fileutil - Procedures implementing some file utilities
This package provides implementations of standard unix utilities.
This command performs purely lexical normalization on the path and returns the changed path as its result. Symbolic links in the path are not resolved.
Examples:
fileutil::lexnormalize /foo/./bar
=> /foo/bar
fileutil::lexnormalize /foo/../bar
=> /bar
This command resolves all symbolic links in the path and returns the changed path as its result. In contrast to the builtin file normalize this command resolves a symbolic link in the last element of the path as well.
A command for the testing of several properties of a path. The properties to test for are specified in codes, either as a list of keywords describing the properties, or as a string where each letter is a shorthand for a property to test. The recognized keywords, shorthands, and associated properties are shown in the list below. The tests are executed in the order given to the command.
The result of the command is a boolean value. It will be true if and only if the path passes all the specified tests. In the case of the path not passing one or more test the first failing test will leave a message in the variable referenced by msgvar, if such is specified. The message will be prefixed with label, if it is specified. Note that the variabled referenced by msgvar is not touched at all if all the tests pass.
file readable
file writable
file exists
file executable
file isfile
file isdirectory
A tcl implementation of the UNIX cat command. Returns the contents of the specified file(s). The arguments are files to read, with interspersed options configuring the process. If there are problems reading any of the files, an error will occur, and no data will be returned.
The options accepted are -encoding, -translation, -eofchar, and --. With the exception of the last all options take a single value as argument, as specified by the tcl builtin command fconfigure. The -- has to be used to terminate option processing before a file if that file's name begins with a dash.
Each file can have its own set of options coming before it, and for anything not specified directly the defaults are inherited from the options of the previous file. The first file inherits the system default for unspecified options.
The command replaces the current contents of the specified file with data, with the process configured by the options. The command accepts the same options as ::fileutil::cat. The specification of a non-existent file is legal and causes the command to create the file (and all required but missing directories).
This command is like ::fileutil::writeFile, except that the previous contents of file are not replaced, but appended to. The command accepts the same options as ::fileutil::cat
This comment is similar to ::fileutil::appendToFile, except that the new data is not appended at the end, but inserted at a specified location within the file. In further contrast this command has to be given the path to an existing file. It will not create a missing file, but throw an error instead.
The specified location at has to be an integer number in the range 0 ... [file size file]. 0 will cause insertion of the new data before the first character of the existing content, whereas [file size file] causes insertion after the last character of the existing content, i.e. appending.
The command accepts the same options as ::fileutil::cat.
This command is the complement to ::fileutil::insertIntoFile, removing n characters from the file, starting at location at. The specified location at has to be an integer number in the range 0 ... [file size file] - n. 0 will cause the removal of the new data to start with the first character of the existing content, whereas [file size file] - n causes the removal of the tail of the existing content, i.e. the truncation of the file.
The command accepts the same options as ::fileutil::cat.
This command is a combination of ::fileutil::removeFromFile and ::fileutil::insertIntoFile. It first removes the part of the contents specified by the arguments at and n, and then inserts data at the given location, effectively replacing the removed by content with data. All constraints imposed on at and n by ::fileutil::removeFromFile and ::fileutil::insertIntoFile are obeyed.
The command accepts the same options as ::fileutil::cat.
This command can be seen as the generic core functionality of ::fileutil::replaceInFile. It first reads the contents of the specified file, then runs the command prefix cmd with that data appended to it, and at last writes the result of that invokation back as the new contents of the file.
If the executed command throws an error the file is not changed.
The command accepts the same options as ::fileutil::cat.
An implementation of the UNIX file command, which uses various heuristics to guess the type of a file. Returns a list specifying as much type information as can be determined about the file, from most general (eg, "binary" or "text") to most specific (eg, "gif"). For example, the return value for a GIF file would be "binary graphic gif". The command will detect the following types of files: directory, empty, binary, text, script (with interpreter), executable elf, executable dos, executable ne, executable pe, graphic gif, graphic jpeg, graphic png, graphic tiff, graphic bitmap, html, xml (with doctype if available), message pgp, binary pdf, text ps, text eps, binary gravity_wave_data_frame, compressed bzip, compressed gzip, compressed zip, compressed tar, audio wave, audio mpeg, and link. It further detects doctools, doctoc, and docidx documentation files, and tklib diagrams.
