
平台论坛博客文库

› 论坛 › 程序设计 › Shell › 【分享】刚写的一个搜索/备份文件的脚本

【分享】刚写的一个搜索/备份文件的脚本 [复制链接]

silon212

白手起家

论坛徽章:: 0

发表于 2010-05-06 12:07 |显示全部楼层

刚写的一个脚本，顺便练练shell编程。

起源：

用 grep 在目录中查找含某个模式的文件，然后拷贝这些匹配的文件到另一个目录下，并保持目录结构（因为做递归查找时，可能子目录中和其它目录中有重名的匹配文件）。

脚本 grepcp.sh 代码：

#!/bin/sh
#
# Description:
# search file by grep specified "pattern" in "source-path", then copy matched
# files to directory "target-dir" by name of "target-name".
#
# Author: zy <silon212@gmail.com>
# Version: 1.0
#
# Return:
# 0: success of copy
# 1: any kind of failure
# 2: not failure but not copy
#
APPNAME=`basename $0`
# ++++++++++ usage () ++++++++++
usage ()
{
echo "Usage:"
echo " $APPNAME [-E -i -R] [-p pattern] [-s source-path] [-t target-dir] [target-name]"
echo " $APPNAME [-h]"
echo -e "\nDescription:"
echo " search file by grep specified \"pattern\" in \"source-path\", then copy matched files to directory \"target-dir\" by name of \"target-name\"."
echo -e "\nOptions:"
echo " [-E -i -R]"
echo " give options to grep, see grep help."
echo -e "\n [-p pattern]"
echo " search pattern to grep."
echo -e "\n [-s source-path]"
echo " specify source path for grep search, may be an existed directory or regular file."
echo " if not specified use current working directory as default."
echo -e "\n [-t target-dir]"
echo " specify target directory to contain \"target-name\", MUST be an existed directory."
echo " if not specified use current working directory as default."
echo -e "\n [target-name]"
echo " the copy destination name, may be an directory or regular file determined by \"source-path\"."
echo " the full destination path is \"target-name\" with prefix \"target-dir\"."
echo " if not specified use the basename of \"source-path\"."
echo -e "\nExample:"
echo " $APPNAME -E -s \\home\\user1\\tc1.txt -p \"[0-9]+\""
echo " grep single file \"\\home\\user1\\tc1.txt\", if it contains pattern \"[0-9]+\" copy it to current directory."
echo ""
echo " $APPNAME -E -s \\home\\user1\\tc1.txt -p \"[0-9]+\" bak.txt"
echo " similar as former, but rename file \"tc1.txt\" to \"bak.txt\"."
echo ""
echo " $APPNAME -E -i -R -t .. -p \"[a-z]([a-z0-9]*[-_]?[a-z0-9]+)*@([a-z0-9]*[-_]?[a-z0-9]+)+[\\.][a-z]{2,3}([\\.][a-z]{2})?\" bak"
echo " search current directory recursively by grep pattern \"[a-z]([a-z0-9]*[-_]?[a-z0-9]+)*@([a-z0-9]*[-_]?[a-z0-9]+)+[\\.][a-z]{2,3}([\\.][a-z]{2})?\" (take for E-mail), then copy matched files to \"../bak\", and keep their directory hierarchy."
echo -e "\nAuthor:\n zy <silon212@gmail.com>"
}
# ++++++++++ yes_or_no () ++++++++++
yes_or_no ()
{
