dcf Read and Write Data in DCF Format

Description

Reads or writes an R object from/to a file in Debian Control File format.

Usage

read.dcf(file, fields = NULL, all = FALSE, keep.white = NULL)

write.dcf(x, file = "", append = FALSE, useBytes = FALSE,
          indent = 0.1 * getOption("width"),
          width = 0.9 * getOption("width"),
          keep.white = NULL)

Arguments

file	either a character string naming a file or a connection. "" indicates output to the console. For read.dcf this can name a compressed file (see gzfile).
fields	Fields to read from the DCF file. Default is to read all fields.
all	a logical indicating whether in case of multiple occurrences of a field in a record, all these should be gathered. If all is false (default), only the last such occurrence is used.
keep.white	a character string with the names of the fields for which whitespace should be kept as is, or NULL (default) indicating that there are no such fields. Coerced to character if possible. For fields where whitespace is not to be kept as is, read.dcf removes leading and trailing whitespace, and write.dcf folds using strwrap.
x	the object to be written, typically a data frame. If not, it is attempted to coerce x to a data frame.
append	logical. If TRUE, the output is appended to the file. If FALSE, any existing file of the name is destroyed.
useBytes	logical to be passed to writeLines(), see there: “for expert use”.
indent	a positive integer specifying the indentation for continuation lines in output entries.
width	a positive integer giving the target column for wrapping lines in the output.

Details

DCF is a simple format for storing databases in plain text files that can easily be directly read and written by humans. DCF is used in various places to store R system information, like descriptions and contents of packages.

The DCF rules as implemented in R are:

1. A database consists of one or more records, each with one or more named fields. Not every record must contain each field. Fields may appear more than once in a record.

2. Regular lines start with a non-whitespace character.

3. Regular lines are of form tag:value, i.e., have a name tag and a value for the field, separated by : (only the first : counts). The value can be empty (i.e., whitespace only).

4. Lines starting with whitespace are continuation lines (to the preceding field) if at least one character in the line is non-whitespace. Continuation lines where the only non-whitespace character is a . are taken as blank lines (allowing for multi-paragraph field values).

5. Records are separated by one or more empty (i.e., whitespace only) lines.

6. Individual lines may not be arbitrarily long; prior to R 3.0.2 the length limit was approximately 8191 bytes per line.

Note that read.dcf(all = FALSE) reads the file byte-by-byte. This allows a ‘DESCRIPTION’ file to be read and only its ASCII fields used, or its Encoding field used to re-encode the remaining fields.

write.dcf does not write NA fields.

Value

The default read.dcf(all = FALSE) returns a character matrix with one row per record and one column per field. Leading and trailing whitespace of field values is ignored unless a field is listed in keep.white. If a tag name is specified in the file, but the corresponding value is empty, then an empty string is returned. If the tag name of a field is specified in fields but never used in a record, then the corresponding value is NA. If fields are repeated within a record, the last one encountered is returned. Malformed lines lead to an error.

For read.dcf(all = TRUE) a data frame is returned, again with one row per record and one column per field. The columns are lists of character vectors for fields with multiple occurrences, and character vectors otherwise.

Note that an empty file is a valid DCF file, and read.dcf will return a zero-row matrix or data frame.

For write.dcf, invisible NULL.

Note

As from R 3.4.0, ‘whitespace’ in all cases includes newlines.

References

https://www.debian.org/doc/debian-policy/ch-controlfields.html.

Note that R does not require encoding in UTF-8, which is a recent Debian requirement. Nor does it use the Debian-specific sub-format which allows comment lines starting with #.

See Also

write.table.

available.packages, which uses read.dcf to read the indices of package repositories.

Examples

## Create a reduced version of the DESCRIPTION file in package 'splines'
x <- read.dcf(file = system.file("DESCRIPTION", package = "splines"),
              fields = c("Package", "Version", "Title"))
write.dcf(x)

## An online DCF file with multiple records
con <- url("https://cran.r-project.org/src/contrib/PACKAGES")
y <- read.dcf(con, all = TRUE)
close(con)
utils::str(y)