@intuned/browser-dev 0.1.8-dev.0 → 0.1.10-dev.0

package/dist/ai/extractStructuredData.js

@@ -15,7 +15,8 @@ var _getSimplifiedHtml = require("../common/getSimplifiedHtml");
 var _hashObject = require("../common/hashObject");
 var _Logger = require("../common/Logger");
 var _helpers = require("../helpers");
-var _xpathMapping = require("../common/xpathMapping");
+var _frame_utils = require("../common/frame_utils");
+var _matching = require("../common/matching/matching");
 const extractStructuredData = async options => {
   if ("content" in options && !("source" in options)) {
     return await extractStructuredDataFromContent(options);
@@ -57,11 +58,7 @@ const extractStructuredData = async options => {
   }
   let cacheKey = "";
   if (validatedData.strategy === "HTML") {
-    const containerHandle = isPageInput ? await pageOrLocator.locator("html").elementHandle() : await pageOrLocator.elementHandle();
-    if (!containerHandle) {
-      throw new Error("No HTML content found in the specified region.");
-    }
-    const simplifiedHtml = await (0, _getSimplifiedHtml.getSimplifiedHtml)(containerHandle);
+    const simplifiedHtml = await (0, _frame_utils.getContentWithNestedIframes)(pageOrLocator, 10000, _getSimplifiedHtml.getSimplifiedHtml);
     if (validatedData.enableCache) {
       cacheKey = (0, _hashObject.hashObject)({
         pageUrl: pageObject.url(),
@@ -76,7 +73,7 @@ const extractStructuredData = async options => {
       }, true);
       const cachedResult = await _cache.cache.get(cacheKey);
       if (validatedData.enableDomMatching && cachedResult && cachedResult.matchesMapping) {
-        const isValid = await (0, _xpathMapping.validateXPathMapping)(pageObject, cachedResult.matchesMapping);
+        const isValid = await (0, _matching.validateMatchesMapping)(pageObject, cachedResult.matchesMapping);
         if (isValid) {
           _Logger.logger.info("Returning cached result with valid DOM matching");
           return cachedResult.result;
@@ -106,9 +103,10 @@ const extractStructuredData = async options => {
       if (!validatedData.enableDomMatching) {
         await _cache.cache.set(cacheKey, result.value.result);
       } else {
+        const domValidationHash = await (0, _matching.createMatchesMapping)(pageObject, result.value.result);
         const resultsToCache = {
           result: result.value.result,
-          matchesMapping: result.value.xpathMapping || {}
+          matchesMapping: domValidationHash
         };
         await _cache.cache.set(cacheKey, resultsToCache);
       }
@@ -132,7 +130,7 @@ const extractStructuredData = async options => {
       }, true);
       const cachedResult = await _cache.cache.get(cacheKey);
       if (validatedData.enableDomMatching && cachedResult && cachedResult.matchesMapping) {
-        const isValid = await (0, _xpathMapping.validateXPathMapping)(pageObject, cachedResult.matchesMapping);
+        const isValid = await (0, _matching.validateMatchesMapping)(pageObject, cachedResult.matchesMapping);
         if (isValid) {
           _Logger.logger.info("Returning cached result with valid DOM matching");
           return cachedResult.result;
@@ -168,9 +166,10 @@ const extractStructuredData = async options => {
       if (!validatedData.enableDomMatching) {
         await _cache.cache.set(cacheKey, result.value.result);
       } else {
+        const domValidationHash = await (0, _matching.createMatchesMapping)(pageObject, result.value.result);
         const resultsToCache = {
           result: result.value.result,
-          matchesMapping: result.value.xpathMapping || {}
+          matchesMapping: domValidationHash
         };
         await _cache.cache.set(cacheKey, resultsToCache);
       }
@@ -178,13 +177,8 @@ const extractStructuredData = async options => {
     return result.value.result;
   }
   if (validatedData.strategy === "MARKDOWN") {
-    const containerHandle = isPageInput ? await pageOrLocator.locator("html").elementHandle() : await pageOrLocator.elementHandle();
-    const html = await (containerHandle === null || containerHandle === void 0 ? void 0 : containerHandle.innerHTML());
-    if (!html) {
-      throw new Error("No HTML content found in the specified region.");
-    }
     const markdown = await (0, _helpers.extractMarkdown)({
-      source: pageObject
+      source: pageOrLocator
     });
     if (validatedData.enableCache) {
       cacheKey = (0, _hashObject.hashObject)({
@@ -200,14 +194,13 @@ const extractStructuredData = async options => {
         })
       }, true);
       const cachedResult = await _cache.cache.get(cacheKey);
-      if (enableDomMatching && cachedResult && cachedResult.matchesMapping) {
-        const cachedXpathMapping = cachedResult.matchesMapping;
-        const isValid = await (0, _xpathMapping.validateXPathMapping)(pageObject, cachedXpathMapping);
+      if (validatedData.enableDomMatching && cachedResult && cachedResult.matchesMapping) {
+        const isValid = await (0, _matching.validateMatchesMapping)(pageObject, cachedResult.matchesMapping);
         if (isValid) {
           _Logger.logger.info("Returning cached result with valid DOM matching");
           return cachedResult.result;
         }
-      } else if (cachedResult && !enableDomMatching) {
+      } else if (cachedResult && !validatedData.enableDomMatching) {
         _Logger.logger.info("Returning cached result");
         return cachedResult;
       }
@@ -229,15 +222,16 @@ const extractStructuredData = async options => {
       throw new Error(result.error.context);
     }
     if (validatedData.enableCache) {
-      if (!enableDomMatching) {
+      if (!validatedData.enableDomMatching) {
         await _cache.cache.set(cacheKey, result.value.result);
-        return result.value.result;
+      } else {
+        const domValidationHash = await (0, _matching.createMatchesMapping)(pageObject, result.value.result);
+        const resultsToCache = {
+          result: result.value.result,
+          matchesMapping: domValidationHash
+        };
+        await _cache.cache.set(cacheKey, resultsToCache);
       }
-      const resultsToCache = {
-        result: result.value.result,
-        matchesMapping: result.value.xpathMapping || {}
-      };
-      await _cache.cache.set(cacheKey, resultsToCache);
     }
     return result.value.result;
   }

package/dist/ai/extractStructuredDataUsingAi.js

@@ -154,7 +154,30 @@ async function extractStructuredDataUsingAi(input) {
       });
       _Logger.logger.info(`Extraction failed,
          Total LLM ${isGateway ? "Cost In Cents" : "Tokens"}: ${accumulatedTokens}`);
-      return (0, _neverthrow.err)(Errors.invalidExtractionResult(error instanceof Error ? error.message : "Unknown error during extraction"));
+      let errorMessage = "Unknown error during extraction";
+      if (error instanceof Error) {
+        errorMessage = error.message;
+        const apiError = error;
+        if (apiError.responseBody) {
+          try {
+            const responseBody = JSON.parse(apiError.responseBody);
+            if (responseBody.error) {
+              if (typeof responseBody.error === "string") {
+                errorMessage = responseBody.error;
+              } else if (responseBody.error.message) {
+                errorMessage = responseBody.error.message;
+              } else {
+                errorMessage = JSON.stringify(responseBody.error);
+              }
+            }
+          } catch {
+            if (typeof apiError.responseBody === "string") {
+              errorMessage = apiError.responseBody;
+            }
+          }
+        }
+      }
+      return (0, _neverthrow.err)(Errors.invalidExtractionResult(errorMessage));
     }
   }
   _Logger.logger.info(`Extraction failed.