@codemation/core-nodes-ocr 0.2.0

package/dist/index.cjs

@@ -0,0 +1,8 @@
+const require_analyzeInvoiceNode = require('./analyzeInvoiceNode-BIw8j_Zb.cjs');
+
+exports.analyzeDocumentNode = require_analyzeInvoiceNode.analyzeDocumentNode;
+exports.analyzeImageNode = require_analyzeInvoiceNode.analyzeImageNode;
+exports.analyzeInvoiceNode = require_analyzeInvoiceNode.analyzeInvoiceNode;
+exports.analyzeWithAzure = require_analyzeInvoiceNode.analyzeWithAzure;
+exports.azureContentUnderstandingCredentialType = require_analyzeInvoiceNode.azureContentUnderstandingCredentialType;
+exports.mapAnalysisResult = require_analyzeInvoiceNode.mapAnalysisResult;

package/dist/index.d.cts

@@ -0,0 +1,199 @@
+import { _ as CredentialSessionFactory, a as NodeActivationId, b as WorkflowId, c as RunnableNodeConfig, d as CredentialAdvancedSectionPresentation, f as CredentialAuthDefinition, g as CredentialMaterialSourceKind, h as CredentialJsonRecord, i as Items, m as CredentialHealthTester, n as TypeToken, o as RunDataSnapshot, p as CredentialFieldSchema, r as Item, s as RunId, u as AnyCredentialType, v as CredentialTypeId, y as NodeId } from "./runtimeTypes-ffl603pJ.cjs";
+import { ZodType } from "zod";
+import { AnalysisResult } from "@azure/ai-content-understanding";
+
+//#region ../core/src/contracts/itemExpr.d.ts
+declare const ITEM_EXPR_BRAND: unique symbol;
+type ItemExprResolvedContext = Readonly<{
+  runId: RunId;
+  workflowId: WorkflowId;
+  nodeId: NodeId;
+  activationId: NodeActivationId;
+  data: RunDataSnapshot;
+}>;
+/**
+ * Context aligned with former {@link ItemInputMapperContext} — use **`data`** to read any completed upstream node.
+ */
+type ItemExprContext = ItemExprResolvedContext;
+type ItemExprArgs<TItemJson = unknown> = Readonly<{
+  item: Item<TItemJson>;
+  itemIndex: number;
+  items: Items<TItemJson>;
+  ctx: ItemExprContext;
+}>;
+type ItemExprCallback<T, TItemJson = unknown> = (args: ItemExprArgs<TItemJson>) => T | Promise<T>;
+type ItemExpr<T, TItemJson = unknown> = Readonly<{
+  readonly [ITEM_EXPR_BRAND]: true;
+  readonly fn: ItemExprCallback<T, TItemJson>;
+}>;
+//#endregion
+//#region ../core/src/contracts/params.d.ts
+type Expr<T, TItemJson = unknown> = ItemExpr<T, TItemJson>;
+type ParamDeep<T, TItemJson = unknown> = Expr<T, TItemJson> | (T extends readonly (infer U)[] ? ReadonlyArray<ParamDeep<U, TItemJson>> : never) | (T extends object ? { [K in keyof T]: ParamDeep<T[K], TItemJson> } : T);
+//#endregion
+//#region ../core/src/authoring/defineNode.types.d.ts
+type ResolvableCredentialType = AnyCredentialType | CredentialTypeId;
+type DefinedNodeCredentialBinding = ResolvableCredentialType | Readonly<{
+  readonly type: ResolvableCredentialType | ReadonlyArray<ResolvableCredentialType>;
+  readonly label?: string;
+  readonly optional?: true;
+  readonly helpText?: string;
+  readonly helpUrl?: string;
+}>;
+type DefinedNodeCredentialBindings = Readonly<Record<string, DefinedNodeCredentialBinding>>;