An implementation of the unix command find. Adapted from the Tcler's Wiki. Takes at most two arguments, the path to the directory to start searching from and a command to use to evaluate interest in each file. The path defaults to ".", i.e. the current directory. The command defaults to the empty string, which means that all files are of interest. The command takes care not to lose itself in infinite loops upon encountering circular link structures. The result of the command is a list containing the paths to the interesting files.
The filtercmd, if specified, is interpreted as a command prefix and one argument is added to it, the name of the file or directory find is currently looking at. Note that this name is not fully qualified. It has to be joined it with the result of pwd to get an absolute filename.
The result of filtercmd is a boolean value that indicates if the current file should be included in the list of interesting files.
Example:
# find .tcl files
package require fileutil
proc is_tcl {name} {return [string match *.tcl $name]}
set tcl_files [fileutil::find . is_tcl]
This command is based upon the TclX command recursive_glob, except that it doesn't allow recursion over more than one directory at a time. It uses ::fileutil::find internally and is thus able to and does follow symbolic links, something the TclX command does not do. First argument is the directory to start the search in, second argument is a list of patterns. The command returns a list of all files reachable through basedir whose names match at least one of the patterns. The options before the pattern-list determine the style of matching, either regexp or glob. glob-style matching is the default if no options are given. Usage of the option -- stops option processing. This allows the use of a leading '-' in the patterns.
The command reads the file filename and executes the script cmd for every line in the file. During the execution of the script the variable var is set to the contents of the current line. The return value of this command is the result of the last invocation of the script cmd or the empty string if the file was empty.
Implementation of grep. Adapted from the Tcler's Wiki. The first argument defines the pattern to search for. This is followed by a list of files to search through. The list is optional and stdin will be used if it is missing. The result of the procedures is a list containing the matches. Each match is a single element of the list and contains filename, number and contents of the matching line, separated by a colons.
The install command is similar in functionality to the install command found on many unix systems, or the shell script distributed with many source distributions (unix/install-sh in the Tcl sources, for example). It copies source, which can be either a file or directory to destination, which should be a directory, unless source is also a single file. The ?-m? option lets the user specify a unix-style mode (either octal or symbolic - see file attributes.
Removes the first n elements from the specified path and returns the modified path. If n is greater than the number of components in path an empty string is returned. The number of components in a given path may be determined by performing llength on the list returned by file split.
If, and only if the path is inside of the directory returned by [pwd] (or the current working directory itself) it is made relative to that directory. In other words, the current working directory is stripped from the path. The possibly modified path is returned as the result of the command. If the current working directory itself was specified for path the result is the string ".".
If, and only of the path is inside of the directory "prefix" (or the prefix directory itself) it is made relative to that directory. In other words, the prefix directory is stripped from the path. The possibly modified path is returned as the result of the command. If the prefix directory itself was specified for path the result is the string ".".
This command ensures that the path is not escaping the directory jail. It always returns an absolute path derived from path which is within jail.
If path is an absolute path and already within jail it is returned unmodified.
An absolute path outside of jail is stripped of its root element and then put into the jail by prefixing it with it. The same happens if path is relative, except that nothing is stripped of it. Before adding the jail prefix the path is lexically normalized to prevent the caller from using .. segments in path to escape the jail.
Implementation of touch. Alter the atime and mtime of the specified files. If -c, do not create files if they do not already exist. If -r, use the atime and mtime from ref_file. If -t, use the integer clock value time. It is illegal to specify both -r and -t. If -a, only change the atime. If -m, only change the mtime.
This command is not available for Tcl versions less than 8.3.
The command returns the path of a directory where the caller can place temporary files, such as "/tmp" on Unix systems. The algorithm we use to find the correct directory is as follows:
The directory set by an invokation of ::fileutil::tempdir with an argument. If this is present it is tried exclusively and none of the following item are tried.
