task.py 10 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315
  1. import logging
  2. import math
  3. import re
  4. from datetime import datetime
  5. from typing import Optional
  6. import uuid
  7. from open_webui.utils.misc import get_last_user_message, get_messages_content
  8. from open_webui.env import SRC_LOG_LEVELS
  9. from open_webui.config import DEFAULT_RAG_TEMPLATE
  10. log = logging.getLogger(__name__)
  11. log.setLevel(SRC_LOG_LEVELS["RAG"])
  12. def get_task_model_id(
  13. default_model_id: str, task_model: str, task_model_external: str, models
  14. ) -> str:
  15. # Set the task model
  16. task_model_id = default_model_id
  17. # Check if the user has a custom task model and use that model
  18. if models[task_model_id]["owned_by"] == "ollama":
  19. if task_model and task_model in models:
  20. task_model_id = task_model
  21. else:
  22. if task_model_external and task_model_external in models:
  23. task_model_id = task_model_external
  24. return task_model_id
  25. def prompt_template(
  26. template: str, user_name: Optional[str] = None, user_location: Optional[str] = None
  27. ) -> str:
  28. # Get the current date
  29. current_date = datetime.now()
  30. # Format the date to YYYY-MM-DD
  31. formatted_date = current_date.strftime("%Y-%m-%d")
  32. formatted_time = current_date.strftime("%I:%M:%S %p")
  33. formatted_weekday = current_date.strftime("%A")
  34. template = template.replace("{{CURRENT_DATE}}", formatted_date)
  35. template = template.replace("{{CURRENT_TIME}}", formatted_time)
  36. template = template.replace(
  37. "{{CURRENT_DATETIME}}", f"{formatted_date} {formatted_time}"
  38. )
  39. template = template.replace("{{CURRENT_WEEKDAY}}", formatted_weekday)
  40. if user_name:
  41. # Replace {{USER_NAME}} in the template with the user's name
  42. template = template.replace("{{USER_NAME}}", user_name)
  43. else:
  44. # Replace {{USER_NAME}} in the template with "Unknown"
  45. template = template.replace("{{USER_NAME}}", "Unknown")
  46. if user_location:
  47. # Replace {{USER_LOCATION}} in the template with the current location
  48. template = template.replace("{{USER_LOCATION}}", user_location)
  49. else:
  50. # Replace {{USER_LOCATION}} in the template with "Unknown"
  51. template = template.replace("{{USER_LOCATION}}", "Unknown")
  52. return template
  53. def replace_prompt_variable(template: str, prompt: str) -> str:
  54. def replacement_function(match):
  55. full_match = match.group(
  56. 0
  57. ).lower() # Normalize to lowercase for consistent handling
  58. start_length = match.group(1)
  59. end_length = match.group(2)
  60. middle_length = match.group(3)
  61. if full_match == "{{prompt}}":
  62. return prompt
  63. elif start_length is not None:
  64. return prompt[: int(start_length)]
  65. elif end_length is not None:
  66. return prompt[-int(end_length) :]
  67. elif middle_length is not None:
  68. middle_length = int(middle_length)
  69. if len(prompt) <= middle_length:
  70. return prompt
  71. start = prompt[: math.ceil(middle_length / 2)]
  72. end = prompt[-math.floor(middle_length / 2) :]
  73. return f"{start}...{end}"
  74. return ""
  75. # Updated regex pattern to make it case-insensitive with the `(?i)` flag
  76. pattern = r"(?i){{prompt}}|{{prompt:start:(\d+)}}|{{prompt:end:(\d+)}}|{{prompt:middletruncate:(\d+)}}"
  77. template = re.sub(pattern, replacement_function, template)
  78. return template
  79. def replace_messages_variable(
  80. template: str, messages: Optional[list[str]] = None
  81. ) -> str:
  82. def replacement_function(match):
  83. full_match = match.group(0)
  84. start_length = match.group(1)
  85. end_length = match.group(2)
  86. middle_length = match.group(3)
  87. # If messages is None, handle it as an empty list
  88. if messages is None:
  89. return ""
  90. # Process messages based on the number of messages required
  91. if full_match == "{{MESSAGES}}":
  92. return get_messages_content(messages)
  93. elif start_length is not None:
  94. return get_messages_content(messages[: int(start_length)])
  95. elif end_length is not None:
  96. return get_messages_content(messages[-int(end_length) :])
  97. elif middle_length is not None:
  98. mid = int(middle_length)
  99. if len(messages) <= mid:
  100. return get_messages_content(messages)
  101. # Handle middle truncation: split to get start and end portions of the messages list
  102. half = mid // 2
  103. start_msgs = messages[:half]
  104. end_msgs = messages[-half:] if mid % 2 == 0 else messages[-(half + 1) :]
  105. formatted_start = get_messages_content(start_msgs)
  106. formatted_end = get_messages_content(end_msgs)
  107. return f"{formatted_start}\n{formatted_end}"
  108. return ""
  109. template = re.sub(
  110. r"{{MESSAGES}}|{{MESSAGES:START:(\d+)}}|{{MESSAGES:END:(\d+)}}|{{MESSAGES:MIDDLETRUNCATE:(\d+)}}",
  111. replacement_function,
  112. template,
  113. )
  114. return template
  115. # {{prompt:middletruncate:8000}}
  116. def rag_template(template: str, context: str, query: str):
  117. if template.strip() == "":
  118. template = DEFAULT_RAG_TEMPLATE
  119. if "[context]" not in template and "{{CONTEXT}}" not in template:
  120. log.debug(
  121. "WARNING: The RAG template does not contain the '[context]' or '{{CONTEXT}}' placeholder."
