test: 补充知识处理、检索与 Agent 回归用例

覆盖切块、向量检索、工具调用和历史预算；模型相关用例使用桩。

backend/tests/repo/test_crm_repo.py

@@ -0,0 +1,219 @@
+"""
+Copyright 2024-2026 ReplyPier
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.
+"""
+
+from datetime import datetime, timedelta, timezone
+
+import pytest
+
+from app.models.crm import CrmConnectionStatus, CrmSyncJobStatus
+from app.models.lead_capture import LeadCaptureResponse
+from app.repositories.crm import (
+    CrmConnectionRepository, CrmSyncJobRepository, MAX_ATTEMPTS,
+)
+
+
+@pytest.fixture
+def conn_repo(db):
+    return CrmConnectionRepository(db)
+
+
+@pytest.fixture
+def job_repo(db):
+    return CrmSyncJobRepository(db)
+
+
+@pytest.fixture
+def test_lead_response(db, test_organization, test_agent, test_customer) -> LeadCaptureResponse:
+    response = LeadCaptureResponse(
+        organization_id=test_organization.id,
+        agent_id=test_agent.id,
+        customer_id=test_customer.id,
+        field_values={"email": "lead@example.com", "name": "Lead Person"},
+        consent=True,
+        qualified=True,
+    )
+    db.add(response)
+    db.commit()
+    db.refresh(response)
+    return response
+
+
+CREDS = {"access_token": "at-1", "refresh_token": "rt-1", "api_domain": None}
+
+
+def _aware(dt: datetime) -> datetime:
+    """SQLite loses tzinfo on refresh; normalize for comparisons (PG keeps it)."""
+    return dt if dt.tzinfo else dt.replace(tzinfo=timezone.utc)
+
+
+def _connect(conn_repo, org_id, provider="hubspot", external_id="hub-123", **kwargs):
+    return conn_repo.create_or_update(
+        organization_id=org_id,
+        provider=provider,
+        external_account_id=external_id,
+        credentials=CREDS,
+        display_name="Acme Portal",
+        **kwargs,
+    )
+
+
+class TestCrmConnectionRepository:
+
+    def test_credentials_round_trip_encrypted(self, conn_repo, test_organization):
+        conn = _connect(conn_repo, test_organization.id)
+        # Stored blob is ciphertext, decrypts back to the original dict.
+        assert "at-1" not in conn.encrypted_credentials
+        assert conn_repo.get_credentials(conn) == CREDS
+
+    def test_create_or_update_is_an_upsert(self, conn_repo, test_organization):
+        first = _connect(conn_repo, test_organization.id)
+        conn_repo.set_status(first, CrmConnectionStatus.EXPIRED.value, last_error="dead token")
+
+        reconnected = _connect(conn_repo, test_organization.id, external_id="hub-456")
+        assert reconnected.id == first.id  # same row, not a duplicate
+        assert reconnected.external_account_id == "hub-456"
+        assert reconnected.status == CrmConnectionStatus.ACTIVE.value
+        assert reconnected.last_error is None
+
+    def test_get_active_ignores_expired(self, conn_repo, test_organization):
+        conn = _connect(conn_repo, test_organization.id)
+        assert conn_repo.get_active(test_organization.id, "hubspot") is not None
+        conn_repo.set_status(conn, CrmConnectionStatus.EXPIRED.value)
+        assert conn_repo.get_active(test_organization.id, "hubspot") is None
+
+    def test_save_credentials_mirrors_expiries(self, conn_repo, test_organization):
+        conn = _connect(conn_repo, test_organization.id, provider="pipedrive",
+                        external_id="pd-1")
+        access_exp = datetime.now(timezone.utc) + timedelta(hours=1)
+        refresh_exp = datetime.now(timezone.utc) + timedelta(days=60)
+        conn = conn_repo.save_credentials(
+            conn, {"access_token": "at-2", "refresh_token": "rt-2"},
+            access_token_expires_at=access_exp,
+            refresh_token_expires_at=refresh_exp,
+        )
+        assert conn_repo.get_credentials(conn)["access_token"] == "at-2"
+        assert conn.access_token_expires_at is not None
+        assert conn.refresh_token_expires_at is not None
+        assert conn.last_refreshed_at is not None
+
+    def test_list_needing_refresh(self, conn_repo, db, test_organization):
+        conn = _connect(conn_repo, test_organization.id, provider="pipedrive",
+                        external_id="pd-2")
+        cutoff = datetime.now(timezone.utc) - timedelta(hours=24)
+        # Freshly connected → not due.
+        assert conn_repo.list_needing_refresh("pipedrive", cutoff) == []
+        conn.last_refreshed_at = datetime.now(timezone.utc) - timedelta(days=2)
+        db.commit()
+        assert [c.id for c in conn_repo.list_needing_refresh("pipedrive", cutoff)] == [conn.id]
+        # Expired connections are never swept.
+        conn_repo.set_status(conn, CrmConnectionStatus.EXPIRED.value)
+        assert conn_repo.list_needing_refresh("pipedrive", cutoff) == []
+
+
+class TestCrmSyncJobRepository:
+
+    def test_enqueue_is_idempotent(self, job_repo, test_organization, test_lead_response):
+        job = job_repo.enqueue(test_organization.id, test_lead_response.id, "hubspot")
