svn pre-commit hook兩三事

原文請猛戳：
http://galoisplusplus.coding.me/blog/2016/03/12/svn-precommit-hook/

提及折騰svn hook這件事仍是在去年年末。我廠向來木有Code Review等Quality Assurance流程，全賴老司機們各類強力輸出。而當時咱們項目的很多老司機要麼去了其餘部門要麼離職了，新來的小鮮肉碼農們彷佛對Version Control一無所知。別的不說，單是commit不寫comment，便讓你們頭疼不已，每次在merge時都得額外花很多時間來搞清楚commit的具體內容。雖然我自從投奔git陣營後基本都是用git svn，對svn生疏已久，但以爲加個svn提交限制應非難事，因此便額外花了點時間寫了個pre-commit hook，對commit的comment中的非空白字符作計數，少於必定字數的不讓提交。萬萬沒想到，當時讓SA大大部署上svn服務器後，就有人commit了一個「再試一下」orz...當時個人心裏是崩潰的...在推上圍脖上吐槽後不久，衆位大神各類支招，例如巨硬的泉哥說再搞個語義分析器666...不過最受用的仍是根爺所提到的用cpplint檢查代碼是否符合編碼規範、是否可以編譯經過。剛好那段時間小鮮肉們提交了一些帶有語法錯誤的lua代碼，偶們又木有review制度，結果不寫程序的策劃大大們svn up下來後——「什麼鬼？腫麼不能運行了？！好拙計啊！AAA，快來看bug啊！BBB，SOS！」——致使別人不得不停下手頭的活、額外花時間排查。因此我便在pre-commit hook裏用luac檢查語法錯誤的代碼，把這種坑隊友的事扼殺在搖籃裏。雖然不及cpplint那麼強大，但也基本夠用了。最近一次改這個腳本是由於前段時間有個小鮮肉作了一張4096x4096的圖集，而cocos2d-x文檔裏寫得清清楚楚，大多數手機所支持的最大紋理尺寸其實只有2048x2048...臥槽，幹得漂亮啊，一上線就形成了很多crash...木有review制度好可啪，在發現這個問題以前其餘人一直絕不知情...因此我仍是乾脆在svn hook裏再加個限制吧～

好了，不碎碎唸了，這個gist即是我折騰的svn hook，主要作了這麼幾項功能：

• commit message非空白長度檢查

• lua語法檢查

• 禁止添加文件名帶空格的文件

• 禁止被配在PROHIBITED_FILES的文件被修改

• 確保圖片尺寸小於2048x2048

#!/bin/sh

# PRE-COMMIT HOOK
#
# The pre-commit hook is invoked before a Subversion txn is
# committed.  Subversion runs this hook by invoking a program
# (script, executable, binary, etc.) named 'pre-commit' (for which
# this file is a template), with the following ordered arguments:
#
#   [1] REPOS-PATH   (the path to this repository)
#   [2] TXN-NAME     (the name of the txn about to be committed)
#
#   [STDIN] LOCK-TOKENS ** the lock tokens are passed via STDIN.
#
#   If STDIN contains the line "LOCK-TOKENS:\n" (the "\n" denotes a
#   single newline), the lines following it are the lock tokens for
#   this commit.  The end of the list is marked by a line containing
#   only a newline character.
#
#   Each lock token line consists of a URI-escaped path, followed
#   by the separator character '|', followed by the lock token string,
#   followed by a newline.
#
# The default working directory for the invocation is undefined, so
# the program should set one explicitly if it cares.
#
# If the hook program exits with success, the txn is committed; but
# if it exits with failure (non-zero), the txn is aborted, no commit
# takes place, and STDERR is returned to the client.   The hook
# program can use the 'svnlook' utility to help it examine the txn.
#
# On a Unix system, the normal procedure is to have 'pre-commit'
# invoke other programs to do the real work, though it may do the
# work itself too.
#
#   ***  NOTE: THE HOOK PROGRAM MUST NOT MODIFY THE TXN, EXCEPT  ***
#   ***  FOR REVISION PROPERTIES (like svn:log or svn:author).   ***
#
#   This is why we recommend using the read-only 'svnlook' utility.
#   In the future, Subversion may enforce the rule that pre-commit
#   hooks should not modify the versioned data in txns, or else come
#   up with a mechanism to make it safe to do so (by informing the
#   committing client of the changes).  However, right now neither
#   mechanism is implemented, so hook writers just have to be careful.
#
# Note that 'pre-commit' must be executable by the user(s) who will
# invoke it (typically the user httpd runs as), and that user must
# have filesystem-level permission to access the repository.
#
# On a Windows system, you should name the hook program
# 'pre-commit.bat' or 'pre-commit.exe',
# but the basic idea is the same.
#
# The hook program typically does not inherit the environment of
# its parent process.  For example, a common problem is for the
# PATH environment variable to not be set to its usual value, so
# that subprograms fail to launch unless invoked via absolute path.
# If you're having unexpected problems with a hook program, the
# culprit may be unusual (or missing) environment variables.
# 
# Here is an example hook script, for a Unix /bin/sh interpreter.
# For more examples and pre-written hooks, see those in
# the Subversion repository at
# http://svn.apache.org/repos/asf/subversion/trunk/tools/hook-scripts/ and
# http://svn.apache.org/repos/asf/subversion/trunk/contrib/hook-scripts/

LOG="/tmp/svn.log"
touch ${LOG}

REPOS="$1"
TXN="$2"
echo "REPOS: $REPOS" > ${LOG}
echo "TXN: $TXN" >> ${LOG}

SVNLOOK=""

