RED-4510: Added generated RedactionLog Json files and optimizations for RulesTest.java for Bamboo

.gitignore

@@ -26,4 +26,3 @@
 **/.DS_Store
 **/classpath-data.json
 **/dependencies-and-licenses-overview.txt
-/redaction-service-v1/redaction-service-server-v1/src/test/resources/RedactionLog/

redaction-service-v1/redaction-service-server-v1/src/test/java/com/iqser/red/service/redaction/v1/server/RulesTest.java

@@ -27,7 +27,6 @@ import java.util.Map;
 import java.util.Set;
 import java.util.stream.Collectors;
 
-import org.apache.commons.io.IOUtils;
 import org.apache.commons.lang3.StringUtils;
 import org.junit.After;
 import org.junit.Before;
@@ -114,10 +113,13 @@ public class RulesTest {
     private static final String IMPORTED_REDACTION = "imported_redaction";
     private static final String PII = "PII";
     private static final String RESOURCES_PATH = "src/test/resources/";
-    private static final String REDACTION_LOG_PATH = RESOURCES_PATH + "RedactionLog/";
+    private static final String REDACTION_LOG_PATH = "RedactionLog/";
+    private static final String REDACTION_LOG_RESOURCES_PATH = RESOURCES_PATH + REDACTION_LOG_PATH;
     private final static String TEST_DOSSIER_TEMPLATE_ID = "123";
     private final static String TEST_DOSSIER_ID = "123";
     private String TEST_FILE_ID = "123";
+    private int counter = 0;
+    private int fileSize = 0;
     private final Map<String, List<String>> dictionary = new HashMap<>();
     private final Map<String, List<String>> dossierDictionary = new HashMap<>();
     private final Map<String, List<String>> falsePositive = new HashMap<>();
@@ -176,7 +178,9 @@ public class RulesTest {
                 .hexColor("#ffe187")
                 .isHint(hintTypeMap.get(DOSSIER_REDACTIONS))
                 .isCaseInsensitive(caseInSensitiveMap.get(DOSSIER_REDACTIONS))
-                .isRecommendation(recommendationTypeMap.get(DOSSIER_REDACTIONS)).rank(rankTypeMap.get(DOSSIER_REDACTIONS)).build()));
+                .isRecommendation(recommendationTypeMap.get(DOSSIER_REDACTIONS))
+                .rank(rankTypeMap.get(DOSSIER_REDACTIONS))
+                .build()));
 
         mockDictionaryCalls(null);
         mockDictionaryCalls(0L);
@@ -196,7 +200,7 @@ public class RulesTest {
 
     /**
      * Generates RedactionLog for given file and saves it here: REDACTION_LOG_PATH.
-     * If the RedactionLog already exists, the generating will be skipped
+     * If the RedactionLog already exists, it will be overwritten
      * Test is ignored, because it's for manual tests.
      */
     @Ignore
@@ -204,20 +208,22 @@ public class RulesTest {
     public void generateRedactionLogForOneFile() {
 
         String fileName = "files/Compounds/31 A14111B - EU AIR3 - MCP Section 1 - Identity of the plant protection product.pdf";
-        generateAndSaveRedactionLog(fileName);
+        fileSize = 1;
+        generateAndSaveRedactionLog(fileName, true);
     }
 
 
     /**
      * Generates RedactionLog for all files and saves it here: REDACTION_LOG_PATH.
-     * If a RedactionLog already exists, the generating will be skipped
+     * If a RedactionLog already exists, the generating for this will be skipped
+     * Please commit generated files, if not the following tests will fail
      */
-    @IfProfileValue(name = "test-groups", value = "rules-test")
     @Test
     public void generateRedactionLogForAllFiles() {
 
         Set<String> files = getFileNames(new HashSet<>(), FileSystems.getDefault().getPath(RESOURCES_PATH));
         System.out.println("Will generate RedactionLog for " + files.size() + " files.");
+        fileSize = files.size();
         files.forEach(this::generateAndSaveRedactionLog);
     }
 
@@ -225,20 +231,19 @@ public class RulesTest {
     /**
      * Analyses file and compares its RedactionLog with saved one from here: REDACTION_LOG_PATH.
      * If RedactionLog Json does not exist, test will fail.
-     * Test is ignored, because it's for manual tests.
      */
-    @Ignore
     @Test
     public void analyseFileAndCompareRedactionLog() {
 
-        String fileName = "files/Compounds/31 A14111B - EU AIR3 - MCP Section 1 - Identity of the plant protection product.pdf";
+        String fileName = "files/RulesTest/SYNGENTA_EFSA_sanitisation_GFL_v1_withHighlights.pdf";
+        fileSize = 1;
         analyseFileAndCompareRedactionLog(fileName);
     }
 
 
     /**
      * Analyses all files and compares its RedactionLog with saved one from here: REDACTION_LOG_PATH.
-     * If RedactionLogs Json does not exist, test will fail.
+     * If RedactionLog Json for one file does not exist, whole test will fail.
      */
     @IfProfileValue(name = "test-groups", value = "rules-test")
     @Test
@@ -246,20 +251,28 @@ public class RulesTest {
 
