Frog215 commited on
Commit
2c7d6c4
·
verified ·
1 Parent(s): 87ad51b

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +299 -0
app.py CHANGED
@@ -93,6 +93,274 @@ def search_court_cases(keyword: str, jurisdiction: str = None,
93
  return error_msg
94
 
95
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
96
  @tool
97
  def check_citation_status(citation_id: str) -> Dict:
98
  """Check the current status and treatment of a legal citation
@@ -158,6 +426,37 @@ def check_citation_status(citation_id: str) -> Dict:
158
  error_msg = f"Unexpected error: {str(e)}"
159
  print(error_msg) # debugging
160
  return {"error": error_msg}
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
161
 
162
 
163
  final_answer = FinalAnswerTool()
 
93
  return error_msg
94
 
95
 
96
+ @tool
97
+ def check_citation_status(citation_id: str) -> Dict:
98
+ """Check the current status and treatment of a legal citation
99
+ Args:
100
+ citation_id: The citation ID (e.g., "403 U.S. 713")
101
+ """
102
+ try:
103
+ # Use the correct API base URL with v4
104
+ api_base = "https://www.courtlistener.com/api/rest/v4/"
105
+
106
+ # API vrne z oklepaji mi rabimo brez!
107
+ if "(" in citation_id:
108
+ citation_id = citation_id.split("(")[0].strip()
109
+
110
+ print(f"Looking up citation: {citation_id}")
111
+
112
+ # API
113
+ headers = {"Authorization": f"Token {COURTLISTENER_API_KEY}"}
114
+
115
+ params = {
116
+ "cite": citation_id,
117
+ "court": "scotus"
118
+ }
119
+
120
+ print(f"Request URL: {api_base}opinions/")
121
+ print(f"Params: {params}")
122
+
123
+ # API request, ki ga pošiljamo COurtListnereju
124
+ response = requests.get(f"{api_base}opinions/", params=params, headers=headers)
125
+ from smolagents import CodeAgent, DuckDuckGoSearchTool, HfApiModel, load_tool, tool
126
+ import datetime
127
+ import requests
128
+ import pytz
129
+ import yaml
130
+ from tools.final_answer import FinalAnswerTool
131
+ from Gradio_UI import GradioUI
132
+ from typing import List, Dict
133
+ from bs4 import BeautifulSoup
134
+
135
+ COURTLISTENER_API_KEY = "INSERT_API_KEY"
136
+ API_BASE = "https://www.courtlistener.com/api/rest/v4/"
137
+
138
+
139
+ @tool
140
+ def search_terms_of_service(company_name: str) -> Dict:
141
+ """Search for a company's Terms of Service link
142
+
143
+ Args:
144
+ company_name: Name of the organization/company to search for
145
+
146
+ Returns:
147
+ Dictionary containing company name and link to their Terms of Service
148
+ """
149
+
150
+ try:
151
+ # API URL
152
+ api_base = "https://05bf-176-76-227-247.ngrok-free.app" # lokalno ga runnamo
153
+
154
+ # API request naredimo
155
+ url = f"{api_base}/company/{company_name}"
156
+
157
+ print(f"Request URL: {url}")
158
+
159
+ response = requests.get(url)
160
+
161
+ print(f"Response status: {response.status_code}")
162
+ print(f"Response body preview: {response.text[:200]}...")
163
+
164
+ if response.status_code == 404: #HTTP error!
165
+ return {"error": f"Company '{company_name}' not found in the database"}
166
+
167
+ response.raise_for_status()
168
+
169
+ data = response.json()
170
+
171
+ # ali obstaja več linkov?
172
+ if not data.get('secondary_links') or len(data['secondary_links']) == 0:
173
+ return {"error": f"No Terms of Service links found for {company_name}"}
174
+
175
+ # returnamo ime in link, lahko jih več
176
+ result = {
177
+ "company_name": data["name"],
178
+ "tos_link": data["secondary_links"][0], # prvi link
179
+ "all_links": data["secondary_links"] # več linkov
180
+ }
181
+
182
+ return result
183
+
184
+ except requests.exceptions.RequestException as e:
185
+ error_msg = f"API Error: {str(e)}"
186
+ if hasattr(e, 'response') and e.response is not None:
187
+ error_msg += f" (Status code: {e.response.status_code})"
188
+ if hasattr(e.response, 'text'):
189
+ error_msg += f" - {e.response.text[:200]}"
190
+ print(error_msg) # debug, da prizna, če se zmoti
191
+ return {"error": error_msg}
192
+
193
+ except Exception as e:
194
+ error_msg = f"Unexpected error: {str(e)}"
195
+ print(error_msg) # debug, da AI prizna, če se zmoti
196
+ return {"error": error_msg}
197
+
198
+ @tool
199
+ def extract_tos_content(company_name: str) -> Dict:
200
+ """
201
+ Search for a company's Terms of Service, access the link, and extract raw content from GitLab repository.
202
+
203
+ Args:
204
+ company_name: Name of the organization/company to search for
205
+
206
+ Returns:
207
+ Dictionary containing company name, link, and extracted content
208
+ """
209
+
210
+ tos_info = search_terms_of_service(company_name)
211
+
212
+ # error
213
+ if "error" in tos_info:
214
+ return tos_info
215
+
216
+ try:
217
+ # link od zgoraj
218
+ tos_link = tos_info["tos_link"]
219
+
220
+ print(f"Accessing TOS link: {tos_link}")
221
+
222
