Normalize image galleries
This commit is contained in:
@@ -114,8 +114,11 @@ _COLOR_MAP = {
|
||||
}
|
||||
|
||||
_IMAGE_FIELD_HINTS = ["image", "images", "media", "gallery", "photo", "pic", "picture", "url", "src", "uri", "ref"]
|
||||
_IMAGE_URL_MARKERS = ["img.classistatic.de", "/images/", "/image/", "jpg", "jpeg", "png", "gif", "bmp", "tiff", "webp"]
|
||||
_MOBILEDE_IMAGE_RULE = "mo-640.jpg"
|
||||
_MOBILEDE_IMAGE_HOST = "img.classistatic.de"
|
||||
_MOBILEDE_IMAGE_PATH_PREFIX = "/api/v1/mo-prod/images/"
|
||||
_MOBILEDE_FULLRES_IMAGE_RULE = "mo-1600"
|
||||
_MOBILEDE_PREVIEW_IMAGE_RULE = "mo-360"
|
||||
_MOBILEDE_IMAGE_ID_RE = re.compile(r"^[0-9a-f]{2}/[0-9a-f-]{8,}$", re.IGNORECASE)
|
||||
_COLOR_VALUE_KEYS = {"ecol", "color", "exteriorcolor", "manufacturercolorname", "vehiclecolor", "paint"}
|
||||
_COUNTRY_VALUE_KEYS = {"country", "countrycode"}
|
||||
_BODY_VALUE_KEYS = {"category", "bodytype", "vehiclecategory", "body"}
|
||||
@@ -633,10 +636,20 @@ class MobileDeMapper:
|
||||
|
||||
@staticmethod
|
||||
def _images_from_listing(raw: dict[str, Any]) -> list[ImageRecord]:
|
||||
urls = MobileDeMapper._extract_image_urls(raw)
|
||||
source_urls = dict.fromkeys(MobileDeMapper._extract_image_urls(raw))
|
||||
return [
|
||||
ImageRecord(fullres_image=url, preview_image=url, order_index=index)
|
||||
for index, url in enumerate(dict.fromkeys(url for url in urls if url))
|
||||
ImageRecord(
|
||||
fullres_image=MobileDeMapper._with_mobilede_image_rule(
|
||||
source_url,
|
||||
_MOBILEDE_FULLRES_IMAGE_RULE,
|
||||
),
|
||||
preview_image=MobileDeMapper._with_mobilede_image_rule(
|
||||
source_url,
|
||||
_MOBILEDE_PREVIEW_IMAGE_RULE,
|
||||
),
|
||||
order_index=index,
|
||||
)
|
||||
for index, source_url in enumerate(source_urls)
|
||||
]
|
||||
|
||||
@staticmethod
|
||||
@@ -648,7 +661,7 @@ class MobileDeMapper:
|
||||
if isinstance(value, str):
|
||||
if parent_hint or MobileDeMapper._looks_like_image_url(value):
|
||||
normalized = MobileDeMapper._normalize_image_url(value)
|
||||
if MobileDeMapper._looks_like_image_url(normalized):
|
||||
if normalized:
|
||||
urls.append(normalized)
|
||||
return urls
|
||||
if isinstance(value, list):
|
||||
@@ -669,12 +682,7 @@ class MobileDeMapper:
|
||||
|
||||
@staticmethod
|
||||
def _looks_like_image_url(value: str) -> bool:
|
||||
text = str(value or "").strip().lower()
|
||||
if not text:
|
||||
return False
|
||||
if not (text.startswith("http://") or text.startswith("https://") or text.startswith("//") or text.startswith("/")):
|
||||
return False
|
||||
return any(marker in text for marker in _IMAGE_URL_MARKERS)
|
||||
return bool(MobileDeMapper._normalize_image_url(value))
|
||||
|
||||
@staticmethod
|
||||
def _normalize_image_url(value: str) -> str:
|
||||
@@ -683,19 +691,35 @@ class MobileDeMapper:
|
||||
return ""
|
||||
if url.startswith("//"):
|
||||
url = f"https:{url}"
|
||||
elif url.startswith(_MOBILEDE_IMAGE_HOST):
|
||||
url = f"https://{url}"
|
||||
elif not (url.startswith("http://") or url.startswith("https://")):
|
||||
url = f"https://{url.lstrip('/')}"
|
||||
return MobileDeMapper._normalize_mobilede_image_rule(url)
|
||||
return ""
|
||||
|
||||
parsed = urlsplit(url)
|
||||
if parsed.netloc.lower() != _MOBILEDE_IMAGE_HOST:
|
||||
return ""
|
||||
if not parsed.path.startswith(_MOBILEDE_IMAGE_PATH_PREFIX):
|
||||
return ""
|
||||
|
||||
image_id = parsed.path[len(_MOBILEDE_IMAGE_PATH_PREFIX):].strip("/")
|
||||
if not _MOBILEDE_IMAGE_ID_RE.fullmatch(image_id):
|
||||
return ""
|
||||
|
||||
query_pairs = [
|
||||
(key, query_value)
|
||||
for key, query_value in parse_qsl(parsed.query, keep_blank_values=True)
|
||||
if key.lower() != "rule"
|
||||
]
|
||||
return urlunsplit(("https", _MOBILEDE_IMAGE_HOST, parsed.path, urlencode(query_pairs), ""))
|
||||
|
||||
@staticmethod
|
||||
def _normalize_mobilede_image_rule(url: str) -> str:
|
||||
def _with_mobilede_image_rule(url: str, rule: str) -> str:
|
||||
parsed = urlsplit(url)
|
||||
if "img.classistatic.de" not in parsed.netloc.lower():
|
||||
return url
|
||||
if "/api/v1/mo-prod/images/" not in parsed.path:
|
||||
return url
|
||||
query_pairs = parse_qsl(parsed.query, keep_blank_values=True)
|
||||
if any(key.lower() == "rule" for key, _value in query_pairs):
|
||||
return url
|
||||
query_pairs.append(("rule", _MOBILEDE_IMAGE_RULE))
|
||||
query_pairs = [
|
||||
(key, query_value)
|
||||
for key, query_value in parse_qsl(parsed.query, keep_blank_values=True)
|
||||
if key.lower() != "rule"
|
||||
]
|
||||
query_pairs.append(("rule", rule))
|
||||
return urlunsplit((parsed.scheme, parsed.netloc, parsed.path, urlencode(query_pairs), parsed.fragment))
|
||||
|
||||
Reference in New Issue
Block a user