def search(self, query, page=None, detailed=False):
page = 0 if page is None else int(page)
if page > len(self._pagtok) - 1:
raise ValueError(
'Parameter \'page\' ({page}) must be between 0 and 12.'.format(
page=page))
pagtok = self._pagtok[page]
data = generate_post_data(0, 0, pagtok)
self.params.update({
'q': quote_plus(query),
'c': 'apps',
})
response = send_request('POST', self._search_url, data, self.params)
soup = BeautifulSoup(response.content, 'lxml', from_encoding='utf8')
if detailed:
apps = self._parse_multiple_apps(response)
else:
apps = [
parse_card_info(app)
for app in soup.select('div[data-uitype="500"]')
]
return apps
def developer(self, developer, results=None, page=None, detailed=False):
"""Sends a POST request and retrieves a list of the developer's
published applications on the Play Store.
:param developer: developer name to retrieve apps from, e.g. 'Disney'
:param results: the number of app results to retrieve
:param page: the page number to retrieve
:param detailed: if True, sends request per app for its full detail
:return: a list of app dictionaries
"""
if not isinstance(developer, basestring) or developer.isdigit():
raise ValueError('Parameter \'developer\' must be the developer name, not the developer id.')
results = s.DEV_RESULTS if results is None else results
page = 0 if page is None else page
page_num = (results // 20) * page
if not 0 <= page_num <= 12:
raise ValueError('Page out of range. (results // 20) * page must be between 0 - 12')
pagtok = self._pagtok[page_num]
url = build_url('developer', developer)
data = generate_post_data(results, 0, pagtok)
response = send_request('POST', url, data, self.params)
if detailed:
apps = self._parse_multiple_apps(response)
else:
soup = BeautifulSoup(response.content, 'lxml', from_encoding='utf8')
apps = [parse_card_info(app)
for app in soup.select('div[data-uitype="500"]')]
return apps
def search(self, query, page=None, detailed=False):
"""Sends a POST request and retrieves a list of applications matching
the query term(s).
:param query: search query term(s) to retrieve matching apps
:param page: the page number to retrieve. Max is 12.
:param detailed: if True, sends request per app for its full detail
:return: a list of apps matching search terms
"""
page = 0 if page is None else int(page)
if page > len(self._pagtok) - 1:
raise ValueError(
"Parameter 'page' ({page}) must be between 0 and 12.".format(
page=page))
pagtok = self._pagtok[page]
data = generate_post_data(0, 0, pagtok)
self.params.update({"q": quote_plus(query), "c": "apps"})
response = send_request("POST", self._search_url, data, self.params)
soup = BeautifulSoup(response.content, "lxml", from_encoding="utf8")
if detailed:
apps = self._parse_multiple_apps(response)
else:
apps = [
parse_cluster_card_info(app)
for app in soup.select("div.Vpfmgd")
]
return apps
def categories(self, ignore_promotions=True):
"""Sends a GET request to the front page (app store base url), parses
and returns a list of all available categories.
"""
categories = {}
response = send_request('GET', s.BASE_URL, params=self.params)
soup = BeautifulSoup(response.content, 'lxml', from_encoding='utf8')
category_links = soup.select(
'div[id*="action-dropdown-children"] a[href*="category"]')
age_query = '?age='
for cat in category_links:
url = urljoin(s.BASE_URL, cat.attrs['href'])
category_id = url.split('/')[-1]
name = cat.string.strip()
if age_query in category_id:
category_id = 'FAMILY'
url = url.split('?')[0]
name = 'Family'
if category_id not in categories:
if ignore_promotions and '/store/apps/category/' not in url:
continue
categories[category_id] = {
'name': name,
'url': url,
'category_id': category_id
}
return categories
def categories(self):
"""Sends a GET request to the front page (app store base url), parses
and returns a list of all available categories.
