@docmana/sdk 0.28.1 → 0.36.0-hitl.2

package/CHANGELOG.md

@@ -1,217 +1,308 @@
-# Changelog
-
+# Changelog
+
+## 0.36.0-hitl.2 - 2026-10-23 (HITL prerelease)
+
+- Add `executionEvents.list()` and typed, non-consuming sequence-paged node
+  history for formulas with or without human review. Add the expanded
+  `verifyExecutionEventCallback` / `parseExecutionEventCallback` while retaining
+  the existing strict review-only helpers and final callback behavior.
+- Publish pilot versions under the explicit `hitl` npm distribution tag;
+  other prereleases use `next` and stable releases use `latest`.
+
+## 0.36.0-hitl.1 - 2026-10-23 (local POC)
+
+- Add `reviews.calculation()` and exported `ReviewCalculation` for non-consuming
+  provisional calculation state, coherent/target revisions, review counts and
+  per-revision node reuse. Preserve legacy null metadata, organization scope,
+  versioned decisions, assignment and final callback behavior.
+
+## 0.35.0-hitl.3 - 2026-10-22 (local POC)
+
+- Add optional `ReviewTask.findings` and exported `ReviewFinding` for explicit,
+  review-scoped field signals. Preserve missing/empty findings and existing
+  corrections, assignment, global confirmation and final callbacks.
+
+## 0.35.0-hitl.2 - 2026-10-22 (local POC)
+
+- Add optional read-only frozen review assessments (node score, description, feedback, explanation and reasoning). No per-field findings or decision/callback change.
+
+## 0.35.0-hitl.1 - 2026-10-22 (local POC)
+
+- Preserve optional native review lifecycle timestamps for Live's human-wait and activity display. No final result fetch, pipeline or callback contract change.
+
+## 0.34.0 - 2026-10-22
+
+- Add `reviews.document(id, docKey, "preview", options)` and exported
+  `DocumentPreviewManifest` for progressive import-time WebP preparation.
+  Existing document streams, review decisions and final callbacks are unchanged.
+
+## 0.33.0 - 2026-10-21
+
+- Add typed review event callbacks and raw-body HMAC signature verification for
+  Consumer's separate review event relay. Existing execution callbacks are unchanged.
+- Add `Docmana.reviews` for organization-scoped queues, task details, versioned
+  decisions and streaming page/content/location reads. Preserve range/cache
+  statuses and safe conflict metadata; never automatically retry decisions.
+- Expose `AWAITING_REVIEW` as a non-terminal execution status.
+
 ## Unreleased
 
+## 0.32.0 - 2026-10-19
+
+- Release the additive `Docmana.authoring` API for scoped formula reads, create/save/validate/publish/history, native archives, sorting, assisted conversations, editor catalog and mapping proposals.
+- Preserve normalized API business error codes and optional POC provenance types. Optional Git/projection write fields are not accepted by DocMana API main; omit them on the ordinary Mongo authoring path.
+- Add revision-pinned `authoring.documentTrialDefinitions` for server-side assisted document trials in DocMana Flows, using existing DocMana API main reads.
+- Document assisted trial execution separately from native manual trials; document-trial reports can identify their executor. No new DocMana API endpoint is required.
+
+### Earlier local POC iteration notes
+
+The earlier 0.29–0.31 authoring iterations are incorporated in 0.32.0. Assisted classification/extraction now uses `documentTrialDefinitions` and the Flows executor; the experimental DocMana API `/extract/test-assisted` route is not required. The multipart `documentTrial` execution route remains manual-only with API main.
+
+### Previously unreleased authoring additions
+
+- Add full compiled prompt upserts, Git projection provenance, the staged state,
+  runtime capabilities and artifact/tooling hashes for traced publication.
+
+- Add organization-scoped document trials through Consumer Edge and native DocMana API, with native revision checks or compiled YAML draft prompts.
+
+- Add typed Git/Mongo authoring sources, Git publication provenance and Mongo divergence state; forward an optional exact Git commit when publishing.
+
+### Editor catalog and conversations
+
+- Add the organization-scoped native editor catalog and JSONata mapping proposal methods.
