feat: add option to disable OCR (#768)

fixes #344
refs #767

docs-core/src/main/java/com/sismics/docs/core/constant/ConfigType.java

@@ -20,6 +20,11 @@ public enum ConfigType {
      */
     GUEST_LOGIN,
 
+    /**
+     * OCR enabled.
+     */
+    OCR_ENABLED,
+
     /**
      * Default language.
      */

docs-core/src/main/java/com/sismics/docs/core/util/ConfigUtil.java

@@ -9,7 +9,6 @@ import java.util.ResourceBundle;
 /**
  * Configuration parameter utilities.
  *
- * @author jtremeaux
  */
 public class ConfigUtil {
     /**
@@ -72,7 +71,21 @@ public class ConfigUtil {
      */
     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;
+        }
+    }
 }

docs-core/src/main/java/com/sismics/docs/core/util/format/ImageFormatHandler.java

@@ -3,6 +3,8 @@ package com.sismics.docs.core.util.format;
 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 @@ import java.nio.file.Path;
 /**
  * Image format handler.
  *
- * @author bgamard
  */
 public class ImageFormatHandler implements FormatHandler {
     /**
@@ -45,7 +46,7 @@ public class ImageFormatHandler implements FormatHandler {
 
     @Override
     public String extractContent(String language, Path file) throws Exception {
-        if (language == null) {
+        if (language == null || !ConfigUtil.getConfigBooleanValue(ConfigType.OCR_ENABLED, true)) {
             return null;
         }
 

docs-core/src/main/java/com/sismics/docs/core/util/format/PdfFormatHandler.java

@@ -2,6 +2,8 @@ package com.sismics.docs.core.util.format;
 
 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 class PdfFormatHandler implements FormatHandler {
         }
 
         // 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)) {

docs-core/src/main/java/com/sismics/util/jpa/DbOpenHelper.java

@@ -182,9 +182,9 @@ abstract class DbOpenHelper {
     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;

docs-core/src/main/resources/config.properties

@@ -1 +1 @@
-db.version=30
+db.version=31

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';

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

docs-web/src/main/java/com/sismics/docs/rest/resource/AppResource.java

@@ -58,9 +58,9 @@ public class AppResource extends BaseResource {
     private static final Logger log = LoggerFactory.getLogger(AppResource.class);
 
     /**
-     * Returns informations about the application.
+     * Returns information about the application.
      *
-     * @api {get} /app Get application informations
+     * @api {get} /app Get application information
      * @apiName GetApp
      * @apiGroup App
      * @apiSuccess {String} current_version API current version
@@ -85,6 +85,7 @@ public class AppResource extends BaseResource {
         String currentVersion = configBundle.getString("api.current_version");
         String minVersion = configBundle.getString("api.min_version");
         Boolean guestLogin = ConfigUtil.getConfigBooleanValue(ConfigType.GUEST_LOGIN);
+        Boolean ocrEnabled = ConfigUtil.getConfigBooleanValue(ConfigType.OCR_ENABLED, true);
         String defaultLanguage = ConfigUtil.getConfigStringValue(ConfigType.DEFAULT_LANGUAGE);
         UserDao userDao = new UserDao();
         DocumentDao documentDao = new DocumentDao();
@@ -98,6 +99,7 @@ public class AppResource extends BaseResource {
                 .add("current_version", currentVersion.replace("-SNAPSHOT", ""))
                 .add("min_version", minVersion)
                 .add("guest_login", guestLogin)
+                .add("ocr_enabled", ocrEnabled)
                 .add("default_language", defaultLanguage)
                 .add("queued_tasks", AppContext.getInstance().getQueuedTaskCount())
                 .add("total_memory", Runtime.getRuntime().totalMemory())
@@ -140,6 +142,34 @@ public class AppResource extends BaseResource {
         return Response.ok().build();
     }
 
