def _get_contacts(site: dict) -> Optional[List[schema.Contact]]:
contacts = []
if site["attributes"]["SitePhone"]:
for phone in normalize_phone(site["attributes"]["SitePhone"]):
contacts.append(phone)
# Contacts seems to be a free text field where people usually enter emails but also sometimes
# other stuff like numbers, hours of operation, etc
if site["attributes"]["Contact"]:
if "@" in site["attributes"]["Contact"]:
contacts.append(
schema.Contact(contact_type="general",
email=site["attributes"]["Contact"]))
else:
contacts.append(
schema.Contact(contact_type="general",
other=site["attributes"]["Contact"]))
url = site["attributes"]["URL"]
if url:
url = url if "http" in url else "https://" + url
URL_RE = re.compile(
r"^((https?):\/\/)(www.)?[a-z0-9]+\.[a-z]+(\/?[a-zA-Z0-9#]+\/?)*$")
valid = URL_RE.match(url)
if valid:
contacts.append(schema.Contact(contact_type="general",
website=url))
if len(contacts) > 0:
return contacts
return None
def _get_contacts(site: dict) -> Optional[List[schema.Contact]]:
contacts = []
if site["attributes"]["Site_Phone"]:
for phone in normalize_phone(site["attributes"]["Site_Phone"]):
contacts.append(phone)
if site["attributes"]["Site_Zotec_Link"]:
contacts.append(
schema.Contact(website=site["attributes"]["Site_Zotec_Link"]))
elif site["attributes"]["Promote_Name"]:
# Sometimes Promote_Name also contains URLs. These are probably worse
# than Site_Zotec_Link, but if they're all that we have we mine as
# well use them
promote_name = site["attributes"]["Promote_Name"]
# Copied from SO: https://stackoverflow.com/questions/3809401/what-is-a-good-regular-expression-to-match-a-url
promote_url_match = re.search(
"https?://(www\\.)?[-a-zA-Z0-9@:%._+~#=]{1,256}\\.[a-zA-Z0-9()]{1,6}\\b([-a-zA-Z0-9()@:%_+.~#?&/=]*)",
promote_name,
)
if promote_url_match:
contacts.append(schema.Contact(website=promote_url_match.string))
if site["attributes"]["Site_Location_Info"]:
contacts.append(
schema.Contact(other=site["attributes"]["Site_Location_Info"]))
if len(contacts) > 0:
return contacts
return None
def _get_contact(site: dict) -> List[schema.Contact]:
contacts = []
for phone in normalize_phone(site["Description"], "general"):
contacts.append(phone)
if contacts:
return contacts
return None
def add_phone(self, phone: Text) -> None:
"""Adds the given phone number to the current site."""
# It's ok to have multiple phone numbers,
# so no need to start fresh if we have a phone number already.
site = self._current_site
site.contact = site.contact or []
site.contact.extend(normalize_phone(phone, contact_type="booking"))
def _get_contacts(site: dict) -> Optional[List[schema.Contact]]:
contacts = []
if site["attributes"]["USER_Contact_Phone"]:
for phone in normalize_phone(
site["attributes"]["USER_Contact_Phone"], contact_type="general"
):
contacts.append(phone)
if site["attributes"]["USER_Contact_Email"]:
email = site["attributes"]["USER_Contact_Email"].replace(" ", "")
if "." not in email:
return
if "/" in email:
split_email = email.split(" / ")
if len(split_email) == 1:
split_email = email.split("/")
if len(split_email) == 1:
return
email = split_email[0]
contacts.append(schema.Contact(contact_type="general", email=email))
if site["attributes"]["USER_Contact_Website"]:
contacts.append(
schema.Contact(
contact_type="general",
website=site["attributes"]["USER_Contact_Website"],
)
)
if len(contacts) > 0:
return contacts
return None
def _get_contacts(site: dict) -> Optional[List[schema.Contact]]:
contacts = []
phones = normalize_phone(site["attributes"]["phone"])
contacts.extend(phones)
website = normalize_url(site["attributes"]["website"])
if website is not None:
contacts.append(
schema.Contact(website=website,
contact_type=schema.ContactType.GENERAL))
if len(contacts) > 0:
return contacts
return None
def _get_contacts(site: dict) -> Optional[List[schema.Contact]]:
contacts = []
if site["attributes"]["phone"]:
for phone in normalize_phone(site["attributes"]["phone"]):
contacts.append(phone)
if site["attributes"]["email"]:
contacts.append(schema.Contact(email=site["attributes"]["email"]))
if site["attributes"]["agencyurl"]:
contacts.append(
schema.Contact(website=site["attributes"]["agencyurl"]))
if len(contacts):
return contacts
return None
def _get_contacts(site: dict) -> Optional[List[schema.Contact]]:
contacts = []
if site["attributes"]["USER_Scheduling_by_Phone"]:
for phone in normalize_phone(
site["attributes"]["USER_Scheduling_by_Phone"]):
contacts.append(phone)
if site["attributes"]["USER_Link_to_Sign_Up"]:
url = site["attributes"]["USER_Link_to_Sign_Up"].strip()
if url is not None and url != "\x08" and url != "-":
url = normalize_url(url)
contacts.append(schema.Contact(website=url))
if len(contacts) > 0:
return contacts
return None
def _get_contacts(site: dict) -> Optional[List[schema.Contact]]:
contacts = []
if site["attributes"]["Phone_Number"]:
for phone in normalize_phone(site["attributes"]["Phone_Number"]):
contacts.append(phone)
# if site["attributes"]["publicEmail"]:
# contacts.append(schema.Contact(email=site["attributes"]["publicEmail"]))
website_contact = _get_website(site)
if website_contact:
contacts.append(website_contact)
if len(contacts) > 0:
return contacts
return None
def _get_contacts(site: dict) -> Optional[List[schema.Contact]]:
location_url = _get_filter(site, "locationUrl", "")
contacts = []
if site["phoneNumber"]:
for phone in normalize_phone(site["phoneNumber"]):
contacts.append(phone)
def cleanup_url(url):
if not url or not url.strip():
return None
if url in ["No website", "Website link", "Booking Registration link"]:
return None
if "@" in url:
# Some of these are email addresses.
# Skipping those for now.
return None
url = re.sub(r"^(http|https):/(\w+)", r"\1://\2", url)
url = re.sub(r"^www.https://", "https://", url)
url = re.sub(r"^https//", "https://", url)
url = re.sub(r"^https:(\w+)", r"https://\1", url)
url = re.sub(r"^https://wynne apothecary.com",
"https://wynneapothecary.com", url)
# workaround until samuelcolvin/pydantic#2778 is merged
url = url.rstrip("#")
if not url.startswith("http"):
url = "http://" + url
return url
if location_url == "healthyguilford.com, conehealth.com/vaccine":
contacts.append(
schema.Contact.create(website="http://healthyguilford.com"))
contacts.append(
schema.Contact.create(website="https://conehealth.com/vaccine"))
# A few sites have "locationUrl" set to something like this:
# `https://myvaccine.fl.gov/ or 866-201-6313`
elif match := re.match(r"^(https://\S+) or (\d\d\d-\d\d\d-\d\d\d\d)$",
location_url):
contacts.append(schema.Contact.create(phone=match[2]))
contacts.append(schema.Contact.create(website=match[1]))
def _get_contacts(site: dict) -> Optional[List[schema.Contact]]:
contacts = []
if "contact-phone" in site:
for phone in normalize_phone(site["contact-phone"],
contact_type="general"):
contacts.append(phone)
# Filter out sites with a url of "/"
if site["url"].startswith("http"):
web_contact = schema.Contact(
contact_type="general",
website=site["url"],
)
contacts.append(web_contact)
if contacts == []:
return None
return contacts
def _get_contacts(site: dict) -> Optional[List[schema.Contact]]:
contacts = []
if site["attributes"]["phone"]:
for phone in normalize_phone(site["attributes"]["phone"]):
contacts.append(phone)
# if site["attributes"]["publicEmail"]:
# contacts.append(schema.Contact(email=site["attributes"]["publicEmail"]))
# there are multiple urls, vaccine, agency, health dept. etc
if site["attributes"]["vaccine_url"]:
url = site["attributes"]["vaccine_url"]
url = sanitize_url(url)
if url:
contacts.append(schema.Contact(website=url))
if len(contacts) > 0:
return contacts
return None
def _get_contacts(site: dict) -> Optional[List[schema.Contact]]:
contacts = []
phones = normalize_phone(site["attributes"]["PublicPhone"])
for phone in phones:
phone = _phone_fixup(phone)
if phone is not None:
contacts.append(
schema.Contact(phone=phone,
contact_type=schema.ContactType.GENERAL))
website = _website_fixup(site["attributes"]["WEBSITE"])
if website is not None:
contacts.append(
schema.Contact(website=website,
contact_type=schema.ContactType.GENERAL))
if len(contacts) > 0:
return contacts
return None
def _get_contacts(site: dict):
ret = []
if site["Appointment Phone"]:
for phone in normalize_phone(site["Appointment Phone"]):
ret.append(phone)
url = site["Web Address"]
