Corrige identidade de artigos e importa histórico parcial com controle

VERSION

@@ -1 +1 @@
-2.6.0
+2.6.1

metrics/legacy_matomo/validation.py

@@ -230,6 +230,7 @@ def validate_manifest(
     progress_callback=None,
     progress_interval=DEFAULT_PROGRESS_INTERVAL,
     stop_controller=None,
+    allow_partial=False,
 ):
     if progress_interval <= 0:
         raise ValueError("Progress interval must be greater than zero.")
@@ -238,6 +239,9 @@ def validate_manifest(
     if manifest.get("migration") != MIGRATION_NAME:
         raise ValueError("Unsupported migration manifest.")
     validate_migration_scope(manifest)
+    missing_days = manifest.get("missing_days", [])
+    if missing_days and not allow_partial:
+        raise ValueError("Partial migration manifest has missing source days.")
 
     collection = manifest["target"]["collection"]
     month = manifest["month"]
@@ -262,6 +266,19 @@ def validate_manifest(
         if value[:7] != month:
             raise ValueError("Empty day outside migration month: %s." % value)
 
+    if missing_days != sorted(set(missing_days)):
+        raise ValueError("Migration missing days must be unique and sorted.")
+    if set(missing_days) & (set(source_days) | set(empty_days)):
+        raise ValueError("Migration missing days overlap available days.")
+    for value in missing_days:
+        date.fromisoformat(value)
+        if value[:7] != month:
+            raise ValueError("Missing day outside migration month: %s." % value)
+
+    missing_day_totals = manifest.get("missing_day_totals", {})
+    if set(missing_day_totals) != set(missing_days):
+        raise ValueError("Migration missing day totals differ from missing days.")
+
     if manifest["counter"].get("month") != month:
         raise ValueError("Counter summary month differs from manifest.")
     if manifest["analytics"].get("source_month") != month:
@@ -291,10 +308,11 @@ def validate_manifest(
         raise ValueError("Counter and analytics totals do not reconcile.")
 
     return {
-        "status": "valid",
+        "status": "partial" if missing_days else "valid",
         "collection": collection,
         "month": month,
         "source_days": len(source_days),
+        "missing_days": missing_days,
         "counter": counter,
         "analytics": analytics,
     }

metrics/management/commands/backfill_counter_month_status.py

@@ -96,7 +96,7 @@ def _from_reports(self, paths, client):
 
             if report.get("status") != "completed":
                 raise CommandError(f"migration report is not completed: {path}")
-            if report.get("validation", {}).get("status") != "valid":
+            if report.get("validation", {}).get("status") not in {"valid", "partial"}:
                 raise CommandError(f"migration report is not valid: {path}")
 
             for dataset in ("counter", "analytics"):

metrics/management/commands/import_legacy_matomo.py

@@ -29,6 +29,7 @@ def add_arguments(self, parser):
         parser.add_argument("--report")
         parser.add_argument("--temporary-directory")
         parser.add_argument("--progress-every", type=int, default=100000)
+        parser.add_argument("--allow-partial", action="store_true")
 
     def _progress(self, dataset, stage, documents):
         self.stdout.write("[%s] %s: %d documents" % (stage, dataset, documents))
@@ -72,6 +73,7 @@ def handle(self, *args, **options):
                 progress_callback=self._progress,
                 progress_interval=options["progress_every"],
                 stop_controller=stop_controller,
+                allow_partial=options["allow_partial"],
             )
             report["validation"] = validation
 

metrics/tests/integration/test_pipelines.py

@@ -1,6 +1,7 @@
 import unittest
 from datetime import datetime
 
+from scielo_usage_counter.url_translator import URLTranslationManager
 from scielo_usage_counter.values import (
     CONTENT_TYPE_ABSTRACT,
     CONTENT_TYPE_FULL_TEXT,
@@ -9,6 +10,7 @@
 
 from metrics.counter.access import accumulation, extraction
 from metrics.counter.access.daily_accumulator import DailyAccessAccumulator
+from metrics.opensearch.keys import document_key
 from metrics.tests.helpers import convert_accumulator
 
 
@@ -124,3 +126,50 @@ def test_opac_article_produces_journal_article_metrics(self):
         self.assertTrue(doc["source_key"].startswith("k1_"))
         self.assertTrue(doc["document_key"].startswith("k1_"))
         self.assertEqual(doc["total_requests"], 1)
+
+    def test_classic_and_opac_urls_share_article_and_session(self):
+        pid_v2 = "S0103-63512021000100001"
+        pid_v3 = "dqLRqnpmnncSmnzMCB8bzPG"
+        translator = URLTranslationManager(
+            sources_metadata=[{"acronym": "neco", "scielo_issn": "0103-6351"}],
+            documents_metadata=[
+                {
+                    "pid_v2": pid_v2,
+                    "pid_v3": pid_v3,
+                    "default_lang": "pt",
+                    "text_langs": ["pt"],
+                    "scielo_issn": "0103-6351",
+                    "publication_year": "2021",
+                }
+            ],
+        )
+        results = DailyAccessAccumulator()
+        urls = (
+            f"/scielo.php?script=sci_arttext&pid={pid_v2}",
+            f"/j/neco/a/{pid_v2}/?format=html",
+            f"/j/neco/a/{pid_v3}/?format=html",
+        )
+
+        for minute, url in enumerate(urls):
+            line = {
+                "url": url,
+                "client_name": "Chrome",
+                "client_version": "120.0",
+                "ip_address": "189.10.20.30",
+                "country_code": "BR",
+                "local_datetime": datetime(2026, 7, 15, 8, 15 + minute, 0),
+            }
+            access = extraction.extract("scl", translator.translate(url))
+            accumulation.accumulate(results, access, line)
+
+        metrics = convert_accumulator(results)
+        self.assertEqual(len(metrics["counter"]), 1)
+        self.assertEqual(len(metrics["analytics"]), 1)
+
+        article = next(iter(metrics["counter"].values()))
+        self.assertEqual(
+            article["document_key"],
+            document_key("scl", "article", pid_v3),
+        )
+        self.assertEqual(article["total_requests"], 3)
+        self.assertEqual(article["unique_requests"], 1)