+    /**
+     * Enable/disable OCR.
+     *
+     * @api {post} /app/ocr Enable/disable OCR
+     * @apiName PostAppOcr
+     * @apiGroup App
+     * @apiParam {Boolean} enabled If true, enable OCR
+     * @apiError (client) ForbiddenError Access denied
+     * @apiPermission admin
+     * @apiVersion 1.5.0
+     *
+     * @param enabled If true, enable OCR
+     * @return Response
+     */
+    @POST
+    @Path("ocr")
+    public Response ocr(@FormParam("enabled") Boolean enabled) {
+        if (!authenticate()) {
+            throw new ForbiddenClientException();
+        }
+        checkBaseFunction(BaseFunction.ADMIN);
+
+        ConfigDao configDao = new ConfigDao();
+        configDao.update(ConfigType.OCR_ENABLED, enabled.toString());
+
+        return Response.ok().build();
+    }
+
     /**
      * General application configuration.
      *

docs-web/src/main/webapp/src/app/docs/controller/settings/SettingsConfig.js

@@ -21,6 +21,15 @@ angular.module('docs').controller('SettingsConfig', function($scope, $rootScope,
     });
   };
 
+  // Enable/disable OCR
+  $scope.changeOcrEnabled = function (enabled) {
+    Restangular.one('app').post('ocr', {
+      enabled: enabled
+    }).then(function () {
+      $scope.app.ocr_enabled = enabled;
+    });
+  };
+
   // Fetch the current theme configuration
   Restangular.one('theme').get().then(function (data) {
     $scope.theme = data;

docs-web/src/main/webapp/src/locale/de.json

@@ -421,6 +421,10 @@
       "smtp_username": "SMTP-Benutzername",
       "smtp_password": "SMTP-Passwort",
       "smtp_updated": "SMTP-Konfiguration erfolgreich aktualisiert",
+      "title_ocr": "Texterkennung (OCR)",
+      "message_ocr": "OCR ist eine Technologie, die es ermöglicht, Text aus Bildern zu extrahieren. Es wird verwendet, um den Inhalt von Bildern zu durchsuchen und zu indizieren. Diese Funktionalität setzt voraus, dass die Tesseract auf Ihrem Server installiert ist.",
+      "enable_ocr": "OCR aktivieren",
+      "disable_ocr": "OCR deaktivieren",
       "webhooks": "Webhooks",
       "webhooks_explain": "Webhooks werden aufgerufen, wenn das angegebene Ereignis eintritt. Die angegebene URL wird mit einer JSON-Payload gepostet, die den Ereignisnamen und die ID der betreffenden Ressource enthält.",
       "webhook_event": "Ereignisse",

docs-web/src/main/webapp/src/locale/en.json

@@ -421,6 +421,10 @@
       "smtp_username": "SMTP username",
       "smtp_password": "SMTP password",
       "smtp_updated": "SMTP configuration updated successfully",
+      "title_ocr": "Optical Character Recognition (OCR)",
+      "message_ocr": "OCR is a feature that extracts text from images and PDF files. This feature requires a working Tesseract installation on the server.",
+      "enable_ocr": "Enable OCR",
+      "disable_ocr": "Disable OCR",
       "webhooks": "Webhooks",
       "webhooks_explain": "Webhooks will be called when the specified event occur. The given URL will be POST-ed with a JSON payload containing the event name and the ID of the concerned resource.",
       "webhook_event": "Event",

docs-web/src/main/webapp/src/partial/docs/settings.config.html

@@ -31,6 +31,17 @@
   </div>
 </form>
 
+<h2>
+  <span translate="settings.config.title_ocr"></span>
+  <span class="label" ng-class="{ 'label-success': app.ocr_enabled, 'label-danger': !app.ocr_enabled }">
+    {{ app.ocr_enabled ? 'enabled' : 'disabled' | translate }}
+</h2>
+<p translate="settings.config.message_ocr" translate-values="{ appName: appName }"></p>
+<div ng-if="app">
+  <button ng-if="!app.ocr_enabled" class="btn btn-primary" ng-click="changeOcrEnabled(true)">{{ 'settings.config.enable_ocr' | translate }}</button>
+  <button ng-if="app.ocr_enabled" class="btn btn-danger" ng-click="changeOcrEnabled(false)">{{ 'settings.config.disable_ocr' | translate }}</button>
+</div>
+
 <h2 translate="settings.config.title_theme"></h2>
 <form class="form-horizontal" name="editColorForm" novalidate>
   <div class="form-group">

