@anyformat/sdk 0.1.0 → 0.2.1

package/dist/index.d.cts

@@ -404,6 +404,10 @@ type BooleanField = {
      * Free-form description shown to the extraction model.
      */
     description: string;
+    /**
+     * Lookup
+     */
+    lookup?: boolean;
     /**
      * Name
      *
@@ -498,6 +502,10 @@ type DateField = {
      * Free-form description shown to the extraction model.
      */
     description: string;
+    /**
+     * Lookup
+     */
+    lookup?: boolean;
     /**
      * Name
      *
@@ -519,6 +527,10 @@ type DatetimeField = {
      * Free-form description shown to the extraction model.
      */
     description: string;
+    /**
+     * Lookup
+     */
+    lookup?: boolean;
     /**
      * Name
      *
@@ -566,6 +578,10 @@ type EnumField = {
      * Enum Options
      */
     enum_options: Array<EnumOption>;
+    /**
+     * Lookup
+     */
+    lookup?: boolean;
     /**
      * Name
      *
@@ -621,6 +637,12 @@ type ExtractNode = {
      * Stable identifier for this node within the graph.
      */
     id: string;
+    /**
+     * Lookup File Uploads
+     *
+     * Inline lookup-file content for the typed create call. The backend uploads each entry to S3 and stores the resulting URI in ``lookup_files``; this field is never persisted in GraphNode.config.
+     */
+    lookup_file_uploads?: Array<LookupFileUpload>;
     /**
      * Lookup Files
      *
@@ -818,6 +840,10 @@ type FloatField = {
      * Free-form description shown to the extraction model.
      */
     description: string;
+    /**
+     * Lookup
+     */
+    lookup?: boolean;
     /**
      * Name
      *
@@ -858,6 +884,10 @@ type IntegerField = {
      * Free-form description shown to the extraction model.
      */
     description: string;
+    /**
+     * Lookup
+     */
+    lookup?: boolean;
     /**
      * Name
      *
@@ -865,6 +895,28 @@ type IntegerField = {
      */
     name: string;
 };
+/**
+ * LookupFileUpload
+ *
+ * Inline lookup-file content carried on the typed create call.
+ *
+ * The backend reads ``filename`` + ``content`` (base64-encoded bytes),
+ * uploads the file to S3 during workflow create, and stores the resulting
+ * URI in ``ExtractNode.lookup_files``. This field is stripped from the
+ * persisted ``GraphNode.config`` — it is create-input only.
+ */
+type LookupFileUpload = {
+    /**
+     * Content
+     *
+     * Base64-encoded file bytes.
+     */
+    content: string;
+    /**
+     * Filename
+     */
+    filename: string;
+};
 /**
  * MultiSelectField
  */
@@ -883,6 +935,10 @@ type MultiSelectField = {
      * Enum Options
      */
     enum_options: Array<EnumOption>;
+    /**
+     * Lookup
+     */
+    lookup?: boolean;
     /**
      * Name
      *
@@ -904,6 +960,10 @@ type ObjectField = {
      * Free-form description shown to the extraction model.
      */
     description: string;
+    /**
+     * Lookup
+     */
+    lookup?: boolean;
     /**
      * Name
      *
@@ -1163,6 +1223,10 @@ type StringField = {
      * Free-form description shown to the extraction model.
      */
     description: string;
+    /**
+     * Lookup
+     */
+    lookup?: boolean;
     /**
      * Name
      *
@@ -1304,22 +1368,25 @@ declare class Run {
 
