敏感关键字匹配在线产品查找

sensitive_keywords_matched.py

#!/usr/bin/env python3
import collections

# 敏感关键字匹配在线产品查找

title_width = 60

FIELD_ITEMID = 0
FIELD_SKU = 1
FIELD_TITLE = 2

Listing = collections.namedtuple('Listing', "ItemID SKU Title")

def parser_listing(line):
    if line and len(line) > 0:
        item_id = line[:12]
        sku = line[13:150].strip()
        title = line[150:].strip().lower()
        return Listing(item_id, sku, title)
    else:
        return None

def matched(keyword, title):
    if not lineOfKeyword or len(lineOfKeyword) == 0 or keyword not in title:
        return False

    for ext_keyword in [keyword+' ', keyword+'-', keyword+', ', keyword+'.', keyword+'/', keyword+';', \
                        ' '+keyword, '-'+keyword, '.'+keyword, '/'+keyword, ';'+keyword]:
        if ext_keyword in title:
            return True

    return False

FILENAME_KEYWORDS = 'C:/Users/zhuqling/Desktop/senstive_keywords.rpt'
FILENAME_LISTING = 'C:/Users/zhuqling/Desktop/listing.rpt'

keywords = []

# 添加关键字
for lineOfKeyword in open(FILENAME_KEYWORDS, encoding='utf8'):
    if lineOfKeyword and len(lineOfKeyword) > 0:
        keywords.append(lineOfKeyword.strip().lower())

print("关键字数目：{0}".format(len(keywords)))

# 循环所有Listing，判断是否关键字匹配
lists = []
for lineOfListing in open(FILENAME_LISTING, encoding='utf8'):
    lists.append(parser_listing(lineOfListing))

print("在线产品数目：{0}".format(len(lists)))

print("匹配结果：")

match_count = 0
for list in lists:
    title = list[FIELD_TITLE]
    #print(title)
    for lineOfKeyword in keywords:
        #print(lineOfKeyword)
        if matched(lineOfKeyword, title):
            print("{0:<20} {1:>14} {2:<24} {3:<{tw}}".format(lineOfKeyword, list[FIELD_ITEMID], list[FIELD_SKU],
                                                           list[FIELD_TITLE] if len(list[FIELD_TITLE]) <=title_width else list[FIELD_TITLE][:title_width-3] + "..." ,
                                                           tw = title_width))
            match_count += 1

print("匹配数目：{0}".format(match_count))