@intuned/browser-dev 0.1.15-dev.0 → 0.1.16-dev.0

package/dist/optimized-extractors/listExtractionHelpers/__tests__/dynamicListExtractor.spec.js

@@ -2,6 +2,7 @@
 
 var _extendedTest = require("../../../common/extendedTest");
 var _dynamicListExtractor = require("../dynamicListExtractor");
+var _ = require("../../");
 var _uuid = require("uuid");
 var _dotenv = require("dotenv");
 var _neverthrow = require("neverthrow");
@@ -249,7 +250,6 @@ _extendedTest.describe.skip("Dynamic List Extractor Caching Tests", () => {
         fullContainerXpath: "/html/body/div[1]"
       }));
       const consoleSpy = _extendedTest.vi.spyOn(_Logger.logger, "debug");
-      const consoleInfoSpy = _extendedTest.vi.spyOn(_Logger.logger, "info");
       const consoleWarnSpy = _extendedTest.vi.spyOn(_Logger.logger, "warn");
       await page.setContent(simpleTemplate);
       const firstResult = await (0, _dynamicListExtractor.dynamicListExtractor)(page, ".books-list", extractionOptions);
@@ -266,4 +266,369 @@ _extendedTest.describe.skip("Dynamic List Extractor Caching Tests", () => {
       console.log("Cache size limit test completed successfully!");
     });
   });
+});
+const FULL_CONTAINER_XPATH = "html[1]/body[1]/div[1]/table[1]/tbody[1]";
+const RELATIVE_CONTAINER_PATH = "div[1]/table[1]/tbody[1]";
+const categoriesEntitySchema = {
+  type: "object",
+  required: ["unspsc"],
+  properties: {
+    unspsc: {
+      type: "string",
+      description: "extract UNSPSC",
+      primary: true
+    },
+    unspsc_description: {
+      type: "string",
+      description: "extract UNSPSC description"
+    }
+  }
+};
+function makeCategoriesOptions(label, searchRegion) {
+  return {
+    itemEntityName: "categories",
+    label,
+    itemEntitySchema: categoriesEntitySchema,
+    strategy: {
+      model: "claude-sonnet-4-20250514",
+      type: "HTML"
+    },
+    variantKey: label,
+    searchRegion: searchRegion
+  };
+}
+const categoriesTemplate = `
+<div class="categories">
+  <table>
+    <tbody>
+      <tr><td>45111901</td><td>Audioconferencing systems</td></tr>
+      <tr><td>81111809</td><td>System installation services</td></tr>
+    </tbody>
+  </table>
+</div>
+`;
+const mockResultValues = [{
+  rowIndex: 0,
+  result: {
+    unspsc: {
+      matchText: "45111901",
+      matchXpath: `${FULL_CONTAINER_XPATH}/tr[1]/td[1]`,
+      matchType: "direct-text",
+      sourceText: "45111901"
+    },
+    unspsc_description: {
+      matchText: "Audioconferencing systems",
+      matchXpath: `${FULL_CONTAINER_XPATH}/tr[1]/td[2]`,
+      matchType: "direct-text",
+      sourceText: "Audioconferencing systems"
+    }
+  }
+}, {
+  rowIndex: 1,
+  result: {
+    unspsc: {
+      matchText: "81111809",
+      matchXpath: `${FULL_CONTAINER_XPATH}/tr[2]/td[1]`,
+      matchType: "direct-text",
+      sourceText: "81111809"
+    },
+    unspsc_description: {
+      matchText: "System installation services",
+      matchXpath: `${FULL_CONTAINER_XPATH}/tr[2]/td[2]`,
+      matchType: "direct-text",
+      sourceText: "System installation services"
+    }
+  }
+}];
+const expectedCategories = [{
+  unspsc: "45111901",
+  unspsc_description: "Audioconferencing systems"
+}, {
+  unspsc: "81111809",
+  unspsc_description: "System installation services"
+}];
+(0, _extendedTest.describe)("dynamicListExtractor - search region cache reuse", () => {
+  (0, _extendedTest.test)("reuses the cache on a second identical call (no AI re-extraction)", async ({
+    page
+  }) => {
+    const label = `categories-search-region-${(0, _uuid.v4)()}`;
+    const cacheModule = await Promise.resolve().then(() => _interopRequireWildcard(require("../../../intunedServices/cache")));
+    const store = new Map();
+    const cacheGetSpy = _extendedTest.vi.spyOn(cacheModule.cache, "get").mockImplementation(async key => store.has(key) ? store.get(key) : null);
+    _extendedTest.vi.spyOn(cacheModule.cache, "set").mockImplementation(async (key, value) => {
+      store.set(key, value);
+    });
+    const runAiExtractionModule = await Promise.resolve().then(() => _interopRequireWildcard(require("../runAiExtraction")));
