Revise Docupipe plan with concrete endpoint workflow

[matthewdavidian]

Motivation

• The previous Docupipe plan was high-level and needed concrete endpoint behavior to guide implementation and testing.
• Provide an async-first, orchestration-focused design that returns raw DocuPipe JSON and leaves schema selection and polling cadence to the application.
• Capture required headers, retry semantics, and practical constraints (URL vs multipart upload) so implementers have clear trade-offs.

Description

• Updated docs/docupipe_integration_plan.md with explicit DocuPipe endpoints (POST /document, GET /document/{id}, POST /v2/standardize/batch, GET /job/{id}, GET /standardization/{id}) and auth/header conventions (X-API-Key, JSON headers).
• Proposed package layout and class names including cDocuPipeClient, cDocuPipeDocumentService, cDocuPipeStandardizationService, cDocuPipeJobService, and docupipeai.h with minimal tDocuPipe* structs that pass through raw JSON and surface orchestration IDs/statuses.
• Defined public async API signatures (e.g., SubmitDocumentFromFile, SubmitDocumentFromUrl, Standardize, GetJob, GetStandardization), a single-retry-on-transient policy, and logging/diagnostics hooks.
• Added phased implementation and verification plan covering client shell, endpoint wiring, parsing/ergonomics, and mocked/integration tests.

Testing

• No automated tests were added or executed for this documentation-only change.
• The updated plan file was reviewed and written to docs/docupipe_integration_plan.md for implementers to follow when adding unit and integration tests in later phases.
• Planned verification includes mocked-response unit tests for submit -> documentId, standardize -> jobId, and job status transitions, plus integration tests with sample documents as described in the plan.

---

Codex Task

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 7f0fb59bd2

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Avoid retrying non-idempotent DocuPipe POST operations

The retry policy currently applies a blind retry to transport failures/429/5xx across the board, which will also hit POST /document and POST /v2/standardize/batch. If the first POST succeeded server-side but the response was lost (timeout/network drop), retrying can create duplicate documents/jobs and corrupt downstream queue state or billing counts. Please scope automatic retries to idempotent reads or require an idempotency key/deduplication strategy for submit/standardize writes.

Useful? React with 👍 / 👎.