         Set<String> files = getFileNames(new HashSet<>(), FileSystems.getDefault().getPath(RESOURCES_PATH));
         System.out.println("Will analyse " + files.size() + " files and compare its RedactionLogs.");
+        fileSize = files.size();
         files.forEach(this::analyseFileAndCompareRedactionLog);
     }
 
 
-    @SneakyThrows
     private void generateAndSaveRedactionLog(String fileName) {
 
-        if (doesRedactionLogExist(fileName)) {
-            System.out.println("Skip generating RedactionLog as Json for " + fileName + " because it already exists.");
+        generateAndSaveRedactionLog(fileName, false);
+    }
+
+
+    @SneakyThrows
+    private void generateAndSaveRedactionLog(String fileName, boolean force) {
+
+        counter++;
+        if (!force && doesRedactionLogExist(fileName)) {
+            System.out.println("(" + counter + "/" + fileSize + ") Skip generating RedactionLog as Json for " + fileName + " because it already exists.");
 
         } else {
             generateTestFileId(fileName);
 
-            System.out.println("Generate RedactionLog as Json for " + fileName + " with fileId " + TEST_FILE_ID);
+            System.out.println("(" + counter + "/" + fileSize + ") Generate RedactionLog as Json for " + fileName + " with fileId " + TEST_FILE_ID);
 
             loadNerForTest();
 
@@ -278,8 +291,9 @@ public class RulesTest {
     @SneakyThrows
     public void analyseFileAndCompareRedactionLog(String fileName) {
 
+        counter++;
         generateTestFileId(fileName);
-        System.out.println("Analyse " + fileName + " with fileId " + TEST_FILE_ID + " and compare it with its saved RedactionLog");
+        System.out.println("(" + counter + "/" + fileSize + ") Analyse " + fileName + " with fileId " + TEST_FILE_ID + " and compare it with its saved RedactionLog");
 
         RedactionLog savedRedactionLog = loadSavedRedactionLog(fileName);
 
@@ -387,7 +401,7 @@ public class RulesTest {
                 assertThat(savedManualChange).isPresent();
             }
 
-            assertThat(savedRedactionLogEntry.get().getEngines()).containsExactly(redactionLogEntry.getEngines().toArray(new Engine[0]));
+            assertThat(savedRedactionLogEntry.get().getEngines()).containsExactlyInAnyOrder(redactionLogEntry.getEngines().toArray(new Engine[0]));
 
             assertThat(savedRedactionLogEntry.get().getReference()).containsAll(redactionLogEntry.getReference());
             assertThat(savedRedactionLogEntry.get().getImportedRedactionIntersections()).containsAll(redactionLogEntry.getImportedRedactionIntersections());
@@ -399,7 +413,7 @@ public class RulesTest {
     private boolean doesRedactionLogExist(String pdfFileName) {
 
         File pdfFile = new File(pdfFileName);
-        String directory = REDACTION_LOG_PATH + pdfFile.getParentFile().getPath();
+        String directory = REDACTION_LOG_RESOURCES_PATH + pdfFile.getParentFile().getPath();
         String fileName = StringUtils.replace(pdfFile.getName(), ".pdf", ".json");
         File file = new File(directory, fileName);
         return file.exists();
@@ -421,7 +435,7 @@ public class RulesTest {
 
         File pdfFile = new File(pdfFileName);
 
-        String directory = REDACTION_LOG_PATH + pdfFile.getParentFile().getPath();
+        String directory = REDACTION_LOG_RESOURCES_PATH + pdfFile.getParentFile().getPath();
         File dr = new File(directory);
         boolean created = dr.mkdirs();
         if (created) {
@@ -441,8 +455,20 @@ public class RulesTest {
     private RedactionLog loadSavedRedactionLog(String pdfFileName) {
 
         File pdfFile = new File(pdfFileName);
-        String directory = REDACTION_LOG_PATH + pdfFile.getParentFile().getPath();
+        String directory = cleanPath(pdfFile.getParentFile().getPath());
+        if (StringUtils.contains(directory, RESOURCES_PATH)) {
+            if (!StringUtils.endsWith(directory, "/")) {
+                directory = directory + "/";
+            }
+            directory = directory + REDACTION_LOG_PATH;
+        } else {
+            if (StringUtils.startsWith(directory, "/")) {
+                directory = StringUtils.substring(directory, 1);
+            }
+            directory = REDACTION_LOG_RESOURCES_PATH + directory;
+        }
         String fileName = StringUtils.replace(pdfFile.getName(), ".pdf", ".json");
+
         File file = new File(directory, fileName);
         assertThat(file).exists();
 
@@ -450,6 +476,31 @@ public class RulesTest {
     }
 
 
+    @SneakyThrows
+    private Set<String> getFileNames(Set<String> fileNames, Path dir) {
+
+        try (DirectoryStream<Path> stream = Files.newDirectoryStream(dir)) {
+            for (Path path : stream) {
+                if (path.toFile().isDirectory()) {
+                    getFileNames(fileNames, path);
+                } else if (StringUtils.endsWith(path.toAbsolutePath().toString(), ".pdf")) {
+                    String absolutePath = cleanPath(path.toAbsolutePath().toString());
+                    int pos = StringUtils.indexOf(absolutePath, RESOURCES_PATH) + RESOURCES_PATH.length();
+
+                    fileNames.add(StringUtils.substring(absolutePath, pos));
+                }
+            }
+        }
+        return fileNames;
+    }
+
+
+    private String cleanPath(String path) {
+
+        return StringUtils.replace(path, "\\", "/");
+    }
+
+
     @SneakyThrows
     private void loadNerForTest() {
 
