工作随笔——pre-commit钩子限制日志长度和提交的文件类型

2014-09-18：解决Subversion edge 的hook中文乱码问题

近期检查SVN时发现备份好的文件体积异常庞大。才跑2个月备份出来的大小就有4G多。仔细查询发现很多很多IDE自动生成的二进制文件。。。我类个去。好吧，开发不自觉就只有强制限制了。

google的时候发现这篇文章：http://blog.csdn.net/clever101/article/details/8560833

自己动手丰衣足食。脚本如下：

#!/bin/bash
export LANG="zh_CN.UTF-8"  #确保中文日志显示正常，便于统计日志
REPOS="$1"
TXN="$2"
#限制日志长度
LENGTH=10

SVNLOOK="/usr/local/csvn/bin/svnlook"
BLACKLIST="target  build  classes  bin/target  .*  *.iml  *.ipr  *.iws  *.class "

function error_exit(){
echo -e "1.Not allowed empty log " >&2
echo -e "2.Logs must be greater than ${LENGTH} characters or chinese" >&2
exit 1
}

function blacklist_exit(){
if [ ! -z "${INBLACKLIST}" ] ;then
echo -e "some files in the blacklist , please cheack again !" >&2
echo -e "blacklist:\n ${BLACKLIST}" >&2
exit 1
fi
}

# 获取字符数量(包括换行符)
TEMP_LENGTH=`${SVNLOOK} log -t "${TXN}" "${REPOS}"|sed 's/\s*$//g'|sed 's/^\s*//g'| wc --chars`
# 统计换行符
TEMP_LINE=`${SVNLOOK} log -t "${TXN}" "${REPOS}"| wc --lines`
# 真实字符数量
LOGMSG_LENGTH=`expr ${TEMP_LENGTH} - ${TEMP_LINE}`
# 文件黑名单过滤
# 过滤文件夹
INBLACKLIST="`${SVNLOOK} changed -t "${TXN}" "${REPOS}" | grep "/target/"`"  #文件夹前后必须有/，否则匹配会有误伤
blacklist_exit
INBLACKLIST="`${SVNLOOK} changed -t "${TXN}" "${REPOS}" | grep "/build/"`"
blacklist_exit
INBLACKLIST="`${SVNLOOK} changed -t "${TXN}" "${REPOS}" | grep "/classes/"`"
blacklist_exit
INBLACKLIST="`${SVNLOOK} changed -t "${TXN}" "${REPOS}" | grep "/bin/target/"`"
blacklist_exit
# 过滤所有已点开头的文件和文件夹
INBLACKLIST="`${SVNLOOK} changed -t "${TXN}" "${REPOS}" | grep "/\."`"
blacklist_exit
# 过滤文件尾缀
INBLACKLIST="`${SVNLOOK} changed -t "${TXN}" "${REPOS}" | grep "\.iml$"`"
blacklist_exit
INBLACKLIST="`${SVNLOOK} changed -t "${TXN}" "${REPOS}" | grep "\.ipr$"`"
blacklist_exit
INBLACKLIST="`${SVNLOOK} changed -t "${TXN}" "${REPOS}" | grep "\.iws$"`"
blacklist_exit
INBLACKLIST="`${SVNLOOK} changed -t "${TXN}" "${REPOS}" | grep "\.class$"`"
blacklist_exit

# 判断日志长度是否满足要求
if [ "${LOGMSG_LENGTH}" -lt ${LENGTH} ];then
error_exit
fi

exit 0

PS:Subversion edge需要在高级管理中开启hooks编码为UTF-8，否则。会导致客户端无法解析返回的日志直接报错。默认为ASCII。

UTF-8 Encoding Assume UTF-8 encoded messages from hooks; default is ASCII.