def compare_revisions(self, old_id, new_id):
if old_id is None or new_id is None:
return ''
from htmldiff import render_html_diff
old_rev = Version.objects.get(id=old_id).content_html
new_rev = Version.objects.get(id=new_id).content_html
return render_html_diff(old_rev, new_rev)
def distance():
data = request.args.get('data', '')
title1 = request.args.get('title1', '')
title2 = request.args.get('title2', '')
data = eval(data)
print data
res = {'title1': title1, 'title2': title2}
nums = data['ids'].values()
for num in nums:
res[u'原差距'] = mongo_spider.sim.distance(data['num'], num)
title_num1 = Simhash(title1).value - sys.maxint
title_num2 = Simhash(title2).value - sys.maxint
res[u'标题差距'] = mongo_spider.sim.distance(title_num1, title_num2)
article = article_mongo.find_one({'_id': data['_id']}, {
'content': 1,
'title': 1
})
content_num1 = Simhash(article['content']).value - sys.maxint
text_num = Simhash(article['title'] + default +
html2text(article['content'])).value - sys.maxint
res['content1'] = article['content']
cursors = article_mongo.find({"_id": {
"$in": data['ids'].keys()
}}, {
'content': 1,
'title': 1
})
for cursor in cursors:
content_num2 = Simhash(cursor['content']).value - sys.maxint
text_num2 = Simhash(cursor['title'] + default +
html2text(cursor['content'])).value - sys.maxint
res[u'正文差距'] = mongo_spider.sim.distance(content_num1, content_num2)
res[u'新差距'] = mongo_spider.sim.distance(text_num2, text_num)
res['content2'] = cursor['content']
d = render_html_diff(res['content1'], res['content2'])
# result = list(d.compare(res['content1'], res['content2']))
# res['res'] = ''.join(result.spilt('+'))
# from pprint import pprint as _pprint
# _pprint(result)
res['d'] = d
return render_template('test/detail.html', data=res)
def _generate_diff_html(self):
if os.path.isfile(self.html_path):
return
tmpl_path = os.path.join(os.path.dirname(__file__), "diff.html")
logging.debug("creating html diff: %s", self.html_path)
diff = htmldiff.render_html_diff(self.old.html, self.new.html)
if '<ins>' not in diff and '<del>' not in diff:
return False
tmpl = jinja2.Template(codecs.open(tmpl_path, "r", "utf8").read())
html = tmpl.render(title=self.new.title,
url=self.old.entry.url,
old_url=self.old.archive_url,
old_time=self.old.created,
new_url=self.new.archive_url,
new_time=self.new.created,
diff=diff)
codecs.open(self.html_path, "w", 'utf8').write(html)
return True
def generate_diff_html(self, path):
if os.path.isfile(path):
logging.error("Diff file already exists: %s", path)
return None
tmpl_path = os.path.join(os.path.dirname(__file__),
"diff_template.html")
if not os.path.isfile(tmpl_path):
logging.error("Failed to find diff template: %s", tmpl_path)
return None
logging.debug("creating html diff: %s", path)
diff = htmldiff.render_html_diff(self.old.html, self.new.html)
if not self.validate_diff(diff):
return None
tmpl = jinja2.Template(codecs.open(tmpl_path, "r", "utf8").read())
html = tmpl.render(title=self.new.title,
url=self.old.entry.url,
old_time=self.old.created,
new_time=self.new.created,
diff=diff)
return html
def view_revisions(request, cat, slug):
news = get_object_or_404(News, slug=slug)
revisions = Version.objects.select_related().filter(news=news) \
.order_by('-created_at')
diff = ''
nb_revisions = revisions.count()
old_id = request.GET.get('old', None)
new_id = request.GET.get('new', None)
if old_id is not None and new_id is not None:
from htmldiff import render_html_diff
old_rev = Version.objects.get(id=old_id).content_html
new_rev = Version.objects.get(id=new_id).content_html
diff = render_html_diff(old_rev, new_rev)
return render(request, 'view_post_revisions.html', {
'post': news, 'post_revisions': revisions,
'nb_revisions': nb_revisions, 'diff': diff
})
def distance():
data = request.args.get('data', '')
title1 = request.args.get('title1', '')
title2 = request.args.get('title2', '')
data = eval(data)
print data
res = {'title1':title1, 'title2':title2}
nums = data['ids'].values()
for num in nums:
res[u'原差距'] = mongo_spider.sim.distance(data['num'], num)
title_num1 = Simhash(title1).value - sys.maxint
title_num2 = Simhash(title2).value - sys.maxint
res[u'标题差距'] = mongo_spider.sim.distance(title_num1, title_num2)
article = article_mongo.find_one({'_id':data['_id']}, {'content':1, 'title':1})
content_num1 = Simhash(article['content']).value - sys.maxint
text_num = Simhash(article['title'] + default + html2text(article['content']) ).value - sys.maxint
res['content1'] = article['content']
cursors = article_mongo.find({"_id":{"$in" : data['ids'].keys()}}, {'content':1, 'title':1})
for cursor in cursors:
content_num2 = Simhash(cursor['content']).value - sys.maxint
text_num2 = Simhash(cursor['title'] + default + html2text(cursor['content'])).value - sys.maxint
res[u'正文差距'] = mongo_spider.sim.distance(content_num1, content_num2)
res[u'新差距'] = mongo_spider.sim.distance(text_num2, text_num )
res['content2'] = cursor['content']
d = render_html_diff(res['content1'], res['content2'])
# result = list(d.compare(res['content1'], res['content2']))
# res['res'] = ''.join(result.spilt('+'))
# from pprint import pprint as _pprint
# _pprint(result)
res['d'] = d
return render_template('test/detail.html', data=res)