@@ -588,24 +639,6 @@ public class RulesTest {
     }
 
 
-    @SneakyThrows
-    private Set<String> getFileNames(Set<String> fileNames, Path dir) {
-
-        try (DirectoryStream<Path> stream = Files.newDirectoryStream(dir)) {
-            for (Path path : stream) {
-                if (path.toFile().isDirectory()) {
-                    getFileNames(fileNames, path);
-                } else if (StringUtils.endsWith(path.toAbsolutePath().toString(), ".pdf")) {
-                    String absolutePath = path.toAbsolutePath().toString();
-                    int pos = StringUtils.indexOf(absolutePath, StringUtils.replace(RESOURCES_PATH, "/", "\\")) + 18;
-                    fileNames.add(StringUtils.substring(absolutePath, pos));
-                }
-            }
-        }
-        return fileNames;
-    }
-
-
     private Type getDictionaryResponse(String type, boolean isDossierDictionary) {
 
         return Type.builder()

redaction-service-v1/redaction-service-server-v1/src/test/resources/RedactionLog/files/Compounds/27 A8637C - EU AIR3 - MCP Section 1 - Identity of the plant protection product.json

@@ -0,0 +1,533 @@
+{
+  "analysisVersion": 1,
+  "analysisNumber": 0,
+  "redactionLogEntry": [
+    {
+      "id": "96c1362cb4ea87461d8f5f141df1912a",
+      "type": "PII",
+      "value": "patrick.gardinal@syngenta.com",
+      "reason": "Personal information found",
+      "matchedRule": 23,
+      "rectangle": false,
+      "legalBasis": "Article 39(e)(3) of Regulation (EC) No 178/2002",
+      "imported": false,
+      "redacted": true,
+      "section": "CP 1.2 Producer of the Plant Protection Product and the Active\nSubstances",
+      "color": [
+        0.4,
+        0.8,
+        1.0
+      ],
+      "positions": [
+        {
+          "topLeft": {
+            "x": 286.92,
+            "y": 368.44
+          },
+          "width": 141.39014,
+          "height": 11.52,
+          "page": 5
+        }
+      ],
+      "sectionNumber": 10,
+      "textBefore": "60 51 E-mail: ",
+      "textAfter": null,
+      "comments": [],
+      "startOffset": 295,
+      "endOffset": 324,
+      "imageHasTransparency": false,
+      "excluded": false,
+      "sourceId": null,
+      "changes": [
+        {
+          "analysisNumber": 0,
+          "type": "ADDED",
+          "dateTime": "2022-07-25T11:12:04.3835183+02:00"
+        }
+      ],
+      "manualChanges": [],
+      "engines": [
+        "RULE"
+      ],
+      "reference": [],
+      "importedRedactionIntersections": [],
+      "dossierDictionaryEntry": false,
+      "localManualRedaction": false,
+      "dictionaryEntry": false,
+      "image": false,
+      "falsePositive": false,
+      "manuallyRemoved": false,
+      "hint": false,
+      "recommendation": false
+    },
+    {
+      "id": "74b528293158d5266ace1eaf9ee2ad80",
+      "type": "PII",
+      "value": "+41 (0) 61 323 60 51",
+      "reason": "Personal information found",
+      "matchedRule": 23,
+      "rectangle": false,
+      "legalBasis": "Article 39(e)(3) of Regulation (EC) No 178/2002",
+      "imported": false,
+      "redacted": true,
+      "section": "Producer of the active substance: Cyprodinil",
+      "color": [
+        0.4,
+        0.8,
+        1.0
+      ],
+      "positions": [
+        {
+          "topLeft": {
+            "x": 286.92,
+            "y": 169.96002
+          },
+          "width": 94.37213,
+          "height": 11.52,
+          "page": 5
+        }
+      ],
+      "sectionNumber": 12,
+      "textBefore": "Gardinal Telephone number: ",
+      "textAfter": " E-mail: patrick.gardinal@syngenta.com",
+      "comments": [],
+      "startOffset": 188,
+      "endOffset": 208,
+      "imageHasTransparency": false,
+      "excluded": false,
+      "sourceId": null,
+      "changes": [
+        {
+          "analysisNumber": 0,
+          "type": "ADDED",
+          "dateTime": "2022-07-25T11:12:04.3835183+02:00"
+        }
+      ],
+      "manualChanges": [],
+      "engines": [
+        "RULE"
+      ],
+      "reference": [],
+      "importedRedactionIntersections": [],
+      "dossierDictionaryEntry": false,
+      "localManualRedaction": false,
+      "dictionaryEntry": false,
+      "image": false,
+      "falsePositive": false,
+      "manuallyRemoved": false,
+      "hint": false,
+      "recommendation": false
+    },
+    {
+      "id": "d282cd8f02e2628c14342ed52d69c517",
+      "type": "PII",
+      "value": "+41 (61) 323 6152",
+      "reason": "Personal information found",
+      "matchedRule": 23,
+      "rectangle": false,
+      "legalBasis": "Article 39(e)(3) of Regulation (EC) No 178/2002",
+      "imported": false,
+      "redacted": true,
+      "section": "CP 1.1 Applicant",
+      "color": [
+        0.4,
+        0.8,
+        1.0
+      ],
+      "positions": [
+        {
+          "topLeft": {
+            "x": 286.92,
+            "y": 610.12
+          },
+          "width": 83.43994,
+          "height": 11.52,
+          "page": 5
+        }
+      ],
+      "sectionNumber": 9,
+      "textBefore": "Micheletti Telephone number: ",