+type DefinedNodeConfigInput<TConfigResolved extends CredentialJsonRecord, TItemJson> = ParamDeep<TConfigResolved, TItemJson>;
+interface DefinedNode<TKey$1 extends string, TConfig extends CredentialJsonRecord, TInputJson, TOutputJson, _TBindings extends DefinedNodeCredentialBindings | undefined = undefined> {
+  readonly kind: "defined-node";
+  readonly key: TKey$1;
+  readonly title: string;
+  readonly description?: string;
+  create<TConfigItemJson = TInputJson>(config: DefinedNodeConfigInput<TConfig, TConfigItemJson>, name?: string, id?: string): RunnableNodeConfig<TInputJson, TOutputJson>;
+  register(context: {
+    registerNode<TValue>(token: TypeToken<TValue>, implementation?: TypeToken<TValue>): void;
+  }): void;
+}
+//#endregion
+//#region src/credentials/azureContentUnderstandingCredential.d.ts
+type AzureContentUnderstandingSession = Readonly<{
+  endpoint: string;
+  apiKey: string;
+}>;
+declare const azureContentUnderstandingCredentialType: Readonly<{
+  definition: Readonly<{
+    typeId: CredentialTypeId;
+    displayName: string;
+    description?: string;
+    publicFields?: ReadonlyArray<CredentialFieldSchema>;
+    secretFields?: ReadonlyArray<CredentialFieldSchema>;
+    advancedSection?: CredentialAdvancedSectionPresentation;
+    supportedSourceKinds?: ReadonlyArray<CredentialMaterialSourceKind>;
+    auth?: CredentialAuthDefinition;
+  }>;
+  createSession: CredentialSessionFactory<{
+    endpoint: unknown;
+  }, {
+    apiKey: unknown;
+  }, Readonly<{
+    endpoint: string;
+    apiKey: string;
+  }>>;
+  test: CredentialHealthTester<{
+    endpoint: unknown;
+  }, {
+    apiKey: unknown;
+  }>;
+}> & {
+  readonly key: string;
+};
+//#endregion
+//#region src/lib/analyzeWithAzure.d.ts
+/** Structured analyzer fields: scalars at leaves; nested objects and arrays preserved. */
+type OcrStructuredFields = Readonly<Record<string, unknown>>;
+/** The output shape returned by all OCR analyzer nodes. */
+type OcrAnalysisOutput = Readonly<{
+  /** Markdown representation of the document content. */
+  content: string;
+  /** Structured fields extracted by the prebuilt analyzer. */
+  fields: OcrStructuredFields;
+}>;
+/**
+ * Analyzes a binary document using an Azure Content Understanding prebuilt analyzer.
+ * Retries on transient failures are handled by the engine via the node's `retryPolicy`.
+ */
+declare function analyzeWithAzure(args: Readonly<{
+  session: AzureContentUnderstandingSession;
+  analyzerId: string;
+  body: Uint8Array;
+  contentType: string;
+}>): Promise<OcrAnalysisOutput>;
+/** @internal Exported for testing — maps a raw AnalysisResult to the node output shape. */
+declare function mapAnalysisResult(result: AnalysisResult): OcrAnalysisOutput;
+//#endregion
+//#region src/nodes/analyzeDocumentNode.d.ts
+type AnalyzeDocumentConfig = Readonly<{
+  /** Key on `item.binary` that holds the document bytes. Default: `"data"`. */
+  binaryField?: string;
