[ZEPPELIN-2956] Downloaded CSV/TSV data will get unexpected division when the column value contains both delimiter and quotation mark.

• Type: Bug
• Status: Open
• Priority: Critical
• Resolution: Unresolved
• Affects Version/s: 0.6.2, 0.7.2, 0.7.3
• Component/s: front-end

Environment

zeppelin 0.7.2
jdk 1.7

• Assignee: Unassigned
• Reporter: bruce xu
• Created: 27/Sep/17 09:47
• Updated: 29/May/18 16:18

Description

After downloading csv/tsv data from zeppelin paragraph, I opened the file through Excel and found that the data is not seperated as expected.

Then I found that's because the colunm value contains both delimiter and quotation mark, so Excel could not recognize and seperate the value correctly.

Here are zeppelin downloaded csv data example:

hello,beijing,i,love,you -> can be well recognized
hello,beijing,i,lo"v"e,you -> can be well recognized
hello,beijing,i,lo"v,"e,you -> can not be well recognized

I met this bug in our production environment, the column value format might be json, and json format data may contain quotation mark and comma.

Issue Links

• Blocked: ZEPPELIN-1803 CSV export doesn't conform to RFC-4180: exported csv is broken in some cases
• is related to: ZEPPELIN-3511 remove old button "Download Data as CSV/TSV"
• links to: GitHub Pull Request #2601