Merge pull request #1022 from misnyo/nyaa

nyaa.si fixed
This commit is contained in:
Adam Tauber 2017-09-07 19:33:07 +02:00 committed by GitHub
commit f4df27fa59
4 changed files with 142 additions and 85 deletions

View file

@ -1,7 +1,7 @@
"""
Nyaa.se (Anime Bittorrent tracker)
Nyaa.si (Anime Bittorrent tracker)
@website http://www.nyaa.se/
@website http://www.nyaa.si/
@provide-api no
@using-api no
@results HTML
@ -12,50 +12,25 @@
from lxml import html
from searx.engines.xpath import extract_text
from searx.url_utils import urlencode
from searx.utils import get_torrent_size, int_or_zero
# engine dependent config
categories = ['files', 'images', 'videos', 'music']
paging = True
# search-url
base_url = 'http://www.nyaa.se/'
base_url = 'http://www.nyaa.si/'
search_url = base_url + '?page=search&{query}&offset={offset}'
# xpath queries
xpath_results = '//table[@class="tlist"]//tr[contains(@class, "tlistrow")]'
xpath_category = './/td[@class="tlisticon"]/a'
xpath_title = './/td[@class="tlistname"]/a'
xpath_torrent_file = './/td[@class="tlistdownload"]/a'
xpath_filesize = './/td[@class="tlistsize"]/text()'
xpath_seeds = './/td[@class="tlistsn"]/text()'
xpath_leeches = './/td[@class="tlistln"]/text()'
xpath_downloads = './/td[@class="tlistdn"]/text()'
# convert a variable to integer or return 0 if it's not a number
def int_or_zero(num):
if isinstance(num, list):
if len(num) < 1:
return 0
num = num[0]
if num.isdigit():
return int(num)
return 0
# get multiplier to convert torrent size to bytes
def get_filesize_mul(suffix):
return {
'KB': 1024,
'MB': 1024 ** 2,
'GB': 1024 ** 3,
'TB': 1024 ** 4,
'KIB': 1024,
'MIB': 1024 ** 2,
'GIB': 1024 ** 3,
'TIB': 1024 ** 4
}[str(suffix).upper()]
xpath_results = '//table[contains(@class, "torrent-list")]//tr[not(th)]'
xpath_category = './/td[1]/a[1]'
xpath_title = './/td[2]/a[last()]'
xpath_torrent_links = './/td[3]/a'
xpath_filesize = './/td[4]/text()'
xpath_seeds = './/td[6]/text()'
xpath_leeches = './/td[7]/text()'
xpath_downloads = './/td[8]/text()'
# do search-request
@ -72,25 +47,32 @@ def response(resp):
dom = html.fromstring(resp.text)
for result in dom.xpath(xpath_results):
# defaults
filesize = 0
magnet_link = ""
torrent_link = ""
# category in which our torrent belongs
category = result.xpath(xpath_category)[0].attrib.get('title')
try:
category = result.xpath(xpath_category)[0].attrib.get('title')
except:
pass
# torrent title
page_a = result.xpath(xpath_title)[0]
title = extract_text(page_a)
# link to the page
href = page_a.attrib.get('href')
href = base_url + page_a.attrib.get('href')
# link to the torrent file
torrent_link = result.xpath(xpath_torrent_file)[0].attrib.get('href')
# torrent size
try:
file_size, suffix = result.xpath(xpath_filesize)[0].split(' ')
file_size = int(float(file_size) * get_filesize_mul(suffix))
except:
file_size = None
for link in result.xpath(xpath_torrent_links):
url = link.attrib.get('href')
if 'magnet' in url:
# link to the magnet
magnet_link = url
else:
# link to the torrent file
torrent_link = url
# seed count
seed = int_or_zero(result.xpath(xpath_seeds))
@ -101,6 +83,14 @@ def response(resp):
# torrent downloads count
downloads = int_or_zero(result.xpath(xpath_downloads))
# let's try to calculate the torrent size
try:
filesize_info = result.xpath(xpath_filesize)[0]
filesize, filesize_multiplier = filesize_info.split()
filesize = get_torrent_size(filesize, filesize_multiplier)
except:
pass
# content string contains all information not included into template
content = 'Category: "{category}". Downloaded {downloads} times.'
content = content.format(category=category, downloads=downloads)
@ -110,8 +100,9 @@ def response(resp):
'content': content,
'seed': seed,
'leech': leech,
'filesize': file_size,
'filesize': filesize,
'torrentfile': torrent_link,
'magnetlink': magnet_link,
'template': 'torrent.html'})
return results