 /** Any extraction field — the discriminated union on `data_type`. */
 type Field = ExtractionSchema["fields"][number];
+type LookupOpt = {
+    lookup?: boolean;
+};
 /**
  * Factory for extraction-schema fields — the TS mirror of the Python
  * `Schema` facade. Each method returns the typed wire model the API
  * validates against; the discriminator (`data_type`) is filled in for you.
  */
 declare const Schema: {
-    readonly string: (name: string, description: string) => StringField;
-    readonly integer: (name: string, description: string) => IntegerField;
-    readonly float: (name: string, description: string) => FloatField;
-    readonly boolean: (name: string, description: string) => BooleanField;
-    readonly date: (name: string, description: string) => DateField;
-    readonly datetime: (name: string, description: string) => DatetimeField;
-    readonly enum: (name: string, description: string, options: EnumOption[]) => EnumField;
-    readonly multiSelect: (name: string, description: string, options: EnumOption[]) => MultiSelectField;
-    readonly object: (name: string, description: string, fields: Field[]) => ObjectField;
-    readonly option: (name: string, description: string) => EnumOption;
+    string: (name: string, description: string, opts?: LookupOpt) => StringField;
+    integer: (name: string, description: string, opts?: LookupOpt) => IntegerField;
+    float: (name: string, description: string, opts?: LookupOpt) => FloatField;
+    boolean: (name: string, description: string, opts?: LookupOpt) => BooleanField;
+    date: (name: string, description: string, opts?: LookupOpt) => DateField;
+    datetime: (name: string, description: string, opts?: LookupOpt) => DatetimeField;
+    enum: (name: string, description: string, options: EnumOption[], opts?: LookupOpt) => EnumField;
+    multiSelect: (name: string, description: string, options: EnumOption[], opts?: LookupOpt) => MultiSelectField;
+    object: (name: string, description: string, fields: Field[], opts?: LookupOpt) => ObjectField;
+    option: (name: string, description: string) => EnumOption;
 };
 
 type ParseMode = "standard" | "agentic";
@@ -1362,6 +1429,8 @@ declare class WorkflowBuilder {
     extract(fields: Field[], opts?: {
         branch?: Branch;
         id?: string;
+        lookupFiles?: string[];
+        lookupSuggestion?: string;
     }): this;
     validate(rules: ValidationRule[], opts?: {
         branch?: Branch;
@@ -1374,7 +1443,7 @@ declare class WorkflowBuilder {
     /** Persist the workflow and return its id. */
     create(): Promise<string>;
     /** Create the workflow, submit a document, and return a pending `Run`. */
-    run(file?: Blob | File | null, opts?: {
+    run(file?: File | null, opts?: {
         text?: string;
     }): Run;
 }

package/dist/index.d.ts

@@ -404,6 +404,10 @@ type BooleanField = {
      * Free-form description shown to the extraction model.
      */
     description: string;
+    /**
+     * Lookup
+     */
+    lookup?: boolean;
     /**
      * Name
      *
@@ -498,6 +502,10 @@ type DateField = {
      * Free-form description shown to the extraction model.
      */
     description: string;
+    /**
+     * Lookup
+     */
+    lookup?: boolean;
     /**
      * Name
      *
@@ -519,6 +527,10 @@ type DatetimeField = {
      * Free-form description shown to the extraction model.
      */
     description: string;
+    /**
+     * Lookup
+     */
+    lookup?: boolean;
     /**
      * Name
      *
@@ -566,6 +578,10 @@ type EnumField = {
      * Enum Options
      */
     enum_options: Array<EnumOption>;
+    /**
+     * Lookup
+     */
+    lookup?: boolean;
     /**
      * Name
      *
@@ -621,6 +637,12 @@ type ExtractNode = {
      * Stable identifier for this node within the graph.
      */
     id: string;
+    /**
+     * Lookup File Uploads
+     *
+     * Inline lookup-file content for the typed create call. The backend uploads each entry to S3 and stores the resulting URI in ``lookup_files``; this field is never persisted in GraphNode.config.
+     */
+    lookup_file_uploads?: Array<LookupFileUpload>;
     /**
      * Lookup Files
      *
@@ -818,6 +840,10 @@ type FloatField = {
      * Free-form description shown to the extraction model.
      */
     description: string;
+    /**
+     * Lookup
+     */
+    lookup?: boolean;
     /**
      * Name
      *
@@ -858,6 +884,10 @@ type IntegerField = {
      * Free-form description shown to the extraction model.
      */
     description: string;
+    /**
+     * Lookup
+     */
+    lookup?: boolean;
     /**
      * Name
      *
@@ -865,6 +895,28 @@ type IntegerField = {
      */
     name: string;
 };
+/**
+ * LookupFileUpload
+ *
+ * Inline lookup-file content carried on the typed create call.
+ *
+ * The backend reads ``filename`` + ``content`` (base64-encoded bytes),
+ * uploads the file to S3 during workflow create, and stores the resulting
+ * URI in ``ExtractNode.lookup_files``. This field is stripped from the
+ * persisted ``GraphNode.config`` — it is create-input only.
+ */
+type LookupFileUpload = {
+    /**
+     * Content
+     *
+     * Base64-encoded file bytes.
+     */
+    content: string;
+    /**
+     * Filename
+     */
+    filename: string;
+};
 /**
  * MultiSelectField
  */
@@ -883,6 +935,10 @@ type MultiSelectField = {
      * Enum Options
      */
     enum_options: Array<EnumOption>;
+    /**
+     * Lookup
+     */
+    lookup?: boolean;
     /**
      * Name
      *
@@ -904,6 +960,10 @@ type ObjectField = {
      * Free-form description shown to the extraction model.
      */
     description: string;
+    /**
+     * Lookup
+     */
+    lookup?: boolean;
     /**
      * Name
      *
@@ -1163,6 +1223,10 @@ type StringField = {
      * Free-form description shown to the extraction model.
      */
     description: string;
+    /**
+     * Lookup
+     */
+    lookup?: boolean;
     /**
      * Name
      *
@@ -1304,22 +1368,25 @@ declare class Run {
 
