diff options
author | Patrick Simianer <simianer@cl.uni-heidelberg.de> | 2013-03-03 12:06:43 +0100 |
---|---|---|
committer | Patrick Simianer <simianer@cl.uni-heidelberg.de> | 2013-03-03 12:06:43 +0100 |
commit | 6b172ef1b1651e6ccf016b26c9ea94fad6df3152 (patch) | |
tree | a62dffda9143a619531be9f96912938659d3a890 /corpus/support/quote-norm.pl | |
parent | 349ee7d5599bb973506c8bbb56926cf9d366b564 (diff) | |
parent | 7deec52e8feb1c908a91224f308e8cbd9a170576 (diff) |
Merge branch 'master' of github.com:pks/cdec-dtrain
Diffstat (limited to 'corpus/support/quote-norm.pl')
-rwxr-xr-x | corpus/support/quote-norm.pl | 14 |
1 files changed, 14 insertions, 0 deletions
diff --git a/corpus/support/quote-norm.pl b/corpus/support/quote-norm.pl index 72b0064d..e4e5055e 100755 --- a/corpus/support/quote-norm.pl +++ b/corpus/support/quote-norm.pl @@ -27,6 +27,20 @@ while(<STDIN>) { s/¨/"/g; s/¡/ ¡ /g; s/¿/ ¿ /g; + + s/ п. п. / п.п. /g; + s/ ст. л. / ст.л. /g; + s/ т. е. / т.е. /g; + s/ т. к. / т.к. /g; + s/ т. ч. / т.ч. /g; + s/ т. д. / т.д. /g; + s/ т. п. / т.п. /g; + s/ и. о. / и.о. /g; + s/ с. г. / с.г. /g; + s/ г. р. / г.р. /g; + s/ т. н. / т.н. /g; + s/ т. ч. / т.ч. /g; + s/ н. э. / н.э. /g; # â<U+0080><U+0099> s/â(\x{80}\x{99}|\x{80}\x{98})/'/g; s/â(\x{80}\x{9c}|\x{80}\x{9d})/"/g; |