chore(corp/data-import): make OR data archive available in env
Change-Id: Idacf42743051eae0cf7010f952a4f91af17ad708 Reviewed-on: https://cl.tvl.fyi/c/depot/+/7892 Reviewed-by: tazjin <tazjin@tvl.su> Tested-by: BuildkiteCI
This commit is contained in:
parent
1a84e3e6d8
commit
ee0c0ee951
1 changed files with 15 additions and 8 deletions
|
@ -9,22 +9,29 @@ let
|
||||||
# mirrored input data from OpenCorpora, as of 2023-01-17.
|
# mirrored input data from OpenCorpora, as of 2023-01-17.
|
||||||
#
|
#
|
||||||
# This data is licensed under CC-BY-SA.
|
# This data is licensed under CC-BY-SA.
|
||||||
inputDataArchive = pkgs.fetchurl {
|
openCorporaArchive = pkgs.fetchurl {
|
||||||
name = "dict.opcorpora.xml.bz";
|
name = "dict.opcorpora.xml.bz";
|
||||||
url = "https://tazj.in/blobs/dict.opcorpora.xml.bz2";
|
url = "https://tazj.in/blobs/opencorpora-20230117.xml.bz2";
|
||||||
sha256 = "04n5g43fkfc93z6xlwf2qfdrfdfl562pc2ivdb3cbbbsy56gkqg6";
|
sha256 = "04n5g43fkfc93z6xlwf2qfdrfdfl562pc2ivdb3cbbbsy56gkqg6";
|
||||||
};
|
};
|
||||||
|
|
||||||
inputData = pkgs.runCommand "dict.opcorpora.xml" { } ''
|
openCorpora = pkgs.runCommand "dict.opcorpora.xml" { } ''
|
||||||
${pkgs.bzip2}/bin/bunzip2 -k -c ${inputDataArchive} > $out
|
${pkgs.bzip2}/bin/bunzip2 -k -c ${openCorporaArchive} > $out
|
||||||
'';
|
'';
|
||||||
|
|
||||||
|
openRussianArchive = pkgs.fetchzip {
|
||||||
|
name = "openrussian-20230117";
|
||||||
|
url = "https://tazj.in/blobs/openrussian-20230117.tar.xz";
|
||||||
|
sha256 = "06jl7i23cx58a0n2626hb82xlzimixvnxp7lxdw0g664kv9bmw25";
|
||||||
|
};
|
||||||
|
|
||||||
# development shell with native deps
|
# development shell with native deps
|
||||||
shell = pkgs.mkShell {
|
shell = pkgs.mkShell {
|
||||||
inherit buildInputs;
|
inherit buildInputs;
|
||||||
|
|
||||||
# make OPENCORPORA_DATA available in the environment
|
# make datasets available in the environment
|
||||||
OPENCORPORA_DATA = inputData;
|
OPENCORPORA_DATA = openCorpora;
|
||||||
|
OPENRUSSIAN_DATA = openRussianArchive;
|
||||||
};
|
};
|
||||||
|
|
||||||
in
|
in
|
||||||
|
@ -33,11 +40,11 @@ lib.fix (self: depot.third_party.naersk.buildPackage {
|
||||||
inherit buildInputs;
|
inherit buildInputs;
|
||||||
|
|
||||||
passthru = depot.nix.readTree.drvTargets {
|
passthru = depot.nix.readTree.drvTargets {
|
||||||
inherit shell inputData;
|
inherit shell openCorpora;
|
||||||
|
|
||||||
# target that actually builds an entire database
|
# target that actually builds an entire database
|
||||||
database = pkgs.runCommand "tvl-russian-db.sqlite" { } ''
|
database = pkgs.runCommand "tvl-russian-db.sqlite" { } ''
|
||||||
${self}/bin/data-import ${inputData} $out
|
${self}/bin/data-import ${openCorpora} $out
|
||||||
'';
|
'';
|
||||||
};
|
};
|
||||||
})
|
})
|
||||||
|
|
Loading…
Reference in a new issue