+
+- Preserve public API business error codes in optional `DocmanaError.apiCode`,
+  without changing existing SDK error categories or retry behavior.
+
+- Add organization-scoped assisted-rule conversations: opening with context and
+  optional transcript, sending messages, and acknowledging approved proposals.
+- Preserve native proposal/context payloads and expired-conversation HTTP 410.
+  Requires Consumer Edge's additive authoring conversation endpoints (DOC-2385).
+
+### Native authoring and archives
+
+- Add `Docmana.authoring` for organization-scoped native formula discovery,
+  creation, draft saving, validation, publication, history and feature flags.
+- Export lossless native authoring types and revision/operation receipt types.
+  Requires Consumer Edge's additive `/v1/authoring` contract (DOC-2385).
 ## 0.28.1 - 2026-10-02
 
 - Add `QUEUED` to the public `ExecutionStatus` type and verify that polling
   continues through `QUEUED` and `INCOMPLETED` before retrieving the completed result.
 
+- Add native archive import/export and formula deletion, with explicit import
+  overwrite confirmation and a 10 MiB archive limit.
+- Add server-side list sorting and formula dates/execution/document totals.
+
 ## 0.28.0 - 2026-09-09
-
-- Expose optional Fraud Detection API build identity on results through
-  `processor.apiVersion`, `processor.gitCommit`, and `processor.buildTime`.
-
-## 0.27.0 - 2026-09-08
-
-- Expose optional physical-document provenance, logical-document identifiers,
-  and page ranges on execution document mappings.
-- Preserve legacy execution mappings unchanged when source provenance is absent.
-
-## 0.26.0 - 2026-09-07
-
-- Expose the exact numeric version that a new flow execution would use through
-  `getFlowMetadata`.
-
-## 0.25.0 - 2026-08-24
-
-- Stream file-path inputs from disk instead of loading them completely into memory.
-- Expose typed `413` and `503` upload errors, preserve `Retry-After`, and retry one
-  multipart admission `503` carrying that header with a fresh request body.
-
-## 0.24.0 - 2026-08-20
-
-- Preserve the exact numeric flow version when mapping raw execution results.
-- Expose each processing step's LLM model and immutable prompt identifier without
-  reintroducing token or cost telemetry.
-- Update vulnerable HTTP and build-tool transitive dependencies to patched versions.
-
-## 0.23.0 - 2026-08-07
-
-- Align the Fraud result with Consumer Edge's functional public contract.
-- Keep verdict, confidence, reasoning, summary, findings, document id, and duration.
-- Remove internal preprocessing, layer telemetry, model usage, token counts, costs,
-  and persistence diagnostics; finding categories remain available through `finding.layer`.
-
-## 0.22.0 - 2026-08-06
-
-- Expose the complete Consumer Edge fraud-analysis telemetry: global totals,
-  preprocessing, verdict, and per-layer duration, token usage, LLM calls, model
-  usage, cost, and errors.
-- Keep model usage Mongo-safe as `modelUsage[]` and continue excluding the
-  sensitive Fraud `document` payload.
-
-## 0.21.0 - 2026-08-05
-
-- Relay the native Fraud job identifier through Consumer Edge while preserving
-  the existing high-level polling workflow.
-- Add local file, named byte, and readable stream inputs to `detectFraud` and
-  `detectFraudAsync` through the existing `FileInput` contract.
-- Stop sending organization correlation on Fraud requests. The deprecated
-  organization arguments remain accepted and ignored for source compatibility.
-- Expose Consumer Edge's BI-friendly Fraud result contract: aggregate `summary`,
-  referenced `findings`, and detection `layers` keyed by their unique names.
-
-## 0.20.0 - 2026-08-04
-
-- Add `detectFraud` to start a Consumer Edge fraud-detection job, poll its
-  status, and return the normalized verdict and findings.
-- Add `detectFraudAsync`, `getFraudDetectionStatus`, and
-  `getFraudDetectionResult` for callers that manage polling themselves.
-- Respect `X-Poll-After` and `Retry-After`, retry transient status failures,
-  and expose typed failure, timeout, and cancellation errors.