The directory named in the TMPDIR environment variable.
The directory named in the TEMP environment variable.
The directory named in the TMP environment variable.
A platform specific location:
"C:\TEMP", "C:\TMP", "\TEMP", and "\TMP" are tried in that order.
The TRASH_FOLDER environment variable is used. This is most likely not correct.
The directories "/tmp", "/var/tmp", and "/usr/tmp" are tried in that order.
The algorithm utilized is mainly that used in the Python standard library. The exception is the first item, the ability to have the search overridden by a user-specified directory.
In this mode the command sets the path as the first and only directory to try as a temp. directory. See the previous item for the use of the set directory. The command returns the empty string.
Invoking this command clears the information set by the last call of [::fileutil::tempdir path]. See the last item too.
The command generates a temporary file name suitable for writing to, and the associated file. The file name will be unique, and the file will be writable and contained in the appropriate system specific temp directory. The name of the file will be returned as the result of the command.
The code was taken from http://wiki.tcl.tk/772, attributed to Igor Volobouev and anon.
The command generates a temporary directory suitable for writing to. The directory name will be unique, and the directory will be writable and contained in the appropriate system specific temp directory. The name of the directory will be returned as the result of the command.
The three options can used to tweak the behaviour of the command:
The initial code for this was supplied by Miguel Martinez Lopez.
This command takes two directory paths, both either absolute or relative and computes the path of dst relative to base. This relative path is returned as the result of the command. As implied in the previous sentence, the command is not able to compute this relationship between the arguments if one of the paths is absolute and the other relative.
Note: The processing done by this command is purely lexical. Symbolic links are not taken into account.
This command takes two file paths, both either absolute or relative and computes the path of dst relative to base, as seen from inside of the base. This is the algorithm how a browser resolves a relative link found in the currently shown file.
The computed relative path is returned as the result of the command. As implied in the previous sentence, the command is not able to compute this relationship between the arguments if one of the paths is absolute and the other relative.
Note: The processing done by this command is purely lexical. Symbolic links are not taken into account.
In this version fileutil::find's broken system for handling symlinks was replaced with one working correctly and properly enumerating all the legal non-cyclic paths under a base directory.
While correct this means that certain pathological directory hierarchies with cross-linked sym-links will now take about O(n**2) time to enumerate whereas the original broken code managed O(n) due to its brokenness.
A concrete example and extreme case is the "/sys" hierarchy under Linux where some hundred devices exist under both "/sys/devices" and "/sys/class" with the two sub-hierarchies linking to the other, generating millions of legal paths to enumerate. The structure, reduced to three devices, roughly looks like
/sys/class/tty/tty0 --> ../../dev/tty0 /sys/class/tty/tty1 --> ../../dev/tty1 /sys/class/tty/tty2 --> ../../dev/tty1 /sys/dev/tty0/bus /sys/dev/tty0/subsystem --> ../../class/tty /sys/dev/tty1/bus /sys/dev/tty1/subsystem --> ../../class/tty /sys/dev/tty2/bus /sys/dev/tty2/subsystem --> ../../class/tty
The command fileutil::find currently has no way to escape this. When having to handle such a pathological hierarchy It is recommended to switch to package fileutil::traverse and the same-named command it provides, and then use the -prefilter option to prevent the traverser from following symbolic links, like so:
package require fileutil::traverse
proc NoLinks {fileName} {
if {[string equal [file type $fileName] link]} {
return 0
}
return 1
}
fileutil::traverse T /sys/devices -prefilter NoLinks
T foreach p {
puts $p
}
T destroy
This document, and the package it describes, will undoubtedly contain bugs and other problems. Please report such in the category fileutil of the Tcllib Trackers. Please also report any ideas for enhancements you may have for either package and/or documentation.
When proposing code changes, please provide unified diffs, i.e the output of diff -u.
Note further that attachments are strongly preferred over inlined patches. Attachments can be made by going to the Edit form of the ticket immediately after its creation, and then using the left-most button in the secondary navigation bar.
Programming tools