VERBOSE=0
while [ $# -gt 0 ]
do
case $1 in
-v)
VERBOSE=1
shift
;;
*)
_PROMPT=$1
shift
;;
esac
done
INVALID_INPUT=1
while [ $INVALID_INPUT -ne 0 ]
do
echo -n "$_PROMPT (y/n)? "
read ANS
INVALID_INPUT=0
ANS=`echo -n "$ANS" | tr '[A-Z]' '[a-z]'`
case "$ANS" in
"y" | "yes")
return 0
;;
"n" | "no")
return 1
;;
*)
if [ $VERBOSE -eq 1 ]; then
echo "invalid answer: $ANS"
fi
INVALID_INPUT=1
;;
esac
done
}
# ++++++++++ main procedure ++++++++++
# 所有选项
ALLOPT="$*"
# echo "ALLOPT is: [$ALLOPT]"
# 设置默认源路径为当前路径
SRC_PATH=`pwd`
# 源路径是否是目录，1 表示目录
SRC_IS_DIR=1
# 设置默认目标目录为当前路径
TARGET_DIR=`pwd`
# 目标名
TARGET_NAME=""
# 实际的目标路径等于目标目录 + 目标名
TARGET_PATH=""
# 传给 grep 的选项
GREP_OPT=""
# 传给 grep 的模式
PATTERN=""
while getopts :EiRp:s:t:h OPT
do
case $OPT in
E)
# echo "-E set"
GREP_OPT="$GREP_OPT -E"
;;
i)
# echo "-i set"
GREP_OPT="$GREP_OPT -i"
;;
R)
# echo "-R set"
GREP_OPT="$GREP_OPT -R"
;;
p)
# echo "-p set"
PATTERN=$OPTARG
;;
s)
# echo "-s set"
SRC_PATH=$OPTARG
;;
t)
# echo "-t set"
TARGET_DIR=$OPTARG
;;
h)
# echo "-h set"
usage;
exit 2;
;;
*)
echo -e "$APPNAME: unknown/invalid option: $ALLOPT\n" >&2
usage;
exit 1;
;;
esac
done
# 让剩余的非选项参数成为拷贝到的目标名
# 目标名必须在所有选项之后
shift `expr $OPTIND - 1`
TARGET_NAME="$@"
# echo "TARGET_NAME: [$TARGET_NAME]"
# 目标名不能包含斜线符
if echo "$TARGET_NAME" | grep "/" >/dev/null 2>&1; then
echo "$APPNAME: target name MUST NOT contain slash (/), exit now."
exit 1
fi
# echo "pattern: [$PATTERN]"
# echo "old SRC_PATH: [$SRC_PATH]"
# echo "old TARGET_DIR: [$TARGET_DIR]"
# 检查源路径
if TEMP=`realpath "$SRC_PATH" 2>/dev/null`; then
SRC_PATH=$TEMP
# echo "full SRC_PATH: [$SRC_PATH]"
if [ -d "$SRC_PATH" ]; then
SRC_IS_DIR=1
else
SRC_IS_DIR=0
fi
else
echo "$APPNAME: failed to get source full path: [$SRC_PATH] (not exist)." >&2
exit 1;
fi
# 检查目标路径
if TEMP=`realpath "$TARGET_DIR" 2>/dev/null`; then
TARGET_DIR=$TEMP
# echo "full TARGET_DIR: [$TARGET_DIR]"
if [ ! -d "$TARGET_DIR" ]; then
echo "$APPNAME: target directory path: [$TARGET_DIR] MUST be a directory." >&2
exit 1;
fi
else
echo "$APPNAME: failed to get target full directory path: [$TARGET_DIR] (not exist)." >&2
exit 1;
fi
# echo "SRC_PATH: [$SRC_PATH]"
# echo "TARGET_DIR: [$TARGET_DIR]"
# 源路径不能是目标路径的前缀
# 这种情况表示：一个目录向其子目录（或其自身）中进行拷贝，e.g.
# /aaa/bbb 目录中内容想拷贝到 /aaa/bbb/ccc 目录中
# 其实只有做递归拷贝时，才会引发问题，而我这里全部都禁止了
if echo "$TARGET_DIR" | egrep "^$SRC_PATH" >/dev/null 2>&1; then
echo "$APPNAME: source path [$SRC_PATH] MUST NOT be prefix of target directory [$TARGET_DIR], it will bring weird operation when copy directory contents into it's sub-directory recursively." >&2
exit 1
fi