+    const runAiExtractionSpy = _extendedTest.vi.spyOn(runAiExtractionModule, "runAiExtraction").mockResolvedValue((0, _neverthrow.ok)({
+      resultValues: mockResultValues,
+      containerPath: RELATIVE_CONTAINER_PATH,
+      fullContainerXpath: FULL_CONTAINER_XPATH,
+      matches: new Map()
+    }));
+    await page.setContent(categoriesTemplate);
+    const searchRegion = page.locator("div.categories");
+    const first = await (0, _dynamicListExtractor.dynamicListExtractor)(page, label, makeCategoriesOptions(label, searchRegion));
+    (0, _extendedTest.expect)(first.isOk()).toBe(true);
+    (0, _extendedTest.expect)(first._unsafeUnwrap()).toEqual(expectedCategories);
+    (0, _extendedTest.expect)(runAiExtractionSpy).toHaveBeenCalledTimes(1);
+    const second = await (0, _dynamicListExtractor.dynamicListExtractor)(page, label, makeCategoriesOptions(label, searchRegion));
+    (0, _extendedTest.expect)(second.isOk()).toBe(true);
+    (0, _extendedTest.expect)(second._unsafeUnwrap()).toEqual(expectedCategories);
+    (0, _extendedTest.expect)(runAiExtractionSpy).toHaveBeenCalledTimes(1);
+    (0, _extendedTest.expect)(cacheGetSpy).toHaveBeenCalledTimes(2);
+    runAiExtractionSpy.mockRestore();
+  });
+  (0, _extendedTest.test)("caches a relative xpath mapping, not an absolute html/body path", async ({
+    page
+  }) => {
+    const label = `categories-mapping-${(0, _uuid.v4)()}`;
+    const cacheModule = await Promise.resolve().then(() => _interopRequireWildcard(require("../../../intunedServices/cache")));
+    const store = new Map();
+    _extendedTest.vi.spyOn(cacheModule.cache, "get").mockImplementation(async key => store.has(key) ? store.get(key) : null);
+    const cacheSetSpy = _extendedTest.vi.spyOn(cacheModule.cache, "set").mockImplementation(async (key, value) => {
+      store.set(key, value);
+    });
+    const runAiExtractionModule = await Promise.resolve().then(() => _interopRequireWildcard(require("../runAiExtraction")));
+    const runAiExtractionSpy = _extendedTest.vi.spyOn(runAiExtractionModule, "runAiExtraction").mockResolvedValue((0, _neverthrow.ok)({
+      resultValues: mockResultValues,
+      containerPath: RELATIVE_CONTAINER_PATH,
+      fullContainerXpath: FULL_CONTAINER_XPATH,
+      matches: new Map()
+    }));
+    await page.setContent(categoriesTemplate);
+    const searchRegion = page.locator("div.categories");
+    await (0, _dynamicListExtractor.dynamicListExtractor)(page, label, makeCategoriesOptions(label, searchRegion));
+    (0, _extendedTest.expect)(cacheSetSpy).toHaveBeenCalledTimes(1);
+    const cached = cacheSetSpy.mock.calls[0][1];
+    const allXpaths = Object.values(cached.matchesMapping).flat().map(entry => entry.xpath);
+    (0, _extendedTest.expect)(allXpaths.length).toBeGreaterThan(0);
+    for (const xpath of allXpaths) {
+      (0, _extendedTest.expect)(xpath).not.toContain("html[1]/body[1]");
+      const resolves = await page.evaluate(({
+        prefix,
+        rel
+      }) => document.evaluate(`${prefix}/${rel}`, document, null, XPathResult.FIRST_ORDERED_NODE_TYPE, null).singleNodeValue !== null, {
+        prefix: FULL_CONTAINER_XPATH,
+        rel: xpath
+      });
+      (0, _extendedTest.expect)(resolves).toBe(true);
+    }
+    runAiExtractionSpy.mockRestore();
+  });
+});
+function row(rowIndexZeroBased, unspsc, description, sources) {
+  const tr = rowIndexZeroBased + 1;
+  return {
+    rowIndex: rowIndexZeroBased,
+    result: {
+      unspsc: {
+        matchText: unspsc,
+        matchXpath: `${FULL_CONTAINER_XPATH}/tr[${tr}]/td[1]`,
+        matchType: "direct-text",
+        sourceText: (sources === null || sources === void 0 ? void 0 : sources.unspsc) ?? unspsc
+      },
+      unspsc_description: {
+        matchText: description,
+        matchXpath: `${FULL_CONTAINER_XPATH}/tr[${tr}]/td[2]`,
+        matchType: "direct-text",
+        sourceText: (sources === null || sources === void 0 ? void 0 : sources.description) ?? description
+      }
+    }
+  };
+}
