feat(KNO-4367): abstract validateExtractedFilePath

src/lib/marshal/email-layout/reader.ts

@@ -1,15 +1,13 @@
-import path from "node:path";
-
 import * as fs from "fs-extra";
 import { set } from "lodash";
 
 import { JsonDataError } from "@/lib/helpers/error";
 import { ParseJsonResult, readJson } from "@/lib/helpers/json";
 import { AnyObj, mapValuesDeep, ObjPath, omitDeep } from "@/lib/helpers/object";
 import {
-  checkIfValidExtractedFilePathFormat,
   FILEPATH_MARKED_RE,
   readExtractedFileSync,
+  validateExtractedFilePath,
 } from "@/lib/marshal/shared/helpers";
 import { EmailLayoutDirContext } from "@/lib/run-context";
 
@@ -63,6 +61,11 @@ const joinExtractedFiles = async (
 ): Promise<JoinExtractedFilesResult> => {
   const errors: JsonDataError[] = [];
 
+  // Tracks each new valid extracted file path seen (rebased to be relative to
+  // layout.json) in the layout json node. Mutated in place, and used
+  // to validate the uniqueness of an extracted path encountered.
+  const uniqueFilePaths = {};
+
   mapValuesDeep(layoutJson, (relpath: string, key: string, parts) => {
     // If not marked with the @suffix, there's nothing to do.
     if (!FILEPATH_MARKED_RE.test(key)) return;
@@ -73,7 +76,10 @@ const joinExtractedFiles = async (
     // Check if the extracted path found at the current field path is valid
     const invalidFilePathError = validateExtractedFilePath(
       relpath,
-      layoutDirCtx,
+      layoutDirCtx.abspath,
+      LAYOUT_JSON,
+      uniqueFilePaths,
+      objPathToFieldStr,
     );
     if (invalidFilePathError) {
       errors.push(invalidFilePathError);
@@ -110,28 +116,3 @@ const joinExtractedFiles = async (
 
   return [layoutJson, errors];
 };
-
-/*
- * Validate the extracted file path based on its format and uniqueness (but not
- * the presence).
- */
-const validateExtractedFilePath = (
-  val: unknown,
-  emailLayoutDirCtx: EmailLayoutDirContext,
-): JsonDataError | undefined => {
-  const layoutJsonPath = path.resolve(emailLayoutDirCtx.abspath, LAYOUT_JSON);
-  // Validate the file path format, and that it is unique per layout.
-  if (
-    !checkIfValidExtractedFilePathFormat(val, layoutJsonPath) ||
-    typeof val !== "string"
-  ) {
-    const error = new JsonDataError(
-      "must be a relative path string to a unique file within the directory",
-      String(val),
-    );
-
-    return error;
-  }
-
-  return undefined;
-};

src/lib/marshal/email-layout/writer.ts

@@ -24,8 +24,8 @@ type CompiledExtractionSettings = Map<ObjKeyOrArrayIdx[], ExtractionSettings>;
 
 /* Traverse a given email layout data and compile extraction settings of every extractable
  * field into a sorted map.
- * 
- * NOTE: Currently we do NOT support content extraction at nested levels for email layouts. 
+ *
+ * NOTE: Currently we do NOT support content extraction at nested levels for email layouts.
  */
 const compileExtractionSettings = (
   emailLayout: EmailLayoutData<WithAnnotation>,

src/lib/marshal/shared/helpers.ts

@@ -79,6 +79,43 @@ export const readExtractedFileSync = (
   return [content!, undefined];
 };
 
+/*
+ * Validate the extracted file path based on its format and uniqueness (but not
+ * the presence).
+ *
+ * Note, the uniqueness check is based on reading from and writing to
+ * uniqueFilePaths, which is MUTATED in place.
+ */
+
+/* eslint-disable max-params */
+export const validateExtractedFilePath = (
+  val: unknown,
+  sourceFileAbspath: string,
+  sourceJson: string,
+  uniqueFilePaths: Record<string, boolean>,
+  objPathToFieldStr: string,
+): JsonDataError | undefined => {
+  const jsonPath = path.resolve(sourceFileAbspath, sourceJson);
+  // Validate the file path format, and that it is unique per entity.
+  if (
+    !checkIfValidExtractedFilePathFormat(val, jsonPath) ||
+    typeof val !== "string" ||
+    val in uniqueFilePaths
+  ) {
+    const error = new JsonDataError(
+      "must be a relative path string to a unique file within the directory",
+      objPathToFieldStr,
+    );
+
+    return error;
+  }
+
+  // Keep track of all the valid extracted file paths that have been seen, so
+  // we can validate each file path's uniqueness as we traverse.
+  uniqueFilePaths[val] = true;
+  return undefined;
+};
+
 /*
  * Validate the file path format of an extracted field. The file path must be:
  *

src/lib/marshal/workflow/reader.ts

@@ -14,9 +14,9 @@ import {
   omitDeep,
 } from "@/lib/helpers/object";
 import {
-  checkIfValidExtractedFilePathFormat,
   FILEPATH_MARKED_RE,
   readExtractedFileSync,
+  validateExtractedFilePath,
 } from "@/lib/marshal/shared/helpers";
 import { WorkflowDirContext } from "@/lib/run-context";
 
@@ -36,41 +36,6 @@ export type WorkflowDirData = WorkflowDirContext & {
 // (e.g. workflow.json, then visual_blocks.json)
 const MAX_EXTRACTION_LEVEL = 2;
 
-/*
- * Validate the extracted file path based on its format and uniqueness (but not
- * the presence).
- *
- * Note, the uniqueness check is based on reading from and writing to
- * uniqueFilePaths, which is MUTATED in place.
- */
-const validateExtractedFilePath = (
-  val: unknown,
-  workflowDirCtx: WorkflowDirContext,
-  uniqueFilePaths: Record<string, boolean>,
-  objPathToFieldStr: string,
-): JsonDataError | undefined => {
-  const workflowJsonPath = path.resolve(workflowDirCtx.abspath, WORKFLOW_JSON);
-
-  // Validate the file path format, and that it is unique per workflow.
-  if (
-    !checkIfValidExtractedFilePathFormat(val, workflowJsonPath) ||
-    typeof val !== "string" ||
-    val in uniqueFilePaths
-  ) {
-    const error = new JsonDataError(
-      "must be a relative path string to a unique file within the directory",
-      objPathToFieldStr,
-    );
-
-    return error;
-  }
-
-  // Keep track of all the valid extracted file paths that have been seen, so
-  // we can validate each file path's uniqueness as we traverse.
-  uniqueFilePaths[val] = true;
-  return undefined;
-};
-
 /*
  * Given a workflow json object, compiles all referenced extracted files from it
  * and returns the updated object with the extracted content joined and inlined.
@@ -148,7 +113,8 @@ const joinExtractedFiles = async (
 
       const invalidFilePathError = validateExtractedFilePath(
         rebasedFilePath,
-        workflowDirCtx,
+        workflowDirCtx.abspath,
+        WORKFLOW_JSON,
         uniqueFilePaths,
         objPathToFieldStr,
       );