Get extraction stats

GET/api/stats

Counts, average quality and latency per extractor over a time window. The per-domain lists (lowQualityDomains, fallbackByDomain) are included only for the admin, or for everyone when sign-in is off and public history is on.

Query parameters

windowstring

How far back to look, as an SQLite date modifier such as -24 hours or -30 days. The log keeps 30 days. A value SQLite cannot parse matches nothing: the response has total 0 and no error.

Default: "-7 days"Example: -30 days

Response

200 · application/json

Extraction stats.

totalinteger

Requests logged in the window, cache hits included. The log keeps 30 days.

windowstring

The time window used.

bySourceobject[]

Requests per extractor, most used first. Absent when total is 0.

lowQualityDomainsobject[]

Up to 20 domains whose fresh (not cached) results scored below 0.4, most frequent first. Only for the admin, or everyone when sign-in is off and public history is on. Absent when total is 0.

fallbackByDomainobject[]

Up to 20 domains where extraction fell back, most frequent first. Counts results from readability-fallback and results whose extractor reason contains fell back or readability thin. Same visibility as lowQualityDomains. Absent when total is 0.

cURLcURLPythonJavaScript

curl --request GET \
  --url 'http://localhost:3000/api/stats?window=-30%20days'

{
  "total": 13,
  "window": "-7 days",
  "bySource": [
    {
      "source": "playwright",
      "count": 5,
      "pct": 38.5,
      "avgQuality": 0.72,
      "avgLen": 2074,
      "avgMs": 4802
    },
    {
      "source": "readability",
      "count": 4,
      "pct": 30.8,
      "avgQuality": 0.89,
      "avgLen": 32030,
      "avgMs": 981
    }
  ],
  "lowQualityDomains": [
    {
      "domain": "www.youtube.com",
      "count": 1,
      "avgQuality": 0.3
    }
  ],
  "fallbackByDomain": []
}