-
|
Hello I expect the second field in the row Quote from manual about Content > Enclosure parameter : I've also tried to use Escape parameter and that was promising but sometimes that breaks the parsing. Thank you very much in advance test_text_file_input_enclosure.csv |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment 3 replies
-
|
In that example the inner quotes look like standard CSV doubled enclosure characters, not a separate escape sequence. So I would first test with the quote character only as the enclosure, and leave Escape empty. Try the Text file input with only these CSV-related settings first: With that convention, this field: "hello this is a ""nice"" comment"is intended to become: Setting Escape to the same character as Enclosure can change how the parser walks through the quoted field, which would explain why rows that previously parsed start getting skipped or shifted. Use Escape only if your source file uses a separate escape style, for example backslash escaping, rather than standard doubled quotes. If empty Escape still skips rows 6-8, the attached CSV/HPL is already a useful repro; adding the exact Hop version and CSV-mode/error-handling settings would make it much easier for maintainers to compare parser behavior. |
Beta Was this translation helpful? Give feedback.
-
|
Thank you for your answer Text file input > Error handling tab is all unchecked (if that what you meant by CSV-mode/error-handling settings) But I'm having another problem with some java version or OS version or ??? I've made this table to explain when it works and when it doesnt. My target is to run the pipeline on the linux server using hop-run I can keep trying with another java version or provider listed in the hop 2.17 documentation |
Beta Was this translation helpful? Give feedback.
-
|
That new stack trace changes the issue: empty Escape is the right CSV setting, but Hop is passing a null escape pattern into I would stop testing Java vendors for now and report it as a Hop bug with the smallest repro:
As a temporary workaround, try setting Escape to a character that never appears in the file rather than quote itself, for example a backslash if the input never uses backslash escapes. That is not as clean as empty Escape, but it may avoid both the NPE and the doubled-quote parsing problem. The important point for maintainers is that empty Escape should not become a null pattern crash; it should be treated as no escape character. |
Beta Was this translation helpful? Give feedback.
-
|
Alright i'm going to submit an issue about the NPE when Escape is empty About the temporary workaround, it doesn't seem to work with I'm marking your first answer as correct |
Beta Was this translation helpful? Give feedback.
In that example the inner quotes look like standard CSV doubled enclosure characters, not a separate escape sequence. So I would first test with the quote character only as the enclosure, and leave Escape empty.
Try the Text file input with only these CSV-related settings first:
With that convention, this field:
"hello this is a ""nice"" comment"is intended to become:
Setting Escape to the same character as Enclosure can change how the parser walks through the quoted field, which would explain why rows that previously parsed start getting skipped or shifted. Use Escape only if your source file uses a separate esca…