Compare commits
No commits in common. "d9344b1b4b271edfc0d5f75db0187146538d6872" and "19d5d1e3ebab381453f3bff1b4cd8e7a3126667d" have entirely different histories.
d9344b1b4b
...
19d5d1e3eb
10
.coveragerc
10
.coveragerc
@ -1,10 +0,0 @@
|
|||||||
[run]
|
|
||||||
source = .
|
|
||||||
branch = true
|
|
||||||
omit =
|
|
||||||
.venv/*
|
|
||||||
|
|
||||||
[report]
|
|
||||||
fail_under = 100
|
|
||||||
show_missing = True
|
|
||||||
skip_covered = True
|
|
4
.gitignore
vendored
4
.gitignore
vendored
@ -4,7 +4,3 @@
|
|||||||
/.idea/
|
/.idea/
|
||||||
# Distribution / packaging:
|
# Distribution / packaging:
|
||||||
*.egg-info/
|
*.egg-info/
|
||||||
# Python cache:
|
|
||||||
__pycache__/
|
|
||||||
# Tests:
|
|
||||||
.coverage
|
|
@ -1,3 +0,0 @@
|
|||||||
#!/usr/bin/env sh
|
|
||||||
|
|
||||||
coverage erase && coverage run -m unittest discover && coverage report
|
|
@ -19,7 +19,7 @@ keywords = webscraping, html, markup, dom, scraper, attributes, tags, stocks, fi
|
|||||||
package_dir =
|
package_dir =
|
||||||
= src
|
= src
|
||||||
packages = find:
|
packages = find:
|
||||||
python_requires = >=3.8
|
python_requires = >=3.7
|
||||||
install_requires =
|
install_requires =
|
||||||
beautifulsoup4
|
beautifulsoup4
|
||||||
aiohttp
|
aiohttp
|
||||||
|
@ -8,7 +8,7 @@ from pathlib import Path
|
|||||||
from . import get_all_data, log
|
from . import get_all_data, log
|
||||||
|
|
||||||
|
|
||||||
async def main() -> None:
|
def main() -> None:
|
||||||
parser = ArgumentParser(description="Scrape all stock symbols")
|
parser = ArgumentParser(description="Scrape all stock symbols")
|
||||||
parser.add_argument(
|
parser.add_argument(
|
||||||
'-v', '--verbose',
|
'-v', '--verbose',
|
||||||
@ -29,7 +29,7 @@ async def main() -> None:
|
|||||||
if args.verbose:
|
if args.verbose:
|
||||||
log.setLevel(logging.DEBUG)
|
log.setLevel(logging.DEBUG)
|
||||||
|
|
||||||
data = await get_all_data(args.sequential)
|
data = asyncio.run(get_all_data(args.sequential))
|
||||||
|
|
||||||
if args.to_file is None:
|
if args.to_file is None:
|
||||||
csv.writer(sys.stdout).writerows(data)
|
csv.writer(sys.stdout).writerows(data)
|
||||||
@ -39,4 +39,4 @@ async def main() -> None:
|
|||||||
|
|
||||||
|
|
||||||
if __name__ == '__main__':
|
if __name__ == '__main__':
|
||||||
asyncio.run(main())
|
main()
|
||||||
|
@ -1,13 +1,12 @@
|
|||||||
import logging
|
from unittest import TestCase
|
||||||
from unittest import IsolatedAsyncioTestCase
|
from unittest.mock import patch, MagicMock, call
|
||||||
from unittest.mock import patch, MagicMock, AsyncMock, call
|
|
||||||
|
|
||||||
from bs4 import BeautifulSoup
|
from bs4 import BeautifulSoup
|
||||||
|
|
||||||
from stocksymbolscraper import scrape
|
from stocksymbolscraper import scrape
|
||||||
|
|
||||||
|
|
||||||
class ScrapeTestCase(IsolatedAsyncioTestCase):
|
class ScrapeTestCase(TestCase):
|
||||||
|
|
||||||
@patch.object(scrape, 'get_single_tr_data')
|
@patch.object(scrape, 'get_single_tr_data')
|
||||||
def test_extract_row_data(self, mock_get_single_tr_data: MagicMock):
|
def test_extract_row_data(self, mock_get_single_tr_data: MagicMock):
|
||||||
@ -54,36 +53,8 @@ class ScrapeTestCase(IsolatedAsyncioTestCase):
|
|||||||
output = scrape.get_str_from_td(test_td)
|
output = scrape.get_str_from_td(test_td)
|
||||||
self.assertEqual(expected_output, output)
|
self.assertEqual(expected_output, output)
|
||||||
|
|
||||||
@patch.object(scrape, 'ClientSession')
|
|
||||||
async def test_soup_from_url(self, mock_session_cls):
|
|
||||||
test_html = '<b>foo</b>'
|
|
||||||
mock_response = MagicMock()
|
|
||||||
mock_response.text = AsyncMock(return_value=test_html)
|
|
||||||
mock_get_return = MagicMock()
|
|
||||||
mock_get_return.__aenter__ = AsyncMock(return_value=mock_response)
|
|
||||||
mock_session_obj = MagicMock()
|
|
||||||
mock_session_obj.get = MagicMock(return_value=mock_get_return)
|
|
||||||
mock_session_cls.return_value = mock_session_obj
|
|
||||||
output = await scrape.soup_from_url('foo')
|
|
||||||
expected_output = BeautifulSoup(test_html, scrape.HTML_PARSER)
|
|
||||||
self.assertEqual(expected_output, output)
|
|
||||||
|
|
||||||
output = await scrape.soup_from_url('foo', mock_session_obj)
|
|
||||||
self.assertEqual(expected_output, output)
|
|
||||||
|
|
||||||
def test_trs_from_page(self):
|
def test_trs_from_page(self):
|
||||||
tr1_text, tr2_text = '<tr>foo</tr>', '<tr>bar</tr>'
|
# Tested function takes URL as argument (GET request is issued)
|
||||||
test_html = f'<div id="marketsindex"><table><tbody>{tr1_text}{tr2_text}</tbody></table></div>'
|
# HTML to be parsed could be substituted
|
||||||
test_soup = BeautifulSoup(test_html, scrape.HTML_PARSER)
|
#
|
||||||
output = scrape.trs_from_page(test_soup)
|
pass
|
||||||
expected_output = test_soup.find_all('tr')
|
|
||||||
self.assertSequenceEqual(expected_output, output)
|
|
||||||
|
|
||||||
logging.disable(logging.CRITICAL)
|
|
||||||
test_html = f'<div id="marketsindex"><table>garbage</table></div>'
|
|
||||||
test_soup = BeautifulSoup(test_html, scrape.HTML_PARSER)
|
|
||||||
with patch.object(scrape, 'open') as mock_open:
|
|
||||||
self.assertRaises(scrape.UnexpectedMarkupError, scrape.trs_from_page, test_soup)
|
|
||||||
mock_open.assert_called_once()
|
|
||||||
mock_open.return_value.__enter__.return_value.write.assert_called_once_with(test_soup.prettify())
|
|
||||||
logging.disable(logging.NOTSET)
|
|
||||||
|
Loading…
x
Reference in New Issue
Block a user