PHDMAFIN commited on
Commit
5d2fdf4
·
verified ·
1 Parent(s): 5eb85c5

Update scraper.py

Browse files
Files changed (1) hide show
  1. scraper.py +11 -4
scraper.py CHANGED
@@ -384,14 +384,21 @@ def sanitize_filename(stem: str) -> str:
384
  return re.sub(r"[^a-zA-Z0-9_\-\.]+", "_", stem).strip("_") or "consulta"
385
 
386
 
387
- def normalize_input_to_url(s: str | None, lang: str = DEFAULT_LANG) -> str | None:
388
  s = (str(s or "")).strip()
389
- if not s or s.lower() == "nan":
390
  return None
 
391
  if s.startswith("http://") or s.startswith("https://"):
392
- if not s.endswith("/"):
393
- s += "/"
394
  return s
 
 
 
 
 
 
395
  s = s.strip("/")
396
  return f"{BASE}/{lang}/arrendar-casas/{s}/"
397
 
 
384
  return re.sub(r"[^a-zA-Z0-9_\-\.]+", "_", stem).strip("_") or "consulta"
385
 
386
 
387
+ def normalize_input_to_url(s):
388
  s = (str(s or "")).strip()
389
+ if not s:
390
  return None
391
+
392
  if s.startswith("http://") or s.startswith("https://"):
393
+ # No agregar "/" si la URL tiene query string (?shape=..., filtros, etc.).
394
+ # Agregar "/" después del query altera el valor del parámetro y puede romper Idealista.
395
  return s
396
+
397
+ s = s.strip("/")
398
+ return f"{BASE}/{LANG}/arrendar-casas/{s}/"
399
+
400
+ s = s.strip("/")
401
+ return f"{BASE}/{LANG}/arrendar-casas/{s}/"
402
  s = s.strip("/")
403
  return f"{BASE}/{lang}/arrendar-casas/{s}/"
404