+ # Gitlab repo
223
+ if "code.europa.eu" in tos_link and "/-/blob/" in tos_link:
224
+ # spremeni '/-/blob/' z '/-/raw/'
225
+ raw_url = tos_link.replace("/-/blob/", "/-/raw/")
226
+
227
+
228
+ if "?" in raw_url:
229
+ raw_url = raw_url.split("?")[0]
230
+
231
+ print(f"Accessing raw content URL: {raw_url}")
232
+
233
+ # request za raw data iz GItlabaa
234
+ response = requests.get(raw_url)
235
+ response.raise_for_status()
236
+
237
+ return {
238
+ "company_name": tos_info["company_name"],
239
+ "tos_link": tos_link,
240
+ "raw_url": raw_url,
241
+ "content": response.text, # cel raw data fajl
242
+ "content_type": "markdown" if raw_url.endswith(".md") else "text"
243
+ }
244
+
245
+ response = requests.get(tos_link)
246
+ response.raise_for_status()
247
+
248
+ soup = BeautifulSoup(response.text, 'html.parser')
249
+
250
+ content = ""
251
+
252
+ pre_element = soup.find('pre')
253
+ if pre_element:
254
+ content = pre_element.get_text()
255
+ else:
256
+ main_content = soup.find('main') or soup.find('article') or soup.find('body')
257
+ if main_content:
258
+ content = main_content.get_text(strip=True)
259
+
260
+ return {
261
+ "company_name": tos_info["company_name"],
262
+ "tos_link": tos_link,
263
+ "content": content,
264
+ "note": "Content extracted from HTML page"
265
+ }
266
+
267
+ except requests.exceptions.RequestException as e:
268
+ error_msg = f"Error accessing TOS link: {str(e)}"
269
+ if hasattr(e, 'response') and e.response is not None:
270
+ error_msg += f" (Status code: {e.response.status_code})"
271
+ if hasattr(e.response, 'text'):
272
+ error_msg += f" - {e.response.text[:200]}"
273
+ print(error_msg)
274
+ return {"error": error_msg}
275
+
276
+ except Exception as e:
277
+ error_msg = f"Unexpected error while processing content: {str(e)}"
278
+ print(error_msg)
279
+ return {"error": error_msg}
280
+
281
+ @tool
282
+ def search_court_cases(keyword: str, jurisdiction: str = None,
283
+ date_range: tuple = None) -> List[Dict]:
284
+
285
+ """Search court cases by keyword and jurisdiction
286
+ Args:
287
+ keyword: Legal concepts or case terms to search
288
+ jurisdiction: Court jurisdiction (e.g., 'scotus' for Supreme Court)
289
+ date_range: Tuple of (start_date, end_date) in UTC
290
+ """
291
+
292
+ try:
293
+ # API
294
+ api_base = "https://www.courtlistener.com/api/rest/v4/"
295
+
296
+ # zgradimo query
297
+ params = {
298
+ "q": keyword,
299
+ "type": "o", # 'o' opinion oz. mnenje
300
+ "page_size": 5
301
+ }
302
+
303
+
304
+ if jurisdiction:
305
+ if jurisdiction.lower() == "scotus":
306
+ params["court"] = "scotus"
307
+ else:
308
+ params["court"] = jurisdiction
309
+
310
+ # Datum in format
311
+ if date_range:
312
+ params["date_filed__gte"] = date_range[0].strftime("%Y-%m-%d")
313
+ params["date_filed__lte"] = date_range[1].strftime("%Y-%m-%d")
314
+
315
+ headers = {"Authorization": f"Token {COURTLISTENER_API_KEY}"}
316
+
317
+ print(f"Request URL: {api_base}search/")
318
+ print(f"Params: {params}")
319
+
320
+ response = requests.get(f"{api_base}search/", params=params, headers=headers)
321
+
322
+ print(f"Response status: {response.status_code}")
323
+ print(f"Response body preview: {response.text[:200]}...")
324
+
325
+ # Handle HTTP errors
326
+ if response.status_code == 403:
327
+ return "Authentication failed: Please check your API key"
328
+
329
+ response.raise_for_status()
330
+
331
+ data = response.json()
332
+
333
+ if 'results' not in data or len(data['results']) == 0:
334
+ return "No matching cases found in the database"
335
+
336
+ cases = []
337
+ for result in data['results']:
338
+ case_info = {
339
+ "case_name": result.get('caseName', result.get('case_name', 'Unknown')),
340
+ "citation": result.get('citation', 'No citation available'),
341
+ "court": result.get('court_full_name', result.get('court', 'Unknown')),
342
+ "date_filed": result.get('date_filed', 'Unknown date'),
343
+ "summary": result.get('text', result.get('snippet', ''))[:200] + '...' if result.get('text') or result.get('snippet') else "No summary available"
344
+ }
345
+ cases.append(case_info)
346
+
347
+ return cases
348
+
349
+ except requests.exceptions.RequestException as e:
350
+ error_msg = f"API Error: {str(e)}"
351
+ if hasattr(e, 'response') and e.response is not None:
352
+ error_msg += f" (Status code: {e.response.status_code})"
353
+ if hasattr(e.response, 'text'):
354
+ error_msg += f" - {e.response.text[:200]}"
355
+ print(error_msg) # debug
356
+ return error_msg
357
+
358
+ except Exception as e:
359
+ error_msg = f"Unexpected error: {str(e)}"
360
+ print(error_msg) # debug
361
+ return error_msg
362
+
363
+
364
  @tool
365
  def check_citation_status(citation_id: str) -> Dict:
366
  """Check the current status and treatment of a legal citation
 
