From f5e8bfb4b1cb7f8e84dbe6cd5967df0e8ef4de7b Mon Sep 17 00:00:00 2001
From: Teodor Sigaev <teodor@sigaev.ru>
Date: Wed, 26 Nov 2003 14:06:16 +0000
Subject: [PATCH] Utility for convert myspell dictionaries to ispell, full
README will be later
---
contrib/tsearch2/my2ispell/Makefile | 47 +++++++++++++++++++++++++++++
contrib/tsearch2/my2ispell/README | 12 ++++++++
2 files changed, 59 insertions(+)
create mode 100644 contrib/tsearch2/my2ispell/Makefile
create mode 100644 contrib/tsearch2/my2ispell/README
diff --git a/contrib/tsearch2/my2ispell/Makefile b/contrib/tsearch2/my2ispell/Makefile
new file mode 100644
index 00000000000..35473cfe812
--- /dev/null
+++ b/contrib/tsearch2/my2ispell/Makefile
@@ -0,0 +1,47 @@
+ZIPFILE=nb_NO
+LANGUAGE=norsk
+
+
+UNZIP=unzip -o
+
+
+all: $(LANGUAGE).dict $(LANGUAGE).aff
+
+$(ZIPFILE).aff: $(ZIPFILE).zip
+ $(UNZIP) $? $@
+ touch $@
+
+
+# 1 Cleanup dictionary
+# 2 remove " symbol
+# 3 add compoundwords controlled flag to word which hasn't it, but
+# has compound only suffixes
+
+$(LANGUAGE).dict: $(ZIPFILE).zip
+ $(UNZIP) $? $(ZIPFILE).dic
+ grep -v -E '^[[:digit:]]+$$' < $(ZIPFILE).dic \
+ | grep -v '\.' \
+ | sed -e 's/"//g' \
+ | perl -pi -e 's|/(\S+)| $$q=$$1; ( $$q=~/[\\_`]/ && $$q!~/z/ ) ? "/$${q}z" : "/$${q}"|e' \
+ | sort \
+ > $@
+
+#just convert affix file
+
+$(LANGUAGE).aff: $(ZIPFILE).aff
+ grep -v -i zyzyzy $(ZIPFILE).aff \
+ | grep -v -i zyzyzy \
+ | perl -pi \
+ -e 's/^COMPOUNDFLAG\s+(\S+)/compoundwords controlled $$1/;' \
+ -e 's/^COMPOUNDMIN\s+(\d+)/compoundmin $$1/;' \
+ -e 's/^PFX\s+(\S+)\s+Y\s+\d+.*$$/ if ( !$$wasprf ) { $$wasprf=1; "prefixes\n\nflag $$1:" } else { "flag $$1:" } /e;' \
+ -e 's/^PFX\s+\S+\s+(\S+)\s+(\S+)\s+(\S+)/ uc(" $$3 > $$2")/e;' \
+ -e 's/^(.*)SFX\s+(\S+)\s+([YN])\s+\d+.*$$/ $$flg=($$3 eq "Y") ? "*" : ""; $$flg="~$$flg" if length $$1; $$q=$$2; $$q="\\$$q" if $$q!~m#[a-zA-Z]#; if ( !$$wassfx ) { $$wassfx=1; "suffixes\n\nflag $$flg$$q:" } else { "flag $$flg$$q:" } /e;' \
+ -e 's/^.*SFX\s+\S+\s+(\S+)\s+(\S+)\s+(\S+)/ uc(" $$3 > ".( ($$1 eq "0") ? "" : "-$$1,").( ($$2 eq "0") ? "" : "$$2") )/e;' \
+ -e 's/^(SET|TRY)/#$$1/' \
+ > $@
+
+clean:
+ rm -rf $(ZIPFILE).aff $(ZIPFILE).dic $(LANGUAGE).dict $(LANGUAGE).aff
+
+
diff --git a/contrib/tsearch2/my2ispell/README b/contrib/tsearch2/my2ispell/README
new file mode 100644
index 00000000000..583936ab42e
--- /dev/null
+++ b/contrib/tsearch2/my2ispell/README
@@ -0,0 +1,12 @@
+Utility for convert MySpell dictionary and affix from
+myspell to ispell format.
+Utility tested on nb_NO.zip and nn_NO.zip from
+OpenOffice (http://lingucomponent.openoffice.org/download_dictionary.html)
+
+usage:
+For example, make norwegian dictionary and affix:
+% cp nb_NO.zip my2ispell
+% cd my2ispell
+% gmake ZIPFILE=nb_NO LANGUAGE=norsk
+
+Author: Teodor Sigaev <teodor@sigaev.ru>
--
GitLab