# lua compiler
LUAC=""
# lua file extension
LUA_EXT="lua"
# png file extension
PNG_EXT="png"

MSG_MIN_CHAR_NUM=3

MAX_PNG_SIZE=2048

PROHIBITED_FILES=(
)

TMP_DIR="/tmp/svn"
if [[ -d ${TMP_DIR} ]]; then
    rm -r ${TMP_DIR}
fi
mkdir -p ${TMP_DIR}

function check_lua_syntax {
local lua_file=$1
echo `${LUAC} ${lua_file} 2>&1`
}

function create_file {
local file_name=$1
# Create tmp file and copy content
tmp_file="${TMP_DIR}/${file_name}"
mkdir -p "$(dirname "${tmp_file}")" && touch "${tmp_file}"
${SVNLOOK} cat -t "${TXN}" "${REPOS}" "${file_name}" > ${tmp_file}
}

# Make sure that the log message contains some text.
commit_msg=`$SVNLOOK log -t "$TXN" "$REPOS" | sed 's/[[:space:]]//g'`
echo ${commit_msg} >> ${LOG}
if [[ `echo ${commit_msg} | wc -c` -lt ${MSG_MIN_CHAR_NUM} ]]; then
    echo "Please write a meaningful comment when committing" 1>&2
    exit 1
fi

changed_info_str=`${SVNLOOK} changed -t "${TXN}" "${REPOS}"`
IFS=$'\n' read -rd '' -a changed_infos <<<"${changed_info_str}"

lua_error_msg=""
png_error_msg=""
for changed_info in "${changed_infos[@]}"; do
    # Prevent commiting file that contains space in its filename
    echo ${changed_info} >> ${LOG}
    operation=`echo ${changed_info} | awk '{print $1}'`
    if [[ ${operation} = "A" ]] && [[ `echo ${changed_info} | awk '{print NF}'` -gt 2 ]]; then
        echo "Please do not commit file that contains space in its filename!" 1>&2
        exit 1
    fi
    file_name=`echo ${changed_info} | awk '{print $2}'`
    echo "operation: ${operation}, file: ${file_name}, ext: ${ext}" >> ${LOG}

    # Check prohibit-commit files
    for prohibited_file in ${PROHIBITED_FILES[@]}; do
        if [[ ${file_name} = ${prohibited_file} ]]; then
            echo "${file_name} is not allowed to be changed!" 1>&2
            exit 1
        fi
    done

    ext=`echo ${file_name} | awk -F"." '{print $NF}'`

    if [[ ${operation} = "U" ]] || [[ ${operation} = "A" ]]; then
        tmp_file="${TMP_DIR}/${file_name}"

        # Check lua syntax
        if [[ ${ext} = ${LUA_EXT} ]]; then
            echo "Check syntax of ${tmp_file}" >> ${LOG}
            create_file ${file_name}
            error_msg=`check_lua_syntax ${tmp_file}`
            if [[ `echo ${error_msg} | sed 's/\n//g'` != "" ]]; then
                lua_error_msg="${lua_error_msg}\n${error_msg}"
            fi
        fi

        # Check file size
        if [[ ${ext} = ${PNG_EXT} ]]; then
            create_file ${file_name}
            png_info=`file ${tmp_file} | sed 's/,//g'`
            png_width=`echo ${png_info} | awk '{print $5}' | bc`
            png_height=`echo ${png_info} | awk '{print $7}' | bc`
            if [[ ${png_width} -gt ${MAX_PNG_SIZE} ]] || [[ ${png_height} -gt ${MAX_PNG_SIZE} ]]; then
                png_error_msg="${png_error_msg}\n${file_name} is too large: ${png_width} x ${png_height}"
            fi
        fi
    fi
done

rm -r ${TMP_DIR}

if [[ ${lua_error_msg} != "" ]] || [[ ${png_error_msg} != "" ]]; then
    if [[ ${lua_error_msg} != "" ]]; then
        echo "lua error: ${lua_error_msg}" >> ${LOG}
        echo "Please fix the error in your lua program:${lua_error_msg}" 1>&2
    fi

    if [[ ${png_error_msg} != "" ]]; then
        echo "png error: ${png_error_msg}" >> ${LOG}
        echo "Please do not commit pictures which are larger than 2048 x 2048:${png_error_msg}" 1>&2
    fi

    exit 1
fi

# Check that the author of this commit has the rights to perform
# the commit on the files and directories being modified.
# commit-access-control.pl "$REPOS" "$TXN" commit-access-control.cfg || exit 1

# All checks passed, so allow the commit.
exit 0

PS. 聰哥說做爲一位開發去搞svn hook怕被公司的其餘開發噴越權管了運維的事，我廠卻是不存在這種問題的——人少活多時間緊，啥活都得攬啊，例如我一位舍友兼同事的大神之前開發手遊但是客戶端、服務器、策劃、切圖全都一人搞定，就差本身出美術圖和特效了。最近我也在像運維同樣寫腳本自動化處理某些項目文件，發現咱們開發各類混亂啊有木有！例如路徑帶空格，個人一些awk腳本就血崩了——我在以上的hook腳本里也有一樣的問題，其實把列計數方式改一下，例如$7改爲${NF-4}，能夠避免這種問題——並且還要把帶空格的路徑用sed替換成轉義後的字符串。不過這種狀況還算好的，還有些人用的單詞有拼寫錯誤，要是全錯了倒也罷了，腳本hold得住，問題是有的對有的錯...偶只能棄療了...