feat(output): separate code and docs into subdirectories

Group copied project files under a dedicated code/ directory
Group generated analysis, manifest, and overview under docs/
This commit is contained in:
2026-04-02 10:36:29 +08:00
parent 5ba2e3217a
commit eef3464257
4 changed files with 56 additions and 27 deletions

View File

@@ -2,6 +2,8 @@
`uipath-explainator` 会按 UiPath 项目入口 `main.xaml` 递归提取 `Invoke Workflow``Invoke VBA`,复制到指定工作区,删除 `<ui:CommentOut>` 里的无效代码,再重新扫描并裁掉不再使用的文件,最后逐个调用 Gemini 生成说明文档。 `uipath-explainator` 会按 UiPath 项目入口 `main.xaml` 递归提取 `Invoke Workflow``Invoke VBA`,复制到指定工作区,删除 `<ui:CommentOut>` 里的无效代码,再重新扫描并裁掉不再使用的文件,最后逐个调用 Gemini 生成说明文档。
输出会自动拆成独立的 `code/``docs/` 两个子目录,避免说明文档和项目文件混在一起。
## 安装 ## 安装
```bash ```bash
@@ -26,7 +28,7 @@ GEMINI_MODEL=gemini-2.5-flash
```bash ```bash
./.venv/bin/python -m uipath_explainator \ ./.venv/bin/python -m uipath_explainator \
/path/to/uipath-project \ /path/to/uipath-project \
--output-dir workspace/project-copy \ --output-dir workspace/project-bundle \
--force --force
``` ```
@@ -40,11 +42,12 @@ GEMINI_MODEL=gemini-2.5-flash
## 输出内容 ## 输出内容
输出目录会保留原项目的相对目录结构,并额外生成 输出目录现在是一个 bundle 目录,里面固定包含
- `manifest.json`: 扫描、剪枝、告警和分析结果清单 - `code/`: 复制出来并经过清理、剪枝后的项目文件,保留原项目相对目录结构
- `OVERVIEW.md`: 总览说明书,除了文件数量,还会解释“初次扫描 -> 清理 Comment Out -> 二次扫描 -> 剪枝 -> Gemini 分析”的整条流水线逻辑 - `docs/manifest.json`: 扫描、剪枝、告警和分析结果清单
- `*.analysis.md`: 每个最终保留文件对应的 Gemini 咨询式说明,固定包含: - `docs/OVERVIEW.md`: 总览说明书,除了文件数量,还会解释“初次扫描 -> 清理 Comment Out -> 二次扫描 -> 剪枝 -> Gemini 分析”的整条流水线逻辑
- `docs/*.analysis.md`: 每个最终保留文件对应的 Gemini 咨询式说明,固定包含:
- 文件定位:这个文件在整个流程中的角色、触发方式、核心作用 - 文件定位:这个文件在整个流程中的角色、触发方式、核心作用
- 流程拆解:按步骤说明它是怎么执行的 - 流程拆解:按步骤说明它是怎么执行的
- 关键逻辑:条件判断、调用关系、边界情况 - 关键逻辑:条件判断、调用关系、边界情况

View File

@@ -11,7 +11,12 @@ from .pipeline import ProjectPipeline
def build_parser() -> ArgumentParser: def build_parser() -> ArgumentParser:
parser = ArgumentParser(description="Extract and explain UiPath project dependencies.") parser = ArgumentParser(description="Extract and explain UiPath project dependencies.")
