scripts: add keyboard-scraper.sh

This script runs keyboard-configuration.config in a chroot.  By running
interesting locales against it, we can obtain keyboard suggestions.

kbds/keyboard-configuration.yaml

@@ -0,0 +1,86 @@
+# File generated by ./scripts/keyboard-scraper.sh, please do not edit manually
+# Based on output from keyboard-configuration.config
+en_US.UTF-8:
+  layout: "us"
+  variant: ""
+en_GB.UTF-8:
+  layout: "gb"
+  variant: ""
+kab_DZ.UTF-8:
+  layout: "dz"
+  variant: "la"
+ca_ES.UTF-8:
+  layout: "es"
+  variant: "cat"
+zh_CN.UTF-8:
+  layout: "cn"
+  variant: ""
+hr_HR.UTF-8:
+  layout: "hr"
+  variant: ""
+cs_CZ.UTF-8:
+  layout: "cz"
+  variant: ""
+ast_ES.UTF-8:
+  layout: "es"
+  variant: "ast"
+be_BY.UTF-8:
+  layout: "us,by"
+  variant: ""
+nl_NL.UTF-8:
+  layout: "us"
+  variant: ""
+fi_FI.UTF-8:
+  layout: "fi"
+  variant: ""
+fr_FR.UTF-8:
+  layout: "fr"
+  variant: "latin9"
+de_DE.UTF-8:
+  layout: "de"
+  variant: ""
+el_GR.UTF-8:
+  layout: "us,gr"
+  variant: ""
+he_IL.UTF-8:
+  layout: "us,il"
+  variant: ""
+hu_HU.UTF-8:
+  layout: "hu"
+  variant: ""
+id_ID.UTF-8:
+  layout: "us"
+  variant: ""
+lv_LV.UTF-8:
+  layout: "lv"
+  variant: ""
+lt_LT.UTF-8:
+  layout: "lt"
+  variant: ""
+pl_PL.UTF-8:
+  layout: "pl"
+  variant: ""
+ru_RU.UTF-8:
+  layout: "us,ru"
+  variant: ""
+sr_RS:
+  layout: "rs,rs"
+  variant: "latin,"
+es_ES.UTF-8:
+  layout: "es"
+  variant: ""
+sv_SE.UTF-8:
+  layout: "se"
+  variant: ""
+bo_IN:
+  layout: "us,cn"
+  variant: ",tib"
+uk_UA.UTF-8:
+  layout: "us,ua"
+  variant: ""
+nb:
+  layout: "us"
+  variant: ""
+oc:
+  layout: "us"
+  variant: ""

scripts/keyboard-scraper.sh

@@ -0,0 +1,74 @@
+#!/bin/bash
+
+# usage: ./scripts/keyboard-scraper.sh
+# setup enough of a chroot to run keyboard-configuration.config
+# and scrape it for keyboard recommendations
+
+set -eu
+
+outfile=kbds/keyboard-configuration.yaml
+langs=$(cat languagelist | cut -d: -f2)
+
+tmpdir=$(mktemp -d -p $(pwd))
+trap "rm -fr $tmpdir" EXIT
+
+if [ ! -f keyboard-configuration.config ] ; then
+    wget https://salsa.debian.org/installer-team/console-setup/-/raw/master/debian/keyboard-configuration.config
+fi
+
+# setup prerequisites for the config script
+mkdir -p $tmpdir/bin
+cp /usr/bin/busybox $tmpdir/bin/sh
+ln $tmpdir/bin/sh $tmpdir/bin/sed
+ln $tmpdir/bin/sh $tmpdir/bin/grep
+
+mkdir -p $tmpdir/usr/share/debconf
+touch $tmpdir/usr/share/debconf/confmodule
+
+mkdir -p $tmpdir/dev
+sudo mknod -m 666 $tmpdir/dev/null c 1 3
+
+cp keyboard-configuration.config $tmpdir
+
+# more prerequisites, and save db_set output for later consumption
+cat > $tmpdir/stage2 <<EOF
+all_kbdnames() { :; }
+db_capb() { :; }
+db_get() { :; }
+keyboard_present() { false; }
+
+dpkg()
+{
+    echo "$(dpkg --print-architecture)"
+}
+
+db_set()
+{
+    echo "\$1" "\$2" >> /db_set.\$LC_CTYPE
+}
+
+# LC_CTYPE is used by the config script to decide which locale we're providing
+# a keyboard suggestion for.
+LC_CTYPE="\$1"
+export LC_CTYPE
+. /keyboard-configuration.config
+EOF
+chmod 755 $tmpdir/stage2
+
+cat > $outfile <<EOF
+# File generated by $0, please do not edit manually
+# Based on output from keyboard-configuration.config
+EOF
+
+for locale in $langs ; do
+    echo "$locale"
+    sudo chroot $tmpdir /bin/sh /stage2 "$locale"
+    # transform interesting db_set values into a simplistic yaml format
+    echo "$locale:" >> $outfile
+    awk '
+        /layoutcode/ {print "  layout: " "\""$2"\""}
+        /variantcode/ {print "  variant: " "\""$2"\""}
+    ' < $tmpdir/db_set.$locale >> $outfile
+done
+
+echo "output to $outfile"