From 7f576b1d329b997630d3b6d0a9881760e58d25c6 Mon Sep 17 00:00:00 2001
From: eknizat <26163420+eknizat@users.noreply.github.com>
Date: Mon, 13 Jul 2020 15:12:25 +0100
Subject: [PATCH] REPO-5235: Use unforked tika (#1081)

* REPO-5235: Use latst unforked tika version from the updated alfresco-data-model
* Recognise new Exif properties and use a previous workaround to set correct tiff:ImageLength and tiff:ImageWidth
---
 pom.xml                                       |  4 ++--
 .../metadata/TikaAutoMetadataExtracter.java   | 20 +++++++------------
 2 files changed, 9 insertions(+), 15 deletions(-)

diff --git a/pom.xml b/pom.xml
index 380f10dd2a..cc88d6a874 100644
--- a/pom.xml
+++ b/pom.xml
@@ -36,7 +36,7 @@
 
         <maven.build.sourceVersion>11</maven.build.sourceVersion>
 
-        <dependency.alfresco-data-model.version>8.132</dependency.alfresco-data-model.version>
+        <dependency.alfresco-data-model.version>8.134</dependency.alfresco-data-model.version>
         <dependency.alfresco-core.version>8.36</dependency.alfresco-core.version>
 
         <dependency.alfresco-legacy-lucene.version>6.2</dependency.alfresco-legacy-lucene.version>
@@ -50,7 +50,7 @@
         <dependency.spring-security.version>5.2.1.RELEASE</dependency.spring-security.version>
         <dependency.httpcomponents.version>4.5.10</dependency.httpcomponents.version>
         <dependency.truezip.version>7.7.10</dependency.truezip.version>
-        <dependency.poi.version>4.1.1</dependency.poi.version>
+        <dependency.poi.version>4.1.2</dependency.poi.version>
         <dependency.ooxml-schemas.version>1.4</dependency.ooxml-schemas.version>
         <dependency.webscripts.version>8.4</dependency.webscripts.version>
         <dependency.opencmis.version>1.0.0</dependency.opencmis.version>
diff --git a/src/main/java/org/alfresco/repo/content/metadata/TikaAutoMetadataExtracter.java b/src/main/java/org/alfresco/repo/content/metadata/TikaAutoMetadataExtracter.java
index 74bec7fe45..63c1ad441a 100644
--- a/src/main/java/org/alfresco/repo/content/metadata/TikaAutoMetadataExtracter.java
+++ b/src/main/java/org/alfresco/repo/content/metadata/TikaAutoMetadataExtracter.java
@@ -68,11 +68,10 @@ public class TikaAutoMetadataExtracter extends TikaPoweredMetadataExtracter
     protected static Log logger = LogFactory.getLog(TikaAutoMetadataExtracter.class);
     private static AutoDetectParser parser;
     private static TikaConfig config;
-    private static String EXIF_IMAGE_HEIGHT_TAG = "Exif Image Height";
-    private static String EXIF_IMAGE_WIDTH_TAG = "Exif Image Width";
+    private static String EXIF_IMAGE_HEIGHT_TAG = "Exif SubIFD:Exif Image Height";
+    private static String EXIF_IMAGE_WIDTH_TAG = "Exif SubIFD:Exif Image Width";
     private static String JPEG_IMAGE_HEIGHT_TAG = "Image Height";
     private static String JPEG_IMAGE_WIDTH_TAG = "Image Width";
-    private static String COMPRESSION_TAG = "Compression";
 
     public static ArrayList<String> SUPPORTED_MIMETYPES;
     private static ArrayList<String> buildMimeTypes(TikaConfig tikaConfig)
@@ -122,24 +121,19 @@ public class TikaAutoMetadataExtracter extends TikaPoweredMetadataExtracter
      */
     @Override
     protected Map<String, Serializable> extractSpecific(Metadata metadata,
-         Map<String, Serializable> properties, Map<String,String> headers) 
+         Map<String, Serializable> properties, Map<String, String> headers)
     {
-        if(MimetypeMap.MIMETYPE_IMAGE_JPEG.equals(metadata.get(Metadata.CONTENT_TYPE)))
+        if (MimetypeMap.MIMETYPE_IMAGE_JPEG.equals(metadata.get(Metadata.CONTENT_TYPE)))
         {
             //check if the image has exif information
-            if(metadata.get(EXIF_IMAGE_WIDTH_TAG) != null
-                    && metadata.get(EXIF_IMAGE_HEIGHT_TAG) != null
-                    && metadata.get(COMPRESSION_TAG) != null)
+            if (metadata.get(EXIF_IMAGE_WIDTH_TAG) != null && metadata.get(EXIF_IMAGE_HEIGHT_TAG) != null)
             {
                 //replace the exif size properties that will be embedded in the node with
                 //the guessed dimensions from Tika
-                putRawValue(TIFF.IMAGE_LENGTH.getName(), extractSize(metadata.get(EXIF_IMAGE_HEIGHT_TAG)), properties);
-                putRawValue(TIFF.IMAGE_WIDTH.getName(), extractSize(metadata.get(EXIF_IMAGE_WIDTH_TAG)), properties);
-                putRawValue(JPEG_IMAGE_HEIGHT_TAG, metadata.get(EXIF_IMAGE_HEIGHT_TAG), properties);
-                putRawValue(JPEG_IMAGE_WIDTH_TAG, metadata.get(EXIF_IMAGE_WIDTH_TAG), properties);
+                putRawValue(TIFF.IMAGE_LENGTH.getName(), extractSize(metadata.get(JPEG_IMAGE_HEIGHT_TAG)), properties);
+                putRawValue(TIFF.IMAGE_WIDTH.getName(), extractSize(metadata.get(JPEG_IMAGE_WIDTH_TAG)), properties);
             }
         }
         return properties;
     }
-    
 }