-
-## 0.19.0 - 2026-08-04
-
-- Add synchronous and asynchronous V2 partial flow execution through Consumer Edge.
-- Use `executionScope`, `flowVersion`, and `classificationNodeNames` in the
-  multipart body, leaving only `organization` in the query string.
-- Return the standard `DocmanaExecutionResult` shape with optional
-  `partialExecution` metadata.
-
-## 0.18.0
-- Expose the optional aggregate `verdict` on `DocmanaExecutionResult`, preserving
-  the `VALID`, `PARTIAL`, `WARNING`, `FAILED`, or `UNKNOWN` value computed by
-  Docmana API and returned by consumer-edge.
-
-## 0.17.0
-- Expose the optional `flowVersion` on `DocmanaExecutionResult`, preserving the
-  exact Docmana flow version used by an execution in polling and callback results.
-
-consu## 0.16.0
-- Scope organization document access to a specific organization: `listOrganizationDocuments`
-  now accepts an `organization` option and `downloadOrganizationDocument(s)` an optional
-  `organization` argument, both forwarded as the `?organization=` query param (relayed to
-  docmana-api as `X-Selected-Organization`). Without it, consumer-edge reads the caller's
-  default organization depot, which leaks another org's documents to callers scoping by site.
-
-## 0.15.0
-- Expose `rejectionCode`, `rejectionMessage`, `documentTypeProximity`, and
-  `documentContentNature` as optional system fields on node results instead of
-  including them in user-defined `data`.
-
-## 0.14.0
-- Add the optional human-readable `label` to every mapped node result while
-  preserving the stable node name as the record key.
-- Preserve the `REJECTED` status returned by classification nodes and expose it
-  in the public `ExecutionStatus` contract.
-- Add document-level `preprocessing` results, including generated steps such as
-  `Document Extraction`, correlated to their document by `doc_key`.
-
-## 0.13.0
-- Add normalized execution metrics to `DocmanaExecutionResult`:
-  `flowId`, `totalDocuments`, `totalPages`, `totalTokens`, `totalMana`, and
-  `totalRunningTime`.
-- Add node-level execution metrics to mapped node results: `runningTime` and
-  `mana`.
-
-## 0.12.0
-- Add `--context <json>` to `docmana flow run`, parsed as a JSON object and
-  passed through to `runFlow` as `RunFlowParams.context`.
-- Return CLI usage errors when `--context` is invalid JSON or is not a JSON
-  object.
-
-## 0.11.0
-- Normalize execution statuses to uppercase SDK values:
-  `PENDING`, `RUNNING`, `INCOMPLETED`, `COMPLETED`, and `FAILED`.
-- Normalize raw API and polling statuses before checking terminal or failed
-  states, so legacy mixed-case values still map consistently.
-- Rename flow metadata `contextSchema` to `jsonContextSchema` in the public
-  `FlowMetadata` shape and README examples.
-
-## 0.10.0
-- Add an optional `organization` to flow selection, emitted as a
-  `?organization=<org>` query param on the request to
-  `docmana-consumer-edge-api`. Purely additive — existing calls are unchanged.
-  - `listFlows({ organization })`, `getFlowMetadata(flowId, { organization })`,
-    and `getFlowDefinition(flowId, { organization })` accept the option.
-  - `RunFlowParams` (and therefore `runFlow` / `runFlowAsync`) gains
-    `organization`, threaded onto the `execute-async` request URL.
-- Expose `organization` on `FlowListItem`: when the consumer-edge tags a flow
-  with its organization, `listFlows()` now carries it through on each item.
-
-## 0.8.1
-- Rename the multipart wire fields sent to `docmana-consumer-edge-api` from
-  `json_context` / `callback_url` to `jsonContext` / `callbackUrl`, matching the
-  edge-api DTOs that dropped their snake_case `@Expose` mappings. Public inputs
-  (`context`, `callbackUrl`) are unchanged.
-
-## 0.8.0
-- Add organization document depot methods and CLI commands:
-  `listOrganizationDocuments()`, `downloadOrganizationDocument()`, and
-  `downloadOrganizationDocuments()`.
-- Add `docmana organization document list/download`.
