parse platenzaak deals

apps/vinyl/src/assets.py

@@ -12,6 +12,7 @@ from dagster_polars.patito import patito_model_to_dagster_type
 from jinja2 import Environment, FileSystemLoader
 from models import Deal
 from partitions import daily_partitions_def, multi_partitions_def
+from platenzaak.parse import parse as parse_platenzaak
 from platenzaak.scrape import scrape as scrape_platenzaak
 from plato.parse import parse as parse_plato
 from plato.scrape import scrape as scrape_plato
@@ -65,26 +66,24 @@ def deals(context: dg.AssetExecutionContext) -> pl.DataFrame:
                     logger.error("Failed to load CSV file!", error=e)
         raise dg.Failure(f"Cannot materialize for the past: {date.date()}")
 
-    if source == "plato":
-        logger.info("Scraping Plato")
-        df = scrape_plato()
-        logger.info("Scraped Plato", rows=len(df), head=df.head().to_markdown())
-        ic(df.columns)
-        return pl.from_pandas(df.assign(**partition_key))
-    if source == "sounds":
-        logger.info("Scraping Sounds")
-        df = scrape_sounds()
-        logger.info("Scraped Sounds", rows=len(df), head=df.head().to_markdown())
-        ic(df.columns)
-        return pl.from_pandas(df.assign(**partition_key))
-    if source == "platenzaak":
-        logger.info("Scraping Platenzaak")
-        df = scrape_platenzaak(logger=logger)
-        logger.info("Scraped Sounds", rows=len(df), head=df.head().to_markdown())
-        ic(df.columns)
-        return pl.from_pandas(df.assign(**partition_key))
+    match source:
+        case "plato":
+            logger.info("Scraping Plato")
+            df = scrape_plato()
+            logger.info("Scraped Plato", rows=len(df), head=df.head().to_markdown())
+        case "sounds":
+            logger.info("Scraping Sounds")
+            df = scrape_sounds()
+            logger.info("Scraped Sounds", rows=len(df), head=df.head().to_markdown())
+        case "platenzaak":
+            logger.info("Scraping Platenzaak")
+            df = scrape_platenzaak(logger=logger)
+            logger.info("Scraped Sounds", rows=len(df), head=df.head().to_markdown())
+        case _:
+            raise ValueError(f"Unknown source: {source}!")
 
-    raise NotImplementedError(f"No implementation for source {source}")
+    ic(df.columns)
+    return pl.from_pandas(df.assign(**partition_key))
 
 
 @asset(
@@ -111,9 +110,10 @@ def cleaned_deals(
             parsed_df = parse_plato(df)
         case "sounds":
             parsed_df = parse_sounds(df)
+        case "platenzaak":
+            parsed_df = parse_platenzaak(df)
         case _:
-            context.log.warning(f"Unknown source: {source}!")
-            return
+            raise ValueError(f"Unknown source: {source}!")
 
     ic(parsed_df.collect_schema())