+  /** MIME type override sent to the analyzer. Falls back to attachment `mimeType` when not set. */
+  contentType?: string;
+  /**
+   * Azure Content Understanding analyzer ID to use.
+   * Defaults to `"prebuilt-document"`. Set this to a custom analyzer ID when you have
+   * a trained model or need a different prebuilt variant.
+   */
+  analyzerId?: string;
+  /** Max bytes the attachment may have before reading. Defaults to 50 MiB. */
+  maxBytes?: number;
+}>;
+declare const analyzeDocumentNode: DefinedNode<"azure-ocr.analyze-document", {
+  binaryField: unknown;
+  contentType: unknown;
+  analyzerId: unknown;
+  maxBytes: unknown;
+}, unknown, Readonly<{
+  content: string;
+  fields: OcrStructuredFields;
+}>, {
+  contentUnderstanding: {
+    type: AnyCredentialType;
+    label: string;
+    helpText: string;
+  };
+}>;
+//#endregion
+//#region src/nodes/analyzeImageNode.d.ts
+type AnalyzeImageConfig = Readonly<{
+  /** Key on `item.binary` that holds the image bytes. Default: `"data"`. */
+  binaryField?: string;
+  /** MIME type override sent to the analyzer. Falls back to attachment `mimeType` when not set. */
+  contentType?: string;
+  /**
+   * Azure Content Understanding analyzer ID to use.
+   * Defaults to `"prebuilt-imageAnalyzer"`. Set this to a custom analyzer ID when you have
+   * a trained model or need a different prebuilt variant.
+   */
+  analyzerId?: string;
+  /** Max bytes the attachment may have before reading. Defaults to 50 MiB. */
+  maxBytes?: number;
+}>;
+declare const analyzeImageNode: DefinedNode<"azure-ocr.analyze-image", {
+  binaryField: unknown;
+  contentType: unknown;
+  analyzerId: unknown;
+  maxBytes: unknown;
+}, unknown, Readonly<{
+  content: string;
+  fields: OcrStructuredFields;
+}>, {
+  contentUnderstanding: {
+    type: AnyCredentialType;
+    label: string;
+    helpText: string;
+  };
+}>;
+//#endregion
+//#region src/nodes/analyzeInvoiceNode.d.ts
+type AnalyzeInvoiceConfig = Readonly<{
+  /** Key on `item.binary` that holds the document bytes. Default: `"data"`. */
+  binaryField?: string;
+  /** MIME type override sent to the analyzer. Falls back to attachment `mimeType` when not set. */
+  contentType?: string;
+  /** Max bytes the attachment may have before reading. Defaults to 50 MiB. */
+  maxBytes?: number;
+}>;
+declare const analyzeInvoiceNode: DefinedNode<"azure-ocr.analyze-invoice", {
+  binaryField: unknown;
+  contentType: unknown;
+  maxBytes: unknown;
+}, unknown, Readonly<{
+  content: string;
+  fields: OcrStructuredFields;
+}>, {
+  contentUnderstanding: {
+    type: AnyCredentialType;
+    label: string;
+    helpText: string;
+  };
+}>;
+//#endregion
+export { AnalyzeDocumentConfig, AnalyzeImageConfig, AnalyzeInvoiceConfig, AzureContentUnderstandingSession, OcrAnalysisOutput, OcrStructuredFields, analyzeDocumentNode, analyzeImageNode, analyzeInvoiceNode, analyzeWithAzure, azureContentUnderstandingCredentialType, mapAnalysisResult };
+//# sourceMappingURL=index.d.cts.map