-- Make `docmana organization document download --output` optional; default output
-  is `./{path}`.
-
-## 0.7.0
-- Add `getFlowMetadata()` / `docmana flow metadata` for projected flow runtime
-  metadata from `docmana-consumer-edge-api`.
-- Add `getFlowDefinition()` / `docmana flow definition` for normalized flow
-  graph definitions without exposing the Docmana ZIP export.
-
-## 0.6.0
-- Add `listOrganizations()` and `docmana organization list` to list organizations
-  available through `docmana-consumer-edge-api`.
-
-## 0.5.0
-- Add `listFlows()` to list flows available through `docmana-consumer-edge-api`,
-  returning `{ id, name, state }[]` with `state: "unknown"` when Docmana omits it.
-- Add `docmana flow list` CLI command, with the same API URL and Bearer token resolution as `docmana flow run`.
-- Rename `RunFlowInput` to `RunFlowParams` and move polling controls to
-  `RunFlowOptions`.
-- Remove the single-file `file` shortcut from flow params; callers now pass
-  `files`.
-- Remove `fileName` from flow params; byte and stream inputs now use
-  `{ data, filename, contentType? }`.
-- Replace `runFlowWithCallback` with `runFlowAsync(..., { callbackUrl })`.
-
-## 0.3.5
-- Add callback execution support: starts a flow and returns `{ executionResultId, fileIds }` **without polling**, registering a `callback_url` so docmana-api POSTs a completion notification when the run finishes. Call `getResult(executionResultId)` on receipt to fetch the mapped result.
-
-## 0.3.3
-- Add `RunFlowParams.context` (business `json_context`): when provided, `runFlow`/`runFlowAsync` include `json_context` in the request body for both `run_flow` and `run_once_flow`. The context is passed through as-is — filtering against the flow's *Start* node properties is performed server-side by docmana-api.
-
-## 0.3.2
-- Add `scoreDescription` to score-node results (Validation, Classification, CrossValidation, Conclusion).
-- Fix: when a target `language` is requested, the **textual** score-node fields `feedback` and `scoreDescription` are now returned from `translations[language]` (falling back to the default values when the translation omits them). The numeric `score` is language-independent and is always taken from the default node. Previously only the dynamic `data` fields were translated while the textual fields stayed in the flow's default language.
-
-## 0.3.1
-- Fix: upload all files in a single `/upload` call (one folder UUID) instead of one call per file. Sending each file separately made Docmana treat them as distinct documents, which breaks multi-document flows ("Node already present" when the same flow node is built per document).
-
-## 0.3.0
-- Relocated mapper logic and tests to `src/mappers/`.
-- Renamed SDK models to `DocmanaX` and API models to `DocmanaApiX` to distinguish raw backend types from clean public SDK types.
-- Pluralized `classification` results into a `classifications` key-value record on document results.
-- Nested dynamic extraction fields inside a dedicated `data` record on node results to prevent property name collisions.
-- Restructured node result interfaces to remove `score` and `feedback` from `DocmanaExtractionResult` and `DocmanaMetadataExtractionResult` (via new `DocmanaScoreNodeResult` interface).
-- Excluded internal fields `result` and `documentType` from mapped SDK results.
-
-## 0.2.2
-- Fix `package-lock.json`.
-
-## 0.2.1
-- 400 responses now throw `DocmanaRequestError` (was `DocmanaAuthError`); add `Incompleted` to `ExecutionStatus`.
-
-## 0.2.0
-- `RunFlowParams.once` selects the ad-hoc `run_once_flow` endpoint (current/draft flow) instead of the published `run_flow`.
-- `DocmanaConfig.headers` are sent on every API request (e.g. `X-Selected-Organization`), without overriding the bearer token or per-request headers.
-- `FileInput` now accepts a plain `Uint8Array` (e.g. pdf-lib output) in addition to path, `Buffer`, and `Stream`.
-- Uploads send a `Content-Type` inferred from the filename extension instead of always `application/octet-stream`.
-
-## 0.1.0
-- Single-call `runFlow` orchestration: upload → run → poll → result.
-- OAuth2 client-credentials auth with in-memory token cache and automatic refresh.