+      "textAfter": " E-mail: sebastien.micheletti@syngenta.com",
+      "comments": [],
+      "startOffset": 164,
+      "endOffset": 181,
+      "imageHasTransparency": false,
+      "excluded": false,
+      "sourceId": null,
+      "changes": [
+        {
+          "analysisNumber": 0,
+          "type": "ADDED",
+          "dateTime": "2022-07-25T11:12:04.3835183+02:00"
+        }
+      ],
+      "manualChanges": [],
+      "engines": [
+        "RULE"
+      ],
+      "reference": [],
+      "importedRedactionIntersections": [],
+      "dossierDictionaryEntry": false,
+      "localManualRedaction": false,
+      "dictionaryEntry": false,
+      "image": false,
+      "falsePositive": false,
+      "manuallyRemoved": false,
+      "hint": false,
+      "recommendation": false
+    },
+    {
+      "id": "245cfd060d4fb078bb211fabfce6d470",
+      "type": "PII",
+      "value": "patrick.gardinal@syngenta.com",
+      "reason": "Personal information found",
+      "matchedRule": 23,
+      "rectangle": false,
+      "legalBasis": "Article 39(e)(3) of Regulation (EC) No 178/2002",
+      "imported": false,
+      "redacted": true,
+      "section": "Producer of the active substance: Cyprodinil",
+      "color": [
+        0.4,
+        0.8,
+        1.0
+      ],
+      "positions": [
+        {
+          "topLeft": {
+            "x": 286.92,
+            "y": 144.64001
+          },
+          "width": 141.39014,
+          "height": 11.52,
+          "page": 5
+        }
+      ],
+      "sectionNumber": 12,
+      "textBefore": "60 51 E-mail: ",
+      "textAfter": null,
+      "comments": [],
+      "startOffset": 217,
+      "endOffset": 246,
+      "imageHasTransparency": false,
+      "excluded": false,
+      "sourceId": null,
+      "changes": [
+        {
+          "analysisNumber": 0,
+          "type": "ADDED",
+          "dateTime": "2022-07-25T11:12:04.3835183+02:00"
+        }
+      ],
+      "manualChanges": [],
+      "engines": [
+        "RULE"
+      ],
+      "reference": [],
+      "importedRedactionIntersections": [],
+      "dossierDictionaryEntry": false,
+      "localManualRedaction": false,
+      "dictionaryEntry": false,
+      "image": false,
+      "falsePositive": false,
+      "manuallyRemoved": false,
+      "hint": false,
+      "recommendation": false
+    },
+    {
+      "id": "67dc43606b8e1489a0594e5b779fa428",
+      "type": "PII",
+      "value": "Patrick Gardinal",
+      "reason": "Personal information found",
+      "matchedRule": 23,
+      "rectangle": false,
+      "legalBasis": "Article 39(e)(3) of Regulation (EC) No 178/2002",
+      "imported": false,
+      "redacted": true,
+      "section": "CP 1.2 Producer of the Plant Protection Product and the Active\nSubstances",
+      "color": [
+        0.4,
+        0.8,
+        1.0
+      ],
+      "positions": [
+        {
+          "topLeft": {
+            "x": 286.92,
+            "y": 419.08
+          },
+          "width": 73.440186,
+          "height": 11.52,
+          "page": 5
+        }
+      ],
+      "sectionNumber": 10,
+      "textBefore": "Basel; Switzerland Contact: ",
+      "textAfter": " Telephone number:",
+      "comments": [],
+      "startOffset": 231,
+      "endOffset": 247,
+      "imageHasTransparency": false,
+      "excluded": false,
+      "sourceId": null,
+      "changes": [
+        {
+          "analysisNumber": 0,
+          "type": "ADDED",
+          "dateTime": "2022-07-25T11:12:04.3835183+02:00"
+        }
+      ],
+      "manualChanges": [],
+      "engines": [
+        "RULE"
+      ],
+      "reference": [],
+      "importedRedactionIntersections": [],
+      "dossierDictionaryEntry": false,
+      "localManualRedaction": false,
+      "dictionaryEntry": false,
+      "image": false,
+      "falsePositive": false,
+      "manuallyRemoved": false,
+      "hint": false,
+      "recommendation": false
+    },
+    {
+      "id": "de79428457282a0781a827b322b5732a",
+      "type": "PII",
+      "value": "sebastien.micheletti@syngenta.com",
+      "reason": "Personal information found",
+      "matchedRule": 23,
+      "rectangle": false,
+      "legalBasis": "Article 39(e)(3) of Regulation (EC) No 178/2002",
+      "imported": false,
+      "redacted": true,
+      "section": "CP 1.1 Applicant",
+      "color": [
+        0.4,
+        0.8,
+        1.0
+      ],
+      "positions": [
+        {
+          "topLeft": {
+            "x": 286.92,
+            "y": 584.8
+          },
+          "width": 161.04922,
+          "height": 10.98,
+          "page": 5
+        }
+      ],
+      "sectionNumber": 9,
+      "textBefore": "323 6152 E-mail: ",
+      "textAfter": null,
+      "comments": [],
+      "startOffset": 190,
+      "endOffset": 223,
+      "imageHasTransparency": false,
+      "excluded": false,
+      "sourceId": null,
+      "changes": [
+        {
+          "analysisNumber": 0,
+          "type": "ADDED",
+          "dateTime": "2022-07-25T11:12:04.3835183+02:00"
+        }
+      ],
+      "manualChanges": [],
+      "engines": [
+        "RULE"
+      ],
+      "reference": [],
