Add convenience helpers and improve documentation

- Add isInProduction() helper for checking status 3 or 4
- Add getErrorSummary() helper for extracting error messages
- Add fetchAllPages() pagination helper for auto-pagination
- Add comprehensive JSDoc to ListResponse interface
- Create ADR-001 documenting decision not to add listAll() method

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
This commit is contained in:
2026-01-16 22:30:36 +01:00
parent 6b2b26f3ea
commit 397cd10a09
4 changed files with 241 additions and 2 deletions

View File

@@ -0,0 +1,112 @@
# ADR-001: No `listAll()` Method in Documents Client
**Status:** Accepted
**Date:** 2026-01-16
**Context:** Response to client feature request (REQ-2 in suggested-improvements.md)
## Context
A client integration team requested a `listAll()` method that would:
1. Return all documents regardless of status in a single call
2. Support filtering by status codes
3. Optionally auto-paginate to fetch all pages
The Binect REST API exposes two distinct endpoints for listing documents:
- `GET /documents` - Returns shippable documents (status 2)
- `GET /documents/errors` - Returns erroneous documents (status 7)
There is no single API endpoint that returns documents across all statuses.
## Decision
We will **not** implement a `listAll()` method on the `DocumentsClient` class.
## Rationale
### 1. Violates Core Design Principle: 1:1 Semantic Mapping
The SDK's architecture mandates a "1:1 semantic mapping to REST endpoints" (CLAUDE.md). A `listAll()` method would need to:
- Call multiple API endpoints (`/documents` and `/documents/errors`)
- Merge and deduplicate results
- Present a unified interface that doesn't exist in the underlying API
This creates abstraction that obscures actual API behavior.
### 2. Violates Transparency Principle
The design constraint "Transparency over abstraction: Developers must reason about actual API calls" means developers should understand exactly which API calls are being made. A `listAll()` method hides the fact that multiple calls occur, making it harder to:
- Debug network issues
- Understand rate limiting implications
- Reason about data freshness (documents may change between calls)
### 3. Incomplete Coverage Cannot Be Fixed
Even if we implemented `listAll()`, we cannot retrieve documents in all statuses:
- Status 1 (IN_PREPARATION) - No list endpoint
- Status 3 (PRODUCTION_QUEUE) - No list endpoint
- Status 4 (PRINTING) - No list endpoint
- Status 5 (SENT) - No list endpoint
- Status 6 (CANCELED) - No list endpoint
A method named `listAll()` that doesn't actually list all documents would be misleading.
### 4. Convenience Layer Alternative Exists
Developers who need to combine results can do so explicitly:
```typescript
import { fetchAllPages } from '@binect/js';
// Explicit composition - developer knows exactly what's happening
const [shippable, erroneous] = await Promise.all([
fetchAllPages((limit, offset) => client.documents.list({ limit, offset })),
fetchAllPages((limit, offset) => client.documents.listErrors({ limit, offset })),
]);
const allAvailable = [...shippable, ...erroneous];
```
This approach:
- Makes the multiple API calls explicit
- Allows parallel execution for better performance
- Gives developers control over error handling per endpoint
- Doesn't pretend to offer functionality the API doesn't support
## Consequences
### Positive
- SDK remains a transparent wrapper over the API
- No false promises about "all documents" when API doesn't support it
- Developers understand the actual API limitations
- Easier to maintain as SDK behavior matches API exactly
### Negative
- Developers must write slightly more code to combine results
- May be perceived as less convenient than competing SDKs
## Alternatives Considered
### Alternative 1: Add `listAll()` to Convenience Layer (helpers.ts)
We considered adding a helper function rather than a method:
```typescript
export async function listAllDocuments(client: BinectClient): Promise<Document[]>
```
**Rejected because:** Even as a helper, the name "listAll" is misleading since it cannot list documents in statuses 1, 3, 4, 5, or 6. A more honest name would be `listShippableAndErroneous()`, which provides little value over explicit composition.
### Alternative 2: Add Status Filter to Existing Methods
We considered adding a status filter parameter to `list()`:
```typescript
await client.documents.list({ status: [2, 7] });
```
**Rejected because:** The API doesn't support this. We would be inventing client-side filtering semantics that don't map to any API capability.
## References
- CLAUDE.md - Design Constraints section
- suggested-improvements.md - REQ-2: Add Method to List All Documents