 /** Any extraction field — the discriminated union on `data_type`. */
 type Field = ExtractionSchema["fields"][number];
+type LookupOpt = {
+    lookup?: boolean;
+};
 /**
  * Factory for extraction-schema fields — the TS mirror of the Python
  * `Schema` facade. Each method returns the typed wire model the API
  * validates against; the discriminator (`data_type`) is filled in for you.
  */
 declare const Schema: {
-    readonly string: (name: string, description: string) => StringField;
-    readonly integer: (name: string, description: string) => IntegerField;
-    readonly float: (name: string, description: string) => FloatField;
-    readonly boolean: (name: string, description: string) => BooleanField;
-    readonly date: (name: string, description: string) => DateField;
-    readonly datetime: (name: string, description: string) => DatetimeField;
-    readonly enum: (name: string, description: string, options: EnumOption[]) => EnumField;
-    readonly multiSelect: (name: string, description: string, options: EnumOption[]) => MultiSelectField;
-    readonly object: (name: string, description: string, fields: Field[]) => ObjectField;
-    readonly option: (name: string, description: string) => EnumOption;
+    string: (name: string, description: string, opts?: LookupOpt) => StringField;
+    integer: (name: string, description: string, opts?: LookupOpt) => IntegerField;
+    float: (name: string, description: string, opts?: LookupOpt) => FloatField;
+    boolean: (name: string, description: string, opts?: LookupOpt) => BooleanField;
+    date: (name: string, description: string, opts?: LookupOpt) => DateField;
+    datetime: (name: string, description: string, opts?: LookupOpt) => DatetimeField;
+    enum: (name: string, description: string, options: EnumOption[], opts?: LookupOpt) => EnumField;
+    multiSelect: (name: string, description: string, options: EnumOption[], opts?: LookupOpt) => MultiSelectField;
+    object: (name: string, description: string, fields: Field[], opts?: LookupOpt) => ObjectField;
+    option: (name: string, description: string) => EnumOption;
 };
 
 type ParseMode = "standard" | "agentic";
@@ -1362,6 +1429,8 @@ declare class WorkflowBuilder {
     extract(fields: Field[], opts?: {
         branch?: Branch;
         id?: string;
+        lookupFiles?: string[];
+        lookupSuggestion?: string;
     }): this;
     validate(rules: ValidationRule[], opts?: {
         branch?: Branch;
@@ -1374,7 +1443,7 @@ declare class WorkflowBuilder {
     /** Persist the workflow and return its id. */
     create(): Promise<string>;
     /** Create the workflow, submit a document, and return a pending `Run`. */
-    run(file?: Blob | File | null, opts?: {
+    run(file?: File | null, opts?: {
         text?: string;
     }): Run;
 }