+      "importedRedactionIntersections": [],
+      "dossierDictionaryEntry": false,
+      "localManualRedaction": false,
+      "dictionaryEntry": false,
+      "image": false,
+      "falsePositive": false,
+      "manuallyRemoved": false,
+      "hint": false,
+      "recommendation": false
+    },
+    {
+      "id": "41340ab190d66ca6dce0ba965ce95f59",
+      "type": "PII",
+      "value": "Patrick Gardinal",
+      "reason": "Personal information found",
+      "matchedRule": 23,
+      "rectangle": false,
+      "legalBasis": "Article 39(e)(3) of Regulation (EC) No 178/2002",
+      "imported": false,
+      "redacted": true,
+      "section": "Producer of the active substance: Cyprodinil",
+      "color": [
+        0.4,
+        0.8,
+        1.0
+      ],
+      "positions": [
+        {
+          "topLeft": {
+            "x": 286.92,
+            "y": 195.28003
+          },
+          "width": 73.440186,
+          "height": 11.52,
+          "page": 5
+        }
+      ],
+      "sectionNumber": 12,
+      "textBefore": "Basel; Switzerland Contact: ",
+      "textAfter": " Telephone number:",
+      "comments": [],
+      "startOffset": 153,
+      "endOffset": 169,
+      "imageHasTransparency": false,
+      "excluded": false,
+      "sourceId": null,
+      "changes": [
+        {
+          "analysisNumber": 0,
+          "type": "ADDED",
+          "dateTime": "2022-07-25T11:12:04.3835183+02:00"
+        }
+      ],
+      "manualChanges": [],
+      "engines": [
+        "RULE"
+      ],
+      "reference": [],
+      "importedRedactionIntersections": [],
+      "dossierDictionaryEntry": false,
+      "localManualRedaction": false,
+      "dictionaryEntry": false,
+      "image": false,
+      "falsePositive": false,
+      "manuallyRemoved": false,
+      "hint": false,
+      "recommendation": false
+    },
+    {
+      "id": "02dd2efb100839e705cdcf6f3141d9d5",
+      "type": "PII",
+      "value": "Sebastien Micheletti",
+      "reason": "Personal information found",
+      "matchedRule": 23,
+      "rectangle": false,
+      "legalBasis": "Article 39(e)(3) of Regulation (EC) No 178/2002",
+      "imported": false,
+      "redacted": true,
+      "section": "CP 1.1 Applicant",
+      "color": [
+        0.4,
+        0.8,
+        1.0
+      ],
+      "positions": [
+        {
+          "topLeft": {
+            "x": 286.92,
+            "y": 634.84
+          },
+          "width": 91.21478,
+          "height": 11.52,
+          "page": 5
+        }
+      ],
+      "sectionNumber": 9,
+      "textBefore": "Basel; Switzerland Contact: ",
+      "textAfter": " Telephone number:",
+      "comments": [],
+      "startOffset": 125,
+      "endOffset": 145,
+      "imageHasTransparency": false,
+      "excluded": false,
+      "sourceId": null,
+      "changes": [
+        {
+          "analysisNumber": 0,
+          "type": "ADDED",
+          "dateTime": "2022-07-25T11:12:04.3835183+02:00"
+        }
+      ],
+      "manualChanges": [],
+      "engines": [
+        "RULE"
+      ],
+      "reference": [],
+      "importedRedactionIntersections": [],
+      "dossierDictionaryEntry": false,
+      "localManualRedaction": false,
+      "dictionaryEntry": false,
+      "image": false,
+      "falsePositive": false,
+      "manuallyRemoved": false,
+      "hint": false,
+      "recommendation": false
+    },
+    {
+      "id": "530c4f47f0afc48432bd849a90918f41",
+      "type": "PII",
+      "value": "+41 (0) 61 323 60 51",
+      "reason": "Personal information found",
+      "matchedRule": 23,
+      "rectangle": false,
+      "legalBasis": "Article 39(e)(3) of Regulation (EC) No 178/2002",
+      "imported": false,
+      "redacted": true,
+      "section": "CP 1.2 Producer of the Plant Protection Product and the Active\nSubstances",
+      "color": [
+        0.4,
+        0.8,
+        1.0
+      ],
+      "positions": [
+        {
+          "topLeft": {
+            "x": 286.92,
+            "y": 393.76
+          },
+          "width": 94.37213,
+          "height": 11.52,
+          "page": 5
+        }
+      ],
+      "sectionNumber": 10,
+      "textBefore": "Gardinal Telephone number: ",
+      "textAfter": " E-mail: patrick.gardinal@syngenta.com",
+      "comments": [],
+      "startOffset": 266,
+      "endOffset": 286,
+      "imageHasTransparency": false,
+      "excluded": false,
+      "sourceId": null,
+      "changes": [
+        {
+          "analysisNumber": 0,
+          "type": "ADDED",
+          "dateTime": "2022-07-25T11:12:04.3835183+02:00"
+        }
+      ],
+      "manualChanges": [],
+      "engines": [
+        "RULE"
+      ],
+      "reference": [],
+      "importedRedactionIntersections": [],
+      "dossierDictionaryEntry": false,
+      "localManualRedaction": false,
+      "dictionaryEntry": false,
+      "image": false,
+      "falsePositive": false,
+      "manuallyRemoved": false,
+      "hint": false,
+      "recommendation": false
+    }
+  ],
+  "legalBasis": [],
+  "dictionaryVersion": 0,
+  "dossierDictionaryVersion": 0,
+  "rulesVersion": 0,
+  "legalBasisVersion": 0
+}

