Question

我有来自csv文件的数据行，存储的第一行是标题。例如：

first line -> [a,b,c,d,e]

second line -> [0,1,2,1,2]

third line -> [4,2,4,1,5]

另外，我有以下格式的数据相关条件字符串：

condition =（（（a = d）OR（a = c））AND（c

输出应仅为第3行。如何评估此条件，并分离所有嵌套的子条件？我正在考虑一个递归函数，通过括号读取，但我的代码中有一个混乱:(。感谢答案，对不起我的英语不好！

PS：我不想使用pandas或csv库 PS2：上述条件只是一个例子，可能有另外更多的嵌套条件，如（（（（（（a = d）AND（c＆gt; e））OR（b = c））AND（e

Answer 1

以下是一个快速解决方案：

from itertools import ifilter

with open('input.csv', 'r') as fi:
    lines = ((rawline, map(int, rawline.split(','))) for rawline in  fi.readlines()[1:])
    results = ifilter(lambda (_, fds): (fds[0] == fds[3] or fds[0] == fds[2]) and (fds[2] < fds[4]), lines)
    for (rawline, _) in results:
        print rawline

input.csv为：

a,b,c,d,e
0,1,2,1,2
4,2,4,1,5

结果输出为：

4,2,4,1,5

更新：更短/更紧凑的实施：

from itertools import ifilter

with open('input.csv', 'r') as fi:
    results = ifilter(
        lambda fds: (fds[0] == fds[3] or fds[0] == fds[2]) and (fds[2] < fds[4]),
        (map(int, rawline.split(',')) for rawline in fi.readlines()[1:]))
    for fields in results:
        print ','.join(map(str,fields))

Answer 2

最好在需求更新/更改时创建新答案

评估字符串格式条件的最快解决方案是使用内置函数eval。这样，您就不必进行繁重/无法负担的解析（lexical analysis和syntactic analysis）

以下是示例代码：

from itertools import ifilter

condition1 = '(((a = d) OR (a = c)) AND (c < e))'

def evalCondition(condition, *args):
    '''
    1) if you have condition format follow python grammar, then you don't need below replacement
    2) assume there is no '>=' or '<=', otherwise, you have to use more sophisticated replacement method e.g. using regular exppression
    '''
    condition = condition.replace('=', '==').replace('OR', 'or').replace('AND', 'and')

    a,b,c,d,e = args
    return eval(condition)

with open('input.csv', 'r') as fi:
    results = ifilter(
        lambda fields: evalCondition(condition1, *fields),
        (map(int, rawline.split(',')) for rawline in fi.readlines()[1:]))
    for fields in results:
        print ','.join(map(str,fields))

input.csv为：

a,b,c,d,e
0,1,2,1,2
4,2,4,1,5

结果输出为：

4,2,4,1,5

基于具有嵌套条件的字符串过滤行

2 个答案: