Befehle

Alle Befehle, Argumente und Optionen von ka 0.10.0, direkt aus der CLI gelesen.

Die Beschreibungen der Befehle, Argumente und Optionen stammen direkt aus der CLI und sind englisch.

Globale Optionen

Diese Optionen gelten für jeden Befehl.

Option	Beschreibung	Standard
`-v, --version`	output the version number
`--corpus <dir>`	corpus directory (default: $OPENKA_CORPUS, else $XDG_DATA_HOME/openka, else ~/.local/share/openka)
`--blobs <dir>`	keep the archived documents here instead of <corpus>/blobs — an existing directory, e.g. on an external drive (default: $OPENKA_BLOBS)
`--timeout <ms>`	timeout per request attempt in milliseconds (a timed-out request is retried once)
`--user-agent <ua>`	User-Agent sent to upstreams
`--max-retries <n>`	retries for a transient 429/503 or a dropped connection (never an over-size response)
`--max-response-bytes <n>`	hard cap on a single response body (default: 134217728, 128 MiB); a document over it is left out of its record, with a warning
`--min-host-interval <ms>`	minimum delay between requests to one host (default: 500); a source's own floor (ka sources show <key>) is never lowered by it
`--max-redirects <n>`	redirects to follow (0 = surface a 3xx as an error)
`--compact`	compact JSON output
`--quiet`	suppress progress output on stderr
`--log-format <format>`	how errors, warnings, notes and progress are written to stderr: text (log4j style: time, level, [topic], message) or jsonl (one JSON object per line: ts, level, topic, msg; `ka sync` adds its event fields); default text

`ka sync`

fetch, extract and store Anfragen from one or more sources

ka sync [options]

Optionen

Option	Beschreibung	Standard
`--source <key[@window]>`	source to sync, repeatable: several run side by side under one corpus lock. A window of its own after @: berlin@2025-01-01..2025-12-31, bund@period=21, bund@2026-01-01..,limit=50 — the shared --since/--until/--period/--limit fill in what it leaves out (baden-wuerttemberg, bayern, berlin, brandenburg, bremen, bund, hamburg, hessen, mecklenburg-vorpommern, niedersachsen, nordrhein-westfalen, parlamentsspiegel, rheinland-pfalz, saarland, sachsen, sachsen-anhalt, schleswig-holstein, thueringen)
`--all`	every source with an adapter of its own (not the parlamentsspiegel aggregator)
`--plan <file>`	run the jobs of a plan file ([[job]] tables: source, since, until, period, limit, ref, retry_failed, only_new, log; see Usage.md)
`--restart`	with --plan, run every job again, also those done in the plan's unfinished round
`--wait`	wait while another run holds the corpus, instead of exiting 3
`--dry-run`	discover only: count the Anfragen and estimate the download, fetching no document and writing nothing
`--since <date>`	only Anfragen dated on or after this date (YYYY-MM-DD)
`--until <date>`	only Anfragen dated on or before this date (YYYY-MM-DD)
`--period <n>`	restrict to one legislative period
`--limit <n>`	stop after this many Anfragen (per source)
`--ref <reference>`	handle only this Anfrage of the window (repeatable), or one it was filed under before; the rest are not asked for
`--retry-failed`	handle only the Anfragen of the window whose last attempt failed (with --ref: those as well)
`--only-new`	skip every Anfrage the corpus holds with all its documents, without a request; handle the rest
`--api-key <key>`	credential for sources that need one (overrides the env var)
`--metadata-only`	download no documents: a new record abstains on qa, a stored one is rebuilt from its archived documents
`--force`	re-extract even when inputs and extractor version are unchanged
`--ignore-robots`	fetch documents from a server whose robots.txt disallows it — your decision; the run warns once per host (the records do not record it)
`--ocr <mode>`	OCR engine for the ocr tier Werte: off, tesseract, tesseract-js
`--ocr-language <lang>`	traineddata language for OCR
`--ocr-version <version>`	require exactly this OCR engine version
`--ocr-traineddata <path>`	traineddata file to hash into the provenance record
`--json`	print the sync report as JSON (an array of reports for several jobs, --all or --plan)
`--log-file <path>`	append the run's events and its other log records to this file as JSON Lines, whatever --log-format is
`--allow-fs <type>`	write to a corpus on this filesystem anyway (repeatable: fat32, exfat); both are refused by default
`--min-free <size>`	free space to keep on the corpus and blob volumes, e.g. 500M or 20G; 0 checks none (default: 1.0 GB)

`ka search`

full-text search over the corpus

ka search [options] [query]

Argumente

[query] optional: search terms; quote a phrase, prefix a term with - to exclude it