Note: May contain some promotions, e.g. "Popular Characters"
"""
categories = {}
strainer = SoupStrainer('a', {'class': 'child-submenu-link'})
response = send_request('GET', s.BASE_URL)
soup = BeautifulSoup(response.content,
'lxml',
from_encoding='utf8',
parse_only=strainer)
category_links = soup.select('a.child-submenu-link')
age = '?age='
for cat in category_links:
url = urljoin(s.BASE_URL, cat.attrs['href'])
category_id = url.split('/')[-1]
name = cat.string.strip()
if age in category_id:
category_id = 'FAMILY'
url = url.split('?')[0]
name = 'Family'
if category_id not in categories:
categories[category_id] = {
'name': name,
'url': url,
'category_id': category_id}
return categories
def categories(self, ignore_promotions=True):
"""Sends a GET request to the front page (app store base url), parses
and returns a list of all available categories.
"""
categories = {}
response = send_request("GET", s.BASE_URL, params=self.params)
soup = BeautifulSoup(response.content, "lxml", from_encoding="utf8")
category_links = soup.select(
'div[id*="action-dropdown-children"] a[href*="category"]')
age_query = "?age="
for cat in category_links:
url = urljoin(s.BASE_URL, cat.attrs["href"])
category_id = url.split("/")[-1]
name = cat.string.strip()
if age_query in category_id:
category_id = "FAMILY"
url = url.split("?")[0]
name = "Family"
if category_id not in categories:
if ignore_promotions and "/store/apps/category/" not in url:
continue
categories[category_id] = {
"name": name,
"url": url,
"category_id": category_id,
}
return categories
def developer(self, developer, results=None, page=None, detailed=False):
if not isinstance(developer, basestring) or developer.isdigit():
raise ValueError(
'Parameter \'developer\' must be the developer name, not the developer id.'
)
results = s.DEV_RESULTS if results is None else results
page = 0 if page is None else page
page_num = (results // 20) * page
if not 0 <= page_num <= 12:
raise ValueError(
'Page out of range. (results // 20) * page must be between 0 - 12'
)
pagtok = self._pagtok[page_num]
url = build_url('developer', developer)
data = generate_post_data(results, 0, pagtok)
response = send_request('POST', url, data, self.params)
if detailed:
apps = self._parse_multiple_apps(response)
else:
soup = BeautifulSoup(response.content,
'lxml',
from_encoding='utf8')
apps = [
parse_card_info(app)
for app in soup.select('div[data-uitype="500"]')
]
return apps
def search(self, query, page=None, detailed=False):
"""Sends a POST request and retrieves a list of applications matching
the query term(s).
:param query: search query term(s) to retrieve matching apps
:param page: the page number to retrieve. Max is 12.
:param detailed: if True, sends request per app for its full detail
:return: a list of apps matching search terms
"""
page = 0 if page is None else int(page)
if page > len(self._pagtok) - 1:
raise ValueError('Parameter \'page\' ({page}) must be between 0 and 12.'.format(
page=page))
pagtok = self._pagtok[page]
data = generate_post_data(0, 0, pagtok)
self.params.update({
'q': quote_plus(query),
'c': 'apps',
})
response = send_request('POST', self._search_url, data, self.params)
soup = BeautifulSoup(response.content, 'lxml', from_encoding='utf8')
if detailed:
apps = self._parse_multiple_apps(response)
else:
apps = [parse_card_info(app)
for app in soup.select('div[data-uitype="500"]')]
return apps
def developer(self, developer, results=None, page=None, detailed=False):
"""Sends a POST request and retrieves a list of the developer's
published applications on the Play Store.
:param developer: developer name to retrieve apps from, e.g. 'Disney'
:param results: the number of app results to retrieve
:param page: the page number to retrieve
:param detailed: if True, sends request per app for its full detail
:return: a list of app dictionaries
"""
if not isinstance(developer, basestring) or developer.isdigit():
raise ValueError('Parameter \'developer\' must be the developer name, not the developer id.')
