gen-data: add infrastructure for data generation

Add also first task 'basic' which generates simple 'pagila'
database.

* generate: New file.
* databases/pagila.sh: New database file.
* tasks/basic/run.sh: New task file.

3 files changed, 166 insertions, 0 deletions

diff --git a/gen-data/databases/pagila.sh b/gen-data/databases/pagila.sh
new file mode 100644
index 0000000..a01d818
--- /dev/null
+++ b/gen-data/databases/pagila.sh
@@ -0,0 +1,39 @@
+create_pagila()
+(
+    testit()
+    {
+        debug "testing database"
+
+        local cmd="psql -tA -d pagila -c \"select city from city where city = 'Banjul';\""
+
+        out="$(admin_cmd "$cmd")"
+        test "$out" = Banjul || return 1
+        test "$(wc -l < pagila_init.log)" -gt 335 || return 1
+
+        test "$({ grep ERROR | wc -l ; } < pagila_init.log)" -lt 2
+    }
+
+    debug "creating DB pagilla"
+    INDENT="$INDENT  "
+
+    pagila="pagila-0.10.1"
+    pagila_tarball="$pagila.zip"
+    pagila_link="http://pgfoundry.org/frs/download.php/1719/$pagila_tarball"
+
+    cached_download $pagila_link
+
+    debug "unzipping tarball"
+    unzip $pagila_tarball &>/dev/null || die "can not unzip pagila"
+
+    pushd $pagila >/dev/null || die "can not switch directory"
+    admin_cmd "createdb pagila --owner postgres" || die "can't create db"
+    { su - postgres -c 'psql -d pagila' < pagila-schema.sql \
+      && su - postgres -c 'psql -d pagila' < pagila-data.sql
+    } &>pagila_init.log || die "can not initialize pagila"
+
+    cp pagila_init.log /tmp
+
+    testit || die "can not test"
+
+    popd >/dev/null || die "can't go back"
+)
diff --git a/gen-data/generate b/gen-data/generate
new file mode 100755
index 0000000..bf9cce3
--- /dev/null
+++ b/gen-data/generate
@@ -0,0 +1,121 @@
+#!/bin/bash
+
+GEN_DATADIR=/var/lib/pgsql/data
+CACHEDIR=/var/cache/dbt
+SRCDIR="$(pwd)"
+DEBUG=1
+CHECK_LOCALE=1
+
+printline() { echo "$INDENT$@" ; }
+
+die() { printline " ERROR $@" ; exit 1 ; }
+
+info() { printline " * $@"; }
+
+debug() { test $DEBUG -eq 1 && printline " ~ $@"; }
+
+admin_cmd()
+{
+    su - postgres -c "$@" || die "can not execute '$@'"
+}
+
+cached_download()
+{
+    local link=$1
+    local bn="$(basename "$1")"
+    local file="$CACHEDIR/$bn"
+
+    if test ! -d "$CACHEDIR"; then
+        mkdir -p "$CACHEDIR" || die "can't create cachedir"
+    fi
+
+    test -e "$file" && cp "$file" ./ && return
+
+    { wget "$link" && cp "$bn" "$CACHEDIR" ; } \
+        || die "can't download"
+}
+
+is_defined_f()
+{
+    declare -f "$1" >/dev/null
+}
+
+run_if_defined()
+{
+    is_defined_f "$1" || return 0
+    "$1"
+}
+
+locale_prereq()
+{
+    local default_locale="en_US.UTF-8"
+    local current_locale="$(cat /etc/locale.conf)"
+    test "$current_locale" = "$default_locale" \
+        || die "bad locale $current_locale, should be $default_locale"
+}
+
+single_task()
+{
+    local task="$1"
+
+    info "running task $task"
+    INDENT="  "
+
+    (
+        . "$SRCDIR/databases/pagila.sh" || die "pagila incl fail"
+        . "$SRCDIR/$task" || die "include fail"
+
+        run_if_defined "hook_start"
+
+        test "$CHECK_LOCALE" -eq 1 && locale_prereq
+
+        debug "initializing database"
+        postgresql-setup initdb &>/dev/null || die "can not initdb"
+
+        debug "starting server"
+        service postgresql start &>/dev/null || die "can't start postgresql"
+
+        tmpdir=$(mktemp -d /tmp/dbt-XXXXXX) || die "can not create temp directory"
+
+        debug "working in $tmpdir"
+        pushd "$tmpdir" >/dev/null || die "can switch to $tmpdir"
+
+        run || die "fail"
+
+        popd >/dev/null || die "can not switch back"
+
+        debug "stopping server"
+        service postgresql stop &>/dev/null || die "can't stop postgresql"
+
+        tarball="$(pwd)/$task_name.tar.gz"
+        tar -czf "$tarball" -C /var/lib/pgsql/ data
+        info "result tarball $tarball"
+        rm -rf "$tmpdir"
+
+        debug "data removal"
+        rm -rf "$GEN_DATADIR"
+    )
+}
+
+generate_tasks()
+{
+    single_task tasks/basic/run.sh
+}
+
+main()
+{
+    test "$UID" -ne 0 && die "run under 'root' user"
+
+    service postgresql status &>/dev/null
+    test $? -ne 3 && die "stop the postgresql server"
+
+    local dbdir="$GEN_DATADIR"
+    test -e "$dbdir" && die "the '$dbdir' should not exist"
+
+    { rpm -qa postgresql && rpm -qa postgresql-server ; } &>/dev/null \
+        || die "PostgreSQL not installed"
+
+    generate_tasks
+}
+
+main
diff --git a/gen-data/tasks/basic/run.sh b/gen-data/tasks/basic/run.sh
new file mode 100644
index 0000000..0118ad3
--- /dev/null
+++ b/gen-data/tasks/basic/run.sh
@@ -0,0 +1,6 @@
+export task_name="basic"
+
+run()
+{
+    create_pagila
+}