tiny-lang-detect

commit

3db712ebeea5d1322b6ef9d9470955a566710262

parent

c2755acb5d4af6c990d63f827e18c83e33e20c69

Author

Tobias Bengfort <tobias.bengfort@posteo.de>

Date

2025-05-06 05:06

convert to shell script

Diffstat

2 files changed, 15 insertions, 18 deletions

diff --git a/Makefile b/Makefile

@@ -1,18 +0,0 @@
    1    -1 .PHONY: all
    2    -1 all: data/wili data/profiles
    3    -1 
    4    -1 data/wili:
    5    -1 	@mkdir -p $@
    6    -1 	wget https://zenodo.org/records/841984/files/wili-2018.zip?download=1 -O /tmp/wili.zip
    7    -1 	unzip /tmp/wili.zip '*_test.txt' -d $@
    8    -1 	@rm /tmp/wili.zip
    9    -1 
   10    -1 data/profiles:
   11    -1 	@mkdir -p $@
   12    -1 	wget https://github.com/DoodleBears/langdetect/archive/refs/heads/master.zip -O /tmp/langdetect.zip
   13    -1 	unzip -j /tmp/langdetect.zip 'langdetect-master/langdetect/profiles/*' -d $@
   14    -1 	@rm /tmp/langdetect.zip
   15    -1 
   16    -1 .PHONY: clean
   17    -1 clean:
   18    -1 	rm -rf data

diff --git a/download_data.sh b/download_data.sh

@@ -0,0 +1,15 @@
   -1     1 #!/bin/sh
   -1     2 
   -1     3 set -e
   -1     4 
   -1     5 rm -rf data
   -1     6 
   -1     7 mkdir -p data/wili
   -1     8 wget 'https://zenodo.org/records/841984/files/wili-2018.zip?download=1' -O /tmp/wili.zip
   -1     9 unzip /tmp/wili.zip '*_test.txt' -d data/wili
   -1    10 rm /tmp/wili.zip
   -1    11 
   -1    12 mkdir -p data/profiles
   -1    13 wget 'https://github.com/DoodleBears/langdetect/archive/refs/heads/master.zip' -O /tmp/langdetect.zip
   -1    14 unzip -j /tmp/langdetect.zip 'langdetect-master/langdetect/profiles/*' -d data/profiles
   -1    15 rm /tmp/langdetect.zip