parser.add_argument("project_dir", type=Path, help="UiPath project root directory") parser.add_argument("project_dir", type=Path, help="UiPath project root directory")
parser.add_argument("--output-dir", type=Path, default=Path("workspace"), help="Copied project output directory") parser.add_argument(
"--output-dir",
type=Path,
default=Path("workspace"),
help="Output bundle directory containing separate code/ and docs/ folders",
)
parser.add_argument("--entry", default="main.xaml", help="Entry XAML file name") parser.add_argument("--entry", default="main.xaml", help="Entry XAML file name")
parser.add_argument("--env-file", type=Path, default=Path(".env"), help="Environment file for Gemini config") parser.add_argument("--env-file", type=Path, default=Path(".env"), help="Environment file for Gemini config")
parser.add_argument("--model", help="Override GEMINI_MODEL") parser.add_argument("--model", help="Override GEMINI_MODEL")
@@ -44,6 +49,8 @@ def main(argv: list[str] | None = None) -> int:
report = pipeline.run(analyzer=analyzer) report = pipeline.run(analyzer=analyzer)
print(f"Output written to: {report.output_root}") print(f"Output written to: {report.output_root}")
print(f"Code written to: {report.code_root}")
print(f"Docs written to: {report.docs_root}")
print(f"Final files: {len(report.final_files)}") print(f"Final files: {len(report.final_files)}")
print(f"Pruned files: {len(report.pruned_files)}") print(f"Pruned files: {len(report.pruned_files)}")
print(f"Warnings: {len(report.warnings)}") print(f"Warnings: {len(report.warnings)}")

View File

@@ -12,6 +12,8 @@ from .scanner import ScanResult, crawl_dependencies, find_entry_file, read_text,
class PipelineReport: class PipelineReport:
project_root: Path project_root: Path
output_root: Path output_root: Path
code_root: Path
docs_root: Path
entry_file: Path entry_file: Path
initial_files: list[Path] initial_files: list[Path]
final_files: list[Path] final_files: list[Path]
@@ -25,6 +27,8 @@ class PipelineReport:
{ {
"project_root": self.project_root.as_posix(), "project_root": self.project_root.as_posix(),
"output_root": self.output_root.as_posix(), "output_root": self.output_root.as_posix(),
"code_root": self.code_root.as_posix(),
"docs_root": self.docs_root.as_posix(),
"entry_file": self.entry_file.as_posix(), "entry_file": self.entry_file.as_posix(),
"initial_files": [item.as_posix() for item in self.initial_files], "initial_files": [item.as_posix() for item in self.initial_files],
"final_files": [item.as_posix() for item in self.final_files], "final_files": [item.as_posix() for item in self.final_files],
@@ -42,6 +46,8 @@ class ProjectPipeline:
def __init__(self, project_root: Path, output_root: Path, entry_name: str, force: bool = False) -> None: def __init__(self, project_root: Path, output_root: Path, entry_name: str, force: bool = False) -> None:
self.project_root = project_root.resolve() self.project_root = project_root.resolve()
self.output_root = output_root.resolve() self.output_root = output_root.resolve()
self.code_root = self.output_root / "code"
self.docs_root = self.output_root / "docs"
self.entry_name = entry_name self.entry_name = entry_name
self.force = force self.force = force
@@ -56,8 +62,8 @@ class ProjectPipeline:
self._copy_files(initial_rel_files) self._copy_files(initial_rel_files)
cleaned_files = self._clean_copied_xaml_files(initial_rel_files) cleaned_files = self._clean_copied_xaml_files(initial_rel_files)
final_scan = crawl_dependencies(self.output_root, self.output_root / entry_rel) final_scan = crawl_dependencies(self.code_root, self.code_root / entry_rel)
final_rel_files = self._relative_files(final_scan, self.output_root) final_rel_files = self._relative_files(final_scan, self.code_root)
pruned_files = self._prune_unused_files(initial_rel_files, final_rel_files) pruned_files = self._prune_unused_files(initial_rel_files, final_rel_files)
analysis_files = self._write_analysis(final_rel_files, analyzer) analysis_files = self._write_analysis(final_rel_files, analyzer)
@@ -66,6 +72,8 @@ class ProjectPipeline:
report = PipelineReport( report = PipelineReport(
project_root=self.project_root, project_root=self.project_root,
output_root=self.output_root, output_root=self.output_root,
code_root=self.code_root,
docs_root=self.docs_root,
entry_file=entry_rel, entry_file=entry_rel,
initial_files=initial_rel_files, initial_files=initial_rel_files,
