重新认识HTTP3xx重定向机制

问题引入

前段时间做数据收集时需要下载网宿cdn的日志进行分析。而网宿对日志下载的接口搞得很复杂，又没有提供相应的sdk，只是提供了一个shell脚本，虽然在ubuntu上使用很方便，但是脚本里面的各种重定向分析非常复杂。故此想对重定向在深入了解一点。

查询网宿日志列表的脚本

#!/bin/sh

TMP_FILE="/tmp/wslog_query_client.log"
#Usage
Usage() {
    echo "wslog_query_client.sh [query_url] [user] [passwd] [start_time] [end_time] [channels]"
    return 0
}
#check input parameters
if [ $# -eq 1 ]; then
    if [ "$1" = "-h" ]; then
        Usage
        exit 0
    else
        Usage
        exit -1
    fi
elif [ $# -ne 6 ]; then
    Usage
    exit -1
fi
#params set
url=$1
user=$2
passwd=`echo $3 | sed 's/&/%26/g' `
start_time=$4
end_time=$5
channels=$6
#access logQuery access API
curl -s -D $TMP_FILE $1
cat $TMP_FILE | grep "HTTP/" | grep "302" > /dev/null
if [ $? -ne 0 ]; then
    exit -2
fi
#redirect to verify url with user and passwd
TMP_URL=`cat $TMP_FILE | grep "Location: "|sed 's/\r//' | awk '{print $2}' | sed 's/http:/https:/'`
TMP_URL="${TMP_URL}?u=$user&p=$passwd&channel=$channels"
curl -s -k -D $TMP_FILE $TMP_URL
cat $TMP_FILE | grep "HTTP/" | grep "302" > /dev/null
if [ $? -ne 0 ]; then
    exit -3
fi
#redirect to query url with start_time, end_time and channels
TMP_URL=`cat $TMP_FILE | grep "Location: "|sed 's/\r//' | awk '{print $2}'`
TMP_URL="