docs-web/src/prod/resources/config.properties

@@ -1,3 +1,3 @@
 api.current_version=${project.version}
 api.min_version=1.0
-db.version=30
+db.version=31

docs-web/src/test/java/com/sismics/docs/rest/TestAppResource.java

@@ -58,6 +58,7 @@ public class TestAppResource extends BaseJerseyTest {
         Assert.assertTrue(totalMemory > 0 && totalMemory > freeMemory);
         Assert.assertEquals(0, json.getJsonNumber("queued_tasks").intValue());
         Assert.assertFalse(json.getBoolean("guest_login"));
+        Assert.assertFalse(json.getBoolean("ocr_enabled"));
         Assert.assertEquals("eng", json.getString("default_language"));
         Assert.assertTrue(json.containsKey("global_storage_current"));
         Assert.assertTrue(json.getJsonNumber("active_user_count").longValue() > 0);
@@ -202,6 +203,37 @@ public class TestAppResource extends BaseJerseyTest {
                         .param("enabled", "false")), JsonObject.class);
     }
 
+    /**
+     * Test the ocr setting
+     */
+    @Test
+    public void testOcrSetting() {
+        // Login admin
+        String adminToken = adminToken();
+
+        // Get OCR configuration
+        JsonObject json = target().path("/app/ocr").request()
+                .cookie(TokenBasedSecurityFilter.COOKIE_NAME, adminToken)
+                .get(JsonObject.class);
+        if (!configInboxChanged) {
+            Assert.assertFalse(json.getBoolean("enabled"));
+        }
+
+        // Change OCR configuration
+        target().path("/app/ocr").request()
+                .cookie(TokenBasedSecurityFilter.COOKIE_NAME, adminToken)
+                .post(Entity.form(new Form()
+                        .param("enabled", "true")
+                ), JsonObject.class);
+        configInboxChanged = true;
+
+        // Get OCR configuration
+        json = target().path("/app/ocr").request()
+                .cookie(TokenBasedSecurityFilter.COOKIE_NAME, adminToken)
+                .get(JsonObject.class);
+        Assert.assertTrue(json.getBoolean("enabled"));
+    }
+
     /**
      * Test SMTP configuration changes.
      */

scripts/dev_setup.sh

@@ -0,0 +1,41 @@
+#!/bin/bash
+
+install_macos_dependencies() {
+  brew install openjdk@11
+  export JAVA_HOME=/usr/local/opt/openjdk@11
+  brew install maven
+  brew install node
+  npm install -g grunt-cli
+  brew install tesseract
+  brew install ffmpeg
+  brew install mediainfo
+}
+
+install_ubuntu_debian_dependencies() {
+  sudo apt update
+  sudo apt install -y openjdk-11-jdk maven nodejs npm tesseract-ocr ffmpeg mediainfo
+  sudo npm install -g grunt-cli
+}
+
+install_fedora_dependencies() {
+  sudo dnf install -y java-11-openjdk-devel maven nodejs npm tesseract ffmpeg mediainfo
+  sudo npm install -g grunt-cli
+}
+
+echo "📥 Installing dependencies"
+if [[ "$OSTYPE" == "darwin"* ]]; then
+  install_macos_dependencies
+elif [[ -f /etc/debian_version ]]; then
+  install_ubuntu_debian_dependencies
+elif [[ -f /etc/fedora-release ]]; then
+  install_fedora_dependencies
+else
+  echo "❔ Unsupported OS. Feel free to contribute!"
+  exit 1
+fi
+
+echo "✅ Dependencies installed."
+
+mvn clean -DskipTests install
+
+echo "You can start the server with 'mvn jetty:run' and then access it at http://localhost:8080/docs-web/src/"