def test_get_url_by_id_correct() -> None:
with CustomTestClient() as c:
url1 = Url(
id="abc123",
url="www.test1.org",
yearmonth="201912",
undesired_url=True,
)
url2 = Url(
id="qwe987",
url="www.test2.net",
yearmonth="200506",
undesired_url=False,
payed_content=True,
)
db.session.add(url1)
db.session.add(url2)
db.session.commit()
res = c.get('/v1/urls/abc123')
data = res.get_json()
assert data["url_id"] == "abc123"
assert data["url"] == "www.test1.org"
assert data["yearmonth"] == "201912"
assert data["undesired_url"] is True
assert data["payed_content"] is False
assert res.status_code == 200
res2 = c.get('/v1/urls/qwe987')
data = res2.get_json()
assert data["url_id"] == "qwe987"
assert data["url"] == "www.test2.net"
assert data["yearmonth"] == "200506"
assert data["undesired_url"] is False
assert data["payed_content"] is True
assert res.status_code == 200
def create_url(dto: CreateUrlDTO) -> None:
assert_valid_yearmonth(dto.yearmonth)
url = Url(id=dto.id,
yearmonth=dto.yearmonth,
url=dto.url,
undesired_url=dto.undesired_url)
url_repo.save(url)
def test_get_articles_by_yearmonth_correct() -> None:
with CustomTestClient() as c:
url_1 = Url(id="qaz987",
url="test",
yearmonth="201910",
undesired_url=False)
article_1 = Article(
id="qaz987",
headline="This is my headline",
body="This is a body text to my article in this test.",
created_at=datetime.utcnow())
url_2 = Url(id="abc123",
url="test2",
yearmonth="201911",
undesired_url=False)
article_2 = Article(
id="abc123",
headline="This is my headline for another article",
body="This is a body text to my other article that i want to find",
created_at=datetime.utcnow())
url_3 = Url(id="uhn768",
url="test3",
yearmonth="201910",
undesired_url=False)
article_3 = Article(id="uhn768",
headline="This",
body="This is a body text.",
created_at=datetime.utcnow())
db.session.add(url_1)
db.session.add(url_2)
db.session.add(url_3)
db.session.add(article_1)
db.session.add(article_2)
db.session.add(article_3)
db.session.commit()
res = c.get(f'/v1/articles/yearmonth/201910')
assert res.status_code == 200
data = res.get_json()
assert data[0]["article_id"] == "qaz987"
assert data[0]["headline"] == "This is my headline"
assert data[0]["body"] == \
"This is a body text to my article in this test."
assert data[0]["created_at"] == article_1.created_at.__str__()
assert data[1]["article_id"] == "uhn768"
assert data[1]["headline"] == "This"
assert data[1]["body"] == "This is a body text."
assert data[1]["created_at"] == article_3.created_at.__str__()
def test_get_urls_by_yearmonth() -> None:
with CustomTestClient() as c:
url1 = Url(
id="abc123",
url="www.test1.org",
yearmonth="201912",
undesired_url=True,
scraped_at=datetime.utcnow()
)
url2 = Url(
id="qwe987",
url="www.test2.net",
yearmonth="201912",
undesired_url=False,
payed_content=False,
)
url3 = Url(
id="try345",
url="another.test3.com",
yearmonth="200503",
undesired_url=False,
payed_content=False,
)
db.session.add(url1)
db.session.add(url2)
db.session.add(url3)
db.session.commit()
res = c.get('/v1/urls/yearmonth/201912')
assert res.status_code == 200
data = res.get_json()
assert len(data) == 2
assert data[0]["url_id"] == "abc123"
assert data[0]["url"] == "www.test1.org"
assert data[0]["yearmonth"] == "201912"
assert data[1]["url_id"] == "qwe987"
assert data[1]["url"] == "www.test2.net"
assert data[1]["yearmonth"] == "201912"
def test_post_article_correct() -> None:
with CustomTestClient() as c:
url = Url(id="abc123",
url="test",
yearmonth="201910",
undesired_url=False)
db.session.add(url)
db.session.commit()
res = c.post('/v1/articles',
data=json.dumps({
"id": "abc123",
"body": "This is my article body text",
"headline": "Breaking News!!"
}),
content_type="application/json")
assert res.status_code == 200
def test_put_urls_unscraped_correct() -> None:
with CustomTestClient() as c:
url = Url(
id="abc123",
url="www.test1.org",
yearmonth="201912"
)
db.session.add(url)
db.session.commit()
assert Url.query.filter_by(id="abc123").first().scraped_at is None
time_scraped = datetime.utcnow()
res = c.put(
'/v1/urls/abc123/unscraped',
data=json.dumps({
"scraped_at": time_scraped.__str__(),
}),
content_type="application/json"
)
assert res.status_code == 200
record = Url.query.filter_by(id="abc123").first()
assert record.scraped_at == time_scraped
def test_get_article_by_id_correct() -> None:
with CustomTestClient() as c:
url = Url(id="qaz987",
url="test",
yearmonth="201910",
undesired_url=False)
article = Article(
id="qaz987",
headline="This is my headline",
body="This is a body text to my article in this test.",
created_at=datetime.utcnow())
db.session.add(url)
db.session.add(article)
db.session.commit()
article_id = "qaz987"
res = c.get(f'/v1/articles/id/{article_id}')
assert res.status_code == 200
data = res.get_json()
assert data["article_id"] == "qaz987"
assert data["headline"] == "This is my headline"
assert data["body"] == \
"This is a body text to my article in this test."
assert data["created_at"] == article.created_at.__str__()
def flag_url_is_scraped(url: Url, scraped_at: datetime) -> None:
url.scraped_at = scraped_at
db.session.commit()