RedactionResult#

class scikitplot.cleanprompt.RedactionResult(text, entries, vault, stats, policy_fingerprint, truncated=False, reported=())[source]#

Output of a redaction pass.

Parameters:
textstr

The redacted text. Safe to transmit.

entriestuple of Entry

Ordered record of every value replaced, sorted by first occurrence.

vaultVault

The label-to-secret mapping needed to restore. Not safe to transmit.

statsStats

Counters for this pass.

policy_fingerprintstr

Stable digest of the policy that produced this result. Restoration checks it, so a vault cannot be applied under a different placeholder grammar than the one that created it.

truncatedbool, default=False

Always False. Present so that callers can assert the pipeline never truncates; a bound is exceeded by raising, not by dropping content.

reportedtuple of Span, default=()

Values left in text because their kind’s action is report (Action), in text order. Empty unless a policy asks for it; they are not hidden, and the text is not safe to transmit while this is non-empty.

Parameters:

See also

scikitplot.cleanprompt._engine.restore

The inverse operation.

entries: tuple[Entry, ...][source]#
property labels: tuple[str, ...]#

Every placeholder label issued, in order.

Type:

Tuple of str

policy_fingerprint: str[source]#
reported: tuple[Span, ...] = ()#
stats: Stats[source]#
summary()[source]#

Return a one-line, secret-free description of this pass.

Returns:
str

A summary naming counts and categories but never the secrets.

Return type:

str

Notes

Developer notes. Upstream printed "Removed sensitive information: " + ", ".join(mapping.keys()), which echoes every secret to the terminal and therefore into shell history and any captured log. This deliberately names categories and counts only.

text: str[source]#
truncated: bool = False#
vault: Any[source]#