In the course of assembling DOSSIER (www.ptf.com/dossier) volumes, I
find myself hand-editing an "executable table of contents" (i.e., a
shell script) to drive the production for each volume. Frequently,
the entries are Unix "man" pages (e.g., "intro", a.out, renice). I
try to keep these in order, but all too often, make mistakes. So, I
wrote some code to detect sequence errors. Here are the rules:
Within each section of the man pages, the entries are listed in
(approximately) ASCII sort order. The exceptions are as follows:
* Each section begins with the "intro" man page.
* Sorting of letters is case-insensitive unless a cross-case match
occurs; in that case, the matching items are sorted in case-
sensitive order.
* Punctuation marks come first, followed by digits and letters.
If two items differ only by a pair of punctuation marks, ASCII
sort order is used to resolve the ambiguity.
The checking is done by some Perl code, within a "shell function".
The shell variables "S" and "T" are, respectively, the current man
page section (e.g., 1) and title (e.g., "intro"). Handling the
"intro" special case is pretty simple, if a bit wonky.
I had to think a bit, however, about how to handle the rest of the
sorting. My first thought was to split both $T and $OldT into lists
of characters, then walk the lists in parallel, checking the rules.
This would certainly have worked, but it didn't seem, erm, Perlish.
So, I came up with the following code, which tickled my funnybone:
tchk () {
#
# tchk - Check $T against $OldT for correct sort order.
#
# Note: Neither $T nor $OldT may contain spaces!
#
# Usage: tchk
if [ $# -ne 0 ]; then
echo "!!! tchk: usage error"
fi
perl -e '
{
$OldS="'"$OldS"'"; $S="'"$S"'";
$OldT="'"$OldT"'"; $T="'"$T"'";
# Handle "intro" weirdness.
if ($S ne $OldS) { # new section
if ($T ne "intro") {
print "!!! missing intro: S=$S, T=$T\n";
}
exit;
}
if ($T eq "intro") {
print "!!! misplaced intro: S=$S, T=$T\n";
exit;
}
$OldT = "" if ($OldT eq "intro");
# Now do simple (:-) matching.
if (lc($T) eq lc($OldT)) { # cross-case match
if ($T lt $OldT) {
print "!!! broken cross-case match: S=$S, T=$T, OldT=$OldT\n";
exit;
}
} else {
($Tst = $T) =~ s|(.)| $1|g;
($OldTst = $OldT) =~ s|(.)| $1|g;
$Tst =~ s| (\d)|1$1|g;
$OldTst =~ s| (\d)|1$1|g;
$Tst =~ s| ([A-Za-z])|2$1|g;
$OldTst =~ s| ([A-Za-z])|2$1|g;
$Tst =~ s| |0|g;
$OldTst =~ s| |0|g;
if ($Tst lt $OldTst) {
print "!!! sequence error: S=$S, T=$T, OldT=$OldT\n";
exit;
}
}
}
'
OldS=$S
OldT=$T
}
--
email: [EMAIL PROTECTED]; phone: +1 650-873-7841
http://www.cfcl.com/rdm - my home page, resume, etc.
http://www.cfcl.com/Meta - The FreeBSD Browser, Meta Project, etc.
http://www.ptf.com/dossier - Prime Time Freeware's DOSSIER series
http://www.ptf.com/tdc - Prime Time Freeware's Darwin Collection