results = s.DEV_RESULTS if results is None else results
page = 0 if page is None else page
page_num = (results // 20) * page
if not 0 <= page_num <= 12:
raise ValueError('Page out of range. (results // 20) * page must be between 0 - 12')
pagtok = self._pagtok[page_num]
url = build_url('developer', developer)
data = generate_post_data(results, 0, pagtok)
response = send_request('POST', url, data, self.params)
if detailed:
apps = self._parse_multiple_apps(response)
else:
soup = BeautifulSoup(response.content, 'lxml', from_encoding='utf8')
apps = [parse_card_info(app)
for app in soup.select('div[data-uitype=500]')]
return apps
def category_items(self, category, detailed=False):
"""
list of apps in category main page
:param detailed:
:param category:
:return:
"""
category_url = build_category_url(category)
response = send_request("GET", category_url)
soup = BeautifulSoup(response.content, 'lxml', from_encoding='utf8')
items_elements = soup.select(self.list_item_selector)
if not len(items_elements):
items_elements = soup.select(self.list_item_promo_selector)
app_ids = [
get_query_params(element['href'])['id'][0]
for element in items_elements
]
if not detailed:
return [{'app_id': app_id} for app_id in app_ids]
return multi_futures_app_request(app_ids=app_ids)
def categories(self, ignore_promotions=True):
categories = {}
strainer = SoupStrainer('ul', {'class': 'submenu-item-wrapper'})
response = send_request('GET', s.BASE_URL, params=self.params)
soup = BeautifulSoup(response.content,
'lxml',
from_encoding='utf8',
parse_only=strainer)
category_links = soup.select('a.child-submenu-link')
category_links += soup.select('a.parent-submenu-link')
age_query = '?age='
for cat in category_links:
url = urljoin(s.BASE_URL, cat.attrs['href'])
category_id = url.split('/')[-1]
name = cat.string.strip()
if age_query in category_id:
category_id = 'FAMILY'
url = url.split('?')[0]
name = 'Family'
if category_id not in categories:
if ignore_promotions and '/store/apps/category/' not in url:
continue
categories[category_id] = {
'name': name,
'url': url,
'category_id': category_id
}
return categories
def cluster_items(self, gsr, detailed=False):
"""
Get cluster page items
https://play.google.com/store/apps/collection/cluster?clp=0g4cChoKFHRvcHNlbGxpbmdfZnJlZV9HQU1FEAcYAw%3D%3D:S:ANO1ljJ_Y5U&gsr=Ch_SDhwKGgoUdG9wc2VsbGluZ19mcmVlX0dBTUUQBxgD:S:ANO1ljL4b8c
:param gsr: cluster id
:return: list of app details
"""
cluster_url = build_cluster_url(gsr=gsr)
response = send_request("GET", cluster_url)
soup = BeautifulSoup(response.content, 'lxml', from_encoding='utf8')
items_elements = soup.select(self.list_item_selector)
if not len(items_elements):
items_elements = soup.select(self.list_item_promo_selector)
app_ids = [
get_query_params(element['href'])['id'][0]
for element in items_elements
]
if not detailed:
return [{'app_id': app_id} for app_id in app_ids]
return multi_futures_app_request(app_ids=app_ids)
def search(self, query, page=None, detailed=False):
"""Sends a POST request and retrieves a list of applications matching
the query term(s).
:param query: search query term(s) to retrieve matching apps
:param page: the page number to retrieve. Max is 12.
:param detailed: if True, sends request per app for its full detail
:return: a list of apps matching search terms
"""
page = 0 if page is None else int(page)
if page > len(self._pagtok) - 1:
raise ValueError('Parameter \'page\' ({page}) must be between 0 and 12.'.format(
page=page))
pagtok = self._pagtok[page]
data = generate_post_data(0, 0, pagtok)
self.params.update({
'q': quote_plus(query),
'c': 'apps',
})
response = send_request('POST', self._search_url, data, self.params)
soup = BeautifulSoup(response.content, 'lxml', from_encoding='utf8')
if detailed:
apps = self._parse_multiple_apps(response)
else:
apps = [parse_card_info(app)
for app in soup.select('div[data-uitype=500]')]
return apps
def collection(
self,
collection_id,
category_id=None,
results=None,
page=None,
age=None,
detailed=False,
):
"""Sends a POST request and fetches a list of applications belonging to
the collection and an optional category.
:param collection_id: the collection id, e.g. 'NEW_FREE'.
:param category_id: (optional) the category id, e.g. 'GAME_ACTION'.
:param results: the number of apps to retrieve at a time.
:param page: page number to retrieve; limitation: page * results <= 500.
