When dumping a table to a CSV file using the /tableToCsv method in the marketplace CSVAdapter API Service, every 101st record is corrupted/lost. The line feed is left off each 100th record, which causes the following record to be appended. Some CSV parsing libraries ignore the extra columns, others throw an error about too many columns. I don’t think it matters what options or tables are used, but for what it is worth, here are the parameters I used for one corrupt example:
{
"filePath": "export/blah.csv",
"tableName": "blah",
"tableOptions": {
"relations": ["bans","routes","zones"],
"excludeProps": ["ownerId", "updated"]
},
"csvOptions": {
"append": false,
"delimiter": "\t",
"eof": false,
"header": true,
"quoted": true
}
}
If you change the pageSize through the tableOptions parameter, say to 20, then the 21st record is hosed.