From edf7492ebdc75599150bf9497f4b58503a2849b1 Mon Sep 17 00:00:00 2001 From: Laurent Mazet Date: Sun, 14 Jun 2020 16:26:24 +0200 Subject: [PATCH] use dictionaries from Aspell website --- dict-dists-extractor/Makefile | 74 +++++++++++++++++++++-------------- 1 file changed, 45 insertions(+), 29 deletions(-) diff --git a/dict-dists-extractor/Makefile b/dict-dists-extractor/Makefile index f13a7ab..2ebba6c 100644 --- a/dict-dists-extractor/Makefile +++ b/dict-dists-extractor/Makefile @@ -6,13 +6,16 @@ LANGS = bg ca da de en eo es fr it nl no pl pt sv uk # to add: ar br cs cy el et eu fi? ga he hr hu hy is lv ro ru sk sl #LANGS = af ar bg br ca cs cy da de el en eo es et eu fi fr ga he hr hu hy is it la lv ms nl no pl pt ro ru sk sl sv tr uk xk +ASPDICT = https://ftp.gnu.org/gnu/aspell/dict/0index.html + LOWS = $(addsuffix .low,$(LANGS)) DICTS = $(addsuffix .dict,$(LANGS)) +DEPEND = Makefile MAKEFLAGS = -s -all: +all: src.mk make $(addprefix check-,$(LANGS)) $(ROOT_DICT): @@ -23,32 +26,45 @@ count: low: $(LOWS) -INSTALL_ASPELL_DICT = apt search aspell-$(1) && \ - ( dpkg -l aspell-$(1) | grep -q ^ii || sudo apt install aspell-$(1) ) -ASPELL_DUMP_MASTER = aspell -d $(1) dump master $(1) | \ - aspell -l $(1) expand | \ +src.mk: + wget $(ASPDICT) -q -O - | \ + awk 'BEGIN { base="'`dirname $(ASPDICT)`'" } \ + // { exit }' > $@ + +include $(wildcard src.mk) + +INSTALL_ASPELL_DICT = tar xjf $(1) && \ + cd $(2) && \ + ./configure && \ + make + +ASPELL_DUMP_MASTER = aspell --dict-dir=$(2) -d $(1) dump master $(1) | \ + aspell --dict-dir=$(2) -l $(1) expand | \ tr -s '[:space:]' '\n' | \ LC_ALL=C sort -S28G -u > $(1).low -%.low: +%.dir: + echo Installing $(@:.dir=) dictionary + $(call INSTALL_ASPELL_DICT,$<,$(<:.tar.bz2=)) + mv $(<:.tar.bz2=) $@ + +%.low: %.dir echo Building list of word $@ - $(call INSTALL_ASPELL_DICT,$(@:.low=)) - $(call ASPELL_DUMP_MASTER,$(@:.low=)) + $(call ASPELL_DUMP_MASTER,$(<:.dir=),./$<) no.low: - echo Building list of word $@ - $(call INSTALL_ASPELL_DICT,no) - $(call ASPELL_DUMP_MASTER,nb) + make nb.low mv nb.low $@ pt.low: - echo Building list of word $@ - $(call INSTALL_ASPELL_DICT,pt-pt) - $(call ASPELL_DUMP_MASTER,pt_PT) + make pt_PT.low mv pt_PT.low $@ -$(ROOT_DICT)/bg.dict: bg.low Makefile +$(ROOT_DICT)/bg.dict: bg.low $(DEPEND) echo Building $@ from $< cat $< | \ grep -v '^.$$' | \ @@ -58,7 +74,7 @@ $(ROOT_DICT)/bg.dict: bg.low Makefile sort | \ uniq > $@ -$(ROOT_DICT)/ca.dict: ca.low Makefile +$(ROOT_DICT)/ca.dict: ca.low $(DEPEND) echo Building $@ from $< cat $< | \ grep -v '^.$$' | \ @@ -74,7 +90,7 @@ $(ROOT_DICT)/ca.dict: ca.low Makefile sort | \ uniq > $@ -$(ROOT_DICT)/da.dict: da.low Makefile +$(ROOT_DICT)/da.dict: da.low $(DEPEND) echo Building $@ from $< cat $< | \ grep -v '^.