+function categoriesAiResult(rows) {
+  return (0, _neverthrow.ok)({
+    resultValues: rows,
+    containerPath: RELATIVE_CONTAINER_PATH,
+    fullContainerXpath: FULL_CONTAINER_XPATH,
+    matches: new Map()
+  });
+}
+function tableHtml(cells) {
+  const rows = cells.map(([a, b]) => `<tr><td>${a}</td><td>${b}</td></tr>`).join("\n      ");
+  return `
+<div class="categories">
+  <table>
+    <tbody>
+      ${rows}
+    </tbody>
+  </table>
+</div>
+`;
+}
+async function setupCategoriesMocks() {
+  const cacheModule = await Promise.resolve().then(() => _interopRequireWildcard(require("../../../intunedServices/cache")));
+  const store = new Map();
+  _extendedTest.vi.spyOn(cacheModule.cache, "get").mockImplementation(async key => store.has(key) ? store.get(key) : null);
+  _extendedTest.vi.spyOn(cacheModule.cache, "set").mockImplementation(async (key, value) => {
+    store.set(key, value);
+  });
+  const aiModule = await Promise.resolve().then(() => _interopRequireWildcard(require("../runAiExtraction")));
+  const aiSpy = _extendedTest.vi.spyOn(aiModule, "runAiExtraction");
+  return {
+    aiSpy,
+    store
+  };
+}
+(0, _extendedTest.describe)("dynamicListExtractor - cache validation edge cases", () => {
+  (0, _extendedTest.afterEach)(() => {
+    _extendedTest.vi.restoreAllMocks();
+  });
+  (0, _extendedTest.test)("invalidates cache when extracted text changes", async ({
+    page
+  }) => {
+    const label = `validation-text-change-${(0, _uuid.v4)()}`;
+    const {
+      aiSpy
+    } = await setupCategoriesMocks();
+    aiSpy.mockResolvedValueOnce(categoriesAiResult([row(0, "45111901", "Audioconferencing systems"), row(1, "81111809", "System installation services")])).mockResolvedValueOnce(categoriesAiResult([row(0, "45111901", "Audioconferencing systems UPDATED"), row(1, "81111809", "System installation services")]));
+    await page.setContent(tableHtml([["45111901", "Audioconferencing systems"], ["81111809", "System installation services"]]));
+    const searchRegion = page.locator("div.categories");
+    const first = await (0, _dynamicListExtractor.dynamicListExtractor)(page, label, makeCategoriesOptions(label, searchRegion));
+    (0, _extendedTest.expect)(first.isOk()).toBe(true);
+    (0, _extendedTest.expect)(aiSpy).toHaveBeenCalledTimes(1);
+    await page.evaluate(() => {
+      const td = document.querySelector("div.categories tbody tr:nth-child(1) td:nth-child(2)");
+      if (td) td.textContent = "Audioconferencing systems UPDATED";
+    });
+    const second = await (0, _dynamicListExtractor.dynamicListExtractor)(page, label, makeCategoriesOptions(label, searchRegion));
+    (0, _extendedTest.expect)(second.isOk()).toBe(true);
+    (0, _extendedTest.expect)(aiSpy).toHaveBeenCalledTimes(2);
+    (0, _extendedTest.expect)(second._unsafeUnwrap()[0]).toEqual({
+      unspsc: "45111901",
+      unspsc_description: "Audioconferencing systems UPDATED"
+    });
+  });
+  (0, _extendedTest.test)("invalidates cache when a row is removed", async ({
+    page
+  }) => {
+    const label = `validation-row-removed-${(0, _uuid.v4)()}`;
+    const {
+      aiSpy
+    } = await setupCategoriesMocks();
+    aiSpy.mockResolvedValueOnce(categoriesAiResult([row(0, "45111901", "Audioconferencing systems"), row(1, "81111809", "System installation services")])).mockResolvedValueOnce(categoriesAiResult([row(0, "45111901", "Audioconferencing systems")]));
+    await page.setContent(tableHtml([["45111901", "Audioconferencing systems"], ["81111809", "System installation services"]]));
+    const searchRegion = page.locator("div.categories");
+    await (0, _dynamicListExtractor.dynamicListExtractor)(page, label, makeCategoriesOptions(label, searchRegion));
+    (0, _extendedTest.expect)(aiSpy).toHaveBeenCalledTimes(1);
+    await page.evaluate(() => {
+      const tr = document.querySelector("div.categories tbody tr:nth-child(2)");
+      tr === null || tr === void 0 || tr.remove();
+    });
+    const second = await (0, _dynamicListExtractor.dynamicListExtractor)(page, label, makeCategoriesOptions(label, searchRegion));