Optionen

Option	Beschreibung	Standard
`--limit <n>`	maximum results (default: 20)
`--offset <n>`	skip this many results
`--snippet`	show a text snippet around the first match
`--like <id>`	semantic search: records similar to this one (needs frozen embeddings)
`--json`	print results as JSON
`--parliament <key>`	restrict to a parliament (repeatable; 17 known, see `ka sources list`)
`--party <name>`	restrict to Anfragen asked by this party (repeatable)
`--year <yyyy>`	restrict to the year the Anfrage was asked (repeatable)
`--period <n>`	restrict to a legislative period (repeatable)
`--from <date>`	asked on or after this date (a record whose question date is unknown is left out)
`--to <date>`	asked on or before this date (a record whose question date is unknown is left out)
`--review-status <status>`	restrict by review status Werte: ok, needs_review, human_verified
`--needs-review`	only records with at least one abstained field

`ka get`

print one record in a machine-readable format

ka get [options] <id>

Argumente

<id> erforderlich: record id, e.g. berlin-19-10006

Optionen

Option	Beschreibung	Standard
`--format <format>`	output format Werte: json, jsonld, csv, md, text
`-o, --out <file>`	write to this file instead of stdout (- = stdout; an existing file needs --force)
`--force`	with --out, replace an existing file

`ka show`

render one record for reading

ka show [options] <id>

Argumente

<id> erforderlich: record id

`ka open`

print the path of a record's archived source document

ka open [options] <id>

Argumente

<id> erforderlich: record id

Optionen

Option	Beschreibung	Standard
`--role <role>`	which document to open when there are several (question_pdf, answer_pdf, combined_pdf, metadata)

`ka verify`

re-run an extraction from the archived bytes and assert identical output

ka verify [options] [id]

Argumente

[id] optional: record id; omit to verify an evenly spaced sample of the corpus

Optionen

Option	Beschreibung	Standard
`--all`	verify every record
`--limit <n>`	how many records to verify when no id is given (default: 25)
`--ocr <mode>`	OCR engine to use for records produced with one Werte: off, tesseract, tesseract-js
`--json`	print results as JSON

`ka review`

work the abstention queue: records the extractor refused to complete

ka review [options]

Optionen

Option	Beschreibung	Standard
`--parliament <key>`	restrict to one parliament (as in search, export and stats)
`--source <key>`	the same as --parliament
`--include-known-gaps`	also the records whose only holes are fields their parliament never provides (ka sources show <key>)
`--limit <n>`	how many records to list (default: 20)
`--mark-verified <id>`	record that a human checked this record against its source
`--group-by <what>`	summarise the queue per source by the kind of field abstained on, with example ids Werte: field
`--json`	print the queue as JSON

`ka reindex`

rebuild the search index and catalog from the stored records

ka reindex [options]

`ka sources`

the source map and its health

`ka sources list`

every parliament, its adapter status and its last sync

ka sources list [options]

Optionen

Option	Beschreibung	Standard
`--json`	print as JSON

`ka sources count`

how many Anfragen each upstream holds, beside how many the corpus has — a request or two per source, no download

ka sources count [options]

Optionen

Option	Beschreibung	Standard
`--source <key>`	count only this source (repeatable; default: every parliament)
`--period <n>`	count one legislative period (DIP can; the Parlamentsspiegel cannot)
`--api-key <key>`	credential for sources that need one (overrides the env var)
`--json`	print as JSON

`ka sources show`

what one source does and what is specific about it

ka sources show [options] <key>

Argumente

<key> erforderlich: source key

`ka reextract`

re-extract stored records from their archived bytes with this build's extractor — no network; then rebuild the index

ka reextract [options] [ids...]

Argumente

[ids…] optional: record ids; or select with --all or the filters

Optionen

Option	Beschreibung	Standard
`--all`	every record in the corpus
`--force`	also the records this build's extractor already stamped
`--dry-run`	say what would change, and write nothing
`--ocr <mode>`	OCR engine for records produced with one Werte: off, tesseract, tesseract-js
`--json`	print the report as JSON
`--parliament <key>`	restrict to a parliament (repeatable; 17 known, see `ka sources list`)
`--party <name>`	restrict to Anfragen asked by this party (repeatable)
`--year <yyyy>`	restrict to the year the Anfrage was asked (repeatable)
`--period <n>`	restrict to a legislative period (repeatable)
`--from <date>`	asked on or after this date (a record whose question date is unknown is left out)
`--to <date>`	asked on or before this date (a record whose question date is unknown is left out)

`ka rm`

remove records from the corpus, with their catalog rows and index postings (takes the corpus lock)

ka rm [options] [ids...]