View file

@ -14,8 +14,8 @@ import re
from lxml import html
from searx.engines.xpath import extract_text
from datetime import datetime
from searx.engines.nyaa import int_or_zero, get_filesize_mul
from searx.url_utils import urlencode
from searx.utils import get_torrent_size, int_or_zero
# engine dependent config
categories = ['files', 'videos', 'music']
@ -76,8 +76,7 @@ def response(resp):
try:
# ('1.228', 'GB')
groups = size_re.match(item).groups()
multiplier = get_filesize_mul(groups[1])
params['filesize'] = int(multiplier * float(groups[0]))
params['filesize'] = get_torrent_size(groups[0], groups[1])
except:
pass
elif item.startswith('Date:'):

View file

@ -290,6 +290,15 @@ def convert_str_to_int(number_str):
return 0
# convert a variable to integer or return 0 if it's not a number
def int_or_zero(num):
if isinstance(num, list):
if len(num) < 1:
return 0
num = num[0]
return convert_str_to_int(num)
def is_valid_lang(lang):
is_abbr = (len(lang) == 2)
if is_abbr:

View file

@ -13,38 +13,92 @@ class TestNyaaEngine(SearxTestCase):
params = nyaa.request(query, dic)
self.assertTrue('url' in params)
self.assertTrue(query in params['url'])
self.assertTrue('nyaa.se' in params['url'])
self.assertTrue('nyaa.si' in params['url'])
def test_response(self):
resp = mock.Mock(text='<html></html>')
self.assertEqual(nyaa.response(resp), [])
html = """
<table class="tlist">
<tbody>
<tr class="trusted tlistrow">
<td class="tlisticon">
<a href="//www.nyaa.se" title="English-translated Anime">
<img src="//files.nyaa.se" alt="English-translated Anime">
</a>
</td>
<td class="tlistname">
<a href="//www.nyaa.se/?page3">
Sample torrent title
</a>
</td>
<td class="tlistdownload">
<a href="//www.nyaa.se/?page_dl" title="Download">
<img src="//files.nyaa.se/www-dl.png" alt="DL">
</a>
</td>
<td class="tlistsize">10 MiB</td>
<td class="tlistsn">1</td>
<td class="tlistln">3</td>
<td class="tlistdn">666</td>
<td class="tlistmn">0</td>
</tr>
</tbody>
<table class="table table-bordered table-hover table-striped torrent-list">
<thead>
<tr>
<th class="hdr-category text-center" style="width:80px;">
<div>Category</div>
</th>
<th class="hdr-name" style="width:auto;">
<div>Name</div>
</th>
<th class="hdr-comments sorting text-center" title="Comments" style="width:50px;">
<a href="/?f=0&amp;c=0_0&amp;q=Death+Parade&amp;s=comments&amp;o=desc"></a>
<i class="fa fa-comments-o"></i>
</th>
<th class="hdr-link text-center" style="width:70px;">
<div>Link</div>
</th>
<th class="hdr-size sorting text-center" style="width:100px;">
<a href="/?f=0&amp;c=0_0&amp;q=Death+Parade&amp;s=size&amp;o=desc"></a>
<div>Size</div>
</th>
<th class="hdr-date sorting_desc text-center" title="In local time" style="width:140px;">
<a href="/?f=0&amp;c=0_0&amp;q=Death+Parade&amp;s=id&amp;o=asc"></a>
<div>Date</div>
</th>
<th class="hdr-seeders sorting text-center" title="Seeders" style="width:50px;">
<a href="/?f=0&amp;c=0_0&amp;q=Death+Parade&amp;s=seeders&amp;o=desc"></a>
<i class="fa fa-arrow-up" aria-hidden="true"></i>
</th>
