, "max_tokens": 500, "temperature": 0.1}
|
36 |
-
timeout=180 # Increased timeout
|
37 |
)
|
38 |
if response.status_code == 200:
|
39 |
return response.json().get("response", "No entities extracted")
|
40 |
else:
|
41 |
-
return f"API Error: {response.status_code}"
|
42 |
except Exception as e:
|
43 |
-
return f"Extraction failed: {str(e)}"
|
|
|
44 |
|
45 |
def find_full_names(search_results: str, entities: str) -> str:
|
|
|
46 |
modal_endpoint = "https://msoaresdiego--mistral-llm-endpoint-fastapi-app.modal.run/generate"
|
47 |
prompt = f"""Based on the search results, find the full names and titles/roles for these entities:
|
48 |
Entities: {entities}
|
@@ -51,64 +60,130 @@ Provide full names with their roles/titles where mentioned."""
|
|
51 |
try:
|
52 |
response = requests.post(
|
53 |
modal_endpoint,
|
54 |
-
json={"prompt": prompt, "max_tokens": 300, "temperature": 0.1}
|
55 |
-
timeout=180 # Increased timeout
|
56 |
)
|
57 |
if response.status_code == 200:
|
58 |
return response.json().get("response", "No full names found")
|
59 |
else:
|
60 |
-
return f"API Error: {response.status_code}"
|
61 |
except Exception as e:
|
62 |
-
return f"Full name extraction failed: {str(e)}"
|
|
|
|
|
|
|
63 |
|
64 |
-
|
65 |
-
|
66 |
if not name.strip():
|
67 |
-
|
|
|
|
|
|
|
|
|
|
|
68 |
|
69 |
try:
|
70 |
-
|
71 |
-
|
72 |
-
|
73 |
-
search_results
|
74 |
-
|
75 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
76 |
|
77 |
-
|
78 |
-
|
79 |
-
|
|
|
|
|
80 |
|
81 |
-
|
82 |
-
progress(
|
83 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
84 |
|
85 |
except Exception as e:
|
86 |
-
|
87 |
-
yield search_results
|
88 |
-
|
89 |
-
|
90 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
91 |
|
92 |
-
# --- Gradio Interface ---
|
93 |
with gr.Blocks(title="Name Research Tool") as demo:
|
94 |
-
gr.Markdown("# Name Research Tool")
|
95 |
gr.Markdown("Enter a business or project name to search for related articles and extract key entities.")
|
96 |
|
97 |
with gr.Row():
|
98 |
name_input = gr.Textbox(label="Name", placeholder="Enter business or project name")
|
99 |
-
|
|
|
|
|
100 |
|
101 |
with gr.Column():
|
102 |
-
output1 = gr.Textbox(label="Search Results", lines=10, max_lines=
|
103 |
output2 = gr.Textbox(label="Extracted Entities", lines=5, max_lines=10)
|
104 |
output3 = gr.Textbox(label="Full Names", lines=5, max_lines=10)
|
105 |
-
status_output = gr.Textbox(label="Status / Progress", lines=1, interactive=False)
|
106 |
|
107 |
-
# Search with progress
|
108 |
search_btn.click(
|
109 |
fn=process_name_with_progress,
|
110 |
inputs=[name_input],
|
111 |
-
outputs=[output1, output2, output3
|
|
|
|
|
|
|
|
|
|
|
|
|
112 |
)
|
113 |
|
114 |
if __name__ == "__main__":
|
|
|
3 |
import time
|
4 |
from duckduckgo_search import DDGS
|
5 |
|
6 |
+
# === Model functions ===
|
7 |
+
|
8 |
def search_articles(name: str) -> str:
|
9 |
+
"""Search for 3 newspaper articles containing the name and keywords using DuckDuckGo"""
|
10 |
keywords = ['founders', 'partners', 'funders', 'owners']
|
11 |
search_query = f'"{name}" ({" OR ".join(keywords)}) site:news'
|
12 |
try:
|
13 |
+
print(f"[DEBUG] Search query: {search_query}")
|
14 |
with DDGS() as ddgs:
|
15 |
results = list(ddgs.text(search_query, max_results=3))
|
16 |
+
print(f"[DEBUG] Raw results: {results}")
|
17 |
if not results:
|
18 |
return f"No articles found for {name}"
|
19 |
+
|
20 |
articles = []
|
21 |
for i, result in enumerate(results, 1):
|
22 |
+
article = f"**{i}. {result.get('title', 'No Title')}**\n"
|
23 |
+
article += f"Source: {result.get('href', 'No URL')}\n"
|
24 |
+
article += f"{result.get('body', 'No Body')}\n"
|
25 |
articles.append(article)
|
26 |
+
|
27 |
return "\n\n".join(articles)
|
28 |
except Exception as e:
|
29 |
+
return f"[ERROR] Search failed: {str(e)}"
|
30 |
+
|
31 |
|
32 |
def extract_entities(search_results: str) -> str:
|
33 |
+
"""Extract entities using Mistral 7B endpoint"""
|
34 |
modal_endpoint = "https://msoaresdiego--mistral-llm-endpoint-fastapi-app.modal.run/generate"
|
35 |
prompt = f"""Extract all person names and organization names from the following text.