:param age: an age range to filter by (only for FAMILY categories)
:param detailed: if True, sends request per app for its full detail
:return: a list of app dictionaries
"""
if collection_id not in COLLECTIONS and not collection_id.startswith(
"promotion"):
raise ValueError("Invalid collection_id '{collection}'.".format(
collection=collection_id))
collection_name = COLLECTIONS.get(collection_id) or collection_id
category = "" if category_id is None else CATEGORIES.get(category_id)
if category is None:
raise ValueError("Invalid category_id '{category}'.".format(
category=category_id))
results = s.NUM_RESULTS if results is None else results
if results > 120:
raise ValueError("Number of results cannot be more than 120.")
page = 0 if page is None else page
if page * results > 500:
raise ValueError(
"Start (page * results) cannot be greater than 500.")
if category.startswith("FAMILY") and age is not None:
self.params["age"] = AGE_RANGE[age]
url = build_collection_url(category, collection_name)
data = generate_post_data(results, page)
response = send_request("POST", url, data, self.params)
if detailed:
apps = self._parse_multiple_apps(response)
else:
soup = BeautifulSoup(response.content,
"lxml",
from_encoding="utf8")
apps = [
parse_card_info(app_card)
for app_card in soup.select('div[data-uitype="500"]')
]
return apps
def test_request_with_params(self):
method = 'GET'
params = {'q': 'google play store'}
response = send_request(method, self.url, params=params)
expected_url = "{base}{params}".format(
base=self.url, params='?q=google+play+store')
self.assertEqual(response.status_code, 200)
self.assertEqual(response.url, expected_url)
def test_request_with_params(self):
method = "GET"
params = {"q": "google play store"}
response = send_request(method, self.url, params=params)
expected_url = "{base}{params}".format(base=self.url,
params="?q=google+play+store")
self.assertEqual(response.status_code, 200)
self.assertEqual(response.url, expected_url)
def reviews(self, app_id, page=1):
"""Sends a POST request and retrieves a list of reviews for
the specified app.
:param app_id: the app to retrieve details from, e.g. 'com.nintendo.zaaa'
:param page: the page number to retrieve; max is 10
:return: a list of reviews
"""
data = {
'reviewType': 0,
'pageNum': page,
'id': app_id,
'reviewSortOrder': 4,
'xhr': 1,
'hl': self.language
}
self.params['authuser'] = '******'
response = send_request('POST', s.REVIEW_URL, data, self.params)
content = response.text
content = content[content.find('[["ecr"'):].strip()
data = json.loads(content)
html = data[0][2]
soup = BeautifulSoup(html, 'lxml', from_encoding='utf8')
reviews = []
for element in soup.select('.single-review'):
review = {}
avatar_style = element.select_one('.author-image').get('style')
if avatar_style:
sheet = cssutils.css.CSSStyleSheet()
sheet.add('tmp { %s }' % avatar_style)
review['author_image'] = list(cssutils.getUrls(sheet))[0]
review_header = element.select_one('.review-header')
review['review_id'] = review_header.get('data-reviewid', '')
review['review_permalink'] = review_header.select_one('.reviews-permalink').get('href')
review['author_name'] = review_header.select_one('.author-name').text
review['review_date'] = review_header.select_one('.review-date').text
curr_rating = review_header.select_one('.current-rating').get('style')
review['current_rating'] = int(int(str(cssutils.parseStyle(curr_rating).width).replace('%', '')) / 20)
body_elem = element.select_one('.review-body')
review_title = body_elem.select_one('.review-title').extract()
body_elem.select_one('.review-link').decompose()
review['review_title'] = review_title.text
review['review_body'] = body_elem.text
reviews.append(review)
return reviews
def suggestions(self, query):
if not query:
raise ValueError("Cannot get suggestions for an empty query.")
self.params.update({
'json': 1,
'c': 0,
'query': query,
})
response = send_request('GET',
self._suggestion_url,
params=self.params)
suggestions = [q['s'] for q in response.json()]
return suggestions
def suggestions(self, query):
"""Sends a GET request and retrieves a list of autocomplete suggestions
matching the query term(s).
:param query: search query term(s) to retrieve autocomplete suggestions
:return: a list of suggested search queries, up to 5
"""
if not query:
raise ValueError("Cannot get suggestions for an empty query.")
params = {'json': 1, 'c': 0, 'hl': 'en', 'gl': 'us', 'query': query}
response = send_request('GET', self._suggestion_url, params=params)
suggestions = [q['s'] for q in response.json()]
return suggestions
def collection(self, collection_id, category_id=None, results=None,
page=None, age=None, detailed=False):
"""Sends a POST request and fetches a list of applications belonging to
the collection and an optional category.
