참고소스 수정본

참고/trafilatura-master/tests/resources/zerolength.cfg

@@ -0,0 +1,50 @@
+# Defines settings for trafilatura (https://github.com/adbar/trafilatura)
+
+[DEFAULT]
+
+# Download
+DOWNLOAD_TIMEOUT = 10
+MAX_FILE_SIZE = 20000000
+MIN_FILE_SIZE = 10
+
+# sleep between requests
+SLEEP_TIME = 0.25
+
+# one line per user-agent
+USER_AGENTS =
+    Firefox
+    Chrome
+    
+# cookie for HTTP requests
+COOKIE = yummy_cookie=choco; tasty_cookie=strawberry
+
+# maximum number of redirects that we will follow
+MAX_REDIRECTS = 2
+
+
+# Extraction
+MIN_EXTRACTED_SIZE = 0
+MIN_EXTRACTED_COMM_SIZE = 0
+MIN_OUTPUT_SIZE = 0
+MIN_OUTPUT_COMM_SIZE = 0
+
+
+# discard documents with too many elements
+MAX_TREE_SIZE = 100
+
+
+# Set to 0 to disable signal
+EXTRACTION_TIMEOUT = 0
+
+
+# Deduplication
+MIN_DUPLCHECK_SIZE = 10
+MAX_REPETITIONS = 3
+
+
+# Extraction option for Htmldate
+EXTENSIVE_DATE_SEARCH = off
+
+
+# URLs in feeds and sitemaps
+EXTERNAL_URLS = off