[FIX] account_statement_import_file: avoid duplicate and slow statement imports

[feg-adhoc]

Problem

Importing a bank statement file can create duplicate bank statements (the same file imported several times). From the user's point of view the import seems to "loop" and never confirm — even though they clicked Import only once.

Root cause

This is a chain of events; the duplication is the important part:

1. Creating the statement can be slow. Two contributors:
   • Odoo core assigns a sequence/name to every account.move created for the statement lines (per-move work that also grows with the journal history).
   • When account_accountant (Enterprise) is installed, account.bank.statement.create() synchronously renders a full PDF of the statement through wkhtmltopdf on every create, unless the context flag skip_pdf_attachment_generation is set. This is very slow and memory-heavy for large statements.
2. If the request takes longer than the reverse proxy / ingress timeout (deployment-specific, outside Odoo's control), the proxy drops the connection.
3. The Odoo web client treats the dropped connection as "connection lost" and automatically re-sends the same RPC when it reconnects.
4. import_file_button is not idempotent: each re-sent request re-runs the whole import and creates another statement. Sheet-based mappings without a unique_import_id column have no per-line dedup, so nothing stops the duplicates.

Net effect: a single user click produces several identical statements and a "looping" UI.

Fix

Two changes, both scoped to the import:

1. Skip the Enterprise synchronous PDF render during import — set skip_pdf_attachment_generation in the statement creation context. The imported source file is already attached to the statement, and the PDF can still be generated manually afterwards. No dependency on account_accountant is added (it is a no-op when that module is absent).

2. Make import_file_button idempotent:

   • a transaction-scoped advisory lock keyed on the file content serializes concurrent imports of the same file — a re-sent request waits for the first one to finish;
   • it then returns the already-imported statement (matched by the attachment checksum) instead of creating a duplicate.

   As a result, a user who triggered a single import never sees an error and never ends up with duplicates, regardless of how the proxy behaves.

Why fix idempotency, and not "why does the import re-run / recompute in the first place?"

This is a fair review question, so to address it up front — we deliberately patch idempotency instead of chasing the underlying slowness first:

• The duplication is a correctness bug that is independent of performance. The trigger is a non-idempotent write action sitting behind a web client that auto-retries on connection loss. Any import slow enough to exceed any proxy timeout will duplicate; making the import faster only changes how often it happens, not whether it can happen. The proxy timeout is deployment-specific and is not something this module controls.
• The slowness does not originate in this module. The per-move sequence assignment / recomputation lives in Odoo core account.move; the synchronous PDF render is Enterprise account_accountant behaviour. Optimising core move creation is a large, risky, cross-module effort — and even a much faster import would still leave the correctness hole open (a short or misconfigured proxy timeout would still duplicate).
• Therefore the correct fix for "a non-idempotent action re-run by an auto-retrying client" is to make the action idempotent and serialized, which is what this PR does. Reducing the underlying import cost is a legitimate but separate concern (Odoo core / a dedicated performance effort) and is orthogonal to the data-integrity fix here.

Note on behaviour

Re-importing the exact same file now returns the already-imported statement instead of creating a new one — consistent with the module's existing dedup intent (unique_import_id). To genuinely re-import a file, remove the previous statement first.

[OCA-git-bot]

Hi @alexis-via,
some modules you are maintaining are being modified, check this out!

[Copilot]

Pull request overview

This PR improves the performance and reliability of bank statement file imports by preventing Odoo Enterprise (account_accountant) from synchronously rendering and attaching a statement PDF during account.bank.statement.create(), which can be slow and resource-intensive for large imports.

Changes:

• Add skip_pdf_attachment_generation to the create context when creating imported bank statements, to avoid Enterprise’s heavy synchronous PDF rendering during import.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.