|
36 |
Format as:
|
|
|
40 |
try:
|
41 |
response = requests.post(
|
42 |
modal_endpoint,
|
43 |
+
json={"prompt": prompt, "max_tokens": 500, "temperature": 0.1}
|
|
|
44 |
)
|
45 |
if response.status_code == 200:
|
46 |
return response.json().get("response", "No entities extracted")
|
47 |
else:
|
48 |
+
return f"[ERROR] API Error: {response.status_code}"
|
49 |
except Exception as e:
|
50 |
+
return f"[ERROR] Extraction failed: {str(e)}"
|
51 |
+
|
52 |
|
53 |
def find_full_names(search_results: str, entities: str) -> str:
|
54 |
+
"""Find full names using Mistral 7B endpoint"""
|
55 |
modal_endpoint = "https://msoaresdiego--mistral-llm-endpoint-fastapi-app.modal.run/generate"
|
56 |
prompt = f"""Based on the search results, find the full names and titles/roles for these entities:
|
57 |
Entities: {entities}
|
|
|
60 |
try:
|
61 |
response = requests.post(
|
62 |
modal_endpoint,
|
63 |
+
json={"prompt": prompt, "max_tokens": 300, "temperature": 0.1}
|
|
|
64 |
)
|
65 |
if response.status_code == 200:
|
66 |
return response.json().get("response", "No full names found")
|
67 |
else:
|
68 |
+
return f"[ERROR] API Error: {response.status_code}"
|
69 |
except Exception as e:
|
70 |
+
return f"[ERROR] Full name extraction failed: {str(e)}"
|
71 |
+
|
72 |
+
|
73 |
+
# === Gradio interface ===
|
74 |
|
75 |
+
def process_name_with_progress(name: str, progress=gr.Progress()):
|
76 |
+
"""Process name with streamed debug updates to help diagnose issues"""
|
77 |
if not name.strip():
|
78 |
+
yield "No name provided", "", ""
|
79 |
+
return
|
80 |
+
|
81 |
+
search_results = ""
|
82 |
+
entities = ""
|
83 |
+
full_names = ""
|
84 |
|
85 |
try:
|
86 |
+
# Step 1: Search
|
87 |
+
progress(0.1, desc="Searching for articles...")
|
88 |
+
search_results += f"[DEBUG] Starting search for: {name}\n"
|
89 |
+
yield search_results, "", ""
|
90 |
+
|
91 |
+
search_start = time.time()
|
92 |
+
articles_output = search_articles(name.strip())
|
93 |
+
search_time = time.time() - search_start
|
94 |
+
|
95 |
+
search_results += f"[DEBUG] Search completed in {search_time:.2f}s\n"
|
96 |
+
search_results += f"{articles_output}\n"
|
97 |
+
yield search_results, "", ""
|
98 |
+
|
99 |
+
# Step 2: Extract entities
|
100 |
+
progress(0.5, desc="Extracting entities...")
|
101 |
+
search_results += "[DEBUG] Starting entity extraction...\n"
|
102 |
+
yield search_results, "[DEBUG] Extracting entities...", ""
|
103 |
|
104 |
+
extract_start = time.time()
|
105 |
+
entities = extract_entities(articles_output)
|
106 |
+
extract_time = time.time() - extract_start
|
107 |
+
search_results += f"[DEBUG] Entity extraction completed in {extract_time:.2f}s\n"
|
108 |
+
yield search_results, entities, ""
|
109 |
|
110 |
+
# Step 3: Full names
|
111 |
+
progress(0.8, desc="Finding full names...")
