add dryrun option to truncateXMLDump()

wikiteam3/dumpgenerator/dump/xmldump/xml_truncate.py

@@ -24,8 +24,12 @@ def addNewline(filename: str) -> None:
         f.write("\n")
 
 
-def truncateXMLDump(filename: str) -> str:
-    """Removes incomplete <page> elements from the end of XML dump files"""
+def truncateXMLDump(filename: str, dryrun: bool = False) -> str:
+    """
+    Removes incomplete <page> elements from the end of XML dump files
+    
+    dryrun: bool - returns the incomplete segment without truncating the file
+    """
 
     with FileReadBackwards(filename, encoding="utf-8") as frb:
         incomplete_segment: str = ""
@@ -36,6 +40,8 @@ def truncateXMLDump(filename: str) -> str:
         while xml_line and "</page>" not in xml_line:
             incomplete_segment = xml_line + incomplete_segment
             xml_line = frb.readline()
+    if dryrun:
+        return incomplete_segment
     incomplete_segment_size = len(incomplete_segment.encode("utf-8"))
     file_size = os.path.getsize(filename)
     if file_size > incomplete_segment_size: