Commit fb595481 authored by Jan Reimes's avatar Jan Reimes
Browse files

🛠️ chore(cli): add new workspace-related CLI options and refactor command arguments

parent 6451d1f4

src/tdoc_crawler/cli/_workspace_commands.py

+25 −15
Original line number Diff line number Diff line
@@ -18,15 +18,27 @@ from tdoc_crawler.cli.args import ( 
    AgendaPatternExcludeOption,

    AgendaPatternOption,

    AutoCrawlSpecsOption,

    DeleteArtifactsOption,

    DeleteLlmWikiOption,

    DryRunOption,

    EndDateOption,

    IncludeInactiveOption,

    LimitOption,

    MdYamlFrontmatterOption,

    ProcessLimitOption,

    ProfileOption,

    ReleaseOption,

    SkipExistingOption,

    SourceKindOption,

    SourcePatternExcludeOption,

    SourcePatternOption,

    StartDateOption,

    TitlePatternExcludeOption,

    TitlePatternOption,

    WorkspaceDeleteForceOption,

    WorkspaceItemsArgument,

    WorkspaceNameOption,

    WorkspaceProcessForceOption,

)

from tdoc_crawler.cli.formatting import TableColumnSpec, print_structured_output

from tdoc_crawler.config import PathConfig, resolve_cache_manager
@@ -119,9 +131,9 @@ def workspace_deactivate() -> None: 
@app.command("delete", help="Delete a workspace and optionally its artifacts.")

def workspace_delete(

    workspace_name: str = typer.Argument(..., help="Workspace name"),

    force: bool = typer.Option(False, "--force", help="Permanently delete workspace and all artifacts"),

    delete_artifacts: bool = typer.Option(False, "--delete-artifacts", help="Delete all workspace artifacts for members"),

    delete_llm_wiki: bool = typer.Option(False, "--delete-llm-wiki", help="Delete the .llm-wiki folder for this workspace"),

    force: WorkspaceDeleteForceOption = False,

    delete_artifacts: DeleteArtifactsOption = False,

    delete_llm_wiki: DeleteLlmWikiOption = False,

) -> None:

    """Permanently delete a workspace and all associated files."""

    normalized = normalize_workspace_name(workspace_name)
@@ -147,7 +159,7 @@ def workspace_delete( 
@app.command("members", help="List workspace members.")

def workspace_members(

    workspace_name: str = typer.Argument(None, help="Workspace name (default: active workspace)"),

    include_inactive: bool = typer.Option(False, "--include-inactive", help="Include inactive members"),

    include_inactive: IncludeInactiveOption = False,

) -> None:

    """List members of a workspace."""

    if workspace_name is None:
@@ -187,14 +199,11 @@ def _should_skip_member( 
@app.command("process", help="Process workspace members.")

def workspace_process(

    workspace_name: str = typer.Argument(None, help="Workspace name (default: active workspace)"),

    force: bool = typer.Option(False, "--force", help="Re-process existing artifacts"),

    limit: int = typer.Option(None, "--limit", help="Limit number of members to process"),

    skip_existing: bool = typer.Option(False, "--skip-existing", help="Skip members that already have artifacts"),

    profile: str = typer.Option(

        DEFAULT_EXTRACTION_PROFILE.value,

        "--profile",

        help="Extraction profile: pdf-only, default, or advanced",

    ),

    force: WorkspaceProcessForceOption = False,

    limit: ProcessLimitOption = None,

    skip_existing: SkipExistingOption = False,

    profile: ProfileOption = DEFAULT_EXTRACTION_PROFILE.value,

    md_yaml_frontmatter: MdYamlFrontmatterOption = True,

) -> None:

    """Extract structured data from all workspace members."""

    if workspace_name is None:
@@ -250,6 +259,7 @@ def workspace_process( 
                profile=extraction_profile,

                force=force,

                release=member.release,

                md_yaml_frontmatter=md_yaml_frontmatter,

            )

            if result_path:

                console.print(f"[green]  Processed {source_id} -> {result_path.name}[/green]")
@@ -290,8 +300,8 @@ def _resolve_spec_release_for_add( 
@app.command("add", help="Add documents to an existing workspace.")

def workspace_add(

    items: WorkspaceItemsArgument = None,

    workspace: str | None = typer.Option(None, "-w", "--workspace", help="Workspace name (default: active workspace)"),

    kind: str = typer.Option("tdoc", "--kind", help="Source kind: tdoc, spec, or other"),

    workspace: WorkspaceNameOption = None,

    kind: SourceKindOption = "tdoc",

    # Filter options for batch-adding from DB query

    agenda: AgendaPatternOption = None,

    agenda_ex: AgendaPatternExcludeOption = None,
@@ -408,7 +418,7 @@ def workspace_add( 
@app.command("clear-invalid", help="Remove members with invalid or missing source paths.")

def workspace_clear_invalid(

    workspace_name: str = typer.Argument(None, help="Workspace name (default: active workspace)"),

    dry_run: bool = typer.Option(False, "--dry-run", help="Show what would be removed without removing"),

    dry_run: DryRunOption = False,

) -> None:

    """Remove members whose source path no longer exists."""

    if workspace_name is None:

src/tdoc_crawler/cli/args.py

+51 −0
Original line number Diff line number Diff line
@@ -156,3 +156,54 @@ NoProgressOption = Annotated[ 
    bool,

    typer.Option("--no-progress", help="Disable progress bar (useful for scripts and CI)"),

]



MdYamlFrontmatterOption = Annotated[

    bool,

    typer.Option("--md-yaml-frontmatter/--no-md-yaml-frontmatter", help="Include YAML frontmatter in generated Markdown"),

]



# Options - Workspaces

WorkspaceDeleteForceOption = Annotated[

    bool,

    typer.Option("--force", help="Permanently delete workspace and all artifacts"),

]

DeleteArtifactsOption = Annotated[

    bool,

    typer.Option("--delete-artifacts", help="Delete all workspace artifacts for members"),

]

DeleteLlmWikiOption = Annotated[

    bool,

    typer.Option("--delete-llm-wiki", help="Delete the .llm-wiki folder for this workspace"),

]

IncludeInactiveOption = Annotated[

    bool,

    typer.Option("--include-inactive", help="Include inactive members"),

]

WorkspaceProcessForceOption = Annotated[

    bool,

    typer.Option("--force", help="Re-process existing artifacts"),

]

ProcessLimitOption = Annotated[

    int | None,

    typer.Option("--limit", help="Limit number of members to process"),

]

SkipExistingOption = Annotated[

    bool,

    typer.Option("--skip-existing", help="Skip members that already have artifacts"),

]

ProfileOption = Annotated[

    str,

    typer.Option("--profile", help="Extraction profile: pdf-only, default, or advanced"),

]

WorkspaceNameOption = Annotated[

    str | None,

    typer.Option("-w", "--workspace", help="Workspace name (default: active workspace)"),

]

SourceKindOption = Annotated[

    str,

    typer.Option("--kind", help="Source kind: tdoc, spec, or other"),

]

DryRunOption = Annotated[

    bool,

    typer.Option("--dry-run", help="Show what would be removed without removing"),

]