+    (0, _extendedTest.expect)(aiSpy).toHaveBeenCalledTimes(2);
+    (0, _extendedTest.expect)(second._unsafeUnwrap()).toHaveLength(1);
+  });
+  (0, _extendedTest.test)("reuses cache for a clean exact-match list", async ({
+    page
+  }) => {
+    const label = `validation-exact-reuse-${(0, _uuid.v4)()}`;
+    const {
+      aiSpy
+    } = await setupCategoriesMocks();
+    aiSpy.mockResolvedValue(categoriesAiResult([row(0, "45111901", "Audioconferencing systems"), row(1, "81111809", "System installation services")]));
+    await page.setContent(tableHtml([["45111901", "Audioconferencing systems"], ["81111809", "System installation services"]]));
+    const searchRegion = page.locator("div.categories");
+    await (0, _dynamicListExtractor.dynamicListExtractor)(page, label, makeCategoriesOptions(label, searchRegion));
+    await (0, _dynamicListExtractor.dynamicListExtractor)(page, label, makeCategoriesOptions(label, searchRegion));
+    (0, _extendedTest.expect)(aiSpy).toHaveBeenCalledTimes(1);
+  });
+  (0, _extendedTest.test)("reuses cache when DOM text has irregular internal whitespace", async ({
+    page
+  }) => {
+    const label = `validation-whitespace-${(0, _uuid.v4)()}`;
+    const {
+      aiSpy
+    } = await setupCategoriesMocks();
+    aiSpy.mockResolvedValue(categoriesAiResult([row(0, "45111901", "Audioconferencing systems"), row(1, "81111809", "System installation services")]));
+    await page.setContent(tableHtml([["45111901", "Audioconferencing  systems"], ["81111809", "System installation services"]]));
+    const searchRegion = page.locator("div.categories");
+    await (0, _dynamicListExtractor.dynamicListExtractor)(page, label, makeCategoriesOptions(label, searchRegion));
+    (0, _extendedTest.expect)(aiSpy).toHaveBeenCalledTimes(1);
+    await (0, _dynamicListExtractor.dynamicListExtractor)(page, label, makeCategoriesOptions(label, searchRegion));
+    (0, _extendedTest.expect)(aiSpy).toHaveBeenCalledTimes(1);
+  });
+  (0, _extendedTest.test)("reuses cache when match was partial (full source text matches)", async ({
+    page
+  }) => {
+    const label = `validation-partial-${(0, _uuid.v4)()}`;
+    const {
+      aiSpy
+    } = await setupCategoriesMocks();
+    aiSpy.mockResolvedValue(categoriesAiResult([row(0, "45111901", "Audioconferencing systems", {
+      description: "Audioconferencing systems and hardware controllers"
+    }), row(1, "81111809", "System installation services", {
+      description: "System installation services - admin"
+    })]));
+    await page.setContent(tableHtml([["45111901", "Audioconferencing systems and hardware controllers"], ["81111809", "System installation services - admin"]]));
+    const searchRegion = page.locator("div.categories");
+    await (0, _dynamicListExtractor.dynamicListExtractor)(page, label, makeCategoriesOptions(label, searchRegion));
+    (0, _extendedTest.expect)(aiSpy).toHaveBeenCalledTimes(1);
+    await (0, _dynamicListExtractor.dynamicListExtractor)(page, label, makeCategoriesOptions(label, searchRegion));
+    (0, _extendedTest.expect)(aiSpy).toHaveBeenCalledTimes(1);
+  });
+});
+(0, _extendedTest.describe)("dynamicListExtractor - single-item container round-trip (real matcher, mocked AI)", () => {
+  (0, _extendedTest.afterEach)(() => {
+    _extendedTest.vi.restoreAllMocks();
+  });
+  (0, _extendedTest.test)("reuses cache for a single-row list where id and name are in sibling cells", async ({
+    page
+  }) => {
+    const label = `single-row-${(0, _uuid.v4)()}`;
+    const listAiModule = await Promise.resolve().then(() => _interopRequireWildcard(require("../utils/extractStructuredListUsingAi")));
+    const aiSpy = _extendedTest.vi.spyOn(listAiModule, "extractStructuredListUsingAi").mockImplementation(async (_entityName, schema) => {
+      const props = Object.keys((schema === null || schema === void 0 ? void 0 : schema.properties) ?? {});
+      return (0, _neverthrow.ok)(props.length <= 1 ? [{
+        vendor: "V00000908"
+      }] : [{
+        vendor: "V00000908",
+        vendor_name: "FLAGPOLES INC."
