Module field_anomaly 

T107 poisoned-data field-level anomaly detector.

Catches the “tarpit / poisoned data / silent 200s” pattern from https://web-scraping-guide.com/#post-extract: targets that return clean 200 responses with subtly wrong field values (price drift, listing reorder, fabricated rows, stale snapshots). The detector observes every field value the pipeline publishes and emits field_anomaly::AnomalyReports when a value is statistically inconsistent with the rolling baseline.

Hidden behind a field-anomaly cargo feature so existing charon users aren’t forced to opt in. T107 poisoned-data field-level anomaly detector.

Catches the “tarpit / poisoned data / silent 200s” pattern from https://web-scraping-guide.com/#post-extract: a target returns clean 200 responses with subtly wrong field values (price drift, listing reorder, fabricated rows, stale snapshots). The detector observes every field value the pipeline publishes and emits AnomalyReports when a value is statistically inconsistent with the rolling baseline.

Two pieces:

• FieldAnomalyDetector — the consumer-owned port trait.
• StatisticalFieldAnomalyDetector — the default adapter implementing price-drift, outlier, listing-reorder, staleness, and cardinality-shift detection.

Hidden behind a field-anomaly cargo feature in the parent crate so existing charon users aren’t forced to opt in.

Structs

AnomalyReport

One anomaly emitted by FieldAnomalyDetector::observe.

FieldPath

Dot-path to a field within a record ("price", "items.0.sku").

SchemaBaseline

Per-field rolling baseline — the statistical summary the detector compares new observations against.

SchemaId

Stable identifier for a schema or data-contract version. Two observations with different SchemaId belong to different baselines — the detector resets state on schema change.

StatisticalFieldAnomalyDetector

Default statistical detector. Rolling-window-based: keeps the last Self::WINDOW_SIZE observations per field per schema and applies the five heuristics in AnomalySignal.

Enums

AnomalyError

Errors raised by FieldAnomalyDetector operations.

AnomalySeverity

Severity of an AnomalyReport.

AnomalySignal

The five anomaly kinds the detector emits.

FieldValue

One observed field value.

Traits

FieldAnomalyDetector

Port trait: observe a field value, baseline reset on schema change.