feat: add option to disable OCR (#768)

fixes #344 refs #767
sismics · Sep 7, 2024 · c2d7f3e · c2d7f3e
1 parent 8f1ff56
commit c2d7f3e
Show file tree

Hide file tree

Showing 16 changed files with 221 additions and 62 deletions.
diff --git a/docs-core/src/main/java/com/sismics/docs/core/constant/ConfigType.java b/docs-core/src/main/java/com/sismics/docs/core/constant/ConfigType.java
@@ -1,9 +1,9 @@
 package com.sismics.docs.core.constant;
 
 /**
- * Configuration parameters. 
+ * Configuration parameters.
  *
- * @author jtremeaux 
+ * @author jtremeaux
  */
 public enum ConfigType {
     /**
@@ -20,6 +20,11 @@ public enum ConfigType {
      */
     GUEST_LOGIN,
 
+    /**
+     * OCR enabled.
+     */
+    OCR_ENABLED,
+
     /**
      * Default language.
      */

diff --git a/docs-core/src/main/java/com/sismics/docs/core/util/ConfigUtil.java b/docs-core/src/main/java/com/sismics/docs/core/util/ConfigUtil.java
@@ -8,13 +8,12 @@
 
 /**
  * Configuration parameter utilities.
- * 
- * @author jtremeaux
+ *
  */
 public class ConfigUtil {
     /**
      * Returns the textual value of a configuration parameter.
-     * 
+     *
      * @param configType Type of the configuration parameter
      * @return Textual value of the configuration parameter
      * @throws IllegalStateException Configuration parameter undefined
@@ -30,7 +29,7 @@ public static String getConfigStringValue(ConfigType configType) {
 
     /**
      * Returns the configuration resource bundle.
-     * 
+     *
      * @return Resource bundle
      */
     public static ResourceBundle getConfigBundle() {
@@ -39,14 +38,14 @@ public static ResourceBundle getConfigBundle() {
 
     /**
      * Returns the integer value of a configuration parameter.
-     * 
+     *
      * @param configType Type of the configuration parameter
      * @return Integer value of the configuration parameter
      * @throws IllegalStateException Configuration parameter undefined
      */
     public static int getConfigIntegerValue(ConfigType configType) {
         String value = getConfigStringValue(configType);
-        
+
         return Integer.parseInt(value);
     }
 
@@ -65,14 +64,28 @@ public static long getConfigLongValue(ConfigType configType) {
 
     /**
      * Returns the boolean value of a configuration parameter.
-     * 
+     *
      * @param configType Type of the configuration parameter
      * @return Boolean value of the configuration parameter
      * @throws IllegalStateException Configuration parameter undefined
      */
     public static boolean getConfigBooleanValue(ConfigType configType) {
         String value = getConfigStringValue(configType);
-
         return Boolean.parseBoolean(value);
     }
+
+    /**
+     * Returns the boolean value of a configuration parameter with a default value.
+     *
+     * @param configType Type of the configuration parameter
+     * @param defaultValue Default value to return if the configuration parameter is undefined
+     * @return Boolean value of the configuration parameter
+     */
+    public static boolean getConfigBooleanValue(ConfigType configType, boolean defaultValue) {
+        try {
+            return getConfigBooleanValue(configType);
+        } catch (IllegalStateException e) {
+            return defaultValue;
+        }
+    }
 }
diff --git a/docs-core/src/main/java/com/sismics/docs/core/util/format/ImageFormatHandler.java b/docs-core/src/main/java/com/sismics/docs/core/util/format/ImageFormatHandler.java
@@ -3,6 +3,8 @@
 import com.google.common.io.Closer;
 import com.sismics.docs.core.constant.Constants;
 import com.sismics.docs.core.util.FileUtil;
+import com.sismics.docs.core.util.ConfigUtil;
+import com.sismics.docs.core.constant.ConfigType;
 import com.sismics.util.mime.MimeType;
 import org.apache.pdfbox.io.MemoryUsageSetting;
 import org.apache.pdfbox.pdmodel.PDDocument;
@@ -22,7 +24,6 @@
 /**
  * Image format handler.
  *
- * @author bgamard
  */
 public class ImageFormatHandler implements FormatHandler {
     /**
@@ -45,7 +46,7 @@ public BufferedImage generateThumbnail(Path file) throws Exception {
 