package/dist/index.d.ts

@@ -0,0 +1,147 @@
+import { d as CredentialAuthDefinition, f as CredentialFieldSchema, g as CredentialTypeId, h as CredentialSessionFactory, l as AnyCredentialType, m as CredentialMaterialSourceKind, p as CredentialHealthTester, r as DefinedNode, u as CredentialAdvancedSectionPresentation } from "./index-C2KJPzqN.js";
+import { AnalysisResult } from "@azure/ai-content-understanding";
+
+//#region src/credentials/azureContentUnderstandingCredential.d.ts
+type AzureContentUnderstandingSession = Readonly<{
+  endpoint: string;
+  apiKey: string;
+}>;
+declare const azureContentUnderstandingCredentialType: Readonly<{
+  definition: Readonly<{
+    typeId: CredentialTypeId;
+    displayName: string;
+    description?: string;
+    publicFields?: ReadonlyArray<CredentialFieldSchema>;
+    secretFields?: ReadonlyArray<CredentialFieldSchema>;
+    advancedSection?: CredentialAdvancedSectionPresentation;
+    supportedSourceKinds?: ReadonlyArray<CredentialMaterialSourceKind>;
+    auth?: CredentialAuthDefinition;
+  }>;
+  createSession: CredentialSessionFactory<{
+    endpoint: unknown;
+  }, {
+    apiKey: unknown;
+  }, Readonly<{
+    endpoint: string;
+    apiKey: string;
+  }>>;
+  test: CredentialHealthTester<{
+    endpoint: unknown;
+  }, {
+    apiKey: unknown;
+  }>;
+}> & {
+  readonly key: string;
+};
+//#endregion
+//#region src/lib/analyzeWithAzure.d.ts
+/** Structured analyzer fields: scalars at leaves; nested objects and arrays preserved. */
+type OcrStructuredFields = Readonly<Record<string, unknown>>;
+/** The output shape returned by all OCR analyzer nodes. */
+type OcrAnalysisOutput = Readonly<{
+  /** Markdown representation of the document content. */
+  content: string;
+  /** Structured fields extracted by the prebuilt analyzer. */
+  fields: OcrStructuredFields;
+}>;
+/**
+ * Analyzes a binary document using an Azure Content Understanding prebuilt analyzer.
+ * Retries on transient failures are handled by the engine via the node's `retryPolicy`.
+ */
+declare function analyzeWithAzure(args: Readonly<{
+  session: AzureContentUnderstandingSession;
+  analyzerId: string;
+  body: Uint8Array;
+  contentType: string;
+}>): Promise<OcrAnalysisOutput>;
+/** @internal Exported for testing — maps a raw AnalysisResult to the node output shape. */
+declare function mapAnalysisResult(result: AnalysisResult): OcrAnalysisOutput;
+//#endregion
+//#region src/nodes/analyzeDocumentNode.d.ts
+type AnalyzeDocumentConfig = Readonly<{
+  /** Key on `item.binary` that holds the document bytes. Default: `"data"`. */
+  binaryField?: string;
+  /** MIME type override sent to the analyzer. Falls back to attachment `mimeType` when not set. */
+  contentType?: string;
+  /**
+   * Azure Content Understanding analyzer ID to use.
+   * Defaults to `"prebuilt-document"`. Set this to a custom analyzer ID when you have
+   * a trained model or need a different prebuilt variant.
+   */
+  analyzerId?: string;
+  /** Max bytes the attachment may have before reading. Defaults to 50 MiB. */
+  maxBytes?: number;
+}>;
+declare const analyzeDocumentNode: DefinedNode<"azure-ocr.analyze-document", {
+  binaryField: unknown;
+  contentType: unknown;
+  analyzerId: unknown;
+  maxBytes: unknown;
+}, unknown, Readonly<{
+  content: string;
+  fields: OcrStructuredFields;
+}>, {
+  contentUnderstanding: {
+    type: AnyCredentialType;
+    label: string;
+    helpText: string;
+  };
+}>;
+//#endregion
+//#region src/nodes/analyzeImageNode.d.ts
+type AnalyzeImageConfig = Readonly<{
+  /** Key on `item.binary` that holds the image bytes. Default: `"data"`. */
+  binaryField?: string;
+  /** MIME type override sent to the analyzer. Falls back to attachment `mimeType` when not set. */
+  contentType?: string;
+  /**
+   * Azure Content Understanding analyzer ID to use.
+   * Defaults to `"prebuilt-imageAnalyzer"`. Set this to a custom analyzer ID when you have
+   * a trained model or need a different prebuilt variant.
+   */
+  analyzerId?: string;
+  /** Max bytes the attachment may have before reading. Defaults to 50 MiB. */
+  maxBytes?: number;
+}>;
+declare const analyzeImageNode: DefinedNode<"azure-ocr.analyze-image", {
+  binaryField: unknown;
+  contentType: unknown;
+  analyzerId: unknown;
+  maxBytes: unknown;
+}, unknown, Readonly<{
+  content: string;
+  fields: OcrStructuredFields;
+}>, {
+  contentUnderstanding: {
+    type: AnyCredentialType;
+    label: string;
+    helpText: string;
+  };
+}>;
+//#endregion
+//#region src/nodes/analyzeInvoiceNode.d.ts
+type AnalyzeInvoiceConfig = Readonly<{
+  /** Key on `item.binary` that holds the document bytes. Default: `"data"`. */
+  binaryField?: string;
+  /** MIME type override sent to the analyzer. Falls back to attachment `mimeType` when not set. */
+  contentType?: string;
+  /** Max bytes the attachment may have before reading. Defaults to 50 MiB. */
+  maxBytes?: number;
+}>;
+declare const analyzeInvoiceNode: DefinedNode<"azure-ocr.analyze-invoice", {
+  binaryField: unknown;
+  contentType: unknown;
+  maxBytes: unknown;
+}, unknown, Readonly<{
+  content: string;
+  fields: OcrStructuredFields;
+}>, {
+  contentUnderstanding: {
+    type: AnyCredentialType;
+    label: string;
+    helpText: string;
+  };
+}>;
+//#endregion
+export { AnalyzeDocumentConfig, AnalyzeImageConfig, AnalyzeInvoiceConfig, AzureContentUnderstandingSession, OcrAnalysisOutput, OcrStructuredFields, analyzeDocumentNode, analyzeImageNode, analyzeInvoiceNode, analyzeWithAzure, azureContentUnderstandingCredentialType, mapAnalysisResult };
+//# sourceMappingURL=index.d.ts.map

