survey-extraction/etl/hubSpotClient/hubspot.py
2025-07-31 14:29:03 +00:00

326 lines
13 KiB
Python

import hubspot
from enum import Enum
from hubspot.crm.deals import PublicObjectSearchRequest
from hubspot.crm.deals.models import SimplePublicObjectInput
from etl.hubSpotClient.types import SubmissionInfoFromDeal
import time
from pydantic import ValidationError
from etl.utils.logger import Logger
import logging
import traceback
from hubspot.crm.objects.notes import SimplePublicObjectInput as NoteInput
from hubspot.crm.associations import BatchInputPublicAssociation, PublicAssociation
import time
class DealStage(Enum):
SURVEYED_COMPLETE_NEEDS_SIGN_OFF = "1617223914"
SURVEYED_NO_ACCESS_NEED_SIGN_OFF = "1617223915"
CUSTOMER_CONTACTED = "888730834"
SURVEYED_COMPLETED_SIGNED_OFF = "1617223916"
FILES_MISSING_FROM_ASSESSOR = "1887736000"
class HubSpotClient():
def __init__(self):
self.access_token = "pat-eu1-064f7f5c-a7d8-4d93-a9b2-b604da6164a6"
self.client = hubspot.Client.create(access_token=self.access_token)
self.logger = Logger(name='HubSpotClient', level=logging.INFO).get_logger()
def get_all_deals(self):
return self.client.crm.deals.get_all()
def get_owner_name_from_id(self, owner_id):
owner = self.client.crm.owners.owners_api.get_by_id(owner_id)
time.sleep(1)
first_name = owner.first_name or ""
last_name = owner.last_name or ""
return f"{first_name} {last_name}".strip()
def get_deal_name_by_id(self, deal_id):
try:
deal = self.client.crm.deals.basic_api.get_by_id(deal_id)
time.sleep(1)
return deal.properties.get("dealname", "No deal name")
except Exception as e:
return "Unknown Deal" # Fallback if the deal name is not found
def get_listings_from_deals_id(self, deals_id):
from hubspot.crm.objects import PublicObjectSearchRequest
found_notes = []
after = None
while True:
# Correct filter for notes associated with the given deal ID
search_request = PublicObjectSearchRequest(
filter_groups=[{
"filters": [{
"propertyName": "associations.deal", # Filter by association to the deal
"operator": "EQ",
"value": deals_id,
}]
}],
properties=["domna_property_id", "owner_property_id", 'national_uprn'], # Properties of the note you need
limit=200,
after=after,
)
# Call the search API
response = self.client.crm.objects.search_api.do_search(object_type="0-420", public_object_search_request=search_request)
time.sleep(1)
# Add the results to the found_notes list
found_notes.extend(response.results)
# Handle pagination if more results are available
if not response.paging or not response.paging.next:
break
after = response.paging.next.after
if found_notes:
return found_notes[0]
return None
def get_domna_and_landlord_id(self, deals_id):
data = self.get_listings_from_deals_id(deals_id)
return data.properties['domna_property_id'], data.properties['owner_property_id'], data.properties.get('national_uprn', '') or ''
def get_notes_from_deals_id(self, deals_id):
from hubspot.crm.objects import PublicObjectSearchRequest
found_notes = []
after = None
while True:
# Correct filter for notes associated with the given deal ID
search_request = PublicObjectSearchRequest(
filter_groups=[{
"filters": [{
"propertyName": "associations.deal", # Filter by association to the deal
"operator": "EQ",
"value": deals_id,
}]
}],
properties=["hs_note_body", "hubspot_owner_id"], # Properties of the note you need
limit=200,
after=after,
)
# Call the search API
response = self.client.crm.objects.search_api.do_search(object_type="notes", public_object_search_request=search_request)
time.sleep(1)
# Add the results to the found_notes list
found_notes.extend(response.results)
# Handle pagination if more results are available
if not response.paging or not response.paging.next:
break
after = response.paging.next.after
all_notes = []
for note in found_notes:
# Extract note content and author information
note_body = note.properties.get("hs_note_body", "No content")
# Collect note details in a dictionary
all_notes.append({
"note_id": note.id,
"note": note_body,
"created_at": note.created_at.strftime("%Y-%m-%d %H:%M:%S"),
})
return all_notes
def get_all_deals_from_stage_id(self, stage_id):
found_deals = []
after = None
while True:
search_request = PublicObjectSearchRequest(
filter_groups=[{
"filters": [{
"propertyName": "dealstage",
"operator": "EQ",
"value": stage_id,
}]
}],
properties=[
"dealname",
"amount",
"hubspot_owner_id",
],
limit=200,
after=after,
)
response = self.client.crm.deals.search_api.do_search(search_request)
time.sleep(1)
found_deals.extend(response.results)
if not response.paging or not response.paging.next:
break
after = response.paging.next.after
all_deals = []
for deal in found_deals:
all_deals.append({
"deal_id": deal.id,
"value": deal.properties["amount"],
"deal_owner": deal.properties.get("hubspot_owner_id"),
})
return all_deals
def get_associations_for_deal(self, deal_id, to_object_type):
"""
Returns a list of associated object IDs of type `to_object_type`
(e.g. "contacts", "companies", "notes", etc.)
