[DRILL-8439] Getting col__ prefix for columns that are not special when extractHeader is enabled

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 1.21.0
• Component/s: Metadata, SQL Parser
• Labels: drill extractHeader

Environment

Enabled extractHeader in the csv config of dfs plugin.
No. of drillbits: Single
OS: Windows

Language

English

• Assignee: Unassigned
• Reporter: Diksha Chaturvedi
• Created: 31/May/23 15:08
• Updated: 07/Jun/23 11:47

Description

As per documentation, Drill appends col_ to the columns that start with a number or special characters.

/**
 * Prefix used to replace non-alphabetic characters at the start of
 * a column name. For example, $foo becomes col_foo. Used
 * because SQL does not allow _foo.
 */
public static final String COLUMN_PREFIX = "col_";

But in my case I'm getting it even for all alphabetical column name.

---

I have the following data in the CSV file,

PRODUCTID	PRODUCTNAME	SUPPLIERID	CATEGORYID	UNIT	PRICE
1	Chais	1	1	10 boxes x 20 bags	18
2	Chang	1	1	24 - 12 oz bottles	19
3	Aniseed Syrup	1	2	12 - 550 ml bottles	10
4	Chef Anton's Cajun Seasoning	2	2	48 - 6 oz jars	22
5	Chef Anton's Gumbo Mix	2	2	36 boxes	21.35

 

While querying on the csv file using following query:

SELECT * FROM dfs.`/var/lib/PRODUCT.csv`

The output is 

---

I know about other criterias like

#UNITS is changed to col_UNITS

FINANCIAL$RECORD is changed to FINANCIAL_RECORD

But what's with PRODUCTID; Why is it changed to col__PRODUCTID_? In this case it has appended extra underscores also.