  122. )
  123. if "<context>" in context and "</context>" in context:
  124. log.debug(
  125. "WARNING: Potential prompt injection attack: the RAG "
  126. "context contains '<context>' and '</context>'. This might be "
  127. "nothing, or the user might be trying to hack something."
  128. )
  129. query_placeholders = []
  130. if "[query]" in context:
  131. query_placeholder = "{{QUERY" + str(uuid.uuid4()) + "}}"
  132. template = template.replace("[query]", query_placeholder)
  133. query_placeholders.append(query_placeholder)
  134. if "{{QUERY}}" in context:
  135. query_placeholder = "{{QUERY" + str(uuid.uuid4()) + "}}"
  136. template = template.replace("{{QUERY}}", query_placeholder)
  137. query_placeholders.append(query_placeholder)
  138. template = template.replace("[context]", context)
  139. template = template.replace("{{CONTEXT}}", context)
  140. template = template.replace("[query]", query)
  141. template = template.replace("{{QUERY}}", query)
  142. for query_placeholder in query_placeholders:
  143. template = template.replace(query_placeholder, query)
  144. return template
  145. def title_generation_template(
  146. template: str, messages: list[dict], user: Optional[dict] = None
  147. ) -> str:
  148. prompt = get_last_user_message(messages)
  149. template = replace_prompt_variable(template, prompt)
  150. template = replace_messages_variable(template, messages)
  151. template = prompt_template(
  152. template,
  153. **(
  154. {"user_name": user.get("name"), "user_location": user.get("location")}
  155. if user
  156. else {}
  157. ),
  158. )
  159. return template
  160. def tags_generation_template(
  161. template: str, messages: list[dict], user: Optional[dict] = None
  162. ) -> str:
  163. prompt = get_last_user_message(messages)
  164. template = replace_prompt_variable(template, prompt)
  165. template = replace_messages_variable(template, messages)
  166. template = prompt_template(
  167. template,
  168. **(
  169. {"user_name": user.get("name"), "user_location": user.get("location")}
  170. if user
  171. else {}
  172. ),
  173. )
  174. return template
  175. def emoji_generation_template(
  176. template: str, prompt: str, user: Optional[dict] = None
  177. ) -> str:
  178. template = replace_prompt_variable(template, prompt)
  179. template = prompt_template(
  180. template,
  181. **(
  182. {"user_name": user.get("name"), "user_location": user.get("location")}
  183. if user
  184. else {}
  185. ),
  186. )
  187. return template
  188. def autocomplete_generation_template(
  189. template: str,
  190. prompt: str,
  191. messages: Optional[list[dict]] = None,
  192. type: Optional[str] = None,
  193. user: Optional[dict] = None,
  194. ) -> str:
  195. template = template.replace("{{TYPE}}", type if type else "")
  196. template = replace_prompt_variable(template, prompt)
  197. template = replace_messages_variable(template, messages)
  198. template = prompt_template(
  199. template,
  200. **(
  201. {"user_name": user.get("name"), "user_location": user.get("location")}
  202. if user
  203. else {}
  204. ),
  205. )
  206. return template
  207. def query_generation_template(
  208. template: str, messages: list[dict], user: Optional[dict] = None
  209. ) -> str:
  210. prompt = get_last_user_message(messages)
  211. template = replace_prompt_variable(template, prompt)
  212. template = replace_messages_variable(template, messages)
  213. template = prompt_template(
  214. template,
  215. **(
  216. {"user_name": user.get("name"), "user_location": user.get("location")}
  217. if user
  218. else {}
  219. ),
  220. )
  221. return template
  222. def moa_response_generation_template(
  223. template: str, prompt: str, responses: list[str]
  224. ) -> str:
  225. def replacement_function(match):
  226. full_match = match.group(0)
  227. start_length = match.group(1)
  228. end_length = match.group(2)
  229. middle_length = match.group(3)
  230. if full_match == "{{prompt}}":
  231. return prompt
  232. elif start_length is not None:
  233. return prompt[: int(start_length)]
  234. elif end_length is not None:
  235. return prompt[-int(end_length) :]
  236. elif middle_length is not None:
  237. middle_length = int(middle_length)
  238. if len(prompt) <= middle_length:
  239. return prompt
  240. start = prompt[: math.ceil(middle_length / 2)]
  241. end = prompt[-math.floor(middle_length / 2) :]
  242. return f"{start}...{end}"
  243. return ""
  244. template = re.sub(
  245. r"{{prompt}}|{{prompt:start:(\d+)}}|{{prompt:end:(\d+)}}|{{prompt:middletruncate:(\d+)}}",
  246. replacement_function,
  247. template,
  248. )
  249. responses = [f'"""{response}"""' for response in responses]
  250. responses = "\n\n".join(responses)
  251. template = template.replace("{{responses}}", responses)
  252. return template
  253. def tools_function_calling_generation_template(template: str, tools_specs: str) -> str:
  254. template = template.replace("{{TOOLS}}", tools_specs)
  255. return template