Loading src/tdoc_crawler/cli/args.py +36 −30 Original line number Diff line number Diff line Loading @@ -12,38 +12,20 @@ from tdoc_crawler.logging import DEFAULT_LEVEL as LOGGING_DEFAULT_LEVEL DEFAULT_VERBOSITY = logging.getLevelName(LOGGING_DEFAULT_LEVEL) CacheDirOption = Annotated[Path | None, typer.Option("--cache-dir", "-c", help="Cache directory", envvar="TDC_CACHE_DIR")] # Arguments TDocIdsArgument = Annotated[list[str] | None, typer.Argument(help="TDoc identifiers to query")] TDocIdArgument = Annotated[str, typer.Argument(help="TDoc identifier to download and open")] CheckoutTDocIdsArgument = Annotated[list[str], typer.Argument(help="TDoc identifier(s) to checkout")] SpecArgument = Annotated[list[str] | None, typer.Argument(help="Spec number(s) to query (dotted or undotted)")] # Options - TDocs/Meetings WorkingGroupOption = Annotated[list[str] | None, typer.Option("--working-group", "-w", help="Filter by working group", envvar="TDC_WORKING_GROUP")] SubgroupOption = Annotated[list[str] | None, typer.Option("--sub-group", "-s", help="Filter by sub-working group", envvar="TDC_SUB_GROUP")] IncrementalOption = Annotated[bool, typer.Option("--incremental/--full", help="Toggle incremental mode")] ClearTDocsOption = Annotated[bool, typer.Option("--clear-tdocs", help="Clear all TDocs before crawling")] ClearSpecsOption = Annotated[bool, typer.Option("--clear-specs", help="Clear all specs before crawling")] ClearDbOption = Annotated[bool, typer.Option("--clear-db", help="Clear all meetings and TDocs before crawling")] CheckoutOption = Annotated[ bool, typer.Option("--checkout/--no-checkout", help="Download and extract metadata results to checkout folder", envvar="TDC_CHECKOUT") ] LimitTDocsOption = Annotated[int | None, typer.Option("--limit-tdocs", help="Limit number of TDocs", envvar="TDC_LIMIT_TDOCS")] LimitMeetingsOption = Annotated[int | None, typer.Option("--limit-meetings", help="Limit meetings overall", envvar="TDC_LIMIT_MEETINGS")] LimitMeetingsPerWgOption = Annotated[ int | None, typer.Option("--limit-meetings-per-wg", help="Limit meetings per working group", envvar="TDC_LIMIT_MEETINGS_PER_WG") ] LimitWgsOption = Annotated[int | None, typer.Option("--limit-wgs", help="Limit number of working groups")] WorkersOption = Annotated[int, typer.Option("--workers", help="Number of parallel subinterpreter workers", envvar="TDC_WORKERS")] OverallTimeoutOption = Annotated[ int | None, typer.Option("--overall-timeout", help="Maximum total crawl duration in seconds (None = unlimited)", envvar="TDC_OVERALL_TIMEOUT"), ] MaxRetriesOption = Annotated[int, typer.Option("--max-retries", help="HTTP retry attempts", envvar="TDC_MAX_RETRIES")] TimeoutOption = Annotated[int, typer.Option("--timeout", help="HTTP timeout seconds", envvar="TDC_TIMEOUT")] VerbosityOption = Annotated[str, typer.Option("--verbosity", "-v", help="Logging verbosity level", envvar="TDC_VERBOSITY")] TDocIdsArgument = Annotated[list[str] | None, typer.Argument(help="TDoc identifiers to query")] OutputFormatOption = Annotated[str, typer.Option("--output", "-o", help="Output format", envvar="TDC_OUTPUT")] FullMetadataOption = Annotated[bool, typer.Option("--full-metadata", help="Fetch full metadata instead of URL only")] UseWhatTheSpecOption = Annotated[ bool, typer.Option("--use-whatthespec/--no-use-whatthespec", help="Use WhatTheSpec API for fetching", envvar="TDC_USE_WHATTHESPEC") ] LimitOption = Annotated[int | None, typer.Option("--limit", "-l", help="Maximum number of rows")] OrderOption = Annotated[str, typer.Option("--order", help="Sort order (asc|desc)")] StartDateOption = Annotated[str | None, typer.Option("--start-date", help="Filter from