:param collection_id: the collection id, e.g. 'NEW_FREE'.
:param category_id: (optional) the category id, e.g. 'GAME_ACTION'.
:param results: the number of apps to retrieve at a time.
:param page: page number to retrieve; limitation: page * results <= 500.
:param age: an age range to filter by (only for FAMILY categories)
:param detailed: if True, sends request per app for its full detail
:return: a list of app dictionaries
"""
if (collection_id not in COLLECTIONS and
not collection_id.startswith('promotion')):
raise ValueError('Invalid collection_id \'{collection}\'.'.format(
collection=collection_id))
collection_name = COLLECTIONS.get(collection_id) or collection_id
category = '' if category_id is None else CATEGORIES.get(category_id)
if category is None:
raise ValueError('Invalid category_id \'{category}\'.'.format(
category=category_id))
results = s.NUM_RESULTS if results is None else results
if results > 120:
raise ValueError('Number of results cannot be more than 120.')
page = 0 if page is None else page
if page * results > 500:
raise ValueError('Start (page * results) cannot be greater than 500.')
if category.startswith('FAMILY') and age is not None:
self.params['age'] = AGE_RANGE[age]
url = build_collection_url(category, collection_name)
data = generate_post_data(results, page)
response = send_request('POST', url, data, self.params)
if detailed:
apps = self._parse_multiple_apps(response)
else:
soup = BeautifulSoup(response.content, 'lxml', from_encoding='utf8')
apps = [parse_card_info(app_card)
for app_card in soup.select('div[data-uitype="500"]')]
return apps
def collection(self,
collection_id,
category_id=None,
results=None,
page=None,
age=None,
detailed=False):
if (collection_id not in COLLECTIONS
and not collection_id.startswith('promotion')):
raise ValueError('Invalid collection_id \'{collection}\'.'.format(
collection=collection_id))
collection_name = COLLECTIONS.get(collection_id) or collection_id
category = '' if category_id is None else CATEGORIES.get(category_id)
if category is None:
raise ValueError('Invalid category_id \'{category}\'.'.format(
category=category_id))
results = s.NUM_RESULTS if results is None else results
if results > 120:
raise ValueError('Number of results cannot be more than 120.')
page = 0 if page is None else page
if page * results > 500:
raise ValueError(
'Start (page * results) cannot be greater than 500.')
if category.startswith('FAMILY') and age is not None:
self.params['age'] = AGE_RANGE[age]
url = build_collection_url(category, collection_name)
data = generate_post_data(results, page)
response = send_request('POST', url, data, self.params)
if detailed:
apps = self._parse_multiple_apps(response)
else:
soup = BeautifulSoup(response.content,
'lxml',
from_encoding='utf8')
apps = [
parse_card_info(app_card)
for app_card in soup.select('div[data-uitype="500"]')
]
return apps
def suggestions(self, query):
"""Sends a GET request and retrieves a list of autocomplete suggestions
matching the query term(s).
:param query: search query term(s) to retrieve autocomplete suggestions
:return: a list of suggested search queries, up to 5
"""
if not query:
raise ValueError("Cannot get suggestions for an empty query.")
self.params.update({"json": 1, "c": 0, "query": query})
response = send_request("GET",
self._suggestion_url,
params=self.params)
suggestions = [q["s"] for q in response.json()]
return suggestions
def details(self, app_id):
url = build_url('details', app_id)
try:
response = send_request('GET', url, params=self.params)
soup = BeautifulSoup(response.content,
'lxml',
from_encoding='utf8')
except requests.exceptions.HTTPError as e:
raise ValueError('Invalid application ID: {app}. {error}'.format(
app=app_id, error=e))
app_json = parse_app_details(soup)
app_json.update({
'app_id': app_id,
'url': url,
})
return app_json
def similar(self, app_id, detailed=False, **kwargs):
url = build_url('similar', app_id)
response = send_request('GET',
url,
params=self.params,
allow_redirects=True)
soup = BeautifulSoup(response.content, 'lxml', from_encoding='utf8')
if detailed:
apps = self._parse_multiple_apps(response)
else:
apps = [
parse_card_info(app)
for app in soup.select('div[data-uitype="500"]')
]
return apps
def similar(self, app_id, detailed=False):
"""Sends a GET request, follows the redirect, and retrieves a list of
applications similar to the specified app.
