mirror of
https://github.com/MidnightCommander/mc
synced 2024-12-23 04:46:55 +03:00
.. | ||
a | ||
arfs | ||
cpio.in | ||
deb.in | ||
extfs.ini | ||
ftplist.in | ||
lha.in | ||
lslR.in | ||
mailfs | ||
patchfs | ||
rar.in | ||
README | ||
rpm | ||
zip.in | ||
zoo.in |
Writing scripts for Midnight Commander's external vfs IMPORTANT NOTE: extfs is not officialy released and fully bug free in 3.0! You have been warned. If you would really like to try it, you can (by typing make install.extfs in the vfs directory). Starting with version 3.1, the Midnight Commander comes with so called extfs, which is one of the virtual filesystems. This system makes it possible to create new virtual filesystems for the GNU MC very easily. Such work has two basic steps: Editing $(libdir)/extfs/extfs.ini. Creating a shell script/program to handle requests. (Note: $(libdir) should be substituted for actual libdir path stored when configured or compiled, like /usr/local/lib/mc or /usr/lib/mc). The first one is very easy: You assign a vfs prefix and vfs extensions to your vfs. Both will be used in vfs pseudoURL names, like if you assign prefix zip and extensions .zip, .ZIP, then URLs will look like zip:anypath/my.zip/some_path/in_the/archive Then you add a line to the end of the [extfs] section: prefix=space_separated_extensions e.g. zip=.zip .ZIP The second one my require some your knowledges of shell/c programming: You have to create a program (with executable permissions) prefix in $(libdir)/extfs (in our example $(libdir)/extfs/zip). * Commands that should be implemented by your shell script ---------------------------------------------------------- $libdir/extfs/prefix command [arguments] * Command: list archivename This command should list the complete archive content in the following format (a little modified ls -l listing): AAAAAAA NNN OOOOOOOO GGGGGGGG SSSSSSSS DATETIME [PATH/]FILENAME [-> [PATH/]FILENAME[/]]] where (things in [] are optional): AAAAAAA is the permission string like in ls -l NNN is the number of links OOOOOOOO is the owner (either UID or name) GGGGGGGG is the group (either GID or name) SSSSSSSS is the file size FILENAME is the filename PATH is the path from the archive's root without the leading slash (/) DATETIME has one of the following formats: Mon DD hh:mm Mon DD YYYY Mon DD YYYY hh:mm MM-DD-YY hh:mm where Mon is a three digit english month name, DD day 1-31, MM month 01-12, YY two digit year, YYYY four digit year, hh hour and mm minute. If the -> [PATH/]FILENAME part is present, it means: If permissions start with an l (ell), then it is the name that symlink points to. (If this PATH starts with a MC vfs prefix, then it is a symlink somewhere to the other virtual filesystem (if you want to specify path from the local root, use local:/path_name instead of /path_name, since /path_name means from root of the archive listed). If permissions do not start with l, but number of links is greater than one, then it says that this file should be a hardlinked with the other file. * Command: copyout archivename storedfilename extractto This should extract from archive archivename the file called storedfilename (possibly with path if not located in archive's root) to file extractto. * Command: copyin archivename storedfilename sourcefile This should add to the archivename the sourcefile with the name storedfilename inside the archive. Important note: archivename in the above examples may not have the extension you are expecting to have, like it may happen that archivename will be something like /tmp/f43513254 or just anything. Some archivers do not like it, so you'll have to find some workaround. --------------------------------------------------------- Don't forget to mark this file executable (chmod 755 ThisFile, for example) This is a skeleton structure of the executable: #!/bin/sh # Command functions mcvfs_list () # $1 is the archive name { # Apply a system command to obtain a list of filenames # For example 'zip -l $1' # Scan each line of the 'list' output, discarding unused information, and # constructing a printable line in a form, described above, that mc can use. # Exit } mcvfs_copyout () # $1 is the archive name # $2 is a name of a file within the archive # $3 is a name of a file within the system (to add from or extract to) { # Apply the system command used to extract one file from the archive # Exit } mcvfs_copyin () # $1 is the archive name # $2 is a name of a file within the archive # $3 is a name of a file within the system (to add from or extract to) { # Apply the system command used to add one file to the archive # Exit } # Command line parser # $1 is the command # $2 is the archive name # $3 is a name of a file within the archive # $4 is a name of a file within the system (to add from or extract to) case "$1" in list) mcvfs_list $2; exit $?;; copyout) mcvfs_copyout $2 $3 $4; exit $?;; copyin) mcvfs_copyin $2 $3 $4; exit $?;; esac # Show an error if this was called with some other command exit 1 --------------------------------------------------------- In constructing these routines, errors will be made, and mc will not display a malformed printing line. That can lead the programmer down many false trails in search of the bug. Since this routine is an executable shell script it can be run from the command line independently of mc, and its output will show on the console or can be redirected to a file. * Putting it to use ---------------------------------------------------------- The file .mc.ext in a home directory, and in mc's user directory (commonly /usr/local/lib/mc), contains instructions for operations on files depending on filename extensions. It is well documented in other files in this distribution, so here are just a few notes specifically on use of the Virtual File System you just built. There are entries in .mc.ext defining a few operations that can be done on a file from an mc panel. Typically they are annotated with a hash mark and a file extension like this: # zip There must be a way to find the file by extension, so the next line does that. In essence it says "identify the string ".zip" or (|) ".ZIP" at the end ($) of a filename": regex/\.(zip|ZIP)$ The operations themselves follow that. They must be indented by at least a space, and a tab works as well. In particular, the Open operation will now use your new virtual file system by cd'ing to it like this: Open=%cd zip:%d/%p This is the line used when a file is highlighted in a panel and the user presses <Enter> or <Return>. The contents of the archive should show just as if they were in a real directory, and can be manipulated as such. The rest of the entry pertains to use of the F3 View key: View=%view{ascii} unzip -v %f And perhaps an optional icon for X: Icon=zip.xpm And perhaps an operation to extract the contents of the file, called from a menu selection: Extract=unzip %f '*' This is just an example. The current entry for .zip files has a menu selection of 'Unzip' which could be used in place of 'Extract'. What goes here depends on what items you have in, or add to, the menu system, and that's another subject. The sum of this is the .mc.ext entry: # zip regex/\.(zip|ZIP)$ Open=%cd zip:%d/%p View=%view{ascii} unzip -v %f Icon=zip.xpm Extract=unzip %f '*' Add an entry like this to the .mc.ext file in a user's home directory, If you want others to have it, add it to the mc.ext file in the mc system directory, often /usr/local/lib/mc/mc.ext. Notice this file is not prepended with a dot. Once all this is done, and things are in their proper places, exit mc if you were using it, and restart it so it picks up the new information. That's all there is to it. The hardest part is making a listing function that sorts the output of a system listing command and turns it into a form that mc can use. Currently awk (or gawk) is used because nearly all systems have it. If another scripting language is available, like perl, that could also be used.