summaryrefslogtreecommitdiff
path: root/doc
diff options
context:
space:
mode:
Diffstat (limited to 'doc')
-rw-r--r--doc/Makefile31
1 files changed, 31 insertions, 0 deletions
diff --git a/doc/Makefile b/doc/Makefile
index 8ecf093..0cd56a4 100644
--- a/doc/Makefile
+++ b/doc/Makefile
@@ -1,3 +1,11 @@
+### settings ###
+CORPUSDIR:=../corpus
+STATSDIR:=..
+WIKIEXTRACTOR:=../3rdparty/wikiextractor/WikiExtractor.py
+OPTROUNDS=100000
+# pin layers, keep hand-optimized numbers, keep top row free
+OPTPINS='0;1;2;0,Bl1;0,Bl2;0,Bl3;0,Bl4;0,Bl5;0,Bl6;0,Bl7;0,Br6;0,Br5;0,Br4;0,Br3;0,Br2;0,Br1;3,Cl4;3,Cl3;3,Cl2;3,Cl1;3,Dl4;3,Dl3;3,Dl2;3,Dl1;3,El5;3,El4;3,El3;3,El2;3,Dl5;3,Cl5;3,El6'
+
all: ar-lulua.xmodmap ar-lulua.svg ar-asmo663.svg ar-linux.svg ar-malas.svg ar-phonetic.svg ar-osman.svg letterfreq.json ar-khorshid.svg
letterfreq.json: ../stats.pickle
@@ -27,3 +35,26 @@ ar-osman.svg:
ar-khorshid.svg:
lulua-render svg -l ar-khorshid $@
+### corpora to stats ###
+# Sorry, but I can’t provide corpus data due to copyright issues
+$(STATSDIR)/stats-bbcarabic.pickle: $(CORPUSDIR)/bbcarabic/raw
+ time find $< -type f | lulua-write bbcarabic ar-lulua > $@
+
+$(STATSDIR)/stats-aljazeera.pickle: $(CORPUSDIR)/aljazeera/raw
+ time find $< -type f | lulua-write aljazeera ar-lulua > $@
+
+$(STATSDIR)/stats-tanzil.pickle: $(CORPUSDIR)/tanzil-quaran/plain.txt.lz
+ echo $< | lulua-write text ar-lulua | lulua-combine-stats > $@
+
+$(STATSDIR)/stats-arwiki.pickle: $(CORPUSDIR)/arwiki/arwiki-20190701-pages-articles.xml.bz2
+ $(WIKIEXTRACTOR) -ns 0 --json -o - $< 2>/dev/null | jq .text | lulua-write json ar-lulua | lulua-combine-stats > $@
+
+$(STATSDIR)/stats.pickle: $(STATSDIR)/stats-arwiki.pickle \
+ $(STATSDIR)/stats-tanzil.pickle $(STATSDIR)/stats-aljazeera.pickle \
+ $(STATSDIR)/stats-bbcarabic.pickle
+ cat $^ | lulua-combine-stats > $@
+
+### optimization ###
+optimized.yaml: $(STATSDIR)/stats.pickle
+ lulua-optimize -n $(OPTROUNDS) -r -p $(OPTPINS) -l ar-lulua < $< > $@
+