Disable the WACZ exporter extension when the downloader middleware or…

… spider middleware is enabled (#16)
q-m · Oct 31, 2024 · e044c9b · e044c9b
1 parent 9ea5f71
commit e044c9b
Showing 1 changed file with 6 additions and 0 deletions.
diff --git a/scrapy_webarchive/extensions.py b/scrapy_webarchive/extensions.py
@@ -53,6 +53,12 @@ def __init__(self, settings: Settings, crawler: Crawler) -> None:
 
         if not self.settings["SW_EXPORT_URI"]:
             raise NotConfigured
+
+        if "scrapy_webarchive.spidermiddlewares.WaczCrawlMiddleware" in settings.getlist('SPIDER_MIDDLEWARES'):
+            raise NotConfigured("You must disable the WaczCrawlMiddleware before you can use this extension.")
+
+        if "scrapy_webarchive.downloadermiddlewares.WaczMiddleware" in settings.getlist('DOWNLOADER_MIDDLEWARES'):
+            raise NotConfigured("You must disable the WaczMiddleware before you can use this extension.")
 
         self.store: FilesStoreProtocol = self._get_store(spider_name=crawler.spider.name)
         self.writer = WarcFileWriter(collection_name=crawler.spider.name)