ISO timestamp", envvar="TDC_START_DATE")] Loading @@ -52,6 +34,14 @@ NoFetchOption = Annotated[ bool, typer.Option("--no-fetch", help="Disable automatic fetching of missing TDocs from portal"), ] IncrementalOption = Annotated[bool, typer.Option("--incremental/--full", help="Toggle incremental mode")] LimitTDocsOption = Annotated[int | None, typer.Option("--limit-tdocs", help="Limit number of TDocs", envvar="TDC_LIMIT_TDOCS")] ClearTDocsOption = Annotated[bool, typer.Option("--clear-tdocs", help="Clear all TDocs before crawling")] OverallTimeoutOption = Annotated[ int | None, typer.Option("--overall-timeout", help="Maximum total crawl duration in seconds (None = unlimited)", envvar="TDC_OVERALL_TIMEOUT"), ] OutputFormatOption = Annotated[str, typer.Option("--output", "-o", help="Output format", envvar="TDC_OUTPUT")] EolUsernameOption = Annotated[str | None, typer.Option("--eol-username", help="ETSI Online account username", envvar="TDC_EOL_USERNAME")] EolPasswordOption = Annotated[str | None, typer.Option("--eol-password", help="ETSI Online account password", envvar="TDC_EOL_PASSWORD")] PromptCredentialsOption = Annotated[ Loading @@ -62,13 +52,12 @@ IncludeWithoutFilesOption = Annotated[ bool, typer.Option("--include-without-files", help="Include meetings without files URLs"), ] FullMetadataOption = Annotated[bool, typer.Option("--full-metadata", help="Fetch full metadata instead of URL only")] TDocIdArgument = Annotated[str, typer.Argument(help="TDoc identifier to download and open")] CheckoutTDocIdsArgument = Annotated[list[str], typer.Argument(help="TDoc identifier(s) to checkout")] ForceOption = Annotated[bool, typer.Option("--force", "-f", help="Re-download even if already checked out")] # Options - Specs ClearSpecsOption = Annotated[bool, typer.Option("--clear-specs", help="Clear all specs before crawling")] SpecOption = Annotated[list[str] | None, typer.Option("--spec", help="Spec number(s) (dotted or undotted)")] SpecArgument = Annotated[list[str] | None, typer.Argument(help="Spec number(s) to query (dotted or undotted)")] SpecFileOption = Annotated[Path | None, typer.Option("--spec-file", help="File with spec numbers")] ReleaseOption = Annotated[ str, Loading @@ -82,4 +71,21 @@ ReleaseOption = Annotated[ ), ] DocOnlyOption = Annotated[bool, typer.Option("--doc-only/--no-doc-only", help="Attempt document-only download")] CheckoutDirOption = Annotated[Path | None, typer.Option("--checkout-dir", help="Spec checkout base directory")] # Options - General/Common CacheDirOption = Annotated[Path | None, typer.Option("--cache-dir", "-c", help="Cache directory", envvar="TDC_CACHE_DIR")] ClearDbOption = Annotated[bool, typer.Option("--clear-db", help="Clear all meetings and TDocs before crawling")] CheckoutOption = Annotated[ bool, typer.Option("--checkout/--no-checkout", help="Download and extract metadata results to checkout folder", envvar="TDC_CHECKOUT") ] WorkersOption = Annotated[int, typer.Option("--workers", help="Number of parallel subinterpreter workers", envvar="TDC_WORKERS")] MaxRetriesOption = Annotated[int, typer.Option("--max-retries", help="HTTP retry attempts", envvar="TDC_MAX_RETRIES")] TimeoutOption = Annotated[int, typer.Option("--timeout", help="HTTP timeout seconds", envvar="TDC_TIMEOUT")] VerbosityOption = Annotated[str, typer.Option("--verbosity", "-v", help="Logging verbosity level", envvar="TDC_VERBOSITY")] UseWhatTheSpecOption = Annotated[ bool, typer.Option("--use-whatthespec/--no-use-whatthespec", help="Use WhatTheSpec API for fetching", envvar="TDC_USE_WHATTHESPEC") ] ForceOption = Annotated[bool, typer.Option("--force", "-f", help="Re-download even if already checked out")] Loading