-- Typed error hierarchy (`DocmanaError`, `DocmanaExecutionError`, …).
-- Flexible file inputs: path, `Buffer`, `Stream`, and multi-file uploads.
-- Dual ESM + CJS distribution with bundled type definitions.
+
+- Expose optional Fraud Detection API build identity on results through
+  `processor.apiVersion`, `processor.gitCommit`, and `processor.buildTime`.
+
+## 0.27.0 - 2026-09-08
+
+- Expose optional physical-document provenance, logical-document identifiers,
+  and page ranges on execution document mappings.
+- Preserve legacy execution mappings unchanged when source provenance is absent.
+
+## 0.26.0 - 2026-09-07
+
+- Expose the exact numeric version that a new flow execution would use through
+  `getFlowMetadata`.
+
+## 0.25.0 - 2026-08-24
+
+- Stream file-path inputs from disk instead of loading them completely into memory.
+- Expose typed `413` and `503` upload errors, preserve `Retry-After`, and retry one
+  multipart admission `503` carrying that header with a fresh request body.
+
+## 0.24.0 - 2026-08-20
+
+- Preserve the exact numeric flow version when mapping raw execution results.
+- Expose each processing step's LLM model and immutable prompt identifier without
+  reintroducing token or cost telemetry.
+- Update vulnerable HTTP and build-tool transitive dependencies to patched versions.
+
+## 0.23.0 - 2026-08-07
+
+- Align the Fraud result with Consumer Edge's functional public contract.
+- Keep verdict, confidence, reasoning, summary, findings, document id, and duration.
+- Remove internal preprocessing, layer telemetry, model usage, token counts, costs,
+  and persistence diagnostics; finding categories remain available through `finding.layer`.
+
+## 0.22.0 - 2026-08-06
+
+- Expose the complete Consumer Edge fraud-analysis telemetry: global totals,
+  preprocessing, verdict, and per-layer duration, token usage, LLM calls, model
+  usage, cost, and errors.
+- Keep model usage Mongo-safe as `modelUsage[]` and continue excluding the
+  sensitive Fraud `document` payload.
+
+## 0.21.0 - 2026-08-05
+
+- Relay the native Fraud job identifier through Consumer Edge while preserving
+  the existing high-level polling workflow.
+- Add local file, named byte, and readable stream inputs to `detectFraud` and
+  `detectFraudAsync` through the existing `FileInput` contract.
+- Stop sending organization correlation on Fraud requests. The deprecated
+  organization arguments remain accepted and ignored for source compatibility.
+- Expose Consumer Edge's BI-friendly Fraud result contract: aggregate `summary`,
+  referenced `findings`, and detection `layers` keyed by their unique names.
+
+## 0.20.0 - 2026-08-04
+
+- Add `detectFraud` to start a Consumer Edge fraud-detection job, poll its
+  status, and return the normalized verdict and findings.
+- Add `detectFraudAsync`, `getFraudDetectionStatus`, and
+  `getFraudDetectionResult` for callers that manage polling themselves.
+- Respect `X-Poll-After` and `Retry-After`, retry transient status failures,
+  and expose typed failure, timeout, and cancellation errors.
+
+## 0.19.0 - 2026-08-04
+
+- Add synchronous and asynchronous V2 partial flow execution through Consumer Edge.
+- Use `executionScope`, `flowVersion`, and `classificationNodeNames` in the
+  multipart body, leaving only `organization` in the query string.
+- Return the standard `DocmanaExecutionResult` shape with optional
+  `partialExecution` metadata.
+
+## 0.18.0
+- Expose the optional aggregate `verdict` on `DocmanaExecutionResult`, preserving
+  the `VALID`, `PARTIAL`, `WARNING`, `FAILED`, or `UNKNOWN` value computed by
+  Docmana API and returned by consumer-edge.
+
+## 0.17.0
+- Expose the optional `flowVersion` on `DocmanaExecutionResult`, preserving the
+  exact Docmana flow version used by an execution in polling and callback results.
