Upload data2/instruction_generation/schemas.py with huggingface_hub

data2/instruction_generation/schemas.py

@@ -0,0 +1,57 @@
+"""
+Pydantic data structure definitions for LLM structured output
+"""
+
+from pydantic import BaseModel, Field
+from typing import List, Optional
+
+
+class READMESummary(BaseModel):
+    """Structured output for README summary"""
+    project_overview: str = Field(description="One-sentence project overview")
+    main_features: str = Field(description="Main features list (bullet points)")
+    dependencies: str = Field(description="Dependencies and environment requirements")
+    how_to_run: str = Field(description="How to run/install")
+    directory_structure: str = Field(description="Directory structure highlights")
+    scientific_computing_related: Optional[str] = Field(
+        default=None, 
+        description="Scientific computing/chemistry related features (if any)"
+    )
+    typical_input_output: Optional[str] = Field(
+        default=None, 
+        description="Typical input/output examples (if any)"
+    )
+    notes: Optional[str] = Field(
+        default=None, 
+        description="Notes or special instructions"
+    )
+
+
+class FunctionInfo(BaseModel):
+    """Function information"""
+    function_name: str = Field(description="Function name")
+    function_start_line: int = Field(description="Function start line number (1-indexed, inclusive)")
+    function_end_line: int = Field(description="Function end line number (1-indexed, inclusive)")
+    function_body: str = Field(description="Complete function code body")
+    doc_start_line: Optional[int] = Field(
+        default=None, 
+        description="Documentation comment start line number (if any, 1-indexed, inclusive)"
+    )
+    doc_end_line: Optional[int] = Field(
+        default=None, 
+        description="Documentation comment end line number (if any, 1-indexed, inclusive)"
+    )
+
+
+class FileParseResult(BaseModel):
+    """Parse result for a single code file"""
+    language: str = Field(description="Programming language (e.g., python, cpp, java)")
+    file_path: str = Field(description="Relative file path")
+    dependencies: List[str] = Field(
+        default_factory=list, 
+        description="File-level dependency list (import/include/use/require, etc.)"
+    )
+    functions: List[FunctionInfo] = Field(
+        default_factory=list, 
+        description="List of all functions in the file"
+    )