最近有意地用Python写一些原来习惯于用Perl写的脚本。发现Perl还是有其自由之处，在处理数值与字符串时有很大的方便。我所要做的工作是简单根据广告类别对模拟实验结果进行平均。 这个任务用Perl编写需要36行代码写成，Python需要43行。即使两者都尽可能一个语句一行，仍然是Python需要的行数多（Perl包括其括号）。 Python的额外工作在于： 1）字典Key的初始化(Python不支持Autovivi) 2) Python分别字符串与数值，两者不能统一处理，经常需要相互转换。3）Python （3.0之前版本）在输出非空格分隔的数据时需要额外处理（Perl的$,与$\变量可方便做这件事）
总结： Perl在字符串与数值文本的处理中还是有其强大之处吧， Python未必就比Perl好！
Python 的问题1）可以通过get方法避免： d[i] = d.get(i,0) + record_num
附两个程序：
#!usr/bin/perl
use strict;
my %dict_all_num;
my %dict_click_num;
my %dict_uniq_click_num;
while (<>) {
chomp;
my ($record_num,$cat,$click_num,$uniq_click_num) = split /,/;
$dict_all_num{$cat} += $record_num;
$dict_click_num{$cat} += $click_num;
$dict_uniq_click_num{$cat} += $uniq_click_num;
}
local $,= ",";
local $\ = "\n";
for my $cat (keys %dict_all_num) {
$dict_all_num{$cat} /= 3;
$dict_click_num{$cat} /= 3;
$dict_uniq_click_num{$cat} /= 3;
print int($dict_all_num{$cat}),
$cat,
int($dict_click_num{$cat}),
int($dict_uniq_click_num{$cat}),
$dict_click_num{$cat} / $dict_all_num{$cat},
$dict_uniq_click_num{$cat} / $dict_all_num{$cat};
}
#!usr/bin/python import sys input_file = sys.argv[1] dict_all_num = {} dict_click_num = {} dict_uniq_click_num = {} for line in open(input_file,'r'): arr = line.rstrip().split(',') record_num = int(arr[0]) cat = arr[1] click_num = int(arr[2]) uniq_click_num = int(arr[3]) if not cat in dict_all_num: dict_all_num[cat] = 0 dict_click_num[cat] = 0 dict_uniq_click_num[cat] = 0 dict_all_num[cat] += record_num dict_click_num[cat] += click_num dict_uniq_click_num[cat] += uniq_click_num for cat in dict_all_num: dict_all_num[cat] /= 3 dict_click_num[cat] /= 3 dict_uniq_click_num[cat] /= 3 lst = [ str(dict_all_num[cat]), cat, str(dict_click_num[cat]), str(dict_uniq_click_num[cat]), str(float(dict_click_num[cat]) / dict_all_num[cat]), str(float(dict_uniq_click_num[cat]) / dict_all_num[cat]) ] output_line = ",".join(lst) print output_line