# Some URLs have multiple schemes.
valid_url = re.match(r"(https?:\/\/)*(.+)", url)
if (url == "http://" or url == "https://" or url == "none" or url == ""
or url.startswith("Please email")):
return ret
elif valid_url is not None:
if valid_url.group(1) is None:
url = valid_url.group(2)
else:
url = f"{valid_url.group(1)}{valid_url.group(2)}"
url = normalize_url(url)
ret.append(schema.Contact(website=url))
else:
logger.warning(f"Unknown, invalid URL: {url}")
return ret
def test_normalize_phone():
assert normalize_phone("") == []
assert normalize_phone("abc") == []
assert normalize_phone("1234") == []
assert normalize_phone("1234567890") == [
] # Not a valid phone-like number.
assert normalize_phone("212 555 1212") == ["(212) 555-1212"]
assert normalize_phone("(212) 555 1212") == ["(212) 555-1212"]
assert normalize_phone("212 555 1212 x17") == ["(212) 555-1212 ext. 17"]
assert normalize_phone("212 555 1212 OPTION 17") == [
"(212) 555-1212 ext. 17"
]
assert normalize_phone("212 555 1212, option 17") == [
"(212) 555-1212 ext. 17"
]
assert normalize_phone("212 555 1212 PRESS 17") == [
"(212) 555-1212 ext. 17"
]
assert normalize_phone("212 555 1212, press 17 to schedule") == [
"(212) 555-1212 ext. 17"
]
assert normalize_phone("212 555 1212 / 212 555 1213") == [
"(212) 555-1212",
"(212) 555-1213",
]
def test_normalize_phone():
assert normalize_phone("") == []
assert normalize_phone("abc") == []
assert normalize_phone("1234") == []
assert normalize_phone("1234567890") == [] # Not a valid phone-like number.
assert normalize_phone("212 555 1212") == [schema.Contact(phone="(212) 555-1212")]
assert normalize_phone("(212) 555 1212") == [schema.Contact(phone="(212) 555-1212")]
assert normalize_phone("212 555 1212 ext17") == [
schema.Contact(phone="(212) 555-1212 ext. 17")
]
assert normalize_phone("212 555 1212 x17") == [
schema.Contact(phone="(212) 555-1212 ext. 17")
]
assert normalize_phone("212 555 1212 OPTION 17") == [
schema.Contact(phone="(212) 555-1212 ext. 17")
]
assert normalize_phone("212 555 1212, option 17") == [
schema.Contact(phone="(212) 555-1212 ext. 17")
]
assert normalize_phone("212 555 1212 PRESS 17") == [
schema.Contact(phone="(212) 555-1212 ext. 17")
]
assert normalize_phone("212 555 1212 press #18") == [
schema.Contact(phone="(212) 555-1212 ext. 18")
]
assert normalize_phone("212 555 1212, press 17 to schedule") == [
schema.Contact(phone="(212) 555-1212 ext. 17")
]
assert normalize_phone("212 555 1212 / 212 555 1213") == [
schema.Contact(phone="(212) 555-1212"),
schema.Contact(phone="(212) 555-1213"),
]
def _get_contacts(site: dict) -> Optional[List[schema.Contact]]:
ret = []
if phone := site["attributes"]["phone"]:
ret.extend(normalize_phone(phone))