$$' | \ @@ -89,7 +105,7 @@ $(ROOT_DICT)/da.dict: da.low Makefile sort | \ uniq > $@ -$(ROOT_DICT)/de.dict: de.low Makefile +$(ROOT_DICT)/de.dict: de.low $(DEPEND) echo Building $@ from $< cat $< | \ grep -v '^.$$' | \ @@ -103,7 +119,7 @@ $(ROOT_DICT)/de.dict: de.low Makefile sort | \ uniq > $@ -$(ROOT_DICT)/en.dict: en.low Makefile +$(ROOT_DICT)/en.dict: en.low $(DEPEND) echo Building $@ from $< cat $< | \ grep -v '^.$$' | \ @@ -119,7 +135,7 @@ $(ROOT_DICT)/en.dict: en.low Makefile sort | \ uniq > $@ -$(ROOT_DICT)/eo.dict: eo.low Makefile +$(ROOT_DICT)/eo.dict: eo.low $(DEPEND) echo Building $@ from $< cat $< | \ grep -v '^.$$' | \ @@ -128,7 +144,7 @@ $(ROOT_DICT)/eo.dict: eo.low Makefile sort | \ uniq > $@ -$(ROOT_DICT)/es.dict: es.low Makefile +$(ROOT_DICT)/es.dict: es.low $(DEPEND) echo Building $@ from $< cat $< | \ grep -v '^.$$' | \ @@ -142,7 +158,7 @@ $(ROOT_DICT)/es.dict: es.low Makefile sort | \ uniq > $@ -$(ROOT_DICT)/fr.dict: fr.low Makefile +$(ROOT_DICT)/fr.dict: fr.low $(DEPEND) echo Building $@ from $< cat $< | \ grep -v '^.$$' | \ @@ -157,7 +173,7 @@ $(ROOT_DICT)/fr.dict: fr.low Makefile sort | \ uniq > $@ -$(ROOT_DICT)/it.dict: it.low Makefile +$(ROOT_DICT)/it.dict: it.low $(DEPEND) echo Building $@ from $< cat $< | \ grep -v '^.$$' | \ @@ -172,7 +188,7 @@ $(ROOT_DICT)/it.dict: it.low Makefile sort | \ uniq > $@ -$(ROOT_DICT)/nl.dict: nl.low Makefile +$(ROOT_DICT)/nl.dict: nl.low $(DEPEND) echo Building $@ from $< cat $< | \ grep -v '^.$$' | \ @@ -188,7 +204,7 @@ $(ROOT_DICT)/nl.dict: nl.low Makefile sort | \ uniq > $@ -$(ROOT_DICT)/no.dict: no.low Makefile +$(ROOT_DICT)/no.dict: no.low $(DEPEND) echo Building $@ from $< cat $< | \ grep -v '^.$$' | \ @@ -204,7 +220,7 @@ $(ROOT_DICT)/no.dict: no.low Makefile sort | \ uniq > $@ -$(ROOT_DICT)/pl.dict: pl.low Makefile +$(ROOT_DICT)/pl.dict: pl.low $(DEPEND) echo Building $@ from $< cat $< | \ grep -v '^.$$' | \ @@ -220,7 +236,7 @@ $(ROOT_DICT)/pl.dict: pl.low Makefile sort | \ uniq > $@ -$(ROOT_DICT)/pt.dict: pt.low Makefile +$(ROOT_DICT)/pt.dict: pt.low $(DEPEND) echo Building $@ from $< cat $< | \ grep -v '^.$$' | \ @@ -234,7 +250,7 @@ $(ROOT_DICT)/pt.dict: pt.low Makefile sort | \ uniq > $@ -$(ROOT_DICT)/sv.dict: sv.low Makefile +$(ROOT_DICT)/sv.dict: sv.low $(DEPEND) echo Building $@ from $< cat $< | \ grep -v '^.$$' | \ @@ -247,7 +263,7 @@ $(ROOT_DICT)/sv.dict: sv.low Makefile sort | \ uniq > $@ -$(ROOT_DICT)/uk.dict: uk.low Makefile +$(ROOT_DICT)/uk.dict: uk.low $(DEPEND) echo Building $@ from $< cat $< | \ grep -v '^.$$' | \ -- 2.30.2