anaconda: support 3rd-party repos. tuna/issues#112

This commit is contained in:
bigeagle 2017-01-09 20:35:15 +08:00
parent c421400d5e
commit a5f8abf769

View File

@ -1,6 +1,5 @@
#!/bin/bash #!/bin/bash
# requires: wget, lftp, jq, python3.5, lxml, pyquery # requires: wget, lftp, jq, python3.5, lxml, pyquery
set -e set -e
set -u set -u
set -o pipefail set -o pipefail
@ -8,13 +7,24 @@ set -o pipefail
_here=`dirname $(realpath $0)` _here=`dirname $(realpath $0)`
HTMLPARSE="${_here}/helpers/anaconda-filelist.py" HTMLPARSE="${_here}/helpers/anaconda-filelist.py"
CONDA_REPO_BASE="${CONDA_REPO_BASE:-"https://repo.continuum.io"}" DEFAULT_CONDA_REPO_BASE="https://repo.continuum.io"
LOCAL_DIR_BASE="${TUNASYNC_WORKING_DIR}/pkgs" DEFAULT_CONDA_CLOUD_BASE="https://conda.anaconda.org"
CONDA_REPO_BASE="${CONDA_REPO_BASE:-$DEFAULT_CONDA_REPO_BASE}"
CONDA_CLOUD_BASE="${CONDA_CLOUD_BASE:-$DEFAULT_CONDA_CLOUD_BASE}"
LOCAL_DIR_BASE="${TUNASYNC_WORKING_DIR}"
TMP_DIR=$(mktemp -d) TMP_DIR=$(mktemp -d)
CONDA_REPOS=("free" "r" "mro" "pro") CONDA_REPOS=("free" "r" "mro" "pro")
CONDA_ARCHES=("linux-64" "linux-32" "linux-armv6l" "linux-armv7l" "linux-ppc64le" "osx-64" "osx-32" "win-64" "win-32") CONDA_ARCHES=("linux-64" "linux-32" "linux-armv6l" "linux-armv7l" "linux-ppc64le" "osx-64" "osx-32" "win-64" "win-32")
CONDA_CLOUD_REPOS=(
"conda-forge/linux-64" "conda-forge/osx-64" "conda-forge/win-64"
"msys2/win-64" "r/linux-64" "r/osx-64" "r/osx-64"
)
EXIT_STATUS=0 EXIT_STATUS=0
EXIT_MSG="" EXIT_MSG=""
@ -32,10 +42,10 @@ function check-and-download() {
function cleanup () { function cleanup () {
echo "cleaning up" echo "cleaning up"
[ -d ${TMP_DIR} ] && { [[ -d ${TMP_DIR} ]] && {
[ -f ${TMP_DIR}/repodata.json ] && rm ${TMP_DIR}/repodata.json [[ -f ${TMP_DIR}/repodata.json ]] && rm ${TMP_DIR}/repodata.json
[ -f ${TMP_DIR}/repodata.json.bz2 ] && rm ${TMP_DIR}/repodata.json.bz2 [[ -f ${TMP_DIR}/repodata.json.bz2 ]] && rm ${TMP_DIR}/repodata.json.bz2
[ -f ${TMP_DIR}/failed ] && rm ${TMP_DIR}/failed [[ -f ${TMP_DIR}/failed ]] && rm ${TMP_DIR}/failed
rmdir ${TMP_DIR} rmdir ${TMP_DIR}
} }
} }
@ -53,6 +63,8 @@ function download-with-checksum() {
wget -q -O ${dest_file} ${pkg_url} && { wget -q -O ${dest_file} ${pkg_url} && {
# two space for md5sum check format # two space for md5sum check format
{ md5sum -c - < <(echo "${pkgmd5} ${dest_file}"); } && downloaded=true || trials=$((trials + 1)) { md5sum -c - < <(echo "${pkgmd5} ${dest_file}"); } && downloaded=true || trials=$((trials + 1))
} || {
trials=$((trials + 1))
} }
if (( $trials > 3 )); then if (( $trials > 3 )); then
return 1 return 1
@ -70,7 +82,6 @@ function sync_installer() {
[[ ! -d "$repo_dir" ]] && mkdir -p "$repo_dir" [[ ! -d "$repo_dir" ]] && mkdir -p "$repo_dir"
cd $repo_dir cd $repo_dir
# lftp "${repo_url}/" -e "mirror --verbose -P 5; bye"
while read -a tokens; do while read -a tokens; do
fname=${tokens[0]} fname=${tokens[0]}
@ -89,25 +100,24 @@ function sync_installer() {
fi fi
download-with-checksum ${pkg_url} ${dest_file} ${pkgmd5} || { download-with-checksum ${pkg_url} ${dest_file} ${pkgmd5} || {
echo "Failed to download ${pkg_url}: checksum mismatch" echo "Failed to download ${pkg_url}: checksum mismatch"
echo ${pkg_url} >> ${TMP_DIR}/failed
EXIT_STATUS=2 EXIT_STATUS=2
EXIT_MSG="some files has bad checksum." EXIT_MSG="some files has bad checksum."
} }
