TOC »
csv-abnf
Description
The csv-abnf library contains procedures for parsing and formatting of comma-separated values (CSV) as described in RFC 4180. There are several differences with the RFC:
- The RFC prescribes CRLF standard network line breaks, but many CSV files have platform-dependent line endings, so this library accepts any sequence of CRs and LFs as a line break.
- The format of header lines is exactly like a regular record and the presence of a header can only be determined from the mime type. This library treats all lines as regular records.
- The formal grammar specifies that fields can contain only certain US ASCII characters, but the specification of the MIME type allows for other character sets. This library allow all characters in fields, except for the field delimiter character, CRs and LFs in unquoted fields.
- According to the RFC, the records all have to have the same length. This library allows variable length records.
- The delimiter character is specified by the user and can be a character other than comma, or an SRFI-14 character set.
See also csv-xml and tabular, a related library for parsing and formatting comma- and delimiter-separated values as well as fixed-width columns, implemented with a different set of combinators.
Library Procedures
- csv-record? Xprocedure
Returns #t if the given object is a csv-record, #f otherwise.
- list->csv-record LISTprocedure
Takes in a list of values and creates a csv-record object.
- csv-record->list CSV-RECORDprocedure
Returns the list of values contained in the given csv-record object.
Parsing Procedures: Preliminaries
The grammar in csv-abnf is built with abnf's abnf-lens component, which makes every rule bidirectional: the same rule produces both the parser behind make-parser below and the printer behind make-format further down (see the abnf library for more on abnf-lens, and the "Bidirectional grammar rules" section below for the rules themselves).
Parsing Procedures: csv-abnf
- make-parser #!optional DELIMITERprocedure
make-parser returns a CSV parsing procedure. Optional argument DELIMITER specifies the field delimiter (comma by default). DELIMITER can be a character, or an SRFI-14 character set. The returned procedure takes a string (or, for backwards compatibility, a list of characters) and returns a list of csv-record objects, one per record:
(<#csv-record (FIELD1 FIELD2 ...)> <#csv-record ...> ...)
(import csv-abnf) (define parse-csv (make-parser #\|)) (parse-csv "a|b|c") (map csv-record->list (parse-csv "a|b|c")) ==> (("a" "b" "c"))
Formatting procedures
- make-format #!optional DELIMITERprocedure
Returns procedures for outputting individual field values, CSV records, and lists of CSV records, where each list is printed on a separate line.
Procedure FORMAT-CELL takes in a value, obtains its string representation via format, and surrounds the string with quotes, if it contains characters that need to be escaped (such as quote characters, the delimiter character, or newlines).
Procedure FORMAT-RECORD takes in a row and returns its string representation, based on the strings produced by FORMAT-CELL and the delimiter character. A row can be given either as a bare list of field values, or as a csv-record (e.g. one obtained back from make-parser) -- there is no need to wrap a plain list with list->csv-record before printing it.
Procedure FORMAT-CSV takes in a list of rows (each a bare list or a csv-record; the two can be freely mixed) and produces a string representation using FORMAT-RECORD, with a trailing CRLF after every record.
Example:
(import csv-abnf) (define-values (fmt-cell fmt-record fmt-csv) (make-format #\;)) (fmt-cell "hello") => "hello" ;; This is quoted because it contains delimiter-characters (fmt-cell "one;two;three") => "\"one;two;three\"" ;; This is quoted because it contains quotes, which are then doubled for escaping (fmt-cell "say \"hi\"") => "\"say \"\"hi\"\"\"" ;; Rows are plain lists -- no list->csv-record wrapping needed (fmt-record '("hi there" "let's say \"hello world\" again" "until we are bored")) => "hi there;\"let's say \"\"hello world\"\" again\";until we are bored" ;; And an example of how to quickly convert a list of lists ;; to a CSV string containing the entire CSV file (fmt-csv '(("one" "two") ("and another \"line\"" "of csv stuff"))) => "one;two\r\n\"and another \"\"line\"\"\";of csv stuff\r\n" ;; csv-record objects (e.g. rows just read back with make-parser) work ;; directly too, with no unwrap/rewrap step: (fmt-csv (map list->csv-record '(("one" "two") ("and another \"line\"" "of csv stuff")))) => "one;two\r\n\"and another \"\"line\"\"\";of csv stuff\r\n"
Bidirectional grammar rules
The three rules below are what make-parser and make-format are each built from. Every rule pairs a parser with a printer. The rules are exported directly, in case you want to use them to embed the CSV grammar as part of a larger abnf-lens grammar.
Each rule is a bp value, used with the bp-parse and bp-print procedures. Both the bp record and these two procedures are documented in the abnf page, along with bi-iso, bi-seq, and the other combinators the rules below are built from.
- bi-field DELIMITERprocedure
A single field: a string, quoted only when it needs to be, i.e. when it contains DELIMITER, a quote character, a CR, or a LF.
- bi-record DELIMITERprocedure
One record: a csv-record whose fields are separated by DELIMITER, with no trailing line ending. Its printer also accepts a raw list of field values, the same as FORMAT-RECORD above.
- bi-csv DELIMITERprocedure
A whole CSV file: a list of rows, each followed by a line ending. Its printer accepts a list whose elements are each either a csv-record or a raw list of field values.
(import csv-abnf abnf-lens) ;; bp-parse returns (list (list matched-value) remaining-stream); ;; (car (car ...)) unwraps it to the list of csv-records. (car (car (bp-parse (bi-csv #\,) "a,b\r\nc,d\r\n" (lambda (s) (error "parse failed" s))))) ;; => (list (list->csv-record '("a" "b")) (list->csv-record '("c" "d"))) (bp-print (bi-csv #\,) '(("a" "b") ("c" "d"))) ;; => "a,b\r\nc,d\r\n"
Repository
https://github.com/iraikov/chicken-csv-abnf
Version History
- 6.2 Ported to CHICKEN 6; added bidirectional operations based on abnf-lens
- 5.3 Added module csv-char-list
- 5.1 utf8-related bug fixes
- 5.0 Compatibility with abnf 6.0 / using utf8 for char operations
- 4.7 Created module csv-string
- 4.6 Ensure unit test script returns proper exit code
- 4.5 Compatibility with CharLex->CoreABNF constructor in abnf
- 4.4 Added cr and lf to set of characters allows in escaped strings
- 4.3 Added csv-record? to list of exported identifiers
- 4.1 Fixes in the handling of escaped strings
- 4.0 Compatibility with abnf 5
- 3.2 Fixes to reflect changes in the regex API in Chicken 4.6.0
- 3.1 Added regex as an explicit dependency
- 3.0 Implemented typeclass interface
- 2.0 Added formatting routines
- 1.0 Initial Release
License
Copyright 2009-2026 Ivan Raikov This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. A full copy of the GPL license can be found at <http://www.gnu.org/licenses/>.