csv.Sniffer.snif doesn't set up the dialect properly for a csv created with dialect=csv.excel_tab and containing quote (") char

[GhislainHivon]

BPO	17829
Files	csv_sniffing_excel_tab.py: Exemple of sniffing csv with dialect=csv.excel_tab and quote in data

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = None
closed_at = None
created_at = <Date 2013-04-24.15:06:00.650>
labels = ['type-bug']
title = 'csv.Sniffer.snif doesn\'t set up the dialect properly for a csv created with dialect=csv.excel_tab and containing quote (") char'
updated_at = <Date 2014-03-28.10:04:37.158>
user = 'https://bugs.python.org/GhislainHivon'

bugs.python.org fields:

activity = <Date 2014-03-28.10:04:37.158>
actor = 'Antoon.Pardon'
assignee = 'none'
closed = False
closed_date = None
closer = None
components = []
creation = <Date 2013-04-24.15:06:00.650>
creator = 'GhislainHivon'
dependencies = []
files = ['30001']
hgrepos = []
issue_num = 17829
keywords = []
message_count = 3.0
messages = ['187709', '214800', '215031']
nosy_count = 3.0
nosy_names = ['GhislainHivon', 'Antoon.Pardon', 'dmi.baranov']
pr_nums = []
priority = 'normal'
resolution = None
stage = None
status = 'open'
superseder = None
type = 'behavior'
url = 'https://bugs.python.org/issue17829'
versions = ['Python 2.7', 'Python 3.2']

[GhislainHivon]

When sniffing the dialect of a file created with the csv module with dialect=csv.excel_tab and one of the row contain a quote ("), the delimiter is set to ' ' instead of '\t'.

[AntoonPardon]

I had a look at this and have the following remarks.

1. the file csv_sniffing_excel_tab.py no longer works with python 3.3. It now produces the folowing traceback:

Traceback (most recent call last):
  File "csv_sniffing_excel_tab.py", line 36, in <module>
    create_file()
  File "csv_sniffing_excel_tab.py", line 23, in create_file
    writer.writerows(test_data)
TypeError: 'str' does not support the buffer interface

2. The problem seems to be in the _guess_quote_and_delimiter method. If you always call _guess_delimiter, the sniffer give the correct result.

3. As far as I understand the problem is the first regular expression:
   (?P<delim>[^\\w\\n"\\'])(?P<space> ?)(?P<quote>["\']).*?(?P=quote)(?P=delim)

Now if we have a line as the following

273:MVREGR1:ByEuPo:"Baryton ""Euphonium"" populaire"

The delim group will match the space, the space group will match nothing the quote group will match " the non-group pattern will match "Euphonium" followed by the quote group matching " again and the delim group matching the space.

And so we get the wrong delimiter.

[AntoonPardon]

I included a patch (against 2.7) that seems to make the test work.

The patch prohibits the delim group to match a space.

[pradeepkumarai]

I included a patch (against 2.7) that seems to make the test work.

The patch prohibits the delim group to match a space.