diff options
author | Karl Berry <karl@freefriends.org> | 2006-12-08 00:25:19 +0000 |
---|---|---|
committer | Karl Berry <karl@freefriends.org> | 2006-12-08 00:25:19 +0000 |
commit | 7ff51640e2fd3e968488d312879dee70458f2926 (patch) | |
tree | 6b9ede61598bd09a713892640d2d41da6913e509 /Master/texmf-dist/doc/eplain/util | |
parent | 8291b74ec6f858e3536a5919b74386a7df1dcc24 (diff) |
eplain 3.1 (5dec06)
git-svn-id: svn://tug.org/texlive/trunk@2630 c570f23f-e606-0410-a88d-b1316a301751
Diffstat (limited to 'Master/texmf-dist/doc/eplain/util')
-rw-r--r-- | Master/texmf-dist/doc/eplain/util/idxuniq | 37 | ||||
-rw-r--r-- | Master/texmf-dist/doc/eplain/util/trimsee | 106 |
2 files changed, 143 insertions, 0 deletions
diff --git a/Master/texmf-dist/doc/eplain/util/idxuniq b/Master/texmf-dist/doc/eplain/util/idxuniq new file mode 100644 index 00000000000..6ee7d225f5c --- /dev/null +++ b/Master/texmf-dist/doc/eplain/util/idxuniq @@ -0,0 +1,37 @@ +#!/usr/bin/awk -f + +# (This file is public domain.) + +# This script filters input lines (which are expected to be a list of +# `\indexentry' lines for MakeIndex) eliminating matching lines. The comparison +# is done without regard to hyperlink label names (in the form `IDX*', where `*' +# stands for a non-negative integer). + +# This has to be done to avoid terms which differ only in hyperlink label names +# embedded into them, because MakeIndex will treat these otherwise identical +# terms as distinct, listing more than once a page number for equivalent terms +# on the same page. + +# `stripped' array is indexed by `\indexentry' lines stripped off the hyperlink +# label name. For each term, we use its stripped version as a key into the +# `stripped' array to increment its element. This records the fact that we have +# seen the term with such key. Also, we add the (full) `\indexentry' line to +# the `terms' array, but only if we have not yet seen a term with such key. + +# The idea was borrowed from Edition 3 of `GAWK: Effective AWK Programming: A +# User's Guide for GNU Awk', which contains the following credit: + +# histsort.awk --- compact a shell history file +# Thanks to Byron Rakitzis for the general idea + +{ + temp = $0 + sub (/{IDX[[:digit:]]+}/, "", temp) + if (stripped[temp]++ == 0) + terms[++count] = $0 +} + +END { + for (i = 1; i <= count; i++) + print terms[i] +} diff --git a/Master/texmf-dist/doc/eplain/util/trimsee b/Master/texmf-dist/doc/eplain/util/trimsee new file mode 100644 index 00000000000..8af5ebe6434 --- /dev/null +++ b/Master/texmf-dist/doc/eplain/util/trimsee @@ -0,0 +1,106 @@ +#!/bin/sh + +# (This file is public domain.) + +help () +{ + cat <<EOF +Usage: trimsee [-i IS] [-o OS] [-s SEE] + trimsee {-h|--help|-v|--version} +Remove commas (or other page list separators) in front +of see / see also commands in the output of MakeIndex. +Input is read from stdin, output is directed to stdout. + +-i IS use IS as a regular expression matching separator + in front of see / see also commands in the input + (default: '$DEFAULT_IS') +-o OS use OS as a separator to replace IS in front of + see / see also commands (default: '$DEFAULT_OS') +-s SEE use SEE as a regular expression matching see / + see also commands (default: '$DEFAULT_SEE') +-h, --help show this help message +-v, --version show version +EOF +} + +check_missing_arg () +{ + if (( $1 < 2 )); then + echo "Missing argument to option '$2'" 1>&2 + exit 1 + fi +} + +VERSION='0.1' + +DEFAULT_SEE='\\indexsee' +DEFAULT_IS=', \+' +DEFAULT_OS=' ' + +SEE=$DEFAULT_SEE +IS=$DEFAULT_IS +OS=$DEFAULT_OS + +while [ $# != 0 ] +do + case "$1" in + -i) + check_missing_arg $# "$1" + IS=$2 + shift + ;; + -o) + check_missing_arg $# "$1" + OS=$2 + shift + ;; + -s) + check_missing_arg $# "$1" + SEE=$2 + shift + ;; + -h | --help) + help + exit 0 + ;; + -v | --version) + echo "trimsee version $VERSION" + exit 0 + ;; + *) + echo "Unknown option '$1'" 1>&2 + exit 1 + ;; + esac + shift +done + + +# The idea is to implement line output delayed by one line. When we +# read next line, we check if it starts with a see/see also entry, and +# if it does, we remove the comma at the end of the previous line +# before printing it. We keep previous lines in the hold buffer. + +sed -n ' + +# Remove all commas in front of see/see also inside each line. +s/'"$IS$SEE/$OS$SEE"'/g + +# If this line does not start with see/see also, we will print the +# previous line intact. NOTE: There are two characters inside the +# brackets: a space and a tab. +/^[ ]*'"$SEE"'/! { x; b PRINT; } + +# This line starts with see/see also, so remove comma at the end of +# the previous line before we print it. +x +s/'"$IS"'$/'"$OS"'/ + +# Print the previous line. +:PRINT +1! p + +# At the end of the input, print the last line. +$ { x; p; } + +' |