language: Add generate-normalize-data maintenance script

Change-Id: I64658cb74d81365a99d6b0a01807324eb5a39320

diff --git a/languages/data/README.md b/languages/data/README.md
deleted file mode 100644 (file)
index 8a13fdc..0000000
--- a/languages/data/README.md
+++ /dev/null
@@ -1,7 +0,0 @@
-## normalize-ar.phpc
-
-Generated by `maintenance/language/generateNormalizerDataAr.php`.
-
-## normalize-ml.php
-
-Generated by `maintenance/language/generateNormalizerDataMl.php`.

diff --git a/maintenance/language/generate-normalize-data b/maintenance/language/generate-normalize-data
new file mode 100755 (executable)
index 0000000..b327c6a
--- /dev/null
+++ b/maintenance/language/generate-normalize-data
@@ -0,0 +1,19 @@
+#!/bin/bash -eu
+
+clear_temp() {
+       rm -f UnicodeData.txt
+}
+
+# Setup
+clear_temp
+export MW_INSTALL_PATH="$(cd "$(dirname "${BASH_SOURCE[0]}")"/../.. && pwd)"
+
+# Generate languages/data/normalize-ar.php
+wget https://unicode.org/Public/6.0.0/ucd/UnicodeData.txt
+php "$MW_INSTALL_PATH/maintenance/language/generateNormalizerDataAr.php" --unicode-data-file UnicodeData.txt
+
+# Generate languages/data/normalize-ml.php
+php "$MW_INSTALL_PATH/maintenance/language/generateNormalizerDataMl.php"
+
+# Teardown
+clear_temp