"""
assoc_resp = self.client.crm.deals.associations_api.get_all(
deal_id=deal_id,
to_object_type=to_object_type
)
return [assoc.id for assoc in assoc_resp.results]
def get_deals_from_deal_stage(self, deal_stage: DealStage):
found_deals = []
after = None
while True:
search_request = PublicObjectSearchRequest(
filter_groups=[{
"filters": [{
"propertyName": "dealstage",
"operator": "EQ",
"value": deal_stage.value,
}]
}],
properties=[
"dealname",
"number_of_wet_rooms_needing_ventilation",
"work_type",
"property_needs_trickle_vents",
"domna_survey_post_sap",
"existing_wall_insulation",
"installer",
"submission_folder",
],
limit=200,
after=after,
)
response = self.client.crm.deals.search_api.do_search(search_request)
found_deals.extend(response.results)
if not response.paging or not response.paging.next:
break
after = response.paging.next.after
all_deals = []
for i,deal in enumerate(found_deals):
domna_id, landlord_id, uprn = self.get_domna_and_landlord_id(deal.id)
try:
deal_name = deal.properties['dealname']
self.logger.info(f"Validating <{deal_name}>")
# input(f"Press enter to verfiy <{deal_name}>")
all_deals.append(SubmissionInfoFromDeal(
deal_id= deal.properties["hs_object_id"],
deal_name=deal.properties["dealname"],
work_type=deal.properties["work_type"],
needs_trickle_ventilation=True if deal.properties.get("property_needs_trickle_vents", "NO").upper() == "YES" else False,
post_sap_score=int(deal.properties["domna_survey_post_sap"]),
existing_wall_insulation=deal.properties.get("existing_wall_insulation") if deal.properties.get("existing_wall_insulation") else "None",
no_of_wet_rooms=int(deal.properties["number_of_wet_rooms_needing_ventilation"]),
installer=deal.properties["installer"],
submission_folder_path = deal.properties["submission_folder"],
landlord_id = landlord_id,
domna_id = domna_id,
uprn = uprn,
))
except Exception as e:
def format_error_note(e):
note_text = "⚠️ <b>Automated Verification Failed:</b><br><br>"
if hasattr(e, "errors") and callable(e.errors):
note_text += "❌ <b>Validation Errors:</b><br>"
for error in e.errors():
loc = error.get('loc', 'N/A')
msg = error.get('msg', 'N/A')
error_type = error.get('type', 'N/A')
error_input = error.get('input', 'N/A')
note_text += (
f"• <b>Field:</b> <code>{loc}</code><br>"
f"&nbsp;&nbsp;- <b>Message:</b> {msg}<br>"
f"&nbsp;&nbsp;- <b>Type:</b> {error_type}<br>"
f"&nbsp;&nbsp;- <b>Input:</b> {error_input}<br><br>"
)
else:
note_text += (
"❗ <b>Non-validation error:</b><br>"
f"<pre>{str(e)}</pre><br>"
)
note_text += (
"🛠️ Please review this error and take necessary actions.<br>"
"Contact <b>Jun-te</b> if help is needed: <b>+44 7519 530 549</b> or via Teams."
)
return note_text
deal_id = deal.properties['hs_object_id']
if hasattr(e, "errors"):
for error in e.errors():
self.add_note_to_deal(deal_id, format_error_note(e))
else:
self.logger.error(f"Non-validation error occurred: {str(e)}", exc_info=True)
self.logger.info(f"Deal name <{deal_name}> moving to 'needs additional information'")
self.move_deals_to_different_stage([deal_id], DealStage.FILES_MISSING_FROM_ASSESSOR.value)
return all_deals
def print_all_pipeline_ids(self):
pipelines = self.client.crm.pipelines.pipelines_api.get_all(object_type="deals")
for pipeline in pipelines.results:
print(f"Pipeline: {pipeline.label}")
for stage in pipeline.stages:
print(f" - Label: {stage.label}")
print(f" ID: {stage.id}")
def move_deals_to_different_stage(self, list_of_deals_id, to_stage_id):
deal_properties = SimplePublicObjectInput(
properties={
"dealstage": to_stage_id
}
)
for deal_id in list_of_deals_id:
self.client.crm.deals.basic_api.update(
deal_id,
simple_public_object_input=deal_properties
)
self.logger.info(f"Deal {deal_id} moved to stage with ID {to_stage_id}.")
def add_note_to_deal(self, deal_id, note_text):
try:
# Generate current time in milliseconds since epoch
hs_timestamp = int(time.time() * 1000)
# Step 1: Create the note with hs_timestamp
note = NoteInput(
properties={
"hs_note_body": note_text,
"hs_timestamp": hs_timestamp # Required field in your HubSpot setup
}
)
created_note = self.client.crm.objects.notes.basic_api.create(note)
note_id = created_note.id
# Step 2: Associate the note to the deal
association = PublicAssociation(
_from=note_id,
to=deal_id,
type="note_to_deal"
)
self.client.crm.associations.batch_api.create(
'notes',
'deals',
batch_input_public_association=BatchInputPublicAssociation(
inputs=[association]
)
)
self.logger.info(f"📝 Note added to deal {deal_id}: {note_text}")
except Exception as e:
self.logger.error(f"❌ Failed to add note to deal {deal_id}: {e}", exc_info=True)