src/tdoc_crawler/cli/args.py +36 −30 Original line number Diff line number Diff line Loading @@ -12,38 +12,20 @@ from tdoc_crawler.logging import DEFAULT_LEVEL as LOGGING_DEFAULT_LEVEL DEFAULT_VERBOSITY = logging.getLevelName(LOGGING_DEFAULT_LEVEL) CacheDirOption = Annotated[Path | None, typer.Option("--cache-dir", "-c", help="Cache directory", envvar="TDC_CACHE_DIR")] # Arguments TDocIdsArgument = Annotated[list[str] | None, typer.Argument(help="TDoc identifiers to query")] TDocIdArgument = Annotated[str, typer.Argument(help="TDoc identifier to download and open")] CheckoutTDocIdsArgument = Annotated[list[str], typer.Argument(help="TDoc identifier(s) to checkout")] SpecArgument = Annotated[list[str] | None, typer.Argument(help="Spec number(s) to query (dotted or undotted)")] # Options - TDocs/Meetings WorkingGroupOption = Annotated[list[str] | None, typer.Option("--working-group", "-w", help="Filter by working group", envvar="TDC_WORKING_GROUP")] SubgroupOption = Annotated[list[str] | None, typer.Option("--sub-group", "-s", help="Filter by sub-working group", envvar="TDC_SUB_GROUP")] IncrementalOption = Annotated[bool, typer.Option("--incremental/--full", help="Toggle incremental mode")] ClearTDocsOption = Annotated[bool, typer.Option("--clear-tdocs", help="Clear all TDocs before crawling")] ClearSpecsOption = Annotated[bool, typer.Option("--clear-specs", help="Clear all specs before crawling")] ClearDbOption = Annotated[bool, typer.Option("--clear-db", help="Clear all meetings and TDocs before crawling")] CheckoutOption = Annotated[ bool, typer.Option("--checkout/--no-checkout", help="Download and extract metadata results to checkout folder", envvar="TDC_CHECKOUT") ] LimitTDocsOption = Annotated[int | None, typer.Option("--limit-tdocs", help="Limit number of TDocs", envvar="TDC_LIMIT_TDOCS")] LimitMeetingsOption = Annotated[int | None, typer.Option("--limit-meetings", help="Limit meetings overall", envvar="TDC_LIMIT_MEETINGS")] LimitMeetingsPerWgOption = Annotated[ int | None, typer.Option("--limit-meetings-per-wg", help="Limit meetings per working group", envvar="TDC_LIMIT_MEETINGS_PER_WG") ] LimitWgsOption = Annotated[int | None, typer.Option("--limit-wgs", help="Limit number of working groups")] WorkersOption = Annotated[int, typer.Option("--workers", help="Number of parallel subinterpreter workers", envvar="TDC_WORKERS")] OverallTimeoutOption = Annotated[ int | None, typer.Option("--overall-timeout", help="Maximum total crawl duration in seconds (None = unlimited)", envvar="TDC_OVERALL_TIMEOUT"), ] MaxRetriesOption = Annotated[int, typer.Option("--max-retries", help="HTTP retry attempts", envvar="TDC_MAX_RETRIES")] TimeoutOption = Annotated[int, typer.Option("--timeout", help="HTTP timeout seconds", envvar="TDC_TIMEOUT")] VerbosityOption = Annotated[str, typer.Option("--verbosity", "-v", help="Logging verbosity level", envvar="TDC_VERBOSITY")] TDocIdsArgument = Annotated[list[str] | None, typer.Argument(help="TDoc identifiers to query")] OutputFormatOption = Annotated[str, typer.Option("--output", "-o", help="Output format", envvar="TDC_OUTPUT")] FullMetadataOption = Annotated[bool, typer.Option("--full-metadata", help="Fetch full metadata instead of URL only")] UseWhatTheSpecOption = Annotated[ bool, typer.Option("--use-whatthespec/--no-use-whatthespec", help="Use WhatTheSpec API for fetching", envvar="TDC_USE_WHATTHESPEC") ] LimitOption = Annotated[int | None, typer.Option("--limit", "-l", help="Maximum number of rows")] OrderOption = Annotated[str, typer.Option("--order", help="Sort order (asc|desc)")] StartDateOption = Annotated[str | None, typer.Option("--start-date", help="Filter from