Making file parser flexible to deprecate pdf parser (#3073)

Co-authored-by: Suchintan <suchintan@users.noreply.github.com>

skyvern-frontend/src/routes/workflows/editor/nodes/FileParserNode/FileParserNode.tsx

@@ -10,6 +10,8 @@ import { useDebugStore } from "@/store/useDebugStore";
 import { cn } from "@/util/utils";
 import { NodeHeader } from "../components/NodeHeader";
 import { useParams } from "react-router-dom";
+import { WorkflowDataSchemaInputGroup } from "@/components/DataSchemaInputGroup/WorkflowDataSchemaInputGroup";
+import { dataSchemaExampleForFileExtraction } from "../types";
 
 function FileParserNode({ id, data }: NodeProps<FileParserNode>) {
   const { updateNodeData } = useReactFlow();
@@ -21,8 +23,17 @@ function FileParserNode({ id, data }: NodeProps<FileParserNode>) {
     urlBlockLabel !== undefined && urlBlockLabel === label;
   const [inputs, setInputs] = useState({
     fileUrl: data.fileUrl,
+    jsonSchema: data.jsonSchema,
   });
 
+  function handleChange(key: string, value: unknown) {
+    if (!data.editable) {
+      return;
+    }
+    setInputs({ ...inputs, [key]: value });
+    updateNodeData(id, { [key]: value });
+  }
+
   const isFirstWorkflowBlock = useIsFirstBlockInWorkflow({ id });
 
   return (
@@ -75,15 +86,19 @@ function FileParserNode({ id, data }: NodeProps<FileParserNode>) {
               nodeId={id}
               value={inputs.fileUrl}
               onChange={(value) => {
-                if (!data.editable) {
-                  return;
-                }
-                setInputs({ ...inputs, fileUrl: value });
-                updateNodeData(id, { fileUrl: value });
+                handleChange("fileUrl", value);
               }}
               className="nopan text-xs"
             />
           </div>
+          <WorkflowDataSchemaInputGroup
+            exampleValue={dataSchemaExampleForFileExtraction}
+            value={inputs.jsonSchema}
+            onChange={(value) => {
+              handleChange("jsonSchema", value);
+            }}
+            suggestionContext={{}}
+          />
         </div>
       </div>
     </div>

skyvern-frontend/src/routes/workflows/editor/nodes/FileParserNode/types.ts

@@ -1,9 +1,11 @@
 import type { Node } from "@xyflow/react";
 import { NodeBaseData } from "../types";
+import { AppNode } from "..";
 import { debuggableWorkflowBlockTypes } from "@/routes/workflows/types/workflowTypes";
 
 export type FileParserNodeData = NodeBaseData & {
   fileUrl: string;
+  jsonSchema: string;
 };
 
 export type FileParserNode = Node<FileParserNodeData, "fileParser">;
@@ -14,5 +16,10 @@ export const fileParserNodeDefaultData: FileParserNodeData = {
   label: "",
   fileUrl: "",
   continueOnFailure: false,
+  jsonSchema: "null",
   model: null,
 } as const;
+
+export function isFileParserNode(node: AppNode): node is FileParserNode {
+  return node.type === "fileParser";
+}

skyvern-frontend/src/routes/workflows/editor/panels/WorkflowNodeLibraryPanel.tsx

@@ -162,20 +162,19 @@ const nodeLibraryItems: Array<{
       />
     ),
     title: "File Parser Block",
-    description: "Parse data from files",
-  },
-  {
-    nodeType: "pdfParser",
-    icon: (
-      <WorkflowBlockIcon
-        workflowBlockType={WorkflowBlockTypes.PDFParser}
-        className="size-6"
-      />
-    ),
-    title: "PDF Parser Block",
-    description: "Extract data from PDF files",
+    description: "Parse PDFs, CSVs, and Excel files",
   },
   // {
+  //   nodeType: "pdfParser",
+  //   icon: (
+  //     <WorkflowBlockIcon
+  //       workflowBlockType={WorkflowBlockTypes.PDFParser}
+  //       className="size-6"
+  //     />
+  //   ),
+  //   title: "PDF Parser Block",
+  //   description: "Extract data from PDF files",
+  // },
   //   nodeType: "upload",
   //   icon: (
   //     <WorkflowBlockIcon

skyvern-frontend/src/routes/workflows/editor/workflowEditorUtils.ts

@@ -56,7 +56,10 @@ import { ParametersState } from "./types";
 import { AppNode, isWorkflowBlockNode, WorkflowBlockNode } from "./nodes";
 import { codeBlockNodeDefaultData } from "./nodes/CodeBlockNode/types";
 import { downloadNodeDefaultData } from "./nodes/DownloadNode/types";
-import { fileParserNodeDefaultData } from "./nodes/FileParserNode/types";
+import {
+  isFileParserNode,
+  fileParserNodeDefaultData,
+} from "./nodes/FileParserNode/types";
 import {
   isLoopNode,
   LoopNode,
@@ -468,6 +471,7 @@ function convertToNode(
         data: {
           ...commonData,
           fileUrl: block.file_url,
+          jsonSchema: JSON.stringify(block.json_schema, null, 2),
         },
       };
     }
@@ -1254,7 +1258,8 @@ function getWorkflowBlock(node: WorkflowBlockNode): BlockYAML {
         ...base,
         block_type: "file_url_parser",
         file_url: node.data.fileUrl,
-        file_type: "csv",
+        file_type: "csv", // Backend will auto-detect based on file extension
+        json_schema: JSONParseSafe(node.data.jsonSchema),
       };
     }
     case "textPrompt": {
@@ -2187,6 +2192,15 @@ function getWorkflowErrors(nodes: Array<AppNode>): Array<string> {
     }
   });
 
+  const fileParserNodes = nodes.filter(isFileParserNode);
+  fileParserNodes.forEach((node) => {
+    try {
+      JSON.parse(node.data.jsonSchema);
+    } catch {
+      errors.push(`${node.data.label}: Data schema is not valid JSON.`);
+    }
+  });
+
   const waitNodes = nodes.filter(isWaitNode);
   waitNodes.forEach((node) => {
     const waitTimeString = node.data.waitInSeconds.trim();

skyvern-frontend/src/routes/workflows/types/workflowTypes.ts

@@ -354,7 +354,8 @@ export type SendEmailBlock = WorkflowBlockBase & {
 export type FileURLParserBlock = WorkflowBlockBase & {
   block_type: "file_url_parser";
   file_url: string;
-  file_type: "csv";
+  file_type: "csv" | "excel" | "pdf";
+  json_schema: Record<string, unknown> | null;
 };
 
 export type ValidationBlock = WorkflowBlockBase & {

skyvern-frontend/src/routes/workflows/types/workflowYamlTypes.ts

@@ -308,7 +308,8 @@ export type SendEmailBlockYAML = BlockYAMLBase & {
 export type FileUrlParserBlockYAML = BlockYAMLBase & {
   block_type: "file_url_parser";
   file_url: string;
-  file_type: "csv";
+  file_type: "csv" | "excel" | "pdf";
+  json_schema?: Record<string, unknown> | null;
 };
 
 export type ForLoopBlockYAML = BlockYAMLBase & {