Contributed by Rahul Dhesi
While I was moderating comp.binaries.ibm.pc, I wanted to provide users a simple way of extracting multipart uuencoded binaries. I added BEGIN and END headers to each part to enclose the uuencoded part and provided users with the following script:
cat $* | sed '/^END/,/^BEGIN/d' | uudecode
This script will accept a list of filenames (in order) provided as command-line arguments. It will also accept concatenated articles as standard input.
This script invokes cat in a very useful way that is well known to expert shell script users but not enough used by most others. This allows the user the choice of either providing command-line arguments or standard input.
The script invokes sed to strip out superfluous headers and trailers, except for headers in the first input file and trailers in the last input file. The final result is that the uuencoded part of the multiple input files is extracted and uudecoded. Each input file (see postings in comp.binaries.ibm.pc) has the following form:
I have lots of other shell stuff, but the above is simplest and has proved useful to several thousand comp.binaries.ibm.pc readers.
This one is pretty obvious but accomplishes a lot. For those who might not understand the use of this command, here is the explanation. A Usenet newsgroup such as comp.binaries.ibm.pc distributes public-domain programs and such. Binaries, the object code created by the compiler, cannot be distributed as news articles unless they are "encoded." A program named uuencode converts the binary to an ASCII representation that can be easily distributed. Furthermore, there are limits on the size of news articles and large binaries are broken up into a series of articles (1 of 3, 2 of 3, 3 of 3, for example). Dhesi would break up the encoded binary into manageable chunks, and then add the BEGIN and END lines to delimit the text that contained encoded binary.
A reader of these articles might save each article in a file. Dhesi's script automates the process of combining these articles and removing extraneous information such as the article header as well as the extra BEGIN and END headers. His script removes lines from the first END up to and including the next BEGIN pattern. It combines all the separate encoded parcels and directs them to uudecode, which converts the ASCII representation to binary.
One has to appreciate the amount of manual editing work avoided by a simple one-line script.