CoreCodeBench-Multi

# Multi Testcases for CoreCodeBench ## File Explanation + `CoreCodeBench_Multi.jsonl` Multi test cases for CoreCodeBench. + `CoreCodeBench_Difficult.jsonl` More difficult version for CoreCodeBench multi test cases . ## Key Explanation | Key | Meaning/Description | | ------------------ | ------------------------------------------------------------ | | **id** | A list of unique identifiers for the functions to be completed, typically in the format `module.path.Class::function`. | | **project** | The name of the project this data is associated with. | | **origin_file** | A list of file paths indicating where each function or method is defined in the source code. | | **test_list** | A list of file paths for test scripts that are related to or used for testing the functions/methods. | | **prob_info** | A list of dictionaries, each containing detailed information about a function or method, such as its starting and ending line numbers, and the actual code (`func_code`). Each dictionary includes: `class_start_lineno`/`class_end_lineno`: Line numbers where the class begins and ends. `func_start_lineno`/`func_end_lineno`: Line numbers where the function begins and ends. `func_code`: The source code of the function as a string. | | **type** | A list indicating the type or category of the functions/methods (e.g., "function_empty"). | | **node** | A list of fully qualified names (with module and class) for each function/method. | | **language** | The programming language used. | | **toolfunc_count** | The number of tool-related functions in the data. | | **func_count** | The total number of atomic functions in the data. | | **pytest_info** | A dictionary with information about pytest test results: `total_num` is the total number of unit tests, while `base_passed_num` is the number of base tests that passed. |# CoreCodeBench 多测试用例集 ## 文件说明 + `CoreCodeBench_Multi.jsonl`：CoreCodeBench 配套多测试用例文件。 + `CoreCodeBench_Difficult.jsonl`：CoreCodeBench 多测试用例的高难度版本。 ## 字段说明 | 字段名 | 含义/描述 | | ------ | --------- | | **id** | 待补全函数的唯一标识符列表，格式通常为 `module.path.Class::function`。 | | **project** | 该数据集关联的项目名称。 | | **origin_file** | 文件路径列表，用于标注每个函数或方法在源代码中的定义位置。 | | **test_list** | 用于测试目标函数/方法的相关测试脚本的文件路径列表。 | | **prob_info** | 字典列表，每个字典包含单个函数或方法的详细信息，例如其起止行号与实际代码（`func_code`）。每个字典包含以下子字段：<br> - `class_start_lineno`/`class_end_lineno`：类的起始与结束行号；<br> - `func_start_lineno`/`func_end_lineno`：函数的起始与结束行号；<br> - `func_code`：以字符串形式存储的函数源代码。 | | **type** | 函数/方法的类型或类别列表（例如 `"function_empty"`）。 | | **node** | 每个函数/方法的全限定名称列表（包含模块与类信息）。 | | **language** | 所用的编程语言。 | | **toolfunc_count** | 数据集中工具相关函数的总数量。 | | **func_count** | 数据集中原子函数的总数量。 | | **pytest_info** | 存储 pytest 测试结果信息的字典：`total_num` 为单元测试总数量，`base_passed_num` 为通过的基础测试数量。

搜集汇总

背景与挑战

背景概述

CoreCodeBench-Multi数据集包含用于CoreCodeBench的多测试用例，提供两个JSONL文件：一个标准版本和一个更难的版本。数据集中包括函数标识、项目信息、源代码路径、测试脚本和详细的函数信息，如代码行数和类型，以及pytest测试结果统计。