Merge pull request #76 from scrapinghub/metadata-normalize

Normalize the spider params schema.

Author: kmike

sh_scrapy/commands/shub_image_info.py

@@ -45,7 +45,7 @@ def run(self, args, opts):
             result['metadata'] = {}
             for spider_name in result['spiders']:
                 spider_cls = self.crawler_process.spider_loader.load(spider_name)
-                metadata_dict = get_spider_metadata(spider_cls)
+                metadata_dict = get_spider_metadata(spider_cls, normalize=True)
                 try:
                     # make sure it's serializable
                     json.dumps(metadata_dict)

tests/test_crawl.py

@@ -343,3 +343,95 @@ class MySpider(Spider):
     if not SPIDER_METADATA_AVAILABLE:
         del expected["metadata"]
     assert data == expected
+
+
+@pytest.mark.skipif(not SPIDER_METADATA_AVAILABLE, reason="scrapy-spider-metadata is not installed")
+def test_image_info_args(tmp_path):
+    project_dir = create_project(tmp_path, spider_text="""
+from enum import Enum
+from scrapy import Spider
+from scrapy_spider_metadata import Args
+from pydantic import BaseModel, Field
+
+class ToolEnum(Enum):
+    spanner = "spanner"
+    wrench = "wrench"
+
+class Parameters(BaseModel):
+    tool: ToolEnum = ToolEnum.spanner
+
+class MySpider(Args[Parameters], Spider):
+    name = "myspider"
+""")
+    out, _ = call_command(project_dir, "shub-image-info")
+    data = json.loads(out)
+    expected = {
+        "project_type": "scrapy",
+        "spiders": ["myspider"],
+        "metadata": {
+            "myspider": {
+                "param_schema": {
+                    "properties": {
+                        "tool": {
+                            "default": "spanner",
+                            "enum": ["spanner", "wrench"],
+                            "title": "Tool",
+                            "type": "string",
+                        },
+                    },
+                    "title": "Parameters",
+                    "type": "object",
+                },
+            },
+        },
+    }
+    if not SPIDER_METADATA_AVAILABLE:
+        del expected["metadata"]
+    assert data == expected
+
+
+@pytest.mark.skipif(not SPIDER_METADATA_AVAILABLE, reason="scrapy-spider-metadata is not installed")
+def test_image_info_args_metadata(tmp_path):
+    project_dir = create_project(tmp_path, spider_text="""
+from enum import Enum
+from scrapy import Spider
+from scrapy_spider_metadata import Args
+from pydantic import BaseModel, Field
+
+class ToolEnum(Enum):
+    spanner = "spanner"
+    wrench = "wrench"
+
+class Parameters(BaseModel):
+    tool: ToolEnum = ToolEnum.spanner
+
+class MySpider(Args[Parameters], Spider):
+    name = "myspider"
+    metadata = {"foo": 42}
+""")
+    out, _ = call_command(project_dir, "shub-image-info")
+    data = json.loads(out)
+    expected = {
+        "project_type": "scrapy",
+        "spiders": ["myspider"],
+        "metadata": {
+            "myspider": {
+                "foo": 42,
+                "param_schema": {
+                    "properties": {
+                        "tool": {
+                            "default": "spanner",
+                            "enum": ["spanner", "wrench"],
+                            "title": "Tool",
+                            "type": "string",
+                        },
+                    },
+                    "title": "Parameters",
+                    "type": "object",
+                },
+            },
+        },
+    }
+    if not SPIDER_METADATA_AVAILABLE:
+        del expected["metadata"]
+    assert data == expected

tox.ini

@@ -10,7 +10,8 @@ deps =
     hubstorage
     packaging
     py36-scrapy16: Scrapy==1.6
-    scrapy-spider-metadata; python_version >= "3.8"
+    scrapy-spider-metadata>=0.1.1; python_version >= "3.8"
+    pydantic>=2; python_version >= "3.8"
 
 commands =
     pytest --verbose --cov=sh_scrapy --cov-report=term-missing --cov-report=html --cov-report=xml {posargs: sh_scrapy tests}