|
112 |
+
search_results += "[DEBUG] Starting full name resolution...\n"
|
113 |
+
yield search_results, entities, "[DEBUG] Resolving full names..."
|
114 |
+
|
115 |
+
names_start = time.time()
|
116 |
+
full_names = find_full_names(articles_output, entities)
|
117 |
+
names_time = time.time() - names_start
|
118 |
+
search_results += f"[DEBUG] Full name extraction completed in {names_time:.2f}s\n"
|
119 |
+
|
120 |
+
progress(1.0, desc="Complete!")
|
121 |
+
yield search_results, entities, full_names
|
122 |
|
123 |
except Exception as e:
|
124 |
+
error_msg = f"[ERROR] {str(e)}"
|
125 |
+
yield search_results + error_msg, entities or error_msg, full_names or error_msg
|
126 |
+
|
127 |
+
|
128 |
+
def process_name_simple(name: str):
|
129 |
+
"""Basic version without progress bar - for isolated testing"""
|
130 |
+
if not name.strip():
|
131 |
+
return "", "", ""
|
132 |
+
|
133 |
+
print(f"Starting process for: {name}")
|
134 |
+
total_start = time.time()
|
135 |
+
|
136 |
+
print("Step 1: Searching articles...")
|
137 |
+
search_start = time.time()
|
138 |
+
search_results = search_articles(name.strip())
|
139 |
+
search_time = time.time() - search_start
|
140 |
+
print(f"Search completed in: {search_time:.2f}s")
|
141 |
+
|
142 |
+
print("Step 2: Extracting entities...")
|
143 |
+
extract_start = time.time()
|
144 |
+
entities = extract_entities(search_results)
|
145 |
+
extract_time = time.time() - extract_start
|
146 |
+
print(f"Entity extraction in: {extract_time:.2f}s")
|
147 |
+
|
148 |
+
print("Step 3: Finding full names...")
|
149 |
+
names_start = time.time()
|
150 |
+
full_names = find_full_names(search_results, entities)
|
151 |
+
names_time = time.time() - names_start
|
152 |
+
print(f"Full name resolution in: {names_time:.2f}s")
|
153 |
+
|
154 |
+
total_time = time.time() - total_start
|
155 |
+
print(f"Total time: {total_time:.2f}s")
|
156 |
+
|
157 |
+
return search_results, entities, full_names
|
158 |
+
|
159 |
+
|
160 |
+
# === Gradio UI ===
|
161 |
|
|
|
162 |
with gr.Blocks(title="Name Research Tool") as demo:
|
163 |
+
gr.Markdown("# 🔎 Name Research Tool")
|
164 |
gr.Markdown("Enter a business or project name to search for related articles and extract key entities.")
|
165 |
|
166 |
with gr.Row():
|
167 |
name_input = gr.Textbox(label="Name", placeholder="Enter business or project name")
|
168 |
+
with gr.Column():
|
169 |
+
search_btn = gr.Button("Search (Real-time)", variant="primary")
|
170 |
+
debug_btn = gr.Button("Search (Debug Mode)", variant="secondary")
|
171 |
|
172 |
with gr.Column():
|
173 |
+
output1 = gr.Textbox(label="Search Results (with debug)", lines=10, max_lines=30)
|
174 |
output2 = gr.Textbox(label="Extracted Entities", lines=5, max_lines=10)
|
175 |
output3 = gr.Textbox(label="Full Names", lines=5, max_lines=10)
|
|
|
176 |
|
|
|
177 |
search_btn.click(
|
178 |
fn=process_name_with_progress,
|
179 |
inputs=[name_input],
|
180 |
+
outputs=[output1, output2, output3]
|
181 |
+
)
|
182 |
+
|
183 |
+
debug_btn.click(
|
184 |
+
fn=process_name_simple,
|
185 |
+
inputs=[name_input],
|
186 |
+
outputs=[output1, output2, output3]
|
187 |
)
|
188 |
|
189 |
if __name__ == "__main__":
|