View File

@@ -5,7 +5,7 @@
* They provide convenient predicates and utilities for common operations.
*/
import { DocumentStatus, type Document, type Sending, type ValidationMessage } from './types.js';
import { DocumentStatus, type Document, type ListResponse, type Sending, type ValidationMessage } from './types.js';
/**
* Helper to get status code from document or sending
@@ -71,6 +71,18 @@ export function isCanceled(doc: Document | Sending): boolean {
return getStatusCode(doc) === DocumentStatus.CANCELED;
}
/**
* Check if a document is in production (status 3 or 4).
* This includes documents in the production queue and those currently printing.
*/
export function isInProduction(doc: Document | Sending): boolean {
const status = getStatusCode(doc);
return (
status === DocumentStatus.PRODUCTION_QUEUE ||
status === DocumentStatus.PRINTING
);
}
/**
* Check if a document is in a terminal state (sent, canceled, or erroneous).
*/
@@ -140,6 +152,30 @@ export function hasWarnings(doc: Document): boolean {
return getWarnings(doc).length > 0;
}
/**
* Get a summary of all error messages from a document.
* Returns undefined if the document has no errors.
*
* @param doc - The document to extract error messages from
* @param separator - Separator between messages (default: '; ')
* @returns Concatenated error messages or undefined if no errors
*
* @example
* ```typescript
* const doc = await client.documents.get(id);
* if (isErroneous(doc)) {
* console.error('Upload failed:', getErrorSummary(doc));
* }
* ```
*/
export function getErrorSummary(doc: Document, separator: string = '; '): string | undefined {
const errors = getErrors(doc);
if (errors.length === 0) {
return undefined;
}
return errors.map((e) => e.message).join(separator);
}
// ============================================================================
// Status Description
// ============================================================================
@@ -272,3 +308,66 @@ export async function waitForShippable(
options
);
}
// ============================================================================
// Pagination Utilities
// ============================================================================
/**
* Options for fetchAllPages helper.
*/
export interface FetchAllPagesOptions {
/** Number of items per page (default: 100) */
pageSize?: number;
/** Abort signal for cancellation */
signal?: AbortSignal;
}
/**
* Fetch all pages from a paginated list endpoint.
* This is an opt-in convenience helper for retrieving complete result sets.
*
* @param fetchPage - Function that fetches a single page given limit and offset
* @param options - Pagination options
* @returns Array of all items from all pages
*
* @example
* ```typescript
* // Fetch all shippable documents
* const allDocs = await fetchAllPages(
* (limit, offset) => client.documents.list({ limit, offset })
* );
*
* // Fetch all erroneous documents with custom page size
* const allErrors = await fetchAllPages(
* (limit, offset) => client.documents.listErrors({ limit, offset }),
* { pageSize: 50 }
* );
* ```
*/
export async function fetchAllPages<T>(
fetchPage: (limit: number, offset: number) => Promise<ListResponse<T>>,
options: FetchAllPagesOptions = {}
): Promise<T[]> {
const pageSize = options.pageSize ?? 100;
const allItems: T[] = [];
let offset = 0;
while (true) {
if (options.signal?.aborted) {
throw new Error('Fetch aborted');
}
const response = await fetchPage(pageSize, offset);
allItems.push(...response.items);
// Check if we've fetched all items
if (response.items.length < pageSize || allItems.length >= response.total) {
break;
}
offset += pageSize;
}
return allItems;
}

View File

@@ -77,12 +77,14 @@ export {
isCanceled,
isTerminal,
isCancelable,
isInProduction,
// Validation helpers
getErrors,
getWarnings,
getInfoMessages,
hasErrors,
hasWarnings,
getErrorSummary,
// Status description
getStatusDescription,
// Base64 utilities
@@ -92,4 +94,7 @@ export {
pollUntil,
waitForShippable,
type PollOptions,
// Pagination utilities
fetchAllPages,
type FetchAllPagesOptions,
} from './helpers.js';

View File

@@ -313,12 +313,35 @@ export interface Document {
}
/**
* List response wrapper
* Response structure for paginated list operations.
*
* @remarks
* The actual data is in the `items` array, not `data` or `results`.
*
* @example
* ```typescript
* const response = await client.documents.list();
*
* // Access the documents array via `items`
* for (const doc of response.items) {
* console.log(doc.filename, doc.status.text);
* }
*
* // Pagination information
* console.log(`Showing ${response.items.length} of ${response.total} documents`);
*
* // Check if more pages exist
* const hasMore = response.offset + response.items.length < response.total;
* ```
*/
export interface ListResponse<T> {
/** Array of items returned by the query */
items: T[];
/** Total number of items matching the query (across all pages) */
total: number;
/** Maximum number of items per page */
limit: number;
/** Number of items skipped (for pagination) */
offset: number;
}