package/dist/index.js

@@ -0,0 +1,3 @@
+import { a as mapAnalysisResult, i as analyzeWithAzure, n as analyzeImageNode, o as azureContentUnderstandingCredentialType, r as analyzeDocumentNode, t as analyzeInvoiceNode } from "./analyzeInvoiceNode-uVwe3GHD.js";
+
+export { analyzeDocumentNode, analyzeImageNode, analyzeInvoiceNode, analyzeWithAzure, azureContentUnderstandingCredentialType, mapAnalysisResult };

package/dist/metadata.json

@@ -0,0 +1,72 @@
+{
+  "schemaVersion": 1,
+  "packageName": "@codemation/core-nodes-ocr",
+  "packageVersion": "0.2.0",
+  "description": "Azure AI Content Understanding (OCR) plugin for Codemation — prebuilt document, invoice, and image analyzers.",
+  "kind": "nodes",
+  "nodes": [
+    {
+      "name": "Analyze Document",
+      "kind": "node",
+      "description": "Runs an Azure Content Understanding document analyzer on a binary attachment and returns markdown text plus structured fields. Defaults to the prebuilt general document analyzer.",
+      "inputPorts": [
+        "main"
+      ],
+      "outputPorts": [
+        "main"
+      ],
+      "credentialRefs": [
+        "azureContentUnderstandingCredentialType"
+      ],
+      "sourcePath": "src\\nodes\\analyzeDocumentNode.ts"
+    },
+    {
+      "name": "Analyze Image",
+      "kind": "node",
+      "description": "Runs an Azure Content Understanding image analyzer on a binary attachment and returns markdown text plus structured fields. Defaults to the prebuilt image analyzer.",
+      "inputPorts": [
+        "main"
+      ],
+      "outputPorts": [
+        "main"
+      ],
+      "credentialRefs": [
+        "azureContentUnderstandingCredentialType"
+      ],
+      "sourcePath": "src\\nodes\\analyzeImageNode.ts"
+    },
+    {
+      "name": "Analyze Invoice",
+      "kind": "node",
+      "description": "Runs the Azure Content Understanding prebuilt invoice analyzer on a binary attachment and returns markdown text plus structured fields.",
+      "inputPorts": [
+        "main"
+      ],
+      "outputPorts": [
+        "main"
+      ],
+      "credentialRefs": [
+        "azureContentUnderstandingCredentialType"
+      ],
+      "sourcePath": "src\\nodes\\analyzeInvoiceNode.ts"
+    }
+  ],
+  "credentials": [
+    {
+      "name": "azure.contentUnderstanding",
+      "description": "Azure AI Content Understanding (endpoint + key) for prebuilt document, invoice, and image analyzers.",
+      "fields": [
+        {
+          "key": "endpoint",
+          "type": "string",
+          "required": true
+        },
+        {
+          "key": "apiKey",
+          "type": "password",
+          "required": true
+        }
+      ]
+    }
+  ]
+}