Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
|
@@ -1,13 +1,10 @@
|
|
| 1 |
import re
|
| 2 |
import requests
|
| 3 |
from bs4 import BeautifulSoup
|
| 4 |
-
# DuckDuckGo Search
|
| 5 |
from duckduckgo_search import DDGS
|
| 6 |
-
# Google search (unofficial scraping library)
|
| 7 |
-
from googlesearch import search as google_search
|
| 8 |
# SerpAPI client for various search engines (Google, Bing, etc.)
|
| 9 |
-
|
| 10 |
-
from serpapi import GoogleSearch as SerpApiClient
|
| 11 |
from rake_nltk import Rake
|
| 12 |
import gradio as gr
|
| 13 |
from transformers import pipeline
|
|
@@ -19,21 +16,18 @@ def extract_keywords(text: str) -> list[str]:
|
|
| 19 |
return rake.get_ranked_phrases()[:5]
|
| 20 |
|
| 21 |
# 2) Search functions
|
| 22 |
-
|
| 23 |
def ddg_search_links(query: str, num: int = 5) -> list[str]:
|
| 24 |
ddgs = DDGS()
|
| 25 |
results = ddgs.text(query, max_results=num)
|
| 26 |
return [r['href'] for r in results]
|
| 27 |
|
| 28 |
-
|
| 29 |
def google_search_links(query: str, num: int = 5) -> list[str]:
|
| 30 |
-
|
| 31 |
-
|
| 32 |
|
| 33 |
def serpapi_search_links(query: str, api_key: str, engine: str = 'bing', num: int = 5) -> list[str]:
|
| 34 |
-
# engine: 'bing', 'google', 'yahoo', etc.
|
| 35 |
params = {"engine": engine, "q": query, "api_key": api_key}
|
| 36 |
-
client =
|
| 37 |
data = client.get_dict()
|
| 38 |
results = data.get('organic_results', [])
|
| 39 |
return [r['link'] for r in results if not r.get('sponsored')][:num]
|
|
@@ -94,7 +88,6 @@ def process(prompt: str, web_enabled: bool, serpapi_key: str) -> str:
|
|
| 94 |
|
| 95 |
# Websearch enabled: model first
|
| 96 |
ans = model_answer(prompt)
|
| 97 |
-
# If model confident, return
|
| 98 |
if not is_uncertain(ans):
|
| 99 |
return ans
|
| 100 |
|
|
@@ -102,7 +95,6 @@ def process(prompt: str, web_enabled: bool, serpapi_key: str) -> str:
|
|
| 102 |
links = []
|
| 103 |
links += google_search_links(query)
|
| 104 |
links += ddg_search_links(query)
|
| 105 |
-
# Bing and Google via SerpAPI
|
| 106 |
if serpapi_key:
|
| 107 |
links += serpapi_search_links(query, serpapi_key, engine='bing')
|
| 108 |
links += serpapi_search_links(query, serpapi_key, engine='google')
|
|
@@ -115,9 +107,6 @@ def process(prompt: str, web_enabled: bool, serpapi_key: str) -> str:
|
|
| 115 |
return summary
|
| 116 |
|
| 117 |
# 8) Gradio UI
|
| 118 |
-
def main(prompt: str, web_enabled: bool, serpapi_key: str) -> str:
|
| 119 |
-
return process(prompt, web_enabled, serpapi_key)
|
| 120 |
-
|
| 121 |
with gr.Blocks() as demo:
|
| 122 |
gr.Markdown("# Intelligente KI mit Multi-Engine-Websuche")
|
| 123 |
with gr.Row():
|
|
@@ -128,9 +117,13 @@ with gr.Blocks() as demo:
|
|
| 128 |
output = gr.Textbox(label="Antwort", lines=10)
|
| 129 |
|
| 130 |
btn.click(
|
| 131 |
-
fn=
|
| 132 |
inputs=[prompt_input, web_switch, serp_input],
|
| 133 |
outputs=output
|
| 134 |
)
|
| 135 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 136 |
demo.launch()
|
|
|
|
| 1 |
import re
|
| 2 |
import requests
|
| 3 |
from bs4 import BeautifulSoup
|
| 4 |
+
# DuckDuckGo Search
|
| 5 |
from duckduckgo_search import DDGS
|
|
|
|
|
|
|
| 6 |
# SerpAPI client for various search engines (Google, Bing, etc.)
|
| 7 |
+
from serpapi import GoogleSearch
|
|
|
|
| 8 |
from rake_nltk import Rake
|
| 9 |
import gradio as gr
|
| 10 |
from transformers import pipeline
|
|
|
|
| 16 |
return rake.get_ranked_phrases()[:5]
|
| 17 |
|
| 18 |
# 2) Search functions
|
|
|
|
| 19 |
def ddg_search_links(query: str, num: int = 5) -> list[str]:
|
| 20 |
ddgs = DDGS()
|
| 21 |
results = ddgs.text(query, max_results=num)
|
| 22 |
return [r['href'] for r in results]
|
| 23 |
|
|
|
|
| 24 |
def google_search_links(query: str, num: int = 5) -> list[str]:
|
| 25 |
+
from googlesearch import search
|
| 26 |
+
return list(search(query, num_results=num, pause=2))
|
| 27 |
|
| 28 |
def serpapi_search_links(query: str, api_key: str, engine: str = 'bing', num: int = 5) -> list[str]:
|
|
|
|
| 29 |
params = {"engine": engine, "q": query, "api_key": api_key}
|
| 30 |
+
client = GoogleSearch(params)
|
| 31 |
data = client.get_dict()
|
| 32 |
results = data.get('organic_results', [])
|
| 33 |
return [r['link'] for r in results if not r.get('sponsored')][:num]
|
|
|
|
| 88 |
|
| 89 |
# Websearch enabled: model first
|
| 90 |
ans = model_answer(prompt)
|
|
|
|
| 91 |
if not is_uncertain(ans):
|
| 92 |
return ans
|
| 93 |
|
|
|
|
| 95 |
links = []
|
| 96 |
links += google_search_links(query)
|
| 97 |
links += ddg_search_links(query)
|
|
|
|
| 98 |
if serpapi_key:
|
| 99 |
links += serpapi_search_links(query, serpapi_key, engine='bing')
|
| 100 |
links += serpapi_search_links(query, serpapi_key, engine='google')
|
|
|
|
| 107 |
return summary
|
| 108 |
|
| 109 |
# 8) Gradio UI
|
|
|
|
|
|
|
|
|
|
| 110 |
with gr.Blocks() as demo:
|
| 111 |
gr.Markdown("# Intelligente KI mit Multi-Engine-Websuche")
|
| 112 |
with gr.Row():
|
|
|
|
| 117 |
output = gr.Textbox(label="Antwort", lines=10)
|
| 118 |
|
| 119 |
btn.click(
|
| 120 |
+
fn=process,
|
| 121 |
inputs=[prompt_input, web_switch, serp_input],
|
| 122 |
outputs=output
|
| 123 |
)
|
| 124 |
|
| 125 |
+
gr.Spacer()
|
| 126 |
+
gr.Markdown("---")
|
| 127 |
+
gr.Markdown("*Hinweis: Suche nur bei aktivierter Websuche und nicht bei verbotenen Phrasen.*")
|
| 128 |
+
|
| 129 |
demo.launch()
|