redaction-service-v1/redaction-service-server-v1/src/test/resources/RedactionLog/files/Cyprodinil/39 Cyprodinil - EU AIR3 - Document F - Notification submitted to the Commission.json

@@ -0,0 +1 @@
+{"analysisVersion":1,"analysisNumber":0,"redactionLogEntry":[{"id":"12ac3d26ce82a77e469c64d56081662d","type":"CBI_author","value":"Micheletti S","reason":"Author found","matchedRule":1,"rectangle":false,"legalBasis":"Article 39(e)(3) of Regulation (EC) No 178/2002","imported":false,"redacted":true,"section":"Table in: COPY OF THE APPLICATION FOR RENEWAL, MADE UNDER COMMISSION","color":[1.0,0.88235295,0.5294118],"positions":[{"topLeft":{"x":214.8,"y":639.64},"width":55.039078,"height":11.52,"page":4}],"sectionNumber":3,"textBefore":"Report: Document F/02, ","textAfter":". (2014). Cyprodinil","comments":[],"startOffset":23,"endOffset":35,"imageHasTransparency":false,"excluded":false,"sourceId":null,"changes":[{"analysisNumber":0,"type":"ADDED","dateTime":1658398559.316208600}],"manualChanges":[],"engines":["DICTIONARY"],"reference":[],"importedRedactionIntersections":[],"dossierDictionaryEntry":false,"recommendation":false,"hint":false,"dictionaryEntry":true,"falsePositive":false,"image":false,"localManualRedaction":false,"manuallyRemoved":false},{"id":"dd614353319e7c5a8cf8b3346dc12fac","type":"CBI_author","value":"Micheletti S","reason":"Author found","matchedRule":1,"rectangle":false,"legalBasis":"Article 39(e)(3) of Regulation (EC) No 178/2002","imported":false,"redacted":true,"section":"COPY OF THE APPLICATION FOR RENEWAL, MADE UNDER COMMISSION","color":[1.0,0.88235295,0.5294118],"positions":[{"topLeft":{"x":306.84,"y":667.12},"width":60.71997,"height":12.0,"page":4}],"sectionNumber":5,"textBefore":"F report (","textAfter":"., 2014).","comments":[],"startOffset":247,"endOffset":259,"imageHasTransparency":false,"excluded":false,"sourceId":null,"changes":[{"analysisNumber":0,"type":"ADDED","dateTime":1658398559.316208600}],"manualChanges":[],"engines":["DICTIONARY"],"reference":[],"importedRedactionIntersections":[],"dossierDictionaryEntry":false,"recommendation":false,"hint":false,"dictionaryEntry":true,"falsePositive":false,"image":false,"localManualRedaction":false,"manuallyRemoved":false}],"legalBasis":[],"dictionaryVersion":0,"dossierDictionaryVersion":0,"rulesVersion":0,"legalBasisVersion":0}

redaction-service-v1/redaction-service-server-v1/src/test/resources/RedactionLog/files/Minimal Examples/270Rotated.json

