change attributeName and add comments

openturing · Sep 17, 2024 · 6b79a53 · 6b79a53
1 parent df59fda
commit 6b79a53
Show file tree

Hide file tree

Showing 2 changed files with 10 additions and 12 deletions.
diff --git a/...nklr/sprinklr-app/src/main/java/com/viglet/turing/connector/sprinklr/utils/FileAsset.java b/...nklr/sprinklr-app/src/main/java/com/viglet/turing/connector/sprinklr/utils/FileAsset.java
@@ -41,17 +41,16 @@ public Map<String, Object> toMapAttributes() {
         String formatedModificationDate = formatter.format(modificationDate);
 
         attributes.put("id", id);
-        attributes.put("filename", filename);
-        attributes.put("ocr_content", ocrContent);
-        // TODO: Change TypeDef in Solr.
-        attributes.put("indexing_date", formatedIndexingDate);
+        attributes.put("title", filename);
+        attributes.put("text", ocrContent);
+        attributes.put("publication_date", formatedIndexingDate);
         attributes.put("modification_date", formatedModificationDate);
         attributes.put("url", url);
-        attributes.put("file_size", fileSize);
+        attributes.put("filesize", fileSize);
         attributes.put("extension", extension);
-        attributes.put("asset_type", assetType);
-        attributes.put("asset_category", assetCategory);
         attributes.put("source_apps", List.of("SPRINKLR"));
+        attributes.put("type", "Static File");
+
         return attributes;
     }
 }
diff --git a/...klr-app/src/main/java/com/viglet/turing/connector/sprinklr/utils/FileAssetsExtractor.java b/...klr-app/src/main/java/com/viglet/turing/connector/sprinklr/utils/FileAssetsExtractor.java
@@ -28,7 +28,6 @@
 /**
  * Extracts File Assets from Sprinklr Knowledge Base search result. Has turing URL and turing API key just to use the
  * OCR API.
- *
  * @author Gabriel F. Gomazako
  * @see FileAsset
  * @since 0.3.9
@@ -55,10 +54,10 @@ public FileAssetsExtractor(String turingUrl, String turingApiKey) {
      * @param searchResult Knowledge Base search API result.
      */
     public List<FileAsset> extractFromLinkedAssets(TurSprinklrSearchResult searchResult) {
-        // Extraindo LinkedAssets do resultado da chamada de API de search do Knowledge Base do Sprinkler
+        // Extracting LinkedAssets from the search API result of Sprinkler Knowledge Base.
         List<TurSprinklrAsset> linkedAssets = searchResult.getLinkedAssets();
 
-        // Se não houver resultados.
+        // If there are no linked assets, return an empty list.
         if (linkedAssets == null || linkedAssets.isEmpty()) {
             return Collections.emptyList();
         }
@@ -73,7 +72,7 @@ public List<FileAsset> extractFromLinkedAssets(TurSprinklrSearchResult searchRes
             String extension = null;
             URL url = null;
             try {
-                // assetId em arquivos é a completa do URL do arquivo
+                // assetId it's the complete URI of the file.
                 URI assetURI = new URI(asset.getAssetId()); // ex: google.com/files/text.pdf.
                 id = assetURI.getPath();// /files/text.pdf
                 id = id.substring(id.lastIndexOf('/') + 1); // text.pdf
@@ -90,7 +89,7 @@ public List<FileAsset> extractFromLinkedAssets(TurSprinklrSearchResult searchRes
 
             File downloadedFile = downloadFile(url);
             String contentFromDownloadedFile = null;
-            // Usa OCR para converter o arquivo para string.
+            // Tries to use turing OCR API to extract content from the downloaded file.
             try {
                 log.info("Sending documento to OCR api in: {}", URI.create(turingUrl).toURL());
                 log.info("file type={}", asset.getAssetType());