CC ?= gcc
SF_MIRROR = easynews
-SCIM_TABLES_VER = 0.5.7
+SCIM_TABLES_VER = 0.5.8
SCIM_PINYIN_VER = 0.5.91
LIBTABE_VER = 0.2.3
all: ZhConversion.php tradphrases.notsure simpphrases.notsure wordlist toHans.dict toHant.dict toCN.dict toTW.dict toHK.dict toSG.dict
-Unihan.txt:
+# Download Unihan database and Traditional Chinese / Simplified Chinese phrases files
+Unihan.zip:
wget -nc ftp://ftp.unicode.org/Public/UNIDATA/Unihan.zip
- unzip -q Unihan.zip
-EZ.txt.in:
+scim-tables-$(SCIM_TABLES_VER).tar.gz:
wget -nc http://$(SF_MIRROR).dl.sourceforge.net/sourceforge/scim/scim-tables-$(SCIM_TABLES_VER).tar.gz
- tar -xzf scim-tables-$(SCIM_TABLES_VER).tar.gz -O scim-tables-$(SCIM_TABLES_VER)/tables/zh/EZ-Big.txt.in > EZ.txt.in
-phrase_lib.txt:
+scim-pinyin-$(SCIM_PINYIN_VER).tar.gz:
wget -nc http://$(SF_MIRROR).dl.sourceforge.net/sourceforge/scim/scim-pinyin-$(SCIM_PINYIN_VER).tar.gz
+
+libtabe-$(LIBTABE_VER).tgz:
+ wget -nc http://$(SF_MIRROR).dl.sourceforge.net/sourceforge/libtabe/libtabe-$(LIBTABE_VER).tgz
+
+# Extract the file from a comressed files
+Unihan.txt: Unihan.zip
+ unzip -oq Unihan.zip
+
+EZ.txt.in: scim-tables-$(SCIM_TABLES_VER).tar.gz
+ tar -xzf scim-tables-$(SCIM_TABLES_VER).tar.gz -O scim-tables-$(SCIM_TABLES_VER)/tables/zh/EZ-Big.txt.in > EZ.txt.in
+
+Wubi.txt.in: scim-tables-$(SCIM_TABLES_VER).tar.gz
+ tar -xzf scim-tables-$(SCIM_TABLES_VER).tar.gz -O scim-tables-$(SCIM_TABLES_VER)/tables/zh/Wubi.txt.in > Wubi.txt.in
+
+Ziranma.txt.in: scim-tables-$(SCIM_TABLES_VER).tar.gz
+ tar -xzf scim-tables-$(SCIM_TABLES_VER).tar.gz -O scim-tables-$(SCIM_TABLES_VER)/tables/zh/Ziranma.txt.in > Ziranma.txt.in
+
+
+phrase_lib.txt: scim-pinyin-$(SCIM_PINYIN_VER).tar.gz
tar -xzf scim-pinyin-$(SCIM_PINYIN_VER).tar.gz -O scim-pinyin-$(SCIM_PINYIN_VER)/data/phrase_lib.txt > phrase_lib.txt
-tsi.src:
- wget -nc http://$(SF_MIRROR).dl.sourceforge.net/sourceforge/libtabe/libtabe-$(LIBTABE_VER).tgz
+tsi.src: libtabe-$(LIBTABE_VER).tgz
tar -xzf libtabe-$(LIBTABE_VER).tgz -O libtabe/tsi-src/tsi.src > tsi.src
+# Make a word list
wordlist: phrase_lib.txt EZ.txt.in tsi.src
iconv -c -f big5 -t utf8 tsi.src | $(SED) 's/# //g' | $(SED) 's/[ ][0-9].*//' > wordlist
$(SED) 's/\(.*\)\t[0-9][0-9]*.*/\1/' phrase_lib.txt | $(SED) '1,5d' >> wordlist
cat simp2trad.manual tmp1 > simp2trad.t
t2s_1tomany.t: trad2simp.t
- $(GREP) -s ".\{19,\}" trad2simp.t | $(SED) 's/U+...../"/' | $(SED) 's/|U+...../"=>"/' | $(SED) 's/|U+.....//g' | $(SED) 's/|/",/' > t2s_1tomany.t
+ $(GREP) -s ".\{19,\}" trad2simp.t | $(SED) 's/U+...../"/' | $(SED) 's/|U+...../"=>"/' | $(SED) 's/|U+.....//g' | $(SED) 's/|/",/' > t2s_1tomany.t
t2s_1to1.t: trad2simp.t s2t_1tomany.t
$(SED) "/.*|.*|.*|.*/d" trad2simp.t | $(SED) 's/U+[0-9a-z][0-9a-z]*/"/' | $(SED) 's/|U+[0-9a-z][0-9a-z]*/"=>"/' | $(SED) 's/|/",/' > t2s_1to1.t
ph.t: phrase_lib.txt
$(SED) 's/[\t0-9a-zA-Z]//g' phrase_lib.txt | $(GREP) "^.\{2,4\}$$" > ph.t
-allsimpphrases.t: ph.t
+Wubi.t: Wubi.txt.in
+ $(SED) '1,/BEGIN_TABLE/d' Wubi.txt.in | colrm 1 8 | $(SED) 's/\t.*//' | $(GREP) "^...*" > Wubi.t
+
+Ziranma.t: Ziranma.txt.in
+ $(SED) '1,/BEGIN_TABLE/d' Ziranma.txt.in | colrm 1 8 | $(SED) 's/\t.*//' | $(GREP) "^...*" > Ziranma.t
+
+
+allsimpphrases.t: t2s_1tomany.t ph.t Wubi.t Ziranma.t
rm -f allsimpphrases.t
+ for i in `cat t2s_1tomany.t | $(SED) 's/.*=>".//' | $(SED) 's/"//g' | $(SED) 's/,/\n/' | $(SED) 's/\(.\)/\1\n/g' | sort | uniq `; do $(GREP) $$i Wubi.t >> allsimpphrases.t; done
+ for i in `cat t2s_1tomany.t | $(SED) 's/.*=>".//' | $(SED) 's/"//g' | $(SED) 's/,/\n/' | $(SED) 's/\(.\)/\1\n/g' | sort | uniq `; do $(GREP) $$i Ziranma.t >> allsimpphrases.t; done
for i in `cat t2s_1tomany.t | $(SED) 's/.*=>".//' | $(SED) 's/"//g' | $(SED) 's/,/\n/' | $(SED) 's/\(.\)/\1\n/g' | sort | uniq `; do $(GREP) $$i ph.t >> allsimpphrases.t; done
simpphrases_2.t: allsimpphrases.t
sort t | uniq > t3
$(DIFF) t3 simpphrases_4.t | $(GREP) ">" | $(SED) 's/> //' > t
mv t simpphrases_4.t
- for i in `cat simpphrases_3.t`; do $(GREP) $$i simpphrases_4.t; done | sort | uniq > t3 || true
+ for i in `cat simpphrases_3.t`; do $(GREP) $$i simpphrases_4.t; done | sort | uniq > t3 || true
$(DIFF) t3 simpphrases_4.t | $(GREP) ">" | $(SED) 's/> //' > t
mv t simpphrases_4.t
rm -f \
Unihan.txt \
EZ.txt.in \
+ Wubi.txt.in \
+ Ziranma.txt.in \
phrase_lib.txt \
tsi.src
# Temporary files and other trash
dépôts / lhc / web / wiklou.git / commitdiff