# echo "GREP_OPT: [$GREP_OPT]"
# echo "PATTERN: [$PATTERN]"
# 如果没有指定目标名，则默认用源路径中的名字
if [ "$TARGET_NAME" = "" ]; then
TARGET_NAME=`basename "$SRC_PATH"`
fi
# 产生拷贝目标全路径
if [ "$TARGET_DIR" = "/" ]; then
TARGET_PATH="/$TARGET_NAME"
else
TARGET_PATH="$TARGET_DIR/$TARGET_NAME"
fi
# echo "TARGET_PATH: [$TARGET_PATH]"
# 目标路径是否已存在
if [ -e "$TARGET_PATH" ]; then
# 禁止向自己进行拷贝
SRC_REAL=`realpath "$SRC_PATH"`
TARGET_REAL=`realpath "$TARGET_PATH"`
if [ "$SRC_REAL" = "$TARGET_REAL" ]; then
echo "$APPNAME: the target and source MUST NOT be the same path [$TARGET_REAL]."
exit 1
fi
# 提示用户是否删除已有路径
if yes_or_no -v "target path [$TARGET_PATH] has existed, do you want to remove it"; then
# echo "u enter yes"
if ! rm -f -R "$TARGET_PATH" >/dev/null 2>&1; then
echo "$APPNAME: failed to remove existed target path [$TARGET_PATH]." >&2
exit 1
fi
else
echo "$APPNAME: you make choice on not removing existed target path [$TARGET_PATH], exit now." >&2
exit 1
fi
fi
# ++++++++++ 单个文件拷贝 ++++++++++
if [ $SRC_IS_DIR -eq 0 ]; then
# 文件含匹配模式
if grep $GREP_OPT "$PATTERN" "$SRC_PATH" >/dev/null 2>&1; then
# echo "it match pattern"
if cp "$SRC_PATH" "$TARGET_PATH"; then
echo -e "\nfinish copy from [$SRC_PATH] to [$TARGET_PATH]."
exit 0
else
echo -e "\nfailed to copy from [$SRC_PATH] to [$TARGET_PATH]."
exit 1
fi
# 文件不含匹配模式
else
echo "source file [$SRC_PATH] not match pattern \"$PATTERN\"."
exit 2
fi
fi
# ++++++++++ 目录拷贝 ++++++++++
MATCH_FILES=`grep $GREP_OPT -l "$PATTERN" "$SRC_PATH"`
# 源目录中没有一个文件匹配，也可能是没开启 grep 的 -R 选项
if [ "$MATCH_FILES" = "" ]; then
echo "source directory [$SRC_PATH] contains no file matching pattern \"$PATTERN\" in grep option:$GREP_OPT."
exit 2
fi
# 创建目标路径失败
if ! mkdir "$TARGET_PATH"; then
echo "$APPNAME: failed to make target path [$TARGET_PATH]." >&2
exit 1
fi
# 进入目标路径（目录）
OLD_WD=`pwd`
cd "$TARGET_PATH"
# i=0
# 开始循环创建目录，并拷贝匹配的文件
# 这里用 for 循环不方便，因为路径中可能含有空白字符（空格）
# for 循环将所有的空白字符（空格 tab LF CR）作为分隔符
echo "$MATCH_FILES" | while read MATCH_FILE; do
# echo "$i, full: $MATCH_FILE"
REL_PATH=`echo "$MATCH_FILE" | awk 'gsub(AWK_SRC_PATH,"") {print $0}' AWK_SRC_PATH="$SRC_PATH/"`
# echo "$i, relative path: [$REL_PATH]"
REL_DIR=`dirname "$REL_PATH"`
# echo "$i, relative dir: [$REL_DIR]"
# echo "{{{"
DIR_PARTS=`echo "$REL_DIR" | awk '{gsub("/","\n"); print $0}'`
echo "$DIR_PARTS" | while read DIR_PART; do
# echo "DIR_PART: $DIR_PART"
if [ ! -e "$DIR_PART" ]; then
mkdir "$DIR_PART"
fi
cd "$DIR_PART"
done
cd "$TARGET_PATH"
cp "$MATCH_FILE" "$REL_PATH"
# echo "}}}"
# i=`expr $i + 1`
done
# 回到原来的工作目录
cd "$OLD_WD"
echo -e "\nfinish copy from [$SRC_PATH] to [$TARGET_PATH]."