wright-way-scraper

wright_way_scraper.WrightWayScraper

WrightWayScraper()

A client for the Petango API, specifically for the Wright-Way Rescue shelter.

Initialize the WrightWayScraper client with the base URL and authentication key.

Source code in wright_way_scraper/client.py

def __init__(self) -> None:
    """Initialize the WrightWayScraper client with the base URL and authentication key."""  # noqa: E501, W505
    self.base_url = "https://ws.petango.com"
    self.auth_key = "io53xfw8b0k2ocet3yb83666507n2168taf513lkxrqe681kf8"

get_animal

get_animal(id_: int | str) -> Animal

Parse the HTML for a single animal and return an Animal Pydantic model.

Parameters:

Name	Type	Description	Default
id_	int | str	The unique Petango identifier for the animal.	required

Raises:

Type	Description
TypeError	If any of the required fields are missing from the HTML.

Returns:

Name	Type	Description
Animal	Animal	A Pydantic model representing an animal available for adoption.

Source code in wright_way_scraper/client.py

def get_animal(self, id_: int | str) -> Animal:
    """Parse the HTML for a single animal and return an Animal Pydantic model.

    Args:
        id_ (int | str): The unique Petango identifier for the animal.

    Raises:
        TypeError: If any of the required fields are missing from the HTML.

    Returns:
        Animal: A Pydantic model representing an animal available for adoption.
    """
    tree = self._get_animal_xml(id_)
    labels = {
        "Name": "AnimalName",
        "Species": "Species",
        "Breed": "PrimaryBreed",
        "Birthdate": "DateOfBirth",
        "Gender": "Sex",
        "Size": "Size",
        "Color": "PrimaryColor",
        "Declawed": "Declawed",
        "Housetrained": "Housetrained",
        "Location": "Location",
        "Intake_Date": "LastIntakeDate",
        "Stage": "Stage",
        "Profile_URL": "Photo1",
    }
    mapping = {"id": id_, "URL": self.get_animal_html_url(id_)}
    for k, v in labels.items():
        elem = tree.find(v)
        if elem is None:
            msg = f"Could not find {v} for animal {id_}"
            raise TypeError(msg)
        mapping[k] = elem.text
    try:
        return Animal(**mapping)
    except ValidationError as e:
        msg = f"Could not validate Animal model for animal {id_}"
        raise ValueError(msg) from e

get_animal_html_url

get_animal_html_url(id_: int | str, auth_key: str | None = None) -> str

Get the URL for the Petango HTML page for a single animal.

This page is embedded in a frame on the Wright-Way Rescue website. The URL is constructed using the Petango ID of the animal.

Parameters:

Name	Type	Description	Default
id_	int | str	The Petango ID of the animal.	required
auth_key	str | None	The authentication key for the Petango API. If None, the client's default key is used. Defaults to None.	None

Returns:

Name	Type	Description
str	str	The URL for the Petango HTML page for the animal.

Source code in wright_way_scraper/client.py

def get_animal_html_url(self, id_: int | str, auth_key: str | None = None) -> str:
    """Get the URL for the Petango HTML page for a single animal.

    This page is embedded in a frame on the Wright-Way Rescue website. The URL is
    constructed using the Petango ID of the animal.

    Args:
        id_ (int | str): The Petango ID of the animal.
        auth_key (str | None, optional): The authentication key for the Petango API.
            If None, the client's default key is used. Defaults to None.

    Returns:
        str: The URL for the Petango HTML page for the animal.
    """
    if auth_key is None:
        auth_key = self.auth_key

    _url = "/webservices/adoptablesearch/wsAdoptableAnimalDetails.aspx?css=http://ws.petango.com/WebServices/adoptablesearch/css/styles.css"
    return f"{self.base_url}{_url}&id={id_}&authkey={auth_key}"

get_animal_ids

get_animal_ids() -> list[int]

Parse the HTML table of all animals and return a list of their Petango ids.

Returns:

Type	Description
list[int]	list[int]: A list of Petango ids for all animals available for adoption.

Source code in wright_way_scraper/client.py

def get_animal_ids(self) -> list[int]:
    """Parse the HTML table of all animals and return a list of their Petango ids.

    Returns:
        list[int]: A list of Petango ids for all animals available for adoption.
    """
    tree = self._get_animals_xml()
    ids = []
    for elem in tree:
        if elem.tag != "{http://www.petango.com/}XmlNode":
            msg = f"Expected '{{http://www.petango.com/}}XmlNode', got '{elem.tag}'"
            raise TypeError(msg)
        if (search_elem := elem.find("adoptableSearch")) is None:
            if elem.keys() == ["{http://www.w3.org/2001/XMLSchema-instance}nil"]:
                continue
            msg = "Could not find 'adoptableSearch' in the XmlNode"
            raise ValueError(msg)

        if (id_elem := search_elem.find("ID")) is None:
            msg = "Could not find 'ID' in the adoptableSearch XmlNode"
            raise ValueError(msg)

        if id_elem.text is None:
            msg = "Could not find text in the ID XmlNode"
            raise ValueError(msg)

        try:
            id_ = int(id_elem.text)
            ids.append(id_)
        except ValueError as e:
            msg = f"Could not cast '{id_elem.text}' to an integer"
            raise ValueError(msg) from e

    return ids

get_animal_xml_url

get_animal_xml_url(id_: int | str, auth_key: str | None = None) -> str

Get the URL for the Petango API to get the details of a single animal.