ISO timestamp", envvar="TDC_START_DATE")] Loading @@ -52,6 +34,14 @@ NoFetchOption = Annotated[ bool, typer.Option("--no-fetch", help="Disable automatic fetching of missing TDocs from portal"), ] IncrementalOption = Annotated[bool, typer.Option("--incremental/--full", help="Toggle incremental mode")] LimitTDocsOption = Annotated[int | None, typer.Option("--limit-tdocs", help="Limit number of TDocs", envvar="TDC_LIMIT_TDOCS")] ClearTDocsOption = Annotated[bool, typer.Option("--clear-tdocs", help="Clear all TDocs before crawling")] OverallTimeoutOption = Annotated[ int | None, typer.Option("--overall-timeout", help="Maximum total crawl duration in seconds (None = unlimited)", envvar="TDC_OVERALL_TIMEOUT"), ] OutputFormatOption = Annotated[str, typer.Option("--output", "-o", help="Output format", envvar="TDC_OUTPUT")] EolUsernameOption = Annotated[str | None, typer.Option("--eol-username", help="ETSI Online account username", envvar="TDC_EOL_USERNAME")] EolPasswordOption = Annotated[str | None, typer.Option("--eol-password", help="ETSI Online account password", envvar="TDC_EOL_PASSWORD")] PromptCredentialsOption = Annotated[ Loading @@ -62,13 +52,12 @@ IncludeWithoutFilesOption = Annotated[ bool, typer.Option("--include-without-files", help="Include meetings without files URLs"), ] FullMetadataOption = Annotated[bool, typer.Option("--full-metadata", help="Fetch full metadata instead of URL only")] TDocIdArgument = Annotated[str, typer.Argument(help="TDoc identifier to download and open")] CheckoutTDocIdsArgument = Annotated[list[str], typer.Argument(help="TDoc identifier(s) to checkout")] ForceOption = Annotated[bool, typer.Option("--force", "-f", help="Re-download even if already checked out")] # Options - Specs ClearSpecsOption = Annotated[bool, typer.Option("--clear-specs", help="Clear all specs before crawling")] SpecOption = Annotated[list[str] | None, typer.Option("--spec", help="Spec number(s) (dotted or undotted)")] SpecArgument = Annotated[list[str] | None, typer.Argument(help="Spec number(s) to query (dotted or undotted)")] SpecFileOption = Annotated[Path | None, typer.Option("--spec-file", help="File with spec numbers")] ReleaseOption = Annotated[ str, Loading @@ -82,4 +71,21 @@ ReleaseOption = Annotated[ ), ] DocOnlyOption = Annotated[bool, typer.Option("--doc-only/--no-doc-only", help="Attempt document-only download")] CheckoutDirOption = Annotated[Path | None, typer.Option("--checkout-dir", help="Spec checkout base directory")] # Options - General/Common CacheDirOption = Annotated[Path | None, typer.Option("--cache-dir", "-c", help="Cache directory", envvar="TDC_CACHE_DIR")] ClearDbOption = Annotated[bool, typer.Option("--clear-db", help="Clear all meetings and TDocs before crawling")] CheckoutOption = Annotated[ bool, typer.Option("--checkout/--no-checkout", help="Download and extract metadata results to checkout folder", envvar="TDC_CHECKOUT") ] WorkersOption = Annotated[int, typer.Option("--workers", help="Number of parallel subinterpreter workers", envvar="TDC_WORKERS")] MaxRetriesOption = Annotated[int, typer.Option("--max-retries", help="HTTP retry attempts", envvar="TDC_MAX_RETRIES")] TimeoutOption = Annotated[int, typer.Option("--timeout", help="HTTP timeout seconds", envvar="TDC_TIMEOUT")] VerbosityOption = Annotated[str, typer.Option("--verbosity", "-v", help="Logging verbosity level", envvar="TDC_VERBOSITY")] UseWhatTheSpecOption = Annotated[ bool, typer.Option("--use-whatthespec/--no-use-whatthespec", help="Use WhatTheSpec API for fetching", envvar="TDC_USE_WHATTHESPEC") ] ForceOption = Annotated[bool, typer.Option("--force", "-f", help="Re-download even if already checked out")]