<th class="hdr-leechers sorting text-center" title="Leechers" style="width:50px;">
<a href="/?f=0&amp;c=0_0&amp;q=Death+Parade&amp;s=leechers&amp;o=desc"></a>
<i class="fa fa-arrow-down" aria-hidden="true"></i>
</th>
<th class="hdr-downloads sorting text-center" title="Completed downloads" style="width:50px;">
<a href="/?f=0&amp;c=0_0&amp;q=Death+Parade&amp;s=downloads&amp;o=desc"></a>
<i class="fa fa-check" aria-hidden="true"></i>
</th>
</tr>
</thead>
<tbody>
<tr class="default">
<td style="padding:0 4px;">
<a href="/?c=1_2" title="Anime - English-translated">
<img src="/static/img/icons/nyaa/1_2.png" alt="Anime - English-translated">
</a>
</td>
<td colspan="2">
<a href="/view/1" title="Sample title 1">Sample title 1</a>
</td>
<td class="text-center" style="white-space: nowrap;">
<a href="/download/1.torrent"><i class="fa fa-fw fa-download"></i></a>
<a href="magnet:?xt=urn:btih:2"><i class="fa fa-fw fa-magnet"></i></a>
</td>
<td class="text-center">723.7 MiB</td>
<td class="text-center" data-timestamp="1503307456" title="1 week 3
days 9 hours 44 minutes 39 seconds ago">2017-08-21 11:24</td>
<td class="text-center" style="color: green;">1</td>
<td class="text-center" style="color: red;">3</td>
<td class="text-center">12</td>
</tr>
<tr class="default">
<td style="padding:0 4px;">
<a href="/?c=1_2" title="Anime - English-translated">
<img src="/static/img/icons/nyaa/1_2.png" alt="Anime - English-translated">
</a>
</td>
<td colspan="2">
<a href="/view/2" title="Sample title 2">Sample title 2</a>
</td>
<td class="text-center" style="white-space: nowrap;">
<a href="magnet:?xt=urn:btih:2"><i class="fa fa-fw fa-magnet"></i></a>
</td>
<td class="text-center">8.2 GiB</td>
<td class="text-center" data-timestamp="1491608400" title="4 months 3
weeks 4 days 19 hours 28 minutes 55 seconds ago">2017-04-08 01:40</td>
<td class="text-center" style="color: green;">10</td>
<td class="text-center" style="color: red;">1</td>
<td class="text-center">206</td>
</tr>
</tbody>
</table>
"""
@ -52,15 +106,19 @@ class TestNyaaEngine(SearxTestCase):
results = nyaa.response(resp)
self.assertEqual(type(results), list)
self.assertEqual(len(results), 1)
self.assertEqual(len(results), 2)
r = results[0]
self.assertTrue(r['url'].find('www.nyaa.se/?page3') >= 0)
self.assertTrue(r['torrentfile'].find('www.nyaa.se/?page_dl') >= 0)
self.assertTrue(r['content'].find('English-translated Anime') >= 0)
self.assertTrue(r['content'].find('Downloaded 666 times.') >= 0)
self.assertTrue(r['url'].find('1') >= 0)
self.assertTrue(r['torrentfile'].find('1.torrent') >= 0)
self.assertTrue(r['content'].find('Anime - English-translated') >= 0)
self.assertTrue(r['content'].find('Downloaded 12 times.') >= 0)
self.assertEqual(r['title'], 'Sample torrent title')
self.assertEqual(r['title'], 'Sample title 1')
self.assertEqual(r['seed'], 1)
self.assertEqual(r['leech'], 3)
self.assertEqual(r['filesize'], 10 * 1024 * 1024)
self.assertEqual(r['filesize'], 723700000)
r = results[1]
self.assertTrue(r['url'].find('2') >= 0)
self.assertTrue(r['magnetlink'].find('magnet:') >= 0)