+        assert job is not None
+        assert job.status == CrmSyncJobStatus.PENDING.value
+        # Same lead + provider → no second job.
+        assert job_repo.enqueue(test_organization.id, test_lead_response.id, "hubspot") is None
+        # Same lead, other provider → its own job.
+        other = job_repo.enqueue(test_organization.id, test_lead_response.id, "pipedrive")
+        assert other is not None and other.id != job.id
+
+    def test_claim_marks_processing_and_respects_due_time(
+            self, job_repo, db, test_organization, test_lead_response):
+        due = job_repo.enqueue(test_organization.id, test_lead_response.id, "hubspot")
+        future = job_repo.enqueue(test_organization.id, test_lead_response.id, "pipedrive")
+        future.next_attempt_at = datetime.now(timezone.utc) + timedelta(hours=1)
+        db.commit()
+
+        claimed = job_repo.claim_batch(limit=10)
+        assert [j.id for j in claimed] == [due.id]
+        assert claimed[0].status == CrmSyncJobStatus.PROCESSING.value
+        assert claimed[0].started_at is not None
+        # Already-claimed jobs are not claimed again.
+        assert job_repo.claim_batch(limit=10) == []
+
+    def test_reclaim_stale_returns_crashed_claims(
+            self, job_repo, db, test_organization, test_lead_response):
+        job = job_repo.enqueue(test_organization.id, test_lead_response.id, "hubspot")
+        job_repo.claim_batch(limit=1)
+        db.refresh(job)
+        job.started_at = datetime.now(timezone.utc) - timedelta(minutes=30)
+        db.commit()
+
+        assert job_repo.reclaim_stale(older_than_minutes=10) == 1
+        db.refresh(job)
+        assert job.status == CrmSyncJobStatus.PENDING.value
+        assert job.started_at is None
+
+    def test_complete_stamps_result_and_connection(
+            self, job_repo, conn_repo, test_organization, test_lead_response):
+        conn = _connect(conn_repo, test_organization.id)
+        job = job_repo.enqueue(test_organization.id, test_lead_response.id, "hubspot")
+        job = job_repo.complete(job, {"action": "created", "contact_id": "42"},
+                                connection_id=conn.id)
+        assert job.status == CrmSyncJobStatus.COMPLETED.value
+        assert job.result["contact_id"] == "42"
+        assert job.connection_id == conn.id
+
+    def test_schedule_retry_backs_off_then_fails_terminally(
+            self, job_repo, test_organization, test_lead_response):
+        job = job_repo.enqueue(test_organization.id, test_lead_response.id, "hubspot")
+        before = datetime.now(timezone.utc)
+        job = job_repo.schedule_retry(job, "HTTP 502")
+        assert job.status == CrmSyncJobStatus.PENDING.value
+        assert job.attempts == 1
+        assert _aware(job.next_attempt_at) > before
+
+        job.attempts = MAX_ATTEMPTS - 1
+        job = job_repo.schedule_retry(job, "HTTP 502 again")
+        assert job.status == CrmSyncJobStatus.FAILED.value
+        assert job.attempts == MAX_ATTEMPTS
+
+    def test_retry_after_wins_over_backoff_when_longer(
+            self, job_repo, test_organization, test_lead_response):
+        job = job_repo.enqueue(test_organization.id, test_lead_response.id, "hubspot")
+        job = job_repo.schedule_retry(job, "HTTP 429", retry_after_seconds=7200)
+        # 7200 > capped backoff (3600) → provider's Retry-After is honored.
+        assert _aware(job.next_attempt_at) >= datetime.now(timezone.utc) + timedelta(seconds=7000)
+
+    def test_skip_pending_for_connection(
+            self, job_repo, test_organization, test_lead_response):
+        job = job_repo.enqueue(test_organization.id, test_lead_response.id, "hubspot")
+        count = job_repo.skip_pending_for_connection(
+            test_organization.id, "hubspot", "app uninstalled")
+        assert count == 1
+        assert job_repo.claim_batch(limit=10) == []
+
+    def test_count_recent_failures(self, job_repo, test_organization, test_lead_response):
+        job = job_repo.enqueue(test_organization.id, test_lead_response.id, "hubspot")
+        job_repo.fail(job, "boom")
+        assert job_repo.count_recent_failures(test_organization.id, "hubspot") == 1
+        assert job_repo.count_recent_failures(test_organization.id, "pipedrive") == 0
+
+    def test_reopen_failed_requeues(self, job_repo, test_organization, test_lead_response):
+        job = job_repo.enqueue(test_organization.id, test_lead_response.id, "hubspot")
+        job.attempts = 7
+        job_repo.fail(job, "no_connection")
+        assert job_repo.reopen_failed(test_organization.id, "hubspot") == 1
+        job_repo.db.refresh(job)
+        assert job.status == CrmSyncJobStatus.PENDING.value
+        assert job.attempts == 0
+        # Other providers untouched.
+        assert job_repo.reopen_failed(test_organization.id, "pipedrive") == 0