This URL is used to get the XML data for a single animal from the Petango API. It is not the URL for the HTML page for the animal.

Parameters:

Name	Type	Description	Default
id_	int | str	The Petango ID of the animal.	required
auth_key	str | None	The authentication key for the Petango API. If None, the client's default key is used. Defaults to None.	None

Returns:

Name	Type	Description
str	str	The URL for the Petango API to get the details of a single animal.

Source code in wright_way_scraper/client.py

def get_animal_xml_url(
    self,
    id_: int | str,
    auth_key: str | None = None,
) -> str:
    """Get the URL for the Petango API to get the details of a single animal.

    This URL is used to get the XML data for a single animal from the Petango API.
    It is not the URL for the HTML page for the animal.

    Args:
        id_ (int | str): The Petango ID of the animal.
        auth_key (str | None, optional): The authentication key for the Petango API.
            If None, the client's default key is used. Defaults to None.

    Returns:
        str: The URL for the Petango API to get the details of a single animal.
    """
    if auth_key is None:
        auth_key = self.auth_key

    # HTTP GET request to the Petango API
    # https://ws.petango.com/webservices/wsAdoption.asmx?op=AdoptableDetails
    _url = "/webservices/wsAdoption.asmx/AdoptableDetails"
    return f"{self.base_url}{_url}?animalID={id_}&authkey={auth_key}"

get_animals

get_animals(ids: list[int] | None = None) -> list[Animal]

Get all animals available for adoption.

Parameters:

Name	Type	Description	Default
ids	list[int] | None	A list of Petango ids for animals available for adoption. If None, the list will be generated by calling self.get_animal_ids().	None

Returns:

Type	Description
list[Animal]	list[Animal]: A list of Pydantic models representing animals available for adoption.

Source code in wright_way_scraper/client.py

def get_animals(self, ids: list[int] | None = None) -> list[Animal]:
    """Get all animals available for adoption.

    Args:
        ids (list[int] | None): A list of Petango ids for animals available for
            adoption. If None, the list will be generated by calling
            self.get_animal_ids().

    Returns:
        list[Animal]: A list of Pydantic models representing animals available for
            adoption.
    """
    if ids is None:
        ids = self.get_animal_ids()

    return [self.get_animal(id_) for id_ in ids]

get_animals_xml_url

get_animals_xml_url(auth_key: str | None = None) -> str

Get the URL for the Petango API to get all animals available for adoption.

This URL is used to get the XML data for all animals available for adoption from the Petango API. It is not the URL for the HTML page for the animals, as seen on the Wright-Way Rescue website.

Parameters:

Name	Type	Description	Default
auth_key	str | None	The authentication key for the Petango API. If None, the client's default key is used. Defaults to None.	None

Returns:

Name	Type	Description
str	str	The URL for the Petango API to get all animals available for adoption.

Source code in wright_way_scraper/client.py

def get_animals_xml_url(self, auth_key: str | None = None) -> str:
    """Get the URL for the Petango API to get all animals available for adoption.

    This URL is used to get the XML data for all animals available for adoption from
    the Petango API. It is not the URL for the HTML page for the animals, as seen on
    the Wright-Way Rescue website.

    Args:
        auth_key (str | None, optional): The authentication key for the Petango API.
            If None, the client's default key is used. Defaults to None.

    Returns:
        str: The URL for the Petango API to get all animals available for adoption.
    """
    if auth_key is None:
        auth_key = self.auth_key

    # HTTP GET request to the Petango API
    # https://ws.petango.com/webservices/wsAdoption.asmx?op=AdoptableSearch
    _url = "/webservices/wsAdoption.asmx/AdoptableSearch"
    _keys = [
        "speciesID",
        "sex",
        "ageGroup",
        "location",
        "site",
        "onHold",
        "orderBy",
        "primaryBreed",
        "secondaryBreed",
        "specialNeeds",
        "noDogs",
        "noCats",
        "noKids",
        "stageID",
    ]
    url = f"{self.base_url}{_url}?authkey={auth_key}"
    for key in _keys:
        url += f"&{key}="
    return url

wright_way_scraper.Animal

Bases: BaseModel

A model representing an animal available for adoption from the Wright-Way Rescue.

Attributes:

Name	Type	Description
id	int	The unique Petango identifier for the animal.
Name	str	The name of the animal.
Species	Literal['Cat', 'Dog']	The species of the animal.
Breed	str	The breed of the animal.
Birthdate	datetime | str | None	The birthdate of the animal.
Gender	Literal['Male', 'Female', 'Unknown']	The gender of the animal.
Size	str	The size of the animal.
Color	str	The color of the animal.
Declawed	Literal['Yes', 'No', 'Unknown']	Whether the animal is declawed.
Housetrained	Literal['Yes', 'No', 'Unknown']	Whether the animal is house-trained.
Location	str	The location of the animal.
Intake_Date	datetime | str	The date the animal was taken in by the shelter.
Stage	Literal['Reserved', 'Available']	The stage of the animal.

Raises:

Type	Description
TypeError	If the Intake_Date is not a datetime object or a string in the format 'mm/dd/yyyy'.
TypeError	If the Birthdate is not a datetime object or a string in the Petango API format.