+      }]);
+    });
+    const cacheModule = await Promise.resolve().then(() => _interopRequireWildcard(require("../../../intunedServices/cache")));
+    const store = new Map();
+    _extendedTest.vi.spyOn(cacheModule.cache, "get").mockImplementation(async key => store.has(key) ? store.get(key) : null);
+    _extendedTest.vi.spyOn(cacheModule.cache, "set").mockImplementation(async (key, value) => {
+      store.set(key, value);
+    });
+    await page.setContent(`
+      <table><tbody>
+        <tr><td>ID: <a href="#">V00000908</a></td><td>FLAGPOLES INC.</td></tr>
+      </tbody></table>
+    `);
+    const options = {
+      label,
+      itemEntityName: "vendors",
+      itemEntitySchema: {
+        type: "object",
+        properties: {
+          vendor: {
+            type: "string",
+            primary: true,
+            description: "the vendor id"
+          },
+          vendor_name: {
+            type: "string",
+            description: "the vendor name"
+          }
+        },
+        required: ["vendor", "vendor_name"]
+      },
+      strategy: {
+        model: "claude-sonnet-4-20250514",
+        type: "HTML"
+      },
+      variantKey: label
+    };
+    const first = await (0, _.extractArrayFromPage)(page, options);
+    (0, _extendedTest.expect)(first).toEqual([{
+      vendor: "V00000908",
+      vendor_name: "FLAGPOLES INC."
+    }]);
+    (0, _extendedTest.expect)(aiSpy).toHaveBeenCalledTimes(2);
+    const second = await (0, _.extractArrayFromPage)(page, options);
+    (0, _extendedTest.expect)(second).toEqual(first);
+    (0, _extendedTest.expect)(aiSpy).toHaveBeenCalledTimes(2);
+  });
 });

package/dist/optimized-extractors/listExtractionHelpers/__tests__/testArrayExtractorFromLocator.spec.js

@@ -2,7 +2,9 @@
 
 var _extendedTest = require("../../../common/extendedTest");
 var _ = require("../..");
+var _neverthrow = require("neverthrow");
 var _uuid = require("uuid");
+function _interopRequireWildcard(e, t) { if ("function" == typeof WeakMap) var r = new WeakMap(), n = new WeakMap(); return (_interopRequireWildcard = function (e, t) { if (!t && e && e.__esModule) return e; var o, i, f = { __proto__: null, default: e }; if (null === e || "object" != typeof e && "function" != typeof e) return f; if (o = t ? n : r) { if (o.has(e)) return o.get(e); o.set(e, f); } for (const t in e) "default" !== t && {}.hasOwnProperty.call(e, t) && ((i = (o = Object.defineProperty) && Object.getOwnPropertyDescriptor(e, t)) && (i.get || i.set) ? o(f, t, i) : f[t] = e[t]); return f; })(e, t); }
 const productListTemplate = `
 <div class="products-container">
   <div class="product-item">
@@ -143,4 +145,45 @@ _extendedTest.describe.skip("Array Extractor Caching Tests", () => {
       console.log("All cache behavior tests completed successfully!");
     });
   });
+});
+(0, _extendedTest.describe)("extractArrayFromLocator - option forwarding", () => {
+  (0, _extendedTest.afterEach)(() => {
+    _extendedTest.vi.restoreAllMocks();
+  });
+  (0, _extendedTest.test)("forwards prompt, apiKey and variantKey to dynamicListExtractor", async ({
+    page
+  }) => {
+    const dleModule = await Promise.resolve().then(() => _interopRequireWildcard(require("../dynamicListExtractor")));
+    const spy = _extendedTest.vi.spyOn(dleModule, "dynamicListExtractor").mockResolvedValue((0, _neverthrow.ok)([]));
+    await page.setContent(`<ul id="list"><li>item one</li></ul>`);
+    await (0, _.extractArrayFromLocator)(page.locator("#list"), {
+      label: "items",
+      itemEntityName: "items",
+      itemEntitySchema: {
+        type: "object",
+        properties: {
+          name: {
+            type: "string",
