Last change
on this file since 3b02b04 was
555c7f8,
checked in by Tomasz Obrebski <to@…>, 12 years ago
|
dodanie mo�liwo�ci szybkiego uaktualniania s�ownika dla lema
|
-
Property mode set to
100755
|
File size:
1.0 KB
|
Line | |
---|
1 | #! /bin/bash |
---|
2 | |
---|
3 | if [ $# -lt 2 ] |
---|
4 | then |
---|
5 | echo "Usage:" |
---|
6 | echo " compdic-update-cats <catfile> <difference> <difference> ..." |
---|
7 | echo "where" |
---|
8 | echo " <catfile> - file containing the list morphosyntactic categories" |
---|
9 | echo " <difference> - diff-format file containing lines to be added to/removed from dictionary" |
---|
10 | exit 0 |
---|
11 | fi |
---|
12 | |
---|
13 | tempdir=`mktemp -d /tmp/compdic.XXXXXX` |
---|
14 | |
---|
15 | cats=$1 |
---|
16 | shift |
---|
17 | |
---|
18 | catplus=$tempdir/plus.cat |
---|
19 | catminus=$tempdir/minus.cat |
---|
20 | cattmp=$tempdir/tmp.cat |
---|
21 | |
---|
22 | touch $catplus |
---|
23 | touch $catminus |
---|
24 | |
---|
25 | while (($#)) |
---|
26 | do |
---|
27 | echo processing $1 ... |
---|
28 | cat $1 | egrep '^>' | sed -r 's/^> *//;s/[[:space:]].*$//' | cut -d ',' -f 2 | sort -u | canonize >> $catplus |
---|
29 | cat $1 | egrep '^<' | sed -r 's/^< *//;s/[[:space:]].*$//' | cut -d ',' -f 2 | sort -u | canonize >> $catminus |
---|
30 | shift |
---|
31 | done |
---|
32 | |
---|
33 | echo updating $cats ... |
---|
34 | # comm -23 $cats $catminus > $cattmp |
---|
35 | # mv ${cats} ${cats}~ |
---|
36 | # cat $cattmp $catplus | sort -u > $cats |
---|
37 | |
---|
38 | cat $cats $catplus | sort -u > $cattmp |
---|
39 | |
---|
40 | mv ${cats} ${cats}~ |
---|
41 | mv ${cattmp} ${cats} |
---|
42 | chmod a+r ${cats} |
---|
43 | |
---|
44 | rm -r $tempdir |
---|
Note: See
TracBrowser
for help on using the repository browser.