:param app_id: the app to retrieve details from, e.g. 'com.nintendo.zaaa'
:param detailed: if True, sends request per app for its full detail
:return: a list of similar apps
"""
url = build_url('similar', app_id)
response = send_request('GET', url, allow_redirects=True)
soup = BeautifulSoup(response.content, 'lxml', from_encoding='utf8')
if detailed:
apps = self._parse_multiple_apps(response)
else:
apps = [self._parse_card_info(app)
for app in soup.select('div[data-uitype=500]')]
return apps
def details(self, app_id):
"""Sends a GET request and parses an application's details.
:param app_id: the app to retrieve details, e.g. 'com.nintendo.zaaa'
:return: a dictionary of app details
"""
url = build_url("details", app_id)
try:
response = send_request("GET", url, params=self.params)
soup = BeautifulSoup(response.content,
"lxml",
from_encoding="utf8")
except requests.exceptions.HTTPError as e:
raise ValueError("Invalid application ID: {app}. {error}".format(
app=app_id, error=e))
app_json = parse_app_details(soup)
app_json.update({"app_id": app_id, "url": url})
return app_json
def suggestions(self, query):
"""Sends a GET request and retrieves a list of autocomplete suggestions
matching the query term(s).
:param query: search query term(s) to retrieve autocomplete suggestions
:return: a list of suggested search queries, up to 5
"""
if not query:
raise ValueError("Cannot get suggestions for an empty query.")
self.params.update({
'json': 1,
'c': 0,
'query': query,
})
response = send_request('GET',
self._suggestion_url,
params=self.params)
suggestions = [q['s'] for q in response.json()]
return suggestions
def details(self, app_id):
"""Sends a GET request and parses an application's details.
:param app_id: the app to retrieve details, e.g. 'com.nintendo.zaaa'
:return: a dictionary of app details
"""
url = build_url('details', app_id)
try:
response = send_request('GET', url, params=self.params)
soup = BeautifulSoup(response.content, 'lxml', from_encoding='utf8')
except requests.exceptions.HTTPError as e:
raise ValueError('Invalid application ID: {app}. {error}'.format(
app=app_id, error=e))
app_json = parse_app_details(soup)
app_json.update({
'app_id': app_id,
'url': url,
})
return app_json
def details(self, app_id):
"""Sends a GET request and parses an application's details.
:param app_id: the app to retrieve details, e.g. 'com.nintendo.zaaa'
:return: a dictionary of app details
"""
url = build_url('details', app_id)
try:
response = send_request('GET', url, params=self.params)
soup = BeautifulSoup(response.content, 'lxml', from_encoding='utf8')
except requests.exceptions.HTTPError as e:
raise ValueError('Invalid application ID: {app}. {error}'.format(
app=app_id, error=e))
app_json = parse_app_details(soup)
app_json.update({
'app_id': app_id,
'url': url,
})
return app_json
def similar(self, app_id, detailed=False, **kwargs):
"""Sends a GET request, follows the redirect, and retrieves a list of
applications similar to the specified app.
:param app_id: app to retrieve details from, e.g. 'com.nintendo.zaaa'
:param detailed: if True, sends request per app for its full detail
:return: a list of similar apps
"""
url = build_url('similar', app_id)
response = send_request('GET',
url,
params=self.params,
allow_redirects=True)
soup = BeautifulSoup(response.content, 'lxml', from_encoding='utf8')
if detailed:
apps = self._parse_multiple_apps(response)
else:
apps = [parse_card_info(app)
for app in soup.select('div[data-uitype="500"]')]
return apps
def category_clusters(self, category):
"""
:param category:
:return:
"""
clusters = {}
cluster_selector = "c-wiz > c-wiz > div > div.Z3lOXb > div.xwY9Zc > a"
category_url = build_category_url(category=category)
response = send_request("GET", category_url)
soup = BeautifulSoup(response.content, 'lxml', from_encoding='utf8')
cluster_elements = soup.select(cluster_selector)
for element in cluster_elements:
title = element.h2.text
gsr = get_query_params(element['href'])['gsr'][0]
clusters[title] = gsr
return clusters
def categories(self, ignore_promotions=True):
"""Sends a GET request to the front page (app store base url), parses
and returns a list of all available categories.