metrics/tests/services/test_legacy_matomo_validation.py

@@ -60,6 +60,28 @@ def test_migration_identity_includes_payload_hash(self):
 
         self.assertIn(self.manifest["counter"]["uncompressed_sha256"], migration_id)
 
+    def test_rejects_partial_manifest_with_missing_days(self):
+        self.manifest["missing_days"] = ["2025-08-02"]
+
+        with self.assertRaisesMessage(ValueError, "missing source days"):
+            validation.validate_manifest(self.manifest)
+
+    def test_accepts_partial_manifest_when_explicit(self):
+        self.manifest["missing_days"] = ["2025-08-02"]
+        self.manifest["missing_day_totals"] = {"2025-08-02": {"nbr": {}}}
+
+        result = validation.validate_manifest(self.manifest, allow_partial=True)
+
+        self.assertEqual(result["status"], "partial")
+        self.assertEqual(result["missing_days"], ["2025-08-02"])
+
+    def test_rejects_missing_day_that_overlaps_source_day(self):
+        self.manifest["missing_days"] = ["2025-08-01"]
+        self.manifest["missing_day_totals"] = {"2025-08-01": {"nbr": {}}}
+
+        with self.assertRaisesMessage(ValueError, "overlap available days"):
+            validation.validate_manifest(self.manifest, allow_partial=True)
+
     def test_rejects_manifest_without_source_days(self):
         self.manifest["source_days"] = []
 

metrics/tests/services/test_month_status_command.py

@@ -6,7 +6,7 @@
 from unittest.mock import patch
 
 from django.core.management import call_command
-from django.test import TestCase
+from django.test import SimpleTestCase, TestCase
 
 from collection.models import Collection
 from metrics.models import DailyMetricJob
@@ -99,3 +99,56 @@ def test_uses_completed_migration_reports(self, client_class, replace_status):
             date(2025, 2, 1),
             (1 << 0) | (1 << 1) | (1 << 27),
         )
+
+
+class PartialMigrationMonthStatusTests(SimpleTestCase):
+    @patch(
+        "metrics.management.commands.backfill_counter_month_status."
+        "replace_month_status"
+    )
+    @patch(
+        "metrics.management.commands.backfill_counter_month_status."
+        "OpenSearchUsageClient"
+    )
+    def test_replaces_partial_month_coverage_from_import_report(
+        self, client_class, replace_status
+    ):
+        with tempfile.TemporaryDirectory() as directory:
+            root = Path(directory)
+            manifest_path = root / "manifest.json"
+            report_path = root / "report.json"
+            manifest = {
+                "month": "2025-02",
+                "source_days": ["2025-02-01"],
+                "empty_days": ["2025-02-02"],
+                "missing_days": ["2025-02-03"],
+                "target": {"collection": "scl"},
+            }
+            report = {
+                "status": "completed",
+                "manifest": str(manifest_path),
+                "collection": "scl",
+                "month": "2025-02",
+                "validation": {
+                    "status": "partial",
+                    "counter": {"documents": 10},
+                    "analytics": {"documents": 20},
+                },
+                "imported": {"counter": 10, "analytics": 20},
+            }
+            manifest_path.write_text(json.dumps(manifest), encoding="utf-8")
+            report_path.write_text(json.dumps(report), encoding="utf-8")
+
+            call_command(
+                "backfill_counter_month_status",
+                report=[str(report_path)],
+                apply=True,
+                stdout=StringIO(),
+            )
+
+        replace_status.assert_called_once_with(
+            client_class.return_value,
+            "scl",
+            date(2025, 2, 1),
+            (1 << 0) | (1 << 1),
+        )

requirements/base.txt

@@ -66,7 +66,7 @@ git+https://github.com/scieloorg/scielo_log_validator@2.0.4#egg=scielo_log_valid
 git+https://github.com/scieloorg/scielo_scholarly_data@v0.1.4#egg=scielo_scholarly_data
 
 # SciELO Usage COUNTER
-git+https://github.com/scieloorg/scielo_usage_counter@2.2.3#egg=scielo_usage_counter
+git+https://github.com/scieloorg/scielo_usage_counter@2.2.4#egg=scielo_usage_counter
 
 # Incremental JSON
 ijson==3.5.1