add main.py

main.py

@@ -1,44 +1,104 @@
-import os
+"""Task 69dd4221f309a98be0006b2e – Structured task card parser.
+
+The script accepts a single natural‑language description of a course task and
+returns a validated Pydantic model instance.  It demonstrates a one‑shot
+prompting pattern with a structured output parser.
+
+Usage:
+    python main.py "<task description>"
+
+The script prints the raw model dump and a short human‑readable summary.
+"""
+
+import sys
+from typing import List
+
 from pydantic import BaseModel, Field
 from langchain_core.prompts import PromptTemplate
 from langchain_openai import ChatOpenAI
 from langchain_core.output_parsers import PydanticOutputParser
 
+# ---------------------------------------------------------------------------
+# 1. Pydantic model
+# ---------------------------------------------------------------------------
 class TaskCard(BaseModel):
-    title: str = Field(..., description="Краткое название задачи")
+    """Compact representation of a course task.
-    subject: str = Field(..., description="Предмет или область")
-    deadline_hint: str = Field(..., description="Краткая подсказка о сроке")
-    deliverable_type: str = Field(..., description="Тип сдачи: отчёт, код, презентация и т.д.")
-    grading_hints: list[str] = Field(..., description="Список критериев оценки")
 
-# LLM configuration – BroJS
+    All fields are optional because the model may not be able to infer every
-llm = ChatOpenAI(
+    piece of information from a very short description.  The parser will
-    model="openai/gpt-oss-20b:free",
+    still return a valid instance – missing values will be ``None``.
-    base_url="https://platform.brojs.ru/jrnl-bh/api/inference/v1",
+    """
-    api_key=os.getenv("JOURNAL_MCP_PAT"),
+
-    temperature=0.5,
+    title: str | None = Field(None, description="Short title of the task")
+    subject: str | None = Field(None, description="Subject or topic of the task")
+    deadline_hint: str | None = Field(
+        None, description="Free‑form hint about the deadline (e.g. 'к пятнице')"
+    )
+    deliverable_type: str | None = Field(
+        None, description="What is expected to be submitted (report, code, etc.)"
+    )
+    grading_hints: List[str] | None = Field(
+        None, description="List of hints mentioned about grading"
     )
 
+# ---------------------------------------------------------------------------
+# 2. Prompt + chain
+# ---------------------------------------------------------------------------
 parser = PydanticOutputParser(pydantic_object=TaskCard)
 
-prompt_template = """\nНиже приведена формулировка задания от преподавателя.\nВаша задача – вернуть данные в формате JSON, соответствующем модели TaskCard.\n{format_instructions}\n\nФормулировка: {input_text}\n"""
+prompt_template = """You are a helper that extracts structured information from a
+course task description.  Return a JSON object that matches the following
+schema:
+
+{schema}
+
+The input description is:
+
+{description}
+
+Respond ONLY with the JSON object.  Do not add any extra text.
+"""
 
 prompt = PromptTemplate(
     template=prompt_template,
-    input_variables=["input_text"],
+    input_variables=["description"],
-    partial_variables={"format_instructions": parser.get_format_instructions()},
+    partial_variables={"schema": parser.get_format_instructions()},
 )
 
+# LLM – use the default OpenAI OSS endpoint via environment variables
+llm = ChatOpenAI(model="openai/gpt-oss-20b:free", temperature=0.2)
+
 chain = prompt | llm | parser
 
-if __name__ == "__main__":
+# ---------------------------------------------------------------------------
-    example = "Сдайте к пятнице мини-отчёт по LangChain: 2 страницы, упор на агентов. Оценка: за полноту и за пример кода."
+# 3. Main entry point
-    result = chain.invoke({"input_text": example})
+# ---------------------------------------------------------------------------
-    print("\n--- Parsed Result ---")
+
-    print(result.model_dump())
+def main() -> None:
+    if len(sys.argv) < 2:
+        print("Usage: python main.py '<task description>'")
+        sys.exit(1)
+
+    description = sys.argv[1]
+    result: TaskCard = chain.invoke({"description": description})
+
+    # Print raw model dump
+    print("\n--- Parsed model ---")
+    print(result.model_dump(indent=2))
+
+    # Human‑readable summary
     print("\n--- Summary ---")
-    print(f"Title: {result.title}")
+    print(f"Title: {result.title or 'N/A'}")
-    print(f"Subject: {result.subject}")
+    print(f"Subject: {result.subject or 'N/A'}")
-    print(f"Deadline hint: {result.deadline_hint}")
+    print(f"Deadline hint: {result.deadline_hint or 'N/A'}")
-    print(f"Deliverable type: {result.deliverable_type}")
+    print(f"Deliverable: {result.deliverable_type or 'N/A'}")
-    print(f"Grading hints: {', '.join(result.grading_hints)}")
+    if result.grading_hints:
+        print("Grading hints:")
+        for hint in result.grading_hints:
+            print(f"- {hint}")
+    else:
+        print("Grading hints: N/A")
+
+
+if __name__ == "__main__":
+    main()