codoc Check Code/Documentation Consistency

Description

Find inconsistencies between actual and documented ‘structure’ of R objects in a package. codoc compares names and optionally also corresponding positions and default values of the arguments of functions. codocClasses and codocData compare slot names of S4 classes and variable names of data sets, respectively.

Usage

codoc(package, dir, lib.loc = NULL,
      use.values = NULL, verbose = getOption("verbose"))
codocClasses(package, lib.loc = NULL)
codocData(package, lib.loc = NULL)

Arguments

package	a character string naming an installed package.
dir	a character string specifying the path to a package's root source directory. This must contain the subdirectories ‘man’ with R documentation sources (in Rd format) and ‘R’ with R code. Only used if package is not given.
lib.loc	a character vector of directory names of R libraries, or NULL. The default value of NULL corresponds to all libraries currently known. The specified library trees are used to search for package.
use.values	if FALSE, do not use function default values when comparing code and docs. Otherwise, compare all default values if TRUE, and only the ones documented in the usage otherwise (default).
verbose	a logical. If TRUE, additional diagnostics are printed.

Details

The purpose of codoc is to check whether the documented usage of function objects agrees with their formal arguments as defined in the R code. This is not always straightforward, in particular as the usage information for methods to generic functions often employs the name of the generic rather than the method.

The following algorithm is used. If an installed package is used, it is loaded (unless it is the base package), after possibly detaching an already loaded version of the package. Otherwise, if the sources are used, the R code files of the package are collected and sourced in a new environment. Then, the usage sections of the Rd files are extracted and parsed ‘as much as possible’ to give the formals documented. For interpreted functions in the code environment, the formals are compared between code and documentation according to the values of the argument use.values. Synopsis sections are used if present; their occurrence is reported if verbose is true.

If a package has a namespace both exported and unexported objects are checked, as well as registered S3 methods. (In the unlikely event of differences the order is exported objects in the package, registered S3 methods and finally objects in the namespace and only the first found is checked.)

Currently, the R documentation format has no high-level markup for the basic ‘structure’ of classes and data sets (similar to the usage sections for function synopses). Variable names for data frames in documentation objects obtained by suitably editing ‘templates’ created by prompt are recognized by codocData and used provided that the documentation object is for a single data frame (i.e., only has one alias). codocClasses analogously handles slot names for classes in documentation objects obtained by editing shells created by promptClass.

Help files named ‘pkgname-defunct.Rd’ for the appropriate pkgname are checked more loosely, as they may have undocumented arguments.

Value

codoc returns an object of class "codoc". Currently, this is a list which, for each Rd object in the package where an inconsistency was found, contains an element with a list of the mismatches (which in turn are lists with elements code and docs, giving the corresponding arguments obtained from the function's code and documented usage).

codocClasses and codocData return objects of class "codocClasses" and "codocData", respectively, with a structure similar to class "codoc".

There are print methods for nicely displaying the information contained in such objects.

Note

The default for use.values has been changed from FALSE to NULL, for R versions 1.9.0 and later.

See Also

undoc, QC