fast parser

2026-04-24 12:39:05 +03:00
parent 3a3222c7dc
commit fc1bea562a
19 changed files with 2482 additions and 92 deletions
--- a/tests/test_fast_client.py
+++ b/tests/test_fast_client.py
@@ -0,0 +1,117 @@
+from __future__ import annotations
+
+import json
+
+from iaai_scraper.browser.fast_client import (
+    DETAIL_MARKER,
+    LISTING_MARKER,
+    build_resizer_images_from_keys,
+    is_challenge_response,
+    parse_listing_page,
+    parse_product_details_vm,
+)
+from iaai_scraper.parsing.fast_mapper import FastCarMapper
+
+
+def test_parse_listing_page_extracts_hidden_payloads() -> None:
+    gbp = {"CurrentPage": 1, "PageSize": 100}
+    vehicles = [
+        {
+            "Id": "45078011~US",
+            "Tenant": "US",
+            "InventoryStatus": "RS",
+            "Currency": "USD",
+            "PreBidIndicator": True,
+        }
+    ]
+    html = (
+        f'<input id="GBPSearchQuery" value="{json.dumps(gbp).replace(chr(34), "&quot;")}" />'
+        f'<input id="VehicleDetails" value="{json.dumps(vehicles).replace(chr(34), "&quot;")}" />'
+        '<input id="ResultCount" value="1" />'
+        '<input id="PageSize" value="100" />'
+        '<input id="CurrentPage" value="1" />'
+    )
+
+    parsed = parse_listing_page(html)
+
+    assert parsed.result_count == 1
+    assert parsed.page_size == 100
+    assert parsed.current_page == 1
+    assert parsed.vehicles[0].inventory_id == "45078011~US"
+    assert parsed.vehicles[0].inventory_status == "RS"
+
+
+def test_parse_product_details_vm_and_map_record() -> None:
+    payload = {
+        "inventoryView": {
+            "attributes": {
+                "Id": "45078011~US",
+                "Year": "2002",
+                "Make": "ACURA",
+                "Model": "RSX",
+                "Series": "BASE",
+                "Currency": "USD",
+                "ODOValue": "219187",
+                "ExteriorColor": "GRAY",
+                "DriveLineTypeDesc": "FWD",
+                "Transmission": "Automatic",
+                "BodyStyleName": "COUPE",
+                "EngineSize": "2.0L I-4",
+                "VehicleGrade": "50",
+                "PrimaryDamageDesc": "NORMAL WEAR & TEAR",
+            },
+            "imageDimensions": {
+                "keys": {
+                    "$values": [
+                        {"k": "45078011~US~I1", "w": 1600, "h": 1200, "i": 0},
+                    ]
+                }
+            },
+        },
+        "auctionInformation": {
+            "prebidInformation": {"decimalHighBidAmount": "600", "highBidAmount": "$600"},
+            "biddingInformation": {"buyNowPrice": "$0"},
+        },
+    }
+    html = f'<script id="ProductDetailsVM" type="application/json">{json.dumps(payload)}</script>'
+
+    parsed = parse_product_details_vm(html)
+    record = FastCarMapper().map_payload_to_record(
+        detail_payload=parsed,
+        vehicle_url="https://www.iaai.com/VehicleDetail/45078011~US",
+    )
+
+    assert record.origin_id == "iaai:45078011~US"
+    assert record.brand == "ACURA"
+    assert record.model == "RSX BASE"
+    assert record.year == 2002
+    assert record.price == 600
+    assert record.drive == "FWD"
+    assert record.gearbox == "AT"
+    assert record.body_type == "COUPE"
+    assert record.engine_volume == 2000
+    assert record.is_damaged is False
+    assert len(record.images) == 1
+
+
+def test_build_resizer_images_from_keys_deduplicates_and_orders() -> None:
+    keys = [
+        {"k": "abc~1", "w": 2000, "h": 1500, "i": 2},
+        {"k": "abc~1", "w": 2000, "h": 1500, "i": 2},
+        {"k": "abc~2", "w": 1800, "h": 1200, "i": 1},
+    ]
+    images = build_resizer_images_from_keys(keys)
+
+    assert len(images) == 2
+    assert images[0]["order_index"] == 1
+    assert "imageKeys=abc~2" in str(images[0]["fullres_image"])
+
+
+def test_is_challenge_response_marker_rules() -> None:
+    assert is_challenge_response(status_code=403, body_text="", expected_marker=None) is True
+    assert is_challenge_response(status_code=200, body_text="<html>blocked</html>", expected_marker=LISTING_MARKER) is True
+    assert is_challenge_response(
+        status_code=200,
+        body_text=f'<html>{DETAIL_MARKER}<script src="/_Incapsula_Resource"></script></html>',
+        expected_marker=DETAIL_MARKER,
+    ) is False