Case metadata¶
A document can carry its court filing's case metadata. In a batch set to choose its policy by case type, the case type decides which Philter policy redacts the document, since court rules differ by case type.
Schema¶
| Field | Type | Values | Stored |
|---|---|---|---|
caseNumber |
string | Up to 100 characters. | Encrypted |
caseType |
string | Up to 64 letters, digits, spaces, hyphens, or underscores, starting with a letter or digit. Matched without regard to case; stored lower-cased. | Plain text |
filer |
string | Up to 200 characters. | Encrypted |
feeStatus |
string | paid, unpaid, or waived. |
Plain text |
ifp |
boolean | Whether the filer proceeds in forma pauperis. | Plain text |
Every field is optional. Values are trimmed, and a blank value is the same as none. The case number and filer can identify people, so they are encrypted at rest like a document's text, and audit entries carry keyed hashes of them, not their values. The case type, fee status, and IFP flag stay readable so they can be matched and shown.
Supplying it¶
- Upload: the Case (optional) section of the upload form applies the values to every file in that upload. See Adding documents.
- REST API: a
caseMetadataobject onPOST /api/v1/ingest. - Data sources: for Local Directory and S3-Compatible sources, a sidecar file beside each document. See Sidecar files.
Invalid metadata is refused before anything is stored: the upload shows the
problem, and the API answers 400 with a problems list.
Sidecar files¶
A sidecar is named after its document with .case.json added, and holds a JSON
object with the fields above:
filings/
motion.pdf
motion.pdf.case.json
{"caseNumber": "1:26-cv-01234", "caseType": "civil", "filer": "Acme Dental LLC",
"feeStatus": "paid", "ifp": false}
Sidecars are never imported as documents. A document without one has no case metadata. A sidecar that is not valid JSON, has an unknown field, or has an invalid value fails that document's import, so it is not queued with the wrong metadata; the import log says what was wrong without quoting the values. Sidecars are up to 16 KB.
OpenSearch, Elasticsearch, and Relational Database sources do not read case metadata; set it on their documents through the API afterwards.
Choosing the policy by case type¶
- Map case types to policies under Admin → General → Case types. Each case
type maps to one policy of the Philter user that redacts batches; mapping a
case type again replaces its policy. Changes are audited as
CASE_TYPE_POLICY_CHANGE. - On the batch, tick Choose each document's policy from its case type, when creating it or with Change… in its Policy column. The Policy column then shows By case type. The batch's own policy is kept, and used again if the setting is turned off.
Each document in such a batch is redacted with the policy mapped to its case
type. A document with no case type, or a case type with no mapping, is not
redacted with another policy: it is marked FAILED with the reason, and appears
under Admin → Ingest Queue. Give it a case type, or map its case type, then
retry it there.
Batches without the setting use their own policy, whatever their documents' case types.
The policy used¶
When Philter redacts a document, Philter Workspace records the policy name and
version Philter reports applying. The review page shows them beside the case
metadata, and
GET /api/v1/documents/{id}/case-metadata
returns them.
Changing it¶
Edit case on the review page, or
PUT /api/v1/documents/{id}/case-metadata,
replaces a document's case metadata, unless the document is finalized or still
waiting to be redacted (PENDING or PROCESSING). A change
does not redact the document again. In a batch that chooses its policy by case
type, the new case type applies when the document is redacted next, such as when
a failed document is retried.
Each change is audited as DOCUMENT_CASE_METADATA_UPDATE, with the previous and
new case type, fee status, and IFP flag, and keyed hashes of a changed case
number or filer.