+
+consu## 0.16.0
+- Scope organization document access to a specific organization: `listOrganizationDocuments`
+  now accepts an `organization` option and `downloadOrganizationDocument(s)` an optional
+  `organization` argument, both forwarded as the `?organization=` query param (relayed to
+  docmana-api as `X-Selected-Organization`). Without it, consumer-edge reads the caller's
+  default organization depot, which leaks another org's documents to callers scoping by site.
+
+## 0.15.0
+- Expose `rejectionCode`, `rejectionMessage`, `documentTypeProximity`, and
+  `documentContentNature` as optional system fields on node results instead of
+  including them in user-defined `data`.
+
+## 0.14.0
+- Add the optional human-readable `label` to every mapped node result while
+  preserving the stable node name as the record key.
+- Preserve the `REJECTED` status returned by classification nodes and expose it
+  in the public `ExecutionStatus` contract.
+- Add document-level `preprocessing` results, including generated steps such as
+  `Document Extraction`, correlated to their document by `doc_key`.
+
+## 0.13.0
+- Add normalized execution metrics to `DocmanaExecutionResult`:
+  `flowId`, `totalDocuments`, `totalPages`, `totalTokens`, `totalMana`, and
+  `totalRunningTime`.
+- Add node-level execution metrics to mapped node results: `runningTime` and
+  `mana`.
+
+## 0.12.0
+- Add `--context <json>` to `docmana flow run`, parsed as a JSON object and
+  passed through to `runFlow` as `RunFlowParams.context`.
+- Return CLI usage errors when `--context` is invalid JSON or is not a JSON
+  object.
+
+## 0.11.0
+- Normalize execution statuses to uppercase SDK values:
+  `PENDING`, `RUNNING`, `INCOMPLETED`, `COMPLETED`, and `FAILED`.
+- Normalize raw API and polling statuses before checking terminal or failed
+  states, so legacy mixed-case values still map consistently.
+- Rename flow metadata `contextSchema` to `jsonContextSchema` in the public
+  `FlowMetadata` shape and README examples.
+
+## 0.10.0
+- Add an optional `organization` to flow selection, emitted as a
+  `?organization=<org>` query param on the request to
+  `docmana-consumer-edge-api`. Purely additive — existing calls are unchanged.
+  - `listFlows({ organization })`, `getFlowMetadata(flowId, { organization })`,
+    and `getFlowDefinition(flowId, { organization })` accept the option.
+  - `RunFlowParams` (and therefore `runFlow` / `runFlowAsync`) gains
+    `organization`, threaded onto the `execute-async` request URL.
+- Expose `organization` on `FlowListItem`: when the consumer-edge tags a flow
+  with its organization, `listFlows()` now carries it through on each item.
+
+## 0.8.1
+- Rename the multipart wire fields sent to `docmana-consumer-edge-api` from
+  `json_context` / `callback_url` to `jsonContext` / `callbackUrl`, matching the
+  edge-api DTOs that dropped their snake_case `@Expose` mappings. Public inputs
+  (`context`, `callbackUrl`) are unchanged.
+
+## 0.8.0
+- Add organization document depot methods and CLI commands:
+  `listOrganizationDocuments()`, `downloadOrganizationDocument()`, and
+  `downloadOrganizationDocuments()`.
+- Add `docmana organization document list/download`.
+- Make `docmana organization document download --output` optional; default output
+  is `./{path}`.
+
+## 0.7.0
+- Add `getFlowMetadata()` / `docmana flow metadata` for projected flow runtime
+  metadata from `docmana-consumer-edge-api`.
+- Add `getFlowDefinition()` / `docmana flow definition` for normalized flow
+  graph definitions without exposing the Docmana ZIP export.
+
+## 0.6.0
+- Add `listOrganizations()` and `docmana organization list` to list organizations
+  available through `docmana-consumer-edge-api`.
+
+## 0.5.0
+- Add `listFlows()` to list flows available through `docmana-consumer-edge-api`,
+  returning `{ id, name, state }[]` with `state: "unknown"` when Docmana omits it.
+- Add `docmana flow list` CLI command, with the same API URL and Bearer token resolution as `docmana flow run`.
+- Rename `RunFlowInput` to `RunFlowParams` and move polling controls to
+  `RunFlowOptions`.