     @Override
     public String extractContent(String language, Path file) throws Exception {
-        if (language == null) {
+        if (language == null || !ConfigUtil.getConfigBooleanValue(ConfigType.OCR_ENABLED, true)) {
             return null;
         }
 

diff --git a/docs-core/src/main/java/com/sismics/docs/core/util/format/PdfFormatHandler.java b/docs-core/src/main/java/com/sismics/docs/core/util/format/PdfFormatHandler.java
@@ -2,6 +2,8 @@
 
 import com.google.common.io.Closer;
 import com.sismics.docs.core.util.FileUtil;
+import com.sismics.docs.core.util.ConfigUtil;
+import com.sismics.docs.core.constant.ConfigType;
 import com.sismics.util.mime.MimeType;
 import org.apache.pdfbox.io.MemoryUsageSetting;
 import org.apache.pdfbox.multipdf.PDFMergerUtility;
@@ -53,7 +55,7 @@ public String extractContent(String language, Path file) {
         }
 
         // No text content, try to OCR it
-        if (language != null && content != null && content.trim().isEmpty()) {
+        if (language != null && content != null && content.trim().isEmpty() && ConfigUtil.getConfigBooleanValue(ConfigType.OCR_ENABLED, true)) {
             StringBuilder sb = new StringBuilder();
             try (InputStream inputStream = Files.newInputStream(file);
                  PDDocument pdfDocument = PDDocument.load(inputStream)) {

diff --git a/docs-core/src/main/java/com/sismics/util/jpa/DbOpenHelper.java b/docs-core/src/main/java/com/sismics/util/jpa/DbOpenHelper.java
@@ -39,7 +39,7 @@ abstract class DbOpenHelper {
     private static final Logger log = LoggerFactory.getLogger(DbOpenHelper.class);
 
     private final JdbcConnectionAccess jdbcConnectionAccess;
-    
+
     private final List<Exception> exceptions = new ArrayList<>();
 
     private Formatter formatter;
@@ -99,7 +99,7 @@ public void open() {
                 onCreate();
                 oldVersion = 0;
             }
-            
+
             // Execute update script
             ResourceBundle configBundle = ConfigUtil.getConfigBundle();
             Integer currentVersion = Integer.parseInt(configBundle.getString("db.version"));
@@ -126,7 +126,7 @@ public void open() {
 
     /**
      * Execute all upgrade scripts in ascending order for a given version.
-     * 
+     *
      * @param version Version number
      * @throws Exception e
      */
@@ -136,7 +136,7 @@ void executeAllScript(final int version) throws Exception {
             return name.matches("dbupdate-" + versionString + "-\\d+\\.sql");
         });
         Collections.sort(fileNameList);
-        
+
         for (String fileName : fileNameList) {
             if (log.isInfoEnabled()) {
                 log.info(MessageFormat.format("Executing script: {0}", fileName));
@@ -145,16 +145,16 @@ void executeAllScript(final int version) throws Exception {
             executeScript(is);
         }
     }
-    
+
     /**
      * Execute a SQL script. All statements must be one line only.
-     * 
+     *
      * @param inputScript Script to execute
      * @throws IOException e
      */
     private void executeScript(InputStream inputScript) throws IOException {
         List<String> lines = CharStreams.readLines(new InputStreamReader(inputScript));
-        
+
         for (String sql : lines) {
             if (Strings.isNullOrEmpty(sql) || sql.startsWith("--")) {
                 continue;
@@ -178,21 +178,21 @@ private void executeScript(InputStream inputScript) throws IOException {
     }
 
     public abstract void onCreate() throws Exception;
-    
+
     public abstract void onUpgrade(int oldVersion, int newVersion) throws Exception;
-    
+
     /**
-     * Returns a List of all Exceptions which occured during the export.
+     * Returns a List of all Exceptions which occurred during the export.
      *
-     * @return A List containig the Exceptions occured during the export
+     * @return A List containing the Exceptions occurred during the export
      */
     public List<?> getExceptions() {
         return exceptions;
     }
 
     /**
      * Format the output SQL statements.
-     * 
+     *
      * @param format True to format
      */
     public void setFormat(boolean format) {

diff --git a/docs-core/src/main/resources/config.properties b/docs-core/src/main/resources/config.properties
@@ -1 +1 @@
-db.version=30
+db.version=31
diff --git a/docs-core/src/main/resources/db/update/dbupdate-031-0.sql b/docs-core/src/main/resources/db/update/dbupdate-031-0.sql
@@ -0,0 +1,7 @@
+-- DBUPDATE-031-0.SQL
+
+-- Insert a new setting for OCR recognition
+insert into T_CONFIG (CFG_ID_C, CFG_VALUE_C) values ('OCR_ENABLED', 'true');
+
+-- Update the database version
+update T_CONFIG set CFG_VALUE_C = '31' where CFG_ID_C = 'DB_VERSION';
diff --git a/docs-web/src/dev/resources/config.properties b/docs-web/src/dev/resources/config.properties
@@ -1,3 +1,3 @@
 api.current_version=${project.version}
 api.min_version=1.0
-db.version=30
+db.version=31