done < <(wget -O- ${repo_url} | $HTMLPARSE) done < <(wget -O- ${repo_url} | $HTMLPARSE)
} }
sync_installer "${CONDA_REPO_BASE}/archive/" "${TUNASYNC_WORKING_DIR}/archive/" function sync_repo () {
sync_installer "${CONDA_REPO_BASE}/miniconda/" "${TUNASYNC_WORKING_DIR}/miniconda/" local repo_url="$1"
local local_dir="$2"
[[ ! -d ${local_dir} ]] && mkdir -p ${local_dir}
repodata_url="${repo_url}/repodata.json"
bz2_repodata_url="${repo_url}/repodata.json.bz2"
for repo in ${CONDA_REPOS[@]}; do
for arch in ${CONDA_ARCHES[@]}; do
PKG_REPO_BASE="${CONDA_REPO_BASE}/pkgs/$repo/$arch"
repodata_url="${PKG_REPO_BASE}/repodata.json"
bz2_repodata_url="${PKG_REPO_BASE}/repodata.json.bz2"
LOCAL_DIR="${LOCAL_DIR_BASE}/$repo/$arch"
[ ! -d ${LOCAL_DIR} ] && mkdir -p ${LOCAL_DIR}
tmp_repodata="${TMP_DIR}/repodata.json" tmp_repodata="${TMP_DIR}/repodata.json"
tmp_bz2_repodata="${TMP_DIR}/repodata.json.bz2" tmp_bz2_repodata="${TMP_DIR}/repodata.json.bz2"
tmp_failed_files="${TMP_DIR}/failed"
check-and-download ${repodata_url} ${tmp_repodata} check-and-download ${repodata_url} ${tmp_repodata}
check-and-download ${bz2_repodata_url} ${tmp_bz2_repodata} check-and-download ${bz2_repodata_url} ${tmp_bz2_repodata}
@ -120,8 +130,8 @@ for repo in ${CONDA_REPOS[@]}; do
pkgsize=${tokens[1]} pkgsize=${tokens[1]}
pkgmd5=${tokens[2]} pkgmd5=${tokens[2]}
pkg_url="${PKG_REPO_BASE}/${pkgfile}" pkg_url="${repo_url}/${pkgfile}"
dest_file="${LOCAL_DIR}/${pkgfile}" dest_file="${local_dir}/${pkgfile}"
if [[ -f ${dest_file} ]]; then if [[ -f ${dest_file} ]]; then
rsize=`stat -c "%s" ${dest_file}` rsize=`stat -c "%s" ${dest_file}`
@ -130,19 +140,44 @@ for repo in ${CONDA_REPOS[@]}; do
continue continue
fi fi
fi fi
download-with-checksum ${pkg_url} ${dest_file} ${pkgmd5} || { download-with-checksum ${pkg_url} ${dest_file} ${pkgmd5} || {
echo "Failed to download ${pkg_url}: checksum mismatch" echo "Failed to download ${pkg_url}: checksum mismatch"
echo ${pkg_url} >> ${tmp_failed_files} echo ${pkg_url} >> ${TMP_DIR}/failed
EXIT_MSG="some files has bad checksum." EXIT_MSG="some files has bad checksum."
} }
done < <(bzip2 -c -d ${tmp_bz2_repodata} | jq -r "${jq_cmd}") done < <(bzip2 -c -d ${tmp_bz2_repodata} | jq -r "${jq_cmd}")
mv -f "${TMP_DIR}/repodata.json" "${LOCAL_DIR}/repodata.json" mv -f "${TMP_DIR}/repodata.json" "${local_dir}/repodata.json"
mv -f "${TMP_DIR}/repodata.json.bz2" "${LOCAL_DIR}/repodata.json.bz2" mv -f "${TMP_DIR}/repodata.json.bz2" "${local_dir}/repodata.json.bz2"
mv -f "${tmp_failed_files}" "${TUNASYNC_WORKING_DIR}/failed_packages.txt" }
sync_installer "${CONDA_REPO_BASE}/archive/" "${LOCAL_DIR_BASE}/archive/"
sync_installer "${CONDA_REPO_BASE}/miniconda/" "${LOCAL_DIR_BASE}/miniconda/"
for repo in ${CONDA_REPOS[@]}; do
for arch in ${CONDA_ARCHES[@]}; do
remote_url="${CONDA_REPO_BASE}/pkgs/$repo/$arch"
local_dir="${LOCAL_DIR_BASE}/pkgs/$repo/$arch"
sync_repo "${remote_url}" "${local_dir}" || true
done done
done done
for repo in ${CONDA_CLOUD_REPOS[@]}; do
remote_url="${CONDA_CLOUD_BASE}/${repo}"
local_dir="${LOCAL_DIR_BASE}/cloud/${repo}"
sync_repo "${remote_url}" "${local_dir}" || true
done
[[ -f ${TMP_DIR}/failed ]] && {
echo "failed to download following packages:"
cat ${TMP_DIR}/failed
mv ${TMP_DIR}/failed ${LOCAL_DIR_BASE}/failed_packages.txt
}
[[ -z $EXIT_MSG ]] || echo $EXIT_MSG [[ -z $EXIT_MSG ]] || echo $EXIT_MSG
exit $EXIT_STATUS exit $EXIT_STATUS