package/dist/index.js

@@ -1,3 +1,7 @@
+// src/builder.ts
+import { readFileSync } from "fs";
+import { basename } from "path";
+
 // src/result.ts
 var Result = class {
   constructor(raw) {
@@ -1030,10 +1034,16 @@ var WorkflowBuilder = class {
       "extract(fields, { branch }) is only valid after classify() or split()"
     );
     const branchId = this.normalizeExtractBranch(opts.branch);
+    const lookupFileUploads = (opts.lookupFiles ?? []).map((filePath) => ({
+      filename: basename(filePath),
+      content: readFileSync(filePath).toString("base64")
+    }));
     const node = {
       type: "extract",
       id: opts.id ?? this.nextId("extract"),
-      extraction_schema: { fields }
+      extraction_schema: { fields },
+      ...lookupFileUploads.length > 0 ? { lookup_file_uploads: lookupFileUploads } : {},
+      ...opts.lookupSuggestion !== void 0 ? { lookup_suggestion: opts.lookupSuggestion } : {}
     };
     this.extracts.push([node, branchId]);
     return this;
@@ -1127,6 +1137,13 @@ var WorkflowBuilder = class {
       !(file === null && opts.text === void 0),
       "run() requires a document \u2014 pass a file or opts.text"
     );
+    if (file !== null) {
+      const filename = file.name;
+      this.require(
+        typeof filename === "string" && filename.length > 0,
+        'run() needs a named file so the server can detect its type \u2014 pass a File, e.g. new File([bytes], "invoice.pdf"), not a bare Blob'
+      );
+    }
     return new Run(this.client, this.build(), file, opts.text ?? null);
   }
 };
@@ -1149,31 +1166,20 @@ var Anyformat = class {
 };
 
 // src/schema.ts
+function withLookup(base, opts) {
+  if (opts?.lookup) return { ...base, lookup: true };
+  return base;
+}
 var Schema = {
-  string: (name, description) => ({ data_type: "string", name, description }),
-  integer: (name, description) => ({ data_type: "integer", name, description }),
-  float: (name, description) => ({ data_type: "float", name, description }),
-  boolean: (name, description) => ({ data_type: "boolean", name, description }),
-  date: (name, description) => ({ data_type: "date", name, description }),
-  datetime: (name, description) => ({ data_type: "datetime", name, description }),
-  enum: (name, description, options) => ({
-    data_type: "enum",
-    name,
-    description,
-    enum_options: options
-  }),
-  multiSelect: (name, description, options) => ({
-    data_type: "multi_select",
-    name,
-    description,
-    enum_options: options
-  }),
-  object: (name, description, fields) => ({
-    data_type: "object",
-    name,
-    description,
-    nested_fields: fields
-  }),
+  string: (name, description, opts) => withLookup({ data_type: "string", name, description }, opts),
+  integer: (name, description, opts) => withLookup({ data_type: "integer", name, description }, opts),
+  float: (name, description, opts) => withLookup({ data_type: "float", name, description }, opts),
+  boolean: (name, description, opts) => withLookup({ data_type: "boolean", name, description }, opts),
+  date: (name, description, opts) => withLookup({ data_type: "date", name, description }, opts),
+  datetime: (name, description, opts) => withLookup({ data_type: "datetime", name, description }, opts),
+  enum: (name, description, options, opts) => withLookup({ data_type: "enum", name, description, enum_options: options }, opts),
+  multiSelect: (name, description, options, opts) => withLookup({ data_type: "multi_select", name, description, enum_options: options }, opts),
+  object: (name, description, fields, opts) => withLookup({ data_type: "object", name, description, nested_fields: fields }, opts),
   option: (name, description) => ({ name, description })
 };
 export {