mirror of
https://github.com/Bunsly/JobSpy.git
synced 2026-03-05 03:54:31 -08:00
Get full indeed description (#70)
This commit is contained in:
@@ -121,7 +121,7 @@ class Country(Enum):
|
||||
# internal for ziprecruiter
|
||||
US_CANADA = ("usa/ca", "www")
|
||||
|
||||
# internal for linkeind
|
||||
# internal for linkedin
|
||||
WORLDWIDE = ("worldwide", "www")
|
||||
|
||||
@property
|
||||
|
||||
@@ -235,24 +235,9 @@ class IndeedScraper(Scraper):
|
||||
if response.status_code not in range(200, 400):
|
||||
return None
|
||||
|
||||
soup = BeautifulSoup(response.text, "html.parser")
|
||||
script_tag = soup.find(
|
||||
"script", text=lambda x: x and "window._initialData" in x
|
||||
)
|
||||
|
||||
if not script_tag:
|
||||
return None
|
||||
|
||||
script_code = script_tag.string
|
||||
match = re.search(r"window\._initialData\s*=\s*({.*?})\s*;", script_code, re.S)
|
||||
|
||||
if not match:
|
||||
return None
|
||||
|
||||
json_string = match.group(1)
|
||||
data = json.loads(json_string)
|
||||
try:
|
||||
job_description = data["jobInfoWrapperModel"]["jobInfoModel"][
|
||||
data = json.loads(response.text)
|
||||
job_description = data["body"]["jobInfoWrapperModel"]["jobInfoModel"][
|
||||
"sanitizedJobDescription"
|
||||
]
|
||||
except (KeyError, TypeError, IndexError):
|
||||
|
||||
Reference in New Issue
Block a user