When loading a multipart CSV resource where one file part does not end with a newline, the last line of that part is concatenated directly with the first data line of the next part, producing malformed rows.
For example, given:
chunk1.csv: id,name\n1,english (no trailing newline)chunk2.csv: id,name\n2,german\n
The resulting byte stream is id,name\n1,english2,german\n, so the parser produces one row {"id": 1, "name": "english2"} instead of two correct rows.
When loading a multipart CSV resource where one file part does not end with a newline, the last line of that part is concatenated directly with the first data line of the next part, producing malformed rows.
For example, given:
chunk1.csv:id,name\n1,english(no trailing newline)chunk2.csv:id,name\n2,german\nThe resulting byte stream is
id,name\n1,english2,german\n, so the parser produces one row{"id": 1, "name": "english2"}instead of two correct rows.