final_files=final_rel_files, final_files=final_rel_files,
@@ -82,12 +90,13 @@ class ProjectPipeline:
if not self.force: if not self.force:
raise FileExistsError(f"Output directory already exists: {self.output_root}") raise FileExistsError(f"Output directory already exists: {self.output_root}")
rmtree(self.output_root) rmtree(self.output_root)
self.output_root.mkdir(parents=True, exist_ok=True) self.code_root.mkdir(parents=True, exist_ok=True)
self.docs_root.mkdir(parents=True, exist_ok=True)
def _copy_files(self, relative_files: list[Path]) -> None: def _copy_files(self, relative_files: list[Path]) -> None:
for relative_path in relative_files: for relative_path in relative_files:
source = self.project_root / relative_path source = self.project_root / relative_path
destination = self.output_root / relative_path destination = self.code_root / relative_path
destination.parent.mkdir(parents=True, exist_ok=True) destination.parent.mkdir(parents=True, exist_ok=True)
copy2(source, destination) copy2(source, destination)
@@ -96,7 +105,7 @@ class ProjectPipeline:
for relative_path in relative_files: for relative_path in relative_files:
if relative_path.suffix.lower() != ".xaml": if relative_path.suffix.lower() != ".xaml":
continue continue
output_file = self.output_root / relative_path output_file = self.code_root / relative_path
original = read_text(output_file) original = read_text(output_file)
updated = strip_comment_out_blocks(original) updated = strip_comment_out_blocks(original)
if updated != original: if updated != original:
@@ -110,7 +119,7 @@ class ProjectPipeline:
for relative_path in initial_files: for relative_path in initial_files:
if relative_path in final_set: if relative_path in final_set:
continue continue
target = self.output_root / relative_path target = self.code_root / relative_path
if target.exists(): if target.exists():
target.unlink() target.unlink()
pruned.append(relative_path) pruned.append(relative_path)
@@ -119,7 +128,7 @@ class ProjectPipeline:
def _cleanup_empty_dirs(self) -> None: def _cleanup_empty_dirs(self) -> None:
directories = sorted( directories = sorted(
[path for path in self.output_root.rglob("*") if path.is_dir()], [path for path in self.code_root.rglob("*") if path.is_dir()],
key=lambda item: len(item.parts), key=lambda item: len(item.parts),
reverse=True, reverse=True,
) )
@@ -134,17 +143,17 @@ class ProjectPipeline:
output_files: list[Path] = [] output_files: list[Path] = []
for relative_path in self._ordered_files(final_files): for relative_path in self._ordered_files(final_files):
content = read_text(self.output_root / relative_path) content = read_text(self.code_root / relative_path)
analysis = analyzer.analyze(relative_path, content) analysis = analyzer.analyze(relative_path, content)
analysis_path = self.output_root / f"{relative_path.as_posix()}.analysis.md" analysis_path = self.docs_root / f"{relative_path.as_posix()}.analysis.md"
analysis_path.parent.mkdir(parents=True, exist_ok=True) analysis_path.parent.mkdir(parents=True, exist_ok=True)
analysis_path.write_text(analysis, encoding="utf-8") analysis_path.write_text(analysis, encoding="utf-8")
output_files.append(Path(f"{relative_path.as_posix()}.analysis.md")) output_files.append(Path(f"{relative_path.as_posix()}.analysis.md"))
return output_files return output_files
def _write_report_files(self, report: PipelineReport) -> None: def _write_report_files(self, report: PipelineReport) -> None:
(self.output_root / "manifest.json").write_text(report.to_json(), encoding="utf-8") (self.docs_root / "manifest.json").write_text(report.to_json(), encoding="utf-8")
(self.output_root / "OVERVIEW.md").write_text(self._build_overview(report), encoding="utf-8") (self.docs_root / "OVERVIEW.md").write_text(self._build_overview(report), encoding="utf-8")
def _build_overview(self, report: PipelineReport) -> str: def _build_overview(self, report: PipelineReport) -> str:
warnings = "\n".join(f"- {item}" for item in report.warnings) or "- 无" warnings = "\n".join(f"- {item}" for item in report.warnings) or "- 无"
@@ -155,6 +164,8 @@ class ProjectPipeline:
- Project Root: `{report.project_root.as_posix()}` - Project Root: `{report.project_root.as_posix()}`
- Output Root: `{report.output_root.as_posix()}` - Output Root: `{report.output_root.as_posix()}`
- Code Root: `{report.code_root.as_posix()}`
- Docs Root: `{report.docs_root.as_posix()}`
- Entry File: `{report.entry_file.as_posix()}` - Entry File: `{report.entry_file.as_posix()}`
- Initial Files: {len(report.initial_files)} - Initial Files: {len(report.initial_files)}
- Final Files: {len(report.final_files)} - Final Files: {len(report.final_files)}
@@ -167,18 +178,18 @@ class ProjectPipeline:
## Processing Logic ## Processing Logic
1. **Initial Scan**: 从入口文件开始递归扫描 `Invoke Workflow` 与 `Invoke VBA`,得到初始依赖集合。这个阶段会把 `Comment Out` 里的引用也一起算进去,因为此时还未清理失效代码。 1. **Initial Scan**: 从入口文件开始递归扫描 `Invoke Workflow` 与 `Invoke VBA`,得到初始依赖集合。这个阶段会把 `Comment Out` 里的引用也一起算进去,因为此时还未清理失效代码。
2. **Copy to Workspace**: 将初始依赖集合完整复制到输出目录,保持相对目录结构不变,确保后续裁剪只发生在工作区副本中。 2. **Copy to Workspace**: 将初始依赖集合完整复制到 `code/` 目录,保持相对目录结构不变,确保后续裁剪只发生在代码副本中。
3. **Comment Cleanup**: 对已复制的 XAML 删除 `<ui:CommentOut>` 代码块,让后续扫描只看到当前真正生效的流程节点。 3. **Comment Cleanup**: 对已复制的 XAML 删除 `<ui:CommentOut>` 代码块,让后续扫描只看到当前真正生效的流程节点。
4. **Rescan After Cleanup**: 以清理后的入口文件重新递归扫描,得到清理后的实际依赖集合。 4. **Rescan After Cleanup**: 以清理后的入口文件重新递归扫描,得到清理后的实际依赖集合。
5. **Prune Unused Files**: 删除首次扫描能到达、但二次扫描已不可达的文件。这些文件通常来自被注释掉的工作流、VBA或清理后失效的调用链。 5. **Prune Unused Files**: 删除 `code/` 目录中首次扫描能到达、但二次扫描已不可达的文件。这些文件通常来自被注释掉的工作流、VBA或清理后失效的调用链。
6. **Gemini Analysis**: 仅针对最终保留文件生成 `*.analysis.md`,让说明文档与实际可执行流程保持一致。 6. **Gemini Analysis**: 仅针对最终保留文件在 `docs/` 目录生成 `*.analysis.md`,让说明文档与实际可执行流程保持一致。
## How To Read This Output ## How To Read This Output
- `Initial Files`: 清理 `<ui:CommentOut>` 前扫描到的所有文件,代表“理论上被引用过”的集合。 - `Initial Files`: 清理 `<ui:CommentOut>` 前扫描到的所有文件,代表“理论上被引用过”的集合。
- `Final Files`: 清理后仍可达的文件,代表“当前实际流程会用到”的集合。 - `Final Files`: 清理后 `code/` 目录中仍可达的文件,代表“当前实际流程会用到”的集合。
- `Cleaned XAML Files`: 被移除 `<ui:CommentOut>` 代码块的文件。 - `Cleaned XAML Files`: 被移除 `<ui:CommentOut>` 代码块的文件。
- `Pruned Files`: 只在失效分支中出现、已从输出目录删除的文件。 - `Pruned Files`: 只在失效分支中出现、已从 `code/` 目录删除的文件。
- `Analysis Files`: Gemini 生成的逐文件说明,重点解释该文件的流程、逻辑、数据与风险点。 - `Analysis Files`: `docs/` 目录下 Gemini 生成的逐文件说明,重点解释该文件的流程、逻辑、数据与风险点。
## Cleaned XAML Files ## Cleaned XAML Files
{cleaned} {cleaned}

View File

@@ -83,6 +83,8 @@ class PipelineTests(unittest.TestCase):
tmp_path = Path(tmp) tmp_path = Path(tmp)
project_root = tmp_path / "project" project_root = tmp_path / "project"
output_root = tmp_path / "workspace" output_root = tmp_path / "workspace"
code_root = output_root / "code"
docs_root = output_root / "docs"
(project_root / "Flows").mkdir(parents=True) (project_root / "Flows").mkdir(parents=True)
(project_root / "Flows" / "Active.xaml").write_text(ACTIVE_XAML, encoding="utf-8") (project_root / "Flows" / "Active.xaml").write_text(ACTIVE_XAML, encoding="utf-8")
(project_root / "Flows" / "Old.xaml").write_text(OLD_XAML, encoding="utf-8") (project_root / "Flows" / "Old.xaml").write_text(OLD_XAML, encoding="utf-8")
@@ -104,15 +106,21 @@ class PipelineTests(unittest.TestCase):
self.assertIn("Scripts/Keep.bas", final_files) self.assertIn("Scripts/Keep.bas", final_files)
self.assertNotIn("Flows/Old.xaml", final_files) self.assertNotIn("Flows/Old.xaml", final_files)
self.assertNotIn("Scripts/Drop.bas", final_files) self.assertNotIn("Scripts/Drop.bas", final_files)
self.assertFalse((output_root / "Flows" / "Old.xaml").exists()) self.assertEqual(report.code_root, code_root.resolve())
self.assertFalse((output_root / "Scripts" / "Drop.bas").exists()) self.assertEqual(report.docs_root, docs_root.resolve())
self.assertTrue((output_root / "Flows" / "Active.xaml.analysis.md").exists()) self.assertFalse((code_root / "Flows" / "Old.xaml").exists())
self.assertFalse((code_root / "Scripts" / "Drop.bas").exists())
self.assertTrue((code_root / "Flows" / "Active.xaml").exists())
self.assertTrue((docs_root / "Flows" / "Active.xaml.analysis.md").exists())
self.assertTrue((docs_root / "manifest.json").exists())
overview = (output_root / "OVERVIEW.md").read_text(encoding="utf-8") overview = (docs_root / "OVERVIEW.md").read_text(encoding="utf-8")
self.assertIn("## Processing Logic", overview) self.assertIn("## Processing Logic", overview)
self.assertIn("Initial Scan", overview) self.assertIn("Initial Scan", overview)
self.assertIn("## How To Read This Output", overview) self.assertIn("## How To Read This Output", overview)
self.assertIn("## Cleaned XAML Files", overview) self.assertIn("## Cleaned XAML Files", overview)
self.assertIn("Code Root", overview)
self.assertIn("Docs Root", overview)
if __name__ == "__main__": if __name__ == "__main__":