+            primary: true,
+            description: "the item name"
+          }
+        },
+        required: ["name"]
+      },
+      strategy: {
+        model: "claude-sonnet-4-20250514",
+        type: "HTML"
+      },
+      prompt: "MY_UNIQUE_PROMPT",
+      apiKey: "MY_API_KEY",
+      variantKey: "MY_VARIANT"
+    });
+    (0, _extendedTest.expect)(spy).toHaveBeenCalledTimes(1);
+    const passedOptions = spy.mock.calls[0][2];
+    (0, _extendedTest.expect)(passedOptions).toMatchObject({
+      prompt: "MY_UNIQUE_PROMPT",
+      apiKey: "MY_API_KEY",
+      variantKey: "MY_VARIANT"
+    });
+  });
 });

package/dist/optimized-extractors/listExtractionHelpers/dynamicListExtractor.js

@@ -59,7 +59,7 @@ async function dynamicListExtractor(page, identifier, options) {
     if (cachedResult.exceedsLimit) {
       _Logger.logger.warn(`Cache key ${extractorInputHash} exceeds cache limit and is not cacheable`);
     } else {
-      const isValid = await (0, _xpathMapping.validateXPathMapping)(page, cachedResult.matchesMapping, cachedResult.containerPath);
+      const isValid = await (0, _xpathMapping.validateXPathMapping)(page, cachedResult.matchesMapping, cachedResult.fullContainerXpath);
       if (isValid) {
         const nonRelatedChildrenCount = cachedResult.nonRelatedChildrenCount;
         const currentChildrenCount = await page.evaluate(fullContainerXpath => {
@@ -127,7 +127,7 @@ async function dynamicListExtractor(page, identifier, options) {
   return (0, _neverthrow.ok)(resultsToReturn);
 }
 function buildXpathsMapping(results) {
-  const containerXpath = results.containerPath;
+  const containerXpath = results.fullContainerXpath;
   const xpathsMapping = {};
   for (const result of results.resultValues) {
     for (const [_key, valueObj] of Object.entries(result.result)) {
@@ -137,7 +137,8 @@ function buildXpathsMapping(results) {
         const relativePath = matchedXpath.replace(containerXpath + "/", "");
         const xpathEntry = {
           xpath: relativePath,
-          matchType: value.matchType
+          matchType: value.matchType,
+          sourceText: value.sourceText
         };
         if (!xpathsMapping[value.matchText]) {
           xpathsMapping[value.matchText] = [];

package/dist/optimized-extractors/listExtractionHelpers/runAiExtraction.js

@@ -21,6 +21,24 @@ var _findTableHeaders = require("../common/findTableHeaders");
 var _Logger = require("../../common/Logger");
 var _utils = require("../common/matching/utils");
 function _interopRequireWildcard(e, t) { if ("function" == typeof WeakMap) var r = new WeakMap(), n = new WeakMap(); return (_interopRequireWildcard = function (e, t) { if (!t && e && e.__esModule) return e; var o, i, f = { __proto__: null, default: e }; if (null === e || "object" != typeof e && "function" != typeof e) return f; if (o = t ? n : r) { if (o.has(e)) return o.get(e); o.set(e, f); } for (const t in e) "default" !== t && {}.hasOwnProperty.call(e, t) && ((i = (o = Object.defineProperty) && Object.getOwnPropertyDescriptor(e, t)) && (i.get || i.set) ? o(f, t, i) : f[t] = e[t]); return f; })(e, t); }
+function getCommonAncestorXpath(xpaths) {
+  if (xpaths.length === 0) return null;
+  const segmented = xpaths.map(xpath => xpath.split("/"));
+  const minLen = Math.min(...segmented.map(segments => segments.length));
+  const common = [];
+  for (let i = 0; i < minLen; i++) {
+    const segment = segmented[0][i];
+    if (segmented.every(segments => segments[i] === segment)) {
+      common.push(segment);
+    } else {
+      break;
+    }
+  }
+  if (common.length > 1 && xpaths.some(xpath => xpath === common.join("/"))) {
