From 6e6f8360dc403bb50a305522a4fc3d665dab8f38 Mon Sep 17 00:00:00 2001 From: Cédric Bonhomme Date: Sat, 29 Nov 2014 11:11:42 +0100 Subject: Do not check at all the content. --- pyaggr3g470r/duplicate.py | 8 +++----- 1 file changed, 3 insertions(+), 5 deletions(-) (limited to 'pyaggr3g470r') diff --git a/pyaggr3g470r/duplicate.py b/pyaggr3g470r/duplicate.py index 6f78fd81..f34e284e 100644 --- a/pyaggr3g470r/duplicate.py +++ b/pyaggr3g470r/duplicate.py @@ -1,12 +1,11 @@ #! /usr/bin/env python #-*- coding: utf-8 -*- +import itertools from datetime import timedelta -import itertools import utils - def compare_documents(feed): """ Compare a list of documents by pair. @@ -15,8 +14,7 @@ def compare_documents(feed): for pair in itertools.combinations(feed.articles, 2): date1 = pair[0].date date2 = pair[1].date - if pair[0].content != "" and \ - utils.clear_string(pair[0].title) == utils.clear_string(pair[1].title) and \ - (date1 - date2) < timedelta(days = 1): + if utils.clear_string(pair[0].title) == utils.clear_string(pair[1].title) and \ + (date1 - date2) < timedelta(days = 1): duplicates.append(pair) return duplicates \ No newline at end of file -- cgit