feat: enhance AI parsing and prompts with company details integration

transaction_parser/parser_benchmark/runner.py

@@ -156,6 +156,7 @@ def _run_ai_parsing(self, file_content: str, file_name: str) -> dict:
             document_schema=self.controller.get_schema(),
             document_data=file_content,
             file_doc_name=file_name,
+            company=self.dataset.company,
         )
         self.log.ai_parse_time = flt(default_timer() - start, self.precision)
 

transaction_parser/public/js/transaction_parser_dialog.js

@@ -52,6 +52,17 @@ async function create_transaction_parser_dialog(transaction_type, list_view) {
 					default: get_default_country(),
 					reqd: 1,
 				},
+				{
+					fieldtype: "Section Break",
+				},
+				{
+					fieldname: "company",
+					label: __("Company"),
+					fieldtype: "Link",
+					options: "Company",
+					default: frappe.defaults.get_user_default("Company"),
+					reqd: 1,
+				},
 			],
 			primary_action_label: __("Submit"),
 			primary_action(values) {

transaction_parser/transaction_parser/__init__.py

@@ -15,7 +15,7 @@
 
 
 @frappe.whitelist()
-def parse(transaction, country, file_url, ai_model=None, page_limit=None):
+def parse(transaction, country, file_url, ai_model=None, page_limit=None, company=None):
     is_enabled()
 
     frappe.has_permission(TRANSACTION_MAP[transaction], "create", throw=True)
@@ -27,6 +27,7 @@ def parse(transaction, country, file_url, ai_model=None, page_limit=None):
         file_url=cstr(file_url),
         ai_model=cstr(ai_model),
         page_limit=cint(page_limit),
+        company=cstr(company) if company else None,
         queue="long",
         now=frappe.conf.developer_mode,
     )

transaction_parser/transaction_parser/ai_integration/parser.py

@@ -36,19 +36,27 @@ def parse(
         document_schema: dict,
         document_data: str,
         file_doc_name: str | None = None,
+        company: str | None = None,
     ) -> dict:
-        messages = self._build_messages(document_type, document_schema, document_data)
+        messages = self._build_messages(
+            document_type, document_schema, document_data, company
+        )
         self.ai_response = self.send_message(
             messages=messages, file_doc_name=file_doc_name
         )
         return self.get_content(self.ai_response)
 
     def _build_messages(
-        self, document_type: str, document_schema: dict, document_data: str
+        self,
+        document_type: str,
+        document_schema: dict,
+        document_data: str,
+        company: str | None = None,
     ) -> tuple:
         """Build the message structure for AI API call."""
+        company_info = self._get_company_info(company) if company else ""
         system_prompt = get_system_prompt(document_schema)
-        user_prompt = get_user_prompt(document_type, document_data)
+        user_prompt = get_user_prompt(document_type, document_data, company_info)
 
         return (
             {
@@ -61,6 +69,23 @@ def _build_messages(
             },
         )
 
+    @staticmethod
+    def _get_company_info(company: str) -> str:
+        """Build a company context string with name and address if available."""
+        from frappe.contacts.doctype.address.address import get_company_address
+        from frappe.utils import strip_html
+
+        info = f"Company: {company}"
+
+        address = get_company_address(company)
+        if address and address.company_address_display:
+            address_text = strip_html(address.company_address_display).strip()
+
+            if address_text:
+                info += f"\nLocated at: {address_text}"
+
+        return info
+
     def send_message(self, messages: tuple, file_doc_name: str | None = None) -> dict:
         """Send messages to AI API and handle the response."""
         log = self._create_log_entry(file_doc_name)

transaction_parser/transaction_parser/ai_integration/prompts.py

@@ -34,10 +34,27 @@ def get_system_prompt(document_schema: dict) -> str:
 {document_schema}"""
 
 
-def get_user_prompt(document_type: str, document_data: str) -> str:
+def get_user_prompt(
+    document_type: str, document_data: str, company_info: str = ""
+) -> str:
     input_doc_type = INPUT_DOCUMENTS.get(document_type, "document")
 
-    return f"""Generate {document_type} for given {input_doc_type} according to above JSON schema.
+    company_context = ""
+    if company_info:
+        if document_type == "Sales Order":
+            role_hint = "Use this to correctly identify the company as the seller/vendor and the other party as the customer/buyer."
+        else:
+            role_hint = "Use this to correctly identify the company as the buyer/recipient and the other party as the vendor/supplier."
+
+        company_context = f"""
+
+This {input_doc_type} is received by the following company:
+{company_info}
+
+{role_hint}
+"""
+
+    return f"""Generate {document_type} for the given {input_doc_type} according to above JSON schema.{company_context}
 Document data is given below:
 {document_data}"""
 

transaction_parser/transaction_parser/controllers/transaction.py

@@ -85,6 +85,7 @@ def _parse_file_content(
             document_schema=schema,
             document_data=content,
             file_doc_name=self.file.name,
+            company=self.company,
         )
 
     ###################################

transaction_parser/transaction_parser/doctype/transaction_parser_party_email/transaction_parser_party_email.json

@@ -50,4 +50,4 @@
  "sort_field": "modified",
  "sort_order": "DESC",
  "states": []
-}
+}

transaction_parser/transaction_parser/overrides/communication.py

@@ -13,6 +13,18 @@ def on_update(doc, method=None):
     if not (settings.enabled and settings.parse_incoming_emails):
         return
 
+    matched_account = next(
+        (
+            row
+            for row in settings.incoming_email_accounts
+            if row.to_email in doc.recipients
+        ),
+        None,
+    )
+
+    if not matched_account:
+        return
+
     if settings.parse_party_emails:
         matched_party_config = next(
             (row for row in settings.party_emails if row.party_email == doc.sender),
@@ -41,21 +53,10 @@ def on_update(doc, method=None):
                 settings,
                 default_user,
                 matched_party_config.party,
+                matched_account.company,
             )
             return
 
-    matched_account = next(
-        (
-            row
-            for row in settings.incoming_email_accounts
-            if row.to_email in doc.recipients
-        ),
-        None,
-    )
-
-    if not matched_account:
-        return
-
     # Attachments are not available when the Communication doc is created.
     # Next time the doc is updated, we will check for attachments,
     # and update the flag `is_processed_by_transaction_parser` accordingly.