@@ -0,0 +1 @@
+{"analysisVersion":1,"analysisNumber":0,"redactionLogEntry":[{"id":"36a5e8909a18a5080dd423afb8efd7a7","type":"PII","value":"Page","reason":"Personal information found","matchedRule":19,"rectangle":false,"legalBasis":"Article 39(e)(3) of Regulation (EC) No 178/2002","imported":false,"redacted":true,"section":"Header","color":[0.4,0.8,1.0],"positions":[{"topLeft":{"x":159.0,"y":320.0},"width":10.208038,"height":-21.999977,"page":1}],"sectionNumber":3,"textBefore":"Product Metabolism and ","textAfter":" 1 of","comments":[],"startOffset":368,"endOffset":372,"imageHasTransparency":false,"excluded":false,"sourceId":null,"changes":[{"analysisNumber":0,"type":"ADDED","dateTime":1658398772.538085900}],"manualChanges":[],"engines":["DICTIONARY"],"reference":[],"importedRedactionIntersections":[],"dossierDictionaryEntry":false,"recommendation":false,"hint":false,"dictionaryEntry":true,"falsePositive":false,"image":false,"localManualRedaction":false,"manuallyRemoved":false},{"id":"dca119c3aadd348bc43c0b700c0075c3","type":"PII","value":"Page","reason":"Personal information found","matchedRule":19,"rectangle":false,"legalBasis":"Article 39(e)(3) of Regulation (EC) No 178/2002","imported":false,"redacted":true,"section":"Header","color":[0.4,0.8,1.0],"positions":[{"topLeft":{"x":100.0,"y":81.0},"width":10.208031,"height":-22.0,"page":1}],"sectionNumber":3,"textBefore":"Pages RAM 465/02 ","textAfter":" 1 Tel:","comments":[],"startOffset":399,"endOffset":403,"imageHasTransparency":false,"excluded":false,"sourceId":null,"changes":[{"analysisNumber":0,"type":"ADDED","dateTime":1658398772.538085900}],"manualChanges":[],"engines":["DICTIONARY"],"reference":[],"importedRedactionIntersections":[],"dossierDictionaryEntry":false,"recommendation":false,"hint":false,"dictionaryEntry":true,"falsePositive":false,"image":false,"localManualRedaction":false,"manuallyRemoved":false},{"id":"ef6e0f0ed413dbd7299d992d9a50840b","type":"CBI_author","value":"Crook","reason":"Author found","matchedRule":1,"rectangle":false,"legalBasis":"Article 39(e)(3) of Regulation (EC) No 178/2002","imported":false,"redacted":true,"section":"","color":[1.0,0.88235295,0.5294118],"positions":[{"topLeft":{"x":374.0,"y":389.0},"width":10.515198,"height":-27.02501,"page":1}],"sectionNumber":5,"textBefore":"Author : SJ ","textAfter":" Analytical Science","comments":[],"startOffset":12,"endOffset":17,"imageHasTransparency":false,"excluded":false,"sourceId":null,"changes":[{"analysisNumber":0,"type":"ADDED","dateTime":1658398772.538085900}],"manualChanges":[],"engines":["DICTIONARY"],"reference":[],"importedRedactionIntersections":[],"dossierDictionaryEntry":false,"recommendation":false,"hint":false,"dictionaryEntry":true,"falsePositive":false,"image":false,"localManualRedaction":false,"manuallyRemoved":false},{"id":"0fac36714dd9d1c745565c6dfe06b9b7","type":"PII","value":"Page","reason":"Personal information found","matchedRule":19,"rectangle":false,"legalBasis":"Article 39(e)(3) of Regulation (EC) No 178/2002","imported":false,"redacted":true,"section":"Header","color":[0.4,0.8,1.0],"positions":[{"topLeft":{"x":514.14215,"y":47.94165},"width":24.39972,"height":11.591161,"page":2}],"sectionNumber":4,"textBefore":"RAM 465/02 ","textAfter":" 2 Summary","comments":[],"startOffset":11,"endOffset":15,"imageHasTransparency":false,"excluded":false,"sourceId":null,"changes":[{"analysisNumber":0,"type":"ADDED","dateTime":1658398772.538085900}],"manualChanges":[],"engines":["DICTIONARY"],"reference":[],"importedRedactionIntersections":[],"dossierDictionaryEntry":false,"recommendation":false,"hint":false,"dictionaryEntry":true,"falsePositive":false,"image":false,"localManualRedaction":false,"manuallyRemoved":false}],"legalBasis":[],"dictionaryVersion":0,"dossierDictionaryVersion":0,"rulesVersion":0,"legalBasisVersion":0}

redaction-service-v1/redaction-service-server-v1/src/test/resources/RedactionLog/files/Minimal Examples/Complex Table.json

@@ -0,0 +1 @@
+{"analysisVersion":1,"analysisNumber":0,"redactionLogEntry":[],"legalBasis":[],"dictionaryVersion":0,"dossierDictionaryVersion":0,"rulesVersion":0,"legalBasisVersion":0}

redaction-service-v1/redaction-service-server-v1/src/test/resources/RedactionLog/files/Minimal Examples/Direction90Rotation0.json

@@ -0,0 +1 @@
+{"analysisVersion":1,"analysisNumber":0,"redactionLogEntry":[{"id":"71770acab1fa63f46c026966890b5231","type":"vertebrate","value":"rabbit","reason":null,"matchedRule":0,"rectangle":false,"legalBasis":null,"imported":false,"redacted":false,"section":"","color":[1.0,0.52156866,0.96862745],"positions":[{"topLeft":{"x":77.73999,"y":537.62},"width":-11.043678,"height":36.630165,"page":1}],"sectionNumber":4,"textBefore":null,"textAfter":null,"comments":[],"startOffset":41,"endOffset":47,"imageHasTransparency":false,"excluded":false,"sourceId":null,"changes":[{"analysisNumber":0,"type":"ADDED","dateTime":1658398329.875800000}],"manualChanges":[],"engines":["DICTIONARY"],"reference":[],"importedRedactionIntersections":[],"dossierDictionaryEntry":false,"recommendation":false,"hint":true,"dictionaryEntry":true,"falsePositive":false,"image":false,"localManualRedaction":false,"manuallyRemoved":false}],"legalBasis":[],"dictionaryVersion":0,"dossierDictionaryVersion":0,"rulesVersion":0,"legalBasisVersion":0}

redaction-service-v1/redaction-service-server-v1/src/test/resources/RedactionLog/files/Minimal Examples/US Syngenta.json

