sidski/helseveileder_webscraper
Archived
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

🚚 rename variables to same language

Browse source 
Co-authored-by: haraldnilsen <harald_998@hotmail.com>
Signed-off-by: Sindre Kjelsrud <kjelsrudsindre@gmail.com>
This commit is contained in:
Sindre Kjelsrud 2024-01-08 14:54:36 +01:00
parent 4bcbf8426b
commit 1a1b6c9737
Signed by untrusted user who does not match committer: sidski
GPG key ID: D2BBDF3EDE6BA9A6
1 changed files with 7 additions and 7 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

14
main.py
View file

@ -46,11 +46,11 @@ def fetch_all_info(url):
response = httpx.get(url, headers=headers) response = httpx.get(url, headers=headers)
html = HTMLParser(response.text) html = HTMLParser(response.text)
# Extracting the question "sporsmal" # Extracting the question
sporsmal = html.css_first('div.article-text').text() questions = html.css_first('div.article-text').text()
# Extracting the answer "svar" and the signature "signature" # Extracting the answer and the signature
svar = "" answers = ""
signature = "" signature = ""
specific_div = html.css_first('.article-text.font-serif.text-base.py-10') specific_div = html.css_first('.article-text.font-serif.text-base.py-10')
if specific_div: if specific_div:
@ -61,16 +61,16 @@ def fetch_all_info(url):
signature_match = re.search(r'(Vennlig hilsen|Med vennlig hilsen|Mvh|Lykke til!)\s*(.*)', full_text, re.IGNORECASE) signature_match = re.search(r'(Vennlig hilsen|Med vennlig hilsen|Mvh|Lykke til!)\s*(.*)', full_text, re.IGNORECASE)
if signature_match: if signature_match:
signature = signature_match.group(2) # This is the signature text signature = signature_match.group(2) # This is the signature text
svar = full_text[:signature_match.start()] # This is the text before the signature answers = full_text[:signature_match.start()] # This is the text before the signature
else: else:
svar = full_text # In case there is no signature answers = full_text # In case there is no signature
# Clean up the signature # Clean up the signature
cleaned_signature = re.sub(r'(Vennlig hilsen|Med vennlig hilsen|Mvh|Lykke til!)\s*', '', signature, flags=re.IGNORECASE) cleaned_signature = re.sub(r'(Vennlig hilsen|Med vennlig hilsen|Mvh|Lykke til!)\s*', '', signature, flags=re.IGNORECASE)
cleaned_signature = re.sub(r'^[\s,]*', '', cleaned_signature).strip() # Remove leading commas and spaces cleaned_signature = re.sub(r'^[\s,]*', '', cleaned_signature).strip() # Remove leading commas and spaces
# Create array of results # Create array of results
result = [sporsmal, svar, cleaned_signature] result = [questions, answers, cleaned_signature]
return result return result