Help with FasterCSV monkey patch


#1

Hello, I’ve been trying to monkey patch FCSV to allow escaped colsep
characters within a field:

FCSV.parse “here\, it is, other fields” #need: [[“here\, it is”,
“other fields”]]
^

So what I tried was modifying the :csv_row regexp on the init_parsers
method

1797: ([^#{esc_quote}#{esc_col_sep}]*) # unquoted fields

which I changed with this:

1797: ((?>[^#{esc_quote}#{esc_col_sep}]*
1798: \#{esc_col_sep}
1799: [^#{esc_quote}#{esc_col_sep}])) # unquoted fields

This didn’t work though :(, though my gut tells me I’m close :wink:

Thanks in advance for any help with this :slight_smile:


#2

On Wed, Jan 21, 2009 at 4:50 PM, Alfredo M. removed_email_address@domain.invalid
wrote:

Hello, I’ve been trying to monkey patch FCSV to allow escaped colsep
characters within a field:

FCSV.parse “here\, it is, other fields” #need: [[“here\, it is”,
“other fields”]]
^

Is there a reason why you can’t use normal CSV quoted text escaping?

FCSV.parse ‘“here, it is”, other fields’
=> [[“here, it is”, " other fields"]]

-greg


#3

Gregory B. wrote:

Is there a reason why you can’t use normal CSV quoted text escaping?

FCSV.parse ‘“here, it is”, other fields’
=> [[“here, it is”, " other fields"]]

-greg

Unfortunately yes, the parsing on the app I’m working on has to be
foolproof, allowing such otherwise-invalid format.


#4

On Jan 21, 2009, at 3:50 PM, Alfredo M. wrote:

Hello, I’ve been trying to monkey patch FCSV to allow escaped colsep
characters within a field:

FCSV.parse “here\, it is, other fields” #need: [[“here\, it is”,
“other fields”]]

My advice is don’t do this, unfortunately. :frowning:

I’ve tried to add this feature to FasterCSV multiple times now. It’s
very hard and I haven’t been able to find a good way to do it for
general cases. I fully admit this is a failing of FasterCSV, it’s
very dependent on the proper CSV format.

You probably have three reasonable choices:

  • Feed FasterCSV a line at a time, rescue the MalformedCSVError, and
    switch strategies on those lines
  • Preprocess all lines to be sure they are valid CSV and then hand
    them off
  • Build your own parser

Sorry I wasn’t more help.

James Edward G. II