426
  error_msg = f"Unexpected error: {str(e)}"
427
  print(error_msg) # debugging
428
  return {"error": error_msg}
429
+
430
+
431
+ final_answer = FinalAnswerTool()
432
+
433
+ #kateri model in kje se povežemo!
434
+ model = HfApiModel(
435
+ max_tokens=2096,
436
+ temperature=0.3,
437
+ model_id='https://pflgm2locj2t89co.us-east-1.aws.endpoints.huggingface.cloud/',
438
+ custom_role_conversions=None,
439
+ )
440
+
441
+ #prompti za agenta
442
+ with open("prompts.yaml", 'r') as stream:
443
+ prompt_templates = yaml.safe_load(stream)
444
+
445
+ #param za agenta
446
+ agent = CodeAgent(
447
+ model=model,
448
+ tools=[final_answer,search_terms_of_service,extract_tos_content], #search_court_cases, check_citation_status], #orodja
449
+ max_steps=6,
450
+ verbosity_level=1,
451
+ grammar=None,
452
+ planning_interval=None,
453
+ name="Legal Research Assistant",
454
+ description="Specializes in US federal and state court decisions analysis",
455
+ prompt_templates=prompt_templates
456
+ )
457
+
458
+ GradioUI(agent).launch() #launch za UI
459
+
460
 
461
 
462
  final_answer = FinalAnswerTool()