+- Remove the single-file `file` shortcut from flow params; callers now pass
+  `files`.
+- Remove `fileName` from flow params; byte and stream inputs now use
+  `{ data, filename, contentType? }`.
+- Replace `runFlowWithCallback` with `runFlowAsync(..., { callbackUrl })`.
+
+## 0.3.5
+- Add callback execution support: starts a flow and returns `{ executionResultId, fileIds }` **without polling**, registering a `callback_url` so docmana-api POSTs a completion notification when the run finishes. Call `getResult(executionResultId)` on receipt to fetch the mapped result.
+
+## 0.3.3
+- Add `RunFlowParams.context` (business `json_context`): when provided, `runFlow`/`runFlowAsync` include `json_context` in the request body for both `run_flow` and `run_once_flow`. The context is passed through as-is — filtering against the flow's *Start* node properties is performed server-side by docmana-api.
+
+## 0.3.2
+- Add `scoreDescription` to score-node results (Validation, Classification, CrossValidation, Conclusion).
+- Fix: when a target `language` is requested, the **textual** score-node fields `feedback` and `scoreDescription` are now returned from `translations[language]` (falling back to the default values when the translation omits them). The numeric `score` is language-independent and is always taken from the default node. Previously only the dynamic `data` fields were translated while the textual fields stayed in the flow's default language.
+
+## 0.3.1
+- Fix: upload all files in a single `/upload` call (one folder UUID) instead of one call per file. Sending each file separately made Docmana treat them as distinct documents, which breaks multi-document flows ("Node already present" when the same flow node is built per document).
+
+## 0.3.0
+- Relocated mapper logic and tests to `src/mappers/`.
+- Renamed SDK models to `DocmanaX` and API models to `DocmanaApiX` to distinguish raw backend types from clean public SDK types.
+- Pluralized `classification` results into a `classifications` key-value record on document results.
+- Nested dynamic extraction fields inside a dedicated `data` record on node results to prevent property name collisions.
+- Restructured node result interfaces to remove `score` and `feedback` from `DocmanaExtractionResult` and `DocmanaMetadataExtractionResult` (via new `DocmanaScoreNodeResult` interface).
+- Excluded internal fields `result` and `documentType` from mapped SDK results.
+
+## 0.2.2
+- Fix `package-lock.json`.
+
+## 0.2.1
+- 400 responses now throw `DocmanaRequestError` (was `DocmanaAuthError`); add `Incompleted` to `ExecutionStatus`.
+
+## 0.2.0
+- `RunFlowParams.once` selects the ad-hoc `run_once_flow` endpoint (current/draft flow) instead of the published `run_flow`.
+- `DocmanaConfig.headers` are sent on every API request (e.g. `X-Selected-Organization`), without overriding the bearer token or per-request headers.
+- `FileInput` now accepts a plain `Uint8Array` (e.g. pdf-lib output) in addition to path, `Buffer`, and `Stream`.
+- Uploads send a `Content-Type` inferred from the filename extension instead of always `application/octet-stream`.
+
+## 0.1.0
+- Single-call `runFlow` orchestration: upload → run → poll → result.
+- OAuth2 client-credentials auth with in-memory token cache and automatic refresh.
+- Typed error hierarchy (`DocmanaError`, `DocmanaExecutionError`, …).
+- Flexible file inputs: path, `Buffer`, `Stream`, and multi-file uploads.
+- Dual ESM + CJS distribution with bundled type definitions.
+
+### Local HITL outline addition
+
+- Add reviews.outline with typed published formula steps, edges and review marks. Existing execution, decision and final callback contracts are unchanged.

package/LICENSE

@@ -1,21 +1,21 @@
-MIT License
-
-Copyright (c) 2026 Docmana
-
-Permission is hereby granted, free of charge, to any person obtaining a copy
-of this software and associated documentation files (the "Software"), to deal
-in the Software without restriction, including without limitation the rights
-to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
-copies of the Software, and to permit persons to whom the Software is
-furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice shall be included in all
-copies or substantial portions of the Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
-OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
-SOFTWARE.
+MIT License
+
+Copyright (c) 2026 Docmana
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.