+    common.pop();
+  }
+  return common.length > 0 ? common.join("/") : null;
+}
 async function runAiExtraction(params) {
   return handleNewAiExtraction(params);
 }
@@ -82,7 +100,8 @@ async function handleNewAiExtraction(params) {
           acc[key] = {
             matchText: bestMatch.matchText,
             matchXpath: bestMatch.matchXpath,
-            matchType: bestMatch.matchType
+            matchType: bestMatch.matchType,
+            sourceText: bestMatch.sourceText
           };
         } else {
           _Logger.logger.debug(`value "${value}" for key "${key}" in row ${i + 1} does not have any matches in the page html, dropped for hallucination protection`);
@@ -100,26 +119,15 @@ async function handleNewAiExtraction(params) {
     })));
     let containerPath = null;
     let fullContainerXpath = null;
-    if (resultValues.length > 0 && resultValues[0].result[primaryPropertyName]) {
-      const primaryXpath = resultValues[0].result[primaryPropertyName].matchXpath;
-      if (primaryXpath) {
-        const parentXpath = await pageAndSearchRegion.page.evaluate(xpath => {
-          var _window$__INTUNED__;
-          const result = document.evaluate(xpath, document, null, XPathResult.FIRST_ORDERED_NODE_TYPE, null);
-          const element = result.singleNodeValue;
-          if (!element || !element.parentElement) return null;
-          if ((_window$__INTUNED__ = window.__INTUNED__) !== null && _window$__INTUNED__ !== void 0 && _window$__INTUNED__.getElementXPath) {
-            return window.__INTUNED__.getElementXPath(element.parentElement);
-          }
-          return null;
-        }, primaryXpath);
-        if (parentXpath) {
-          fullContainerXpath = parentXpath;
-          if (hasSearchRegionContainer) {
-            containerPath = await (0, _getRelativeContainerXpathSelector.getRelativeContainerXpathSelector)(pageAndSearchRegion.searchRegion, parentXpath);
-          } else {
-            containerPath = parentXpath;
-          }
+    if (resultValues.length > 0) {
+      const propertyXpaths = Object.values(resultValues[0].result).map(value => value.matchXpath).filter(xpath => !!xpath);
+      const itemContainerXpath = getCommonAncestorXpath(propertyXpaths);
+      if (itemContainerXpath) {
+        fullContainerXpath = itemContainerXpath;
+        if (hasSearchRegionContainer) {
+          containerPath = await (0, _getRelativeContainerXpathSelector.getRelativeContainerXpathSelector)(pageAndSearchRegion.searchRegion, itemContainerXpath);
+        } else {
+          containerPath = itemContainerXpath;
         }
       }
     }
@@ -240,7 +248,8 @@ async function splitDomAndExtractData({
         acc[key] = {
           matchText: bestMatch.matchText,
           matchXpath: bestMatch.matchXpath,
-          matchType: bestMatch.matchType
+          matchType: bestMatch.matchType,
+          sourceText: bestMatch.sourceText
         };
       } else {
         _Logger.logger.debug(`value "${value}" for key "${key}" in row ${i + 1} does not have any matches in the item's html, dropped for hallucination protection`);

package/dist/optimized-extractors/objectExtractionHelpers/AIExtractors.js

@@ -59,7 +59,8 @@ async function runAIExtraction(pageAndSearchRegion, extractionInfo, snapshot, st
       xpathMapping[propertyName] = {
         matchXpath: bestMatch.matchXpath,
         matchText: bestMatch.matchText,
-        matchType: bestMatch.matchType
+        matchType: bestMatch.matchType,
+        sourceText: bestMatch.sourceText
       };
     } else {
       _Logger.logger.debug(`Property ${propertyName} not found in the page HTML, dropped for hallucination protection`);