| 1 |
7431
|
aaronmk
|
#!/bin/sh
|
| 2 |
|
|
# Translates a Redmine HTML page to a thesaurus
|
| 3 |
|
|
# Usage: self <page
|
| 4 |
|
|
|
| 5 |
|
|
sedEreFlag="$(test "$(uname)" = Darwin && echo E || echo r)"
|
| 6 |
|
|
|
| 7 |
|
|
sed () { "$(which sed)" -"$sedEreFlag" "$@";}
|
| 8 |
|
|
|
| 9 |
7442
|
aaronmk
|
ambigTerm=
|
| 10 |
|
|
ambigPrefix=
|
| 11 |
7431
|
aaronmk
|
term=
|
| 12 |
7442
|
aaronmk
|
sed -n 's/^.*<h([1-4])[^>]*><a href="#[^>]+>([^<]+).*$/\1 \2/p'\
|
| 13 |
7431
|
aaronmk
|
|while read -r line; do
|
| 14 |
|
|
set -- $line # split using IFS
|
| 15 |
|
|
level="$1" name="$2"
|
| 16 |
7442
|
aaronmk
|
|
| 17 |
|
|
# Handle synonyms
|
| 18 |
7445
|
aaronmk
|
if test "$level" = 3; then echo "$name,$term"
|
| 19 |
7442
|
aaronmk
|
else
|
| 20 |
|
|
term="$name"
|
| 21 |
|
|
|
| 22 |
|
|
# Handle ambiguous terms
|
| 23 |
|
|
termName="${term#\?}"
|
| 24 |
|
|
if test "$termName" != "$term"; then # ambiguous term (starts with ?)
|
| 25 |
|
|
ambigTerm="$term"
|
| 26 |
|
|
ambigPrefix="${termName}_"
|
| 27 |
|
|
elif test -n "$ambigTerm" -a "${term#$ambigPrefix}" != "$term"; then
|
| 28 |
|
|
# alternative of ambiguous term (starts with $ambigTerm)
|
| 29 |
7445
|
aaronmk
|
echo "$ambigTerm,$term"
|
| 30 |
7442
|
aaronmk
|
else # term not related to ambiguous terms
|
| 31 |
|
|
ambigTerm= ambigPrefix= # clear any ambiguous term in effect
|
| 32 |
|
|
fi
|
| 33 |
7431
|
aaronmk
|
fi
|
| 34 |
|
|
done
|
