Upload data2/instruction_generation/prompts/function_extract.txt with huggingface_hub

data2/instruction_generation/prompts/function_extract.txt

@@ -0,0 +1,48 @@
+You are a professional code analysis tool. Please analyze the following code file and extract file-level dependencies and all function information.
+
+Project Context:
+{project_context}
+
+Code file path: {file_path}
+Programming language: {language}
+
+Below is the code file content:
+
+```{language}
+{code_content}
+```
+
+Requirements:
+1. Extract file-level dependencies: Find all import/include/use/require statements, list external packages, modules, and libraries (only external dependencies, excluding relative imports within the same project)
+2. Extract all functions:
+   - Function name
+   - Function start line number and end line number (1-indexed, inclusive)
+   - Complete function body (including function definition and body)
+   - Function documentation comment start and end line numbers (if any)
+3. For nested functions, also extract them (each independent function should be one record)
+4. If the code file is not function-structured (e.g., scripts, configuration files, etc.), the dependency list can be empty, and the functions list can also be empty
+
+Please output the result according to the following JSON Schema (must use JSON, do not include any other text):
+
+{{
+  "language": "Programming language (e.g., python, cpp, java, etc.)",
+  "file_path": "Relative file path",
+  "dependencies": ["dependency1", "dependency2", ...],
+  "functions": [
+    {{
+      "function_name": "function name",
+      "function_start_line": start_line_number,
+      "function_end_line": end_line_number,
+      "function_body": "complete function code body",
+      "doc_start_line": documentation start line number (if any, otherwise null),
+      "doc_end_line": documentation end line number (if any, otherwise null)
+    }},
+    ...
+  ]
+}}
+
+Note:
+- Output only the JSON object, do not include any prefix or suffix text
+- Line numbers must be accurate for subsequent code extraction
+- function_body should be the complete function code (from def/function/... to the end of the function)
+- If the file is too large to analyze completely, please note this in comments