"""
categories = {}
strainer = SoupStrainer('ul', {'class': 'submenu-item-wrapper'})
response = send_request('GET', s.BASE_URL, params=self.params)
soup = BeautifulSoup(response.content,
'lxml',
from_encoding='utf8',
parse_only=strainer)
category_links = soup.select('a.child-submenu-link')
category_links += soup.select('a.parent-submenu-link')
age_query = '?age='
for cat in category_links:
url = urljoin(s.BASE_URL, cat.attrs['href'])
category_id = url.split('/')[-1]
name = cat.string.strip()
if age_query in category_id:
category_id = 'FAMILY'
url = url.split('?')[0]
name = 'Family'
if category_id not in categories:
if ignore_promotions and '/store/apps/category/' not in url:
continue
categories[category_id] = {
'name': name,
'url': url,
'category_id': category_id}
return categories
def similar(self, app_id, detailed=False, **kwargs):
"""Sends a GET request, follows the redirect, and retrieves a list of
applications similar to the specified app.
:param app_id: app to retrieve details from, e.g. 'com.nintendo.zaaa'
:param detailed: if True, sends request per app for its full detail
:return: a list of similar apps
"""
url = build_url("similar", app_id)
response = send_request("GET",
url,
params=self.params,
allow_redirects=True)
soup = BeautifulSoup(response.content, "lxml", from_encoding="utf8")
if detailed:
apps = self._parse_multiple_apps(response)
else:
apps = [
parse_cluster_card_info(app)
for app in soup.select("div.Vpfmgd")
]
return apps
def reviews(self, app_id_list, page=0):
#print("In Scraper - reviews def \n")
reviews_adder = []
for n in range(len(app_id_list)):
#app_id=app_id_list[n]
#print(app_id)
data = {
'reviewType': 0,
'pageNum': page,
'id': app_id_list[n],
'reviewSortOrder': 4,
'xhr': 1,
'hl': self.language
}
self.params['authuser'] = '******'
#print('before send request')
#print(app_id_list[n])
response = send_request('POST', s.REVIEW_URL, data, self.params)
content = response.text
content = content[content.find('[["ecr"'):].strip()
data = json.loads(content)
#print(data)
html = data[0][2]
soup = BeautifulSoup(html, 'lxml', from_encoding='utf8')
#print(soup)
reviews = []
for element in soup.select('.single-review'):
#print('Inside single review')
review = {}
#print("In Scraper - reviews def- rev_app_id:: \n")
#print(app_id)
review['rev_app_id'] = app_id_list[n]
avatar_style = element.select_one('.author-image').get('style')
#print(avatar_style)
if avatar_style:
sheet = cssutils.css.CSSStyleSheet()
sheet.add('tmp { %s }' % avatar_style)
review['author_image'] = list(cssutils.getUrls(sheet))[0]
review_header = element.select_one('.review-header')
review['review_id'] = review_header.get('data-reviewid', '')
review['review_permalink'] = review_header.select_one(
'.reviews-permalink').get('href')
review['author_name'] = review_header.select_one(
'.author-name').text
review['review_date'] = review_header.select_one(
'.review-date').text
curr_rating = review_header.select_one('.current-rating').get(
'style')
review['current_rating'] = int(
int(
str(cssutils.parseStyle(curr_rating).width).replace(
'%', '')) / 20)
body_elem = element.select_one('.review-body')
review_title = body_elem.select_one('.review-title').extract()
body_elem.select_one('.review-link').decompose()
review['review_title'] = review_title.text
review['review_body'] = body_elem.text
reviews.append(review)
reviews_adder.append(review)
data = ''
return reviews_adder
def test_send_normal_request(self):
method = 'GET'
response = send_request(method, self.url)
self.assertEqual(response.status_code, 200)
self.assertEqual(response.url, self.url)
def test_send_normal_request(self):
method = "GET"
response = send_request(method, self.url)
self.assertEqual(response.status_code, 200)
self.assertEqual(response.url, self.url)
