Frog215 commited on
Commit
cc3e866
·
verified ·
1 Parent(s): 29c850e

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +79 -2
app.py CHANGED
@@ -70,7 +70,84 @@ def search_terms_of_service(company_name: str) -> Dict:
70
  print(error_msg) # debug, da AI prizna, če se zmoti
71
  return {"error": error_msg}
72
 
73
-
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
74
 
75
  @tool
76
  def search_court_cases(keyword: str, jurisdiction: str = None,
@@ -239,7 +316,7 @@ with open("prompts.yaml", 'r') as stream:
239
  #param za agenta
240
  agent = CodeAgent(
241
  model=model,
242
- tools=[final_answer,search_terms_of_service], #search_court_cases, check_citation_status], #orodja
243
  max_steps=6,
244
  verbosity_level=1,
245
  grammar=None,
 
70
  print(error_msg) # debug, da AI prizna, če se zmoti
71
  return {"error": error_msg}
72
 
73
+ @tool
74
+ def extract_tos_content(company_name: str) -> Dict:
75
+ """
76
+ Search for a company's Terms of Service, access the link, and extract content.
77
+
78
+ Args:
79
+ company_name: Name of the organization/company to search for
80
+
81
+ Returns:
82
+ Dictionary containing company name, link, and extracted content
83
+ """
84
+
85
+ # First use the existing function to get the TOS link
86
+ tos_info = search_terms_of_service(company_name)
87
+
88
+ # If there was an error in the search function, return that error
89
+ if "error" in tos_info:
90
+ return tos_info
91
+
92
+ try:
93
+ # Get the link from the search results
94
+ tos_link = tos_info["tos_link"]
95
+
96
+ print(f"Accessing TOS link: {tos_link}")
97
+
98
+ # Request the content from the TOS link
99
+ response = requests.get(tos_link)
100
+ response.raise_for_status()
101
+
102
+ # Parse the HTML content
103
+ soup = BeautifulSoup(response.text, 'html.parser')
104
+
105
+ # Extract repository content - assuming it's in a specific element
106
+ # You may need to adjust these selectors based on the actual page structure
107
+ repo_content = ""
108
+ repo_element = soup.find('div', class_='repository-content')
109
+ if repo_element:
110
+ repo_content = repo_element.get_text(strip=True)
111
+
112
+ # Extract TOS content - assuming it's in a specific element
113
+ # You may need to adjust these selectors based on the actual page structure
114
+ tos_content = ""
115
+ tos_element = soup.find('div', id='terms-of-service') or soup.find('div', class_='terms-of-service')
116
+ if tos_element:
117
+ tos_content = tos_element.get_text(strip=True)
118
+
119
+ # If we couldn't find specific elements, try to get the main content
120
+ if not repo_content and not tos_content:
121
+ main_content = soup.find('main') or soup.find('body')
122
+ if main_content:
123
+ main_content_text = main_content.get_text(strip=True)
124
+ return {
125
+ "company_name": tos_info["company_name"],
126
+ "tos_link": tos_link,
127
+ "content": main_content_text[:5000], # Limit content length
128
+ "note": "Could not identify specific repository or TOS sections"
129
+ }
130
+
131
+ return {
132
+ "company_name": tos_info["company_name"],
133
+ "tos_link": tos_link,
134
+ "repository_content": repo_content[:2500] if repo_content else "No repository content found",
135
+ "tos_content": tos_content[:2500] if tos_content else "No Terms of Service content found"
136
+ }
137
+
138
+ except requests.exceptions.RequestException as e:
139
+ error_msg = f"Error accessing TOS link: {str(e)}"
140
+ if hasattr(e, 'response') and e.response is not None:
141
+ error_msg += f" (Status code: {e.response.status_code})"
142
+ if hasattr(e.response, 'text'):
143
+ error_msg += f" - {e.response.text[:200]}"
144
+ print(error_msg)
145
+ return {"error": error_msg}
146
+
147
+ except Exception as e:
148
+ error_msg = f"Unexpected error while processing content: {str(e)}"
149
+ print(error_msg)
150
+ return {"error": error_msg}
151
 
152
  @tool
153
  def search_court_cases(keyword: str, jurisdiction: str = None,
 
316
  #param za agenta
317
  agent = CodeAgent(
318
  model=model,
319
+ tools=[final_answer,search_terms_of_service,extract_tos_content], #search_court_cases, check_citation_status], #orodja
320
  max_steps=6,
321
  verbosity_level=1,
322
  grammar=None,