零宽正则表达式实例.

正则表达式实例.
下面介绍grep 的零宽断言和sed 的单行多段提取。
零宽断言(zero-length-assertion)又叫环视(look around)，是正则表达式的精华所在.
零宽断言是匹配一个位置。这个位置由一个字符串表达，但这个字符串并不进入匹配的结果中。所以叫零宽。
^匹配行首，$匹配行尾，就是只匹配一个位置，零宽断言可以匹配任意一个位置，这个位置由匹配表达式决定。
这个位置由特定的匹配表达式表示(?=...)(?!...)(?<=...)(?<!...)
其中...是要匹配的表达式，不是三个点。 ?=,?!,?<=,?<! 表示它是零宽，不进入匹配结果。

1.(?=...)
正向前视(positive look ahead),
寻找一个位置点, 这个位置点的右侧与表达式...相匹配;
2.(?!...)
逆向前视(negative look ahead)
寻找一个位置点, 这个位置点的右侧与表达式...不相匹配;
3.(?<=...)
正向后视(positive look behind)
寻找一个位置点, 这个位置点的左侧与表达式...相匹配;
4.(?<!...)
逆向后视(negative look behind)
寻找一个位置点, 这个位置点的左侧与表达式...不相匹配;


举个例子就明白了:
cat train.txt
"train_no":"5l0000G10241","station_train_code":" G102","start_station_telecode":"AOH","start_station_name":" 上海虹桥 ","end_station_telecode":"VNP","end_station_name":" 北京南","from_station_telecode":"AOH","from_station_name":"上海虹桥","to_station_telecode":"VNP","to_station_name":"北京南","start_time":" 06:43","arrive_time":" 12:17","day_difference":"0","train_class_name":"","lishi":"05:34","canWebBuy":"Y","lishiValue":"334","yp_info":"O055300630M0933000649174800008","control_train_day":"20300303","start_train_date":"20150718","seat_feature":"O3M393","yp_ex":"O0M090","train_seat_feature":"3","seat_types":"OM9","location_code":"H1","from_station_no":"01","to_station_no":"09","control_day":59,"sale_time":"1330","is_support_card":"1","note":"","gg_num":"--","gr_num":"--","qt_num":"--","rw_num":"--","rz_num":"--","tz_num":"--","wz_num":"--","yb_num":"--","yw_num":"--","yz_num":"--","ze_num":"630","zy_num":"64","swz_num":"8"

a. 我们要提取列车车次 G102
它具有以下特征：
1. 左侧为:station_train_code":"
2. 右侧为:"
3. 匹配的部分是任意字符，但不能包含"

用grep 写的正则表达式如下: 用到了2个环视:
#cat train.txt | grep -oP '(?<=station_train_code":")[^"]+(?=")'
G102

b. 我们想提取G102,AOH,上海虹桥,北京南几个字符

#cat train.txt | grep -oP '(?<=station_train_code":")[^"]+(?=").*(?<=start_station_name":")[^"]+(?=")'
G102","start_station_telecode":"AOH","start_station_name":"上海虹桥
start_station_telecode, start_station_name 并不是我们想要的，但是由于它们被.* 所匹配，所以也被输出了。
有没有办法输出匹配中的一部分内容呢？ 有，这就是反向引用，用\1,\2, ...\9 代替前面()中的部分。

3. sed 登场，从一行中提取多段内容(最多9段)。把待提取的匹配的内容用括号括起来。然后引用
$ cat train.txt|sed -r 's/.*station_train_code":"([^"]+).*start_station_name":"([^"]+).*end_station_name":"([^"]+).*start_time":"([^"]+).*arrive_time":"([^"]+).*ze_num":"([^"]+).*zy_num":"([^"]+).*swz_num":"([^"]+).*/\1 \2 \3 \4 \5 \6 \7 \8/'
G102 上海虹桥  北京南 06:43 12:17 630 64 8

4. 若果一行中想提取超过9个字段，就只能编程序了。支持正则的最好，不支持就split and search 了。都不算复杂。