From 0a875d4d1a5b5daf53ff11fe2e4c0c1fed5d335a Mon Sep 17 00:00:00 2001
From: bigeagle <justin.w.xd@gmail.com>
Date: Mon, 5 Dec 2016 12:41:45 +0800
Subject: [PATCH] rewritten anaconda installer mirror script

---
 anaconda.sh                  | 66 ++++++++++++++++++++++++++----------
 helpers/anaconda-filelist.py | 37 ++++++++++++++++++++
 helpers/tf-xml-filelist.py   |  2 +-
 3 files changed, 86 insertions(+), 19 deletions(-)
 create mode 100755 helpers/anaconda-filelist.py

diff --git a/anaconda.sh b/anaconda.sh
index bc53fad..8549881 100755
--- a/anaconda.sh
+++ b/anaconda.sh
@@ -1,11 +1,14 @@
 #!/bin/bash
-# requires: wget, lftp, jq
-#
+# requires: wget, lftp, jq, python3.5, lxml, pyquery
 
 set -e
+set -u
 set -o pipefail
 
-CONDA_REPO_BASE=${CONDA_REPO_BASE:-"http://repo.continuum.io"}
+_here=`dirname $(realpath $0)`
+HTMLPARSE="${_here}/helpers/anaconda-filelist.py"
+
+CONDA_REPO_BASE="${CONDA_REPO_BASE:-"https://repo.continuum.io"}"
 LOCAL_DIR_BASE="${TUNASYNC_WORKING_DIR}/pkgs"
 TMP_DIR=$(mktemp -d)
 
@@ -37,6 +40,44 @@ trap cleanup EXIT
 
 echo ${TMP_DIR}
 
+
+function sync_installer() {
+	repo_url="$1"
+	repo_dir="$2"
+
+	[[ ! -d "$repo_dir" ]] && mkdir -p "$repo_dir"
+	cd $repo_dir
+	# lftp "${repo_url}/" -e "mirror --verbose -P 5; bye"
+	
+	while read -a tokens; do
+		fname=${tokens[0]}
+		pkgmd5=${tokens[2]}
+
+		dest_file="${repo_dir}${fname}"
+		pkg_url="${repo_url}${fname}"
+		pkgsize=`curl --head -s ${pkg_url} | grep 'Content-Length' | awk '{print $2}' | tr -d '\r'`
+		
+		declare downloaded=false
+		if [[ -f ${dest_file} ]]; then
+			rsize=`stat -c "%s" ${dest_file}`
+			if (( ${rsize} == ${pkgsize} )); then
+				downloaded=true
+				echo "Skipping ${fname}, size ${pkgsize}"
+			fi
+		fi
+		while [[ $downloaded != true ]]; do
+			echo "downloading ${pkg_url}"
+			wget -q -O ${dest_file} ${pkg_url} && {
+				# two space for md5sum check format
+				{ md5sum -c - < <(echo "${pkgmd5} ${dest_file}"); } && downloaded=true
+			}
+		done
+	done < <(wget -O- ${repo_url} | $HTMLPARSE)
+}
+
+sync_installer "${CONDA_REPO_BASE}/archive/" "${TUNASYNC_WORKING_DIR}/archive/"
+sync_installer "${CONDA_REPO_BASE}/miniconda/" "${TUNASYNC_WORKING_DIR}/miniconda/"
+
 for repo in ${CONDA_REPOS[@]}; do
 	for arch in ${CONDA_ARCHES[@]}; do
 		PKG_REPO_BASE="${CONDA_REPO_BASE}/pkgs/$repo/$arch"
@@ -62,18 +103,18 @@ for repo in ${CONDA_REPOS[@]}; do
 			dest_file="${LOCAL_DIR}/${pkgfile}"
 			
 			declare downloaded=false
-			if [ -f ${dest_file} ]; then
+			if [[ -f ${dest_file} ]]; then
 				rsize=`stat -c "%s" ${dest_file}`
-				if [ ${rsize} -eq ${pkgsize} ]; then
+				if (( ${rsize} == ${pkgsize} )); then
 					downloaded=true
 					echo "Skipping ${pkgfile}, size ${pkgsize}"
 				fi
 			fi
-			while [ $downloaded != true ]; do
+			while [[ $downloaded != true ]]; do
 				echo "downloading ${pkg_url}"
 				wget -q -O ${dest_file} ${pkg_url} && {
 					# two space for md5sum check format
-					echo "${pkgmd5}  ${dest_file}" | md5sum -c - && downloaded=true 
+					{ md5sum -c - < <(echo "${pkgmd5} ${dest_file}"); } && downloaded=true
 				}
 			done
 		done
@@ -83,14 +124,3 @@ for repo in ${CONDA_REPOS[@]}; do
 	done
 done
 
-function sync_installer() {
-	repo_url="$1"
-	repo_dir="$2"
-
-	[ ! -d "$repo_dir" ] && mkdir -p "$repo_dir"
-	cd $repo_dir
-	lftp "${repo_url}/" -e "mirror --verbose -P 5; bye"
-}
-
-sync_installer "${CONDA_REPO_BASE}/archive/" "${TUNASYNC_WORKING_DIR}/archive/"
-sync_installer "${CONDA_REPO_BASE}/miniconda/" "${TUNASYNC_WORKING_DIR}/miniconda/"
diff --git a/helpers/anaconda-filelist.py b/helpers/anaconda-filelist.py
new file mode 100755
index 0000000..bdcf06f
--- /dev/null
+++ b/helpers/anaconda-filelist.py
@@ -0,0 +1,37 @@
+#!/usr/bin/env python3
+from datetime import datetime
+from pyquery import PyQuery as pq
+
+
+def get_filelist(htmlstring):
+    d = pq(htmlstring)
+    for tr in d('table').find('tr'):
+        tds = pq(tr).find('td')
+        if len(tds) != 4:
+            continue
+        fname = tds[0].find('a').text
+        mdate = tds[2].text
+        md5 = tds[3].text
+        ts = datetime.strptime(mdate, "%Y-%m-%d %H:%M:%S").strftime("%s")
+        yield (fname, ts, md5)
+
+
+if __name__ == "__main__":
+    import argparse
+    import fileinput
+
+    parser = argparse.ArgumentParser()
+    parser.add_argument("htmlfile", nargs='?', default="-")
+    args = parser.parse_args()
+
+    if args.htmlfile == "-":
+        htmlstring = '\n'.join([line for line in fileinput.input()])
+    else:
+        with open(args.htmlfile) as f:
+            htmlstring = f.read()
+
+    for file_record in get_filelist(htmlstring):
+        print("\t".join(file_record))
+
+
+# vim: ts=4 sw=4 sts=4 expandtab
diff --git a/helpers/tf-xml-filelist.py b/helpers/tf-xml-filelist.py
index d534900..68557d0 100755
--- a/helpers/tf-xml-filelist.py
+++ b/helpers/tf-xml-filelist.py
@@ -1,7 +1,7 @@
 #!/usr/bin/env python3
-import argparse
 import xml.etree.ElementTree as ET
 
+
 def get_filelist(xmlstring):
     r = ET.fromstring(xmlstring)
     ns = {