Argumente

[ids…] optional: record ids; or select with the filters

Optionen

Option	Beschreibung	Standard
`--documents`	also remove their archived documents that no remaining record refers to
`--orphaned-documents`	remove every archived document no record refers to, and no record
`--move-to <dir>`	move the files under this directory (records/, blobs/) instead of deleting them
`--dry-run`	say what would be removed, and change nothing
`--json`	print the report as JSON
`--parliament <key>`	restrict to a parliament (repeatable; 17 known, see `ka sources list`)
`--party <name>`	restrict to Anfragen asked by this party (repeatable)
`--year <yyyy>`	restrict to the year the Anfrage was asked (repeatable)
`--period <n>`	restrict to a legislative period (repeatable)
`--from <date>`	asked on or after this date (a record whose question date is unknown is left out)
`--to <date>`	asked on or before this date (a record whose question date is unknown is left out)

`ka export`

export the corpus (or a selection of it) in bulk

ka export [options]

Optionen

Option	Beschreibung	Standard
`--format <format>`	output format Werte: csv, jsonl, jsonld
`--limit <n>`	maximum records to export (most relevant first with --query)
`--parliament <key>`	restrict to a parliament (repeatable; 17 known, see `ka sources list`)
`--party <name>`	restrict to Anfragen asked by this party (repeatable)
`--year <yyyy>`	restrict to the year the Anfrage was asked (repeatable)
`--period <n>`	restrict to a legislative period (repeatable)
`--from <date>`	asked on or after this date (a record whose question date is unknown is left out)
`--to <date>`	asked on or before this date (a record whose question date is unknown is left out)
`--query <terms>`	restrict to records matching these search terms
`-o, --out <file>`	write to this file instead of stdout (- = stdout; an existing file needs --force)
`--force`	with --out, replace an existing file

`ka feed`

an Atom feed of the newest matching Anfragen

ka feed [options]

Optionen

Option	Beschreibung	Standard
`--limit <n>`	entries in the feed
`--title <text>`	feed title (default: OpenKA — Kleine Anfragen)
`--id <url>`	feed id / self link (default: urn:openka:feed)
`--parliament <key>`	restrict to a parliament (repeatable; 17 known, see `ka sources list`)
`--party <name>`	restrict to Anfragen asked by this party (repeatable)
`--year <yyyy>`	restrict to the year the Anfrage was asked (repeatable)
`--period <n>`	restrict to a legislative period (repeatable)
`--from <date>`	asked on or after this date (a record whose question date is unknown is left out)
`--to <date>`	asked on or before this date (a record whose question date is unknown is left out)
`--query <terms>`	restrict to records matching these search terms
`-o, --out <file>`	write to this file instead of stdout (- = stdout; an existing file needs --force)
`--force`	with --out, replace an existing file

`ka schema`

print the JSON Schema of the canonical record

ka schema [options]

`ka stats`

what is in this corpus: coverage, completeness, extractor versions, disk use, and breakdowns with --by

ka stats [options]

Optionen

Option	Beschreibung	Standard
`--by <dimension>`	break the records down by party, ministry, month, year, period, parliament; twice for a cross-tab (--by party --by year)
`--no-disk`	leave out what the corpus takes on disk (one stat per file)
`--json`	print as JSON
`--parliament <key>`	restrict to a parliament (repeatable; 17 known, see `ka sources list`)
`--party <name>`	restrict to Anfragen asked by this party (repeatable)
`--year <yyyy>`	restrict to the year the Anfrage was asked (repeatable)
`--period <n>`	restrict to a legislative period (repeatable)
`--from <date>`	asked on or after this date (a record whose question date is unknown is left out)
`--to <date>`	asked on or before this date (a record whose question date is unknown is left out)

`ka doctor`

check the corpus: filesystem, free space, lock, catalog against records, macOS ._* files (exit 3 on a problem)

ka doctor [options]

Optionen

Option	Beschreibung	Standard
`--fix`	remove the macOS ._* and .DS_Store files from the corpus (takes the corpus lock)
`--orphaned-documents`	also count the archived documents no record refers to (reads every record)
`--json`	print the diagnosis as JSON
`--allow-fs <type>`	write to a corpus on this filesystem anyway (repeatable: fat32, exfat); both are refused by default
`--min-free <size>`	free space to keep on the corpus and blob volumes, e.g. 500M or 20G; 0 checks none (default: 1.0 GB)

`ka status`

what a running sync is doing — progress, rate, time left — or how the last one ended

ka status [options]

Optionen

Option	Beschreibung	Standard
`--json`	print the status as JSON
`--watch`	look again every 5 s until no sync runs
`--stalled-after <duration>`	exit 1 when a running sync has not moved for this long (90s, 10m, 2h), or its process is gone