ocr-service/main.py

@@ -20,9 +20,11 @@ import pypdfium2 as pdfium
 from fastapi import FastAPI, Form, Header, HTTPException, UploadFile
 from fastapi.responses import StreamingResponse
 from PIL import Image
+from prometheus_client import REGISTRY
 from prometheus_fastapi_instrumentator import Instrumentator
 
 from confidence import apply_confidence_markers, get_threshold
+from metrics import OcrMetrics, build_metrics
 from spell_check import correct_text, load_spell_checker
 from engines import kraken as kraken_engine
 from engines import surya as surya_engine
@@ -38,6 +40,8 @@ logger = logging.getLogger(__name__)
 
 _models_ready = False
 
+metrics: OcrMetrics = build_metrics(REGISTRY)
+
 ALLOWED_PDF_HOSTS = set(
     h.strip() for h in os.getenv("ALLOWED_PDF_HOSTS", "minio,localhost,127.0.0.1").split(",")
 )

ocr-service/metrics.py

@@ -0,0 +1,92 @@
+"""Prometheus metric definitions for the OCR service.
+
+`build_metrics(registry)` returns a fresh `OcrMetrics` instance bound to the
+given `CollectorRegistry`. Production code calls it once at module load with
+the default `REGISTRY`; tests pass a per-test `CollectorRegistry()` to keep
+counter values isolated between cases (decision #3 on issue #652).
+"""
+
+from __future__ import annotations
+
+from dataclasses import dataclass
+
+from prometheus_client import CollectorRegistry, Counter, Gauge, Histogram
+
+
+@dataclass(frozen=True)
+class OcrMetrics:
+    """Container for every custom OCR metric.
+
+    Counters and gauges are immutable references to `prometheus_client`
+    instances. Mutating them (`.inc()`, `.observe()`, `.set()`) is safe;
+    rebinding the field on the dataclass is not — use `build_metrics` to get
+    a new container.
+    """
+
+    ocr_jobs_total: Counter
+    ocr_pages_total: Counter
+    ocr_skipped_pages_total: Counter
+    ocr_words_total: Counter
+    ocr_illegible_words_total: Counter
+    ocr_processing_seconds: Histogram
+    ocr_training_runs_total: Counter
+    ocr_model_accuracy: Gauge
+    ocr_models_ready: Gauge
+
+
+def build_metrics(registry: CollectorRegistry) -> OcrMetrics:
+    """Create one OcrMetrics instance bound to `registry`."""
+    return OcrMetrics(
+        ocr_jobs_total=Counter(
+            "ocr_jobs_total",
+            "Number of OCR jobs processed, labelled by engine and script type.",
+            ["engine", "script_type"],
+            registry=registry,
+        ),
+        ocr_pages_total=Counter(
+            "ocr_pages_total",
+            "Number of pages successfully OCR'd, labelled by engine.",
+            ["engine"],
+            registry=registry,
+        ),
+        ocr_skipped_pages_total=Counter(
+            "ocr_skipped_pages_total",
+            "Number of pages skipped because the OCR engine raised.",
+            registry=registry,
+        ),
+        ocr_words_total=Counter(
+            "ocr_words_total",
+            "Number of words recognized across all OCR blocks.",
+            registry=registry,
+        ),
+        ocr_illegible_words_total=Counter(
+            "ocr_illegible_words_total",
+            "Number of words below the confidence threshold "
+            "(replaced with [unleserlich]).",
+            registry=registry,
+        ),
+        ocr_processing_seconds=Histogram(
+            "ocr_processing_seconds",
+            "OCR processing time per page (streaming) or per document (non-streaming).",
+            ["engine"],
+            registry=registry,
+        ),
+        ocr_training_runs_total=Counter(
+            "ocr_training_runs_total",
+            "Number of training runs, labelled by kind (recognition|segmentation) "
+            "and outcome (success|error).",
+            ["kind", "outcome"],
+            registry=registry,
+        ),
+        ocr_model_accuracy=Gauge(
+            "ocr_model_accuracy",
+            "Latest model accuracy reported by a successful training run.",
+            ["kind"],
+            registry=registry,
+        ),
+        ocr_models_ready=Gauge(
+            "ocr_models_ready",
+            "1 once the lifespan startup has finished loading models, 0 before.",
+            registry=registry,
+        ),
+    )

ocr-service/test_metrics.py

@@ -9,8 +9,19 @@ from unittest.mock import AsyncMock, patch
 import pytest
 from httpx import ASGITransport, AsyncClient
 from PIL import Image
+from prometheus_client import CollectorRegistry
 
 from main import app
+from metrics import build_metrics
+
+
+@pytest.fixture
+def fresh_metrics(monkeypatch):
+    """Replace the module-level `main.metrics` with one bound to a fresh registry."""
+    registry = CollectorRegistry()
+    test_metrics = build_metrics(registry)
+    monkeypatch.setattr("main.metrics", test_metrics)
+    return test_metrics
 
 
 @pytest.mark.asyncio
@@ -56,3 +67,27 @@ async def test_metrics_includes_http_request_metrics_after_ocr_call():
     body = metrics_response.text
     assert "http_requests_total" in body
     assert "http_request_duration_seconds" in body
+
+
+def test_build_metrics_registers_all_custom_metrics_on_given_registry():
+    """`build_metrics` returns an OcrMetrics bound to the supplied registry."""
+    registry = CollectorRegistry()
+    metrics = build_metrics(registry)
+
+    metric_names = {m.name for m in registry.collect()}
+    expected = {
+        "ocr_jobs",
+        "ocr_pages",
+        "ocr_skipped_pages",
+        "ocr_words",
+        "ocr_illegible_words",
+        "ocr_processing_seconds",
+        "ocr_training_runs",
+        "ocr_model_accuracy",
+        "ocr_models_ready",
+    }
+    assert expected <= metric_names, f"missing: {expected - metric_names}"
+
+    # A second registry yields a separate container — no shared state.
+    other_metrics = build_metrics(CollectorRegistry())
+    assert metrics is not other_metrics