时间戳

首页 > 解决方案 > Couldn't parse csv file if value contains from '

yandex - Couldn't parse csv file if value contains from '

问题描述

Here is demo DDL:

CREATE TABLE X (
    eventdate        Date default toDate(now()),
    f1               String,
    f2               String,
    f3               String
) ENGINE = MergeTree(eventdate, (f1, f2, f3), 8192)

When I try to import data, clickhous fails with error:

$echo "US,'Ewa Beach, Test" | clickhouse-client --database default --query "insert into X(f1, f2, f3) format CSV"

Code: 27. DB::Exception: Cannot parse input: expected , at end of
stream.: (at row 1)

Here is my click house version(it time of writing it's latest version, if I am not mistaking):

$ clickhouse-client --version
ClickHouse client version 1.1.54385.

The problem is ' symbol in the value of field f2. Please, any workaround to make it works?

Maybe is there some option to force using a only double quote as a string delimiter, and consider single quote as plain symbol without special meaning? Please, any solution is welcome.

I just tested with same values but as tsv, and it works. But unfortunately I cannot migrate to tsv right now. I need to find out some workaround for csv to make it work with signle quote in field value.

P.S.

Interesting that if left only two fields, it parses csv correctly.

CREATE TABLE X (
    eventdate        Date default toDate(now()),
    f1               String,
    f2               String
) ENGINE = MergeTree(eventdate, (f1, f2), 8192)

The following query workds:

$echo "US,'Ewa Beach" | clickhouse-client --database default --query "insert into X(f1, f2) format CSV"

Here is related issue on the GitHub

标签： yandexclickhouse

解决方案

我刚刚遇到了同样的问题。我创建了一个 PR https://github.com/yandex/ClickHouse/pull/2574

有了它，您可以指定使用format_allow_csv_single_quote=0以 a 开头的字符串字段解析 csv 数据'

编辑：

现在已经合并了。

推荐阅读