Pandas Overzealous ParserWarning when using dialect param on read_csv

Code Sample

import csv
import pandas

csv.register_dialect('mydialect', delimiter='|', escapechar='\\', doublequote=False, quoting=csv.QUOTE_NONE)

pandas.read_csv('example.csv', dialect='mydialect')

Problem description

I believe that this warning was added to handle conflicts between any dialect-provided csv parsing params and the equivalent params passed individually to read_csv(). However, I don't think it is working as intended. Notice, in my example above I only specify these values by way of the dialect, yet I nonetheless receive these warnings:

ParserWarning: Conflicting values for 'doublequote': 'True' was provided, but the dialect specifies '0'. Using the dialect-specified value.
ParserWarning: Conflicting values for 'escapechar': 'None' was provided, but the dialect specifies '\'. Using the dialect-specified value.
ParserWarning: Conflicting values for 'quoting': '0' was provided, but the dialect specifies '3'. Using the dialect-specified value.
ParserWarning: Conflicting values for 'delimiter': ',' was provided, but the dialect specifies '|'. Using the dialect-specified value.

The reason, I believe, is that the check that tests for conflicts in these values is effectively comparing against the default values when no value is specified as a kwarg. This means anytime I am specifying a value via the dialect which is different from the default this warning is raised.

See:

https://github.com/pandas-dev/pandas/blob/199e52518bd4b5dfc7f8b0652b53b1679c80364a/pandas/io/parsers.py#L744

and

https://github.com/pandas-dev/pandas/blob/199e52518bd4b5dfc7f8b0652b53b1679c80364a/pandas/io/parsers.py#L750

Expected Output

No warnings unless there is a conflict between a dialect value and that of a param explicitly passed to read_csv.

Comment From: kylekeppler

Seconded. Seeing the same thing when I create a custom dialect.

Comment From: phofl

This works and has tests