@@ -0,0 +1 @@
+{"analysisVersion":1,"analysisNumber":0,"redactionLogEntry":[{"id":"64eb5ff4ea1db8665fb6e57c5369b411","type":"PII","value":"Page","reason":"Personal information found","matchedRule":19,"rectangle":false,"legalBasis":"Article 39(e)(3) of Regulation (EC) No 178/2002","imported":false,"redacted":true,"section":"Text in table","color":[0.4,0.8,1.0],"positions":[{"topLeft":{"x":154.48782,"y":654.18},"width":9.992157,"height":19.882175,"page":1}],"sectionNumber":8,"textBefore":null,"textAfter":" 1 of","comments":[],"startOffset":54,"endOffset":58,"imageHasTransparency":false,"excluded":false,"sourceId":null,"changes":[{"analysisNumber":0,"type":"ADDED","dateTime":1658398731.072100800}],"manualChanges":[],"engines":["DICTIONARY"],"reference":[],"importedRedactionIntersections":[],"dossierDictionaryEntry":false,"recommendation":false,"hint":false,"dictionaryEntry":true,"falsePositive":false,"image":false,"localManualRedaction":false,"manuallyRemoved":false}],"legalBasis":[],"dictionaryVersion":0,"dossierDictionaryVersion":0,"rulesVersion":0,"legalBasisVersion":0}

redaction-service-v1/redaction-service-server-v1/src/test/resources/RedactionLog/files/Minimal Examples/foo.json

@@ -0,0 +1 @@
+{"analysisVersion":1,"analysisNumber":0,"redactionLogEntry":[{"id":"9f6837120752b85c20e27d4cb199bd51","type":"CBI_author","value":"B. Foo","reason":"Author found","matchedRule":1,"rectangle":false,"legalBasis":"Article 39(e)(3) of Regulation (EC) No 178/2002","imported":false,"redacted":true,"section":"Header","color":[1.0,0.88235295,0.5294118],"positions":[{"topLeft":{"x":92.520004,"y":712.96},"width":31.589584,"height":12.231649,"page":1}],"sectionNumber":3,"textBefore":null,"textAfter":" F. Bar","comments":[],"startOffset":0,"endOffset":6,"imageHasTransparency":false,"excluded":false,"sourceId":null,"changes":[{"analysisNumber":0,"type":"ADDED","dateTime":1658398909.230209400}],"manualChanges":[],"engines":["DICTIONARY"],"reference":[],"importedRedactionIntersections":[],"dossierDictionaryEntry":false,"recommendation":false,"hint":false,"dictionaryEntry":true,"falsePositive":false,"image":false,"localManualRedaction":false,"manuallyRemoved":false},{"id":"fb930b7f3acf65a74b0b59fee32cc6fa","type":"CBI_author","value":"F. Bar","reason":"Author found","matchedRule":1,"rectangle":false,"legalBasis":"Article 39(e)(3) of Regulation (EC) No 178/2002","imported":false,"redacted":true,"section":"Header","color":[1.0,0.88235295,0.5294118],"positions":[{"topLeft":{"x":125.92424,"y":712.96},"width":29.137009,"height":12.231649,"page":1}],"sectionNumber":3,"textBefore":"B. Foo ","textAfter":" Foo, B.","comments":[],"startOffset":7,"endOffset":13,"imageHasTransparency":false,"excluded":false,"sourceId":null,"changes":[{"analysisNumber":0,"type":"ADDED","dateTime":1658398909.230209400}],"manualChanges":[],"engines":["DICTIONARY"],"reference":[],"importedRedactionIntersections":[],"dossierDictionaryEntry":false,"recommendation":false,"hint":false,"dictionaryEntry":true,"falsePositive":false,"image":false,"localManualRedaction":false,"manuallyRemoved":false},{"id":"435224747b87ed87b60ab363e6610fba","type":"CBI_author","value":"Foo","reason":"Author found","matchedRule":1,"rectangle":false,"legalBasis":"Article 39(e)(3) of Regulation (EC) No 178/2002","imported":false,"redacted":true,"section":"Header","color":[1.0,0.88235295,0.5294118],"positions":[{"topLeft":{"x":92.520004,"y":686.8},"width":18.338272,"height":12.231649,"page":1}],"sectionNumber":3,"textBefore":"Foo F. Bar ","textAfter":", B.","comments":[],"startOffset":14,"endOffset":17,"imageHasTransparency":false,"excluded":false,"sourceId":null,"changes":[{"analysisNumber":0,"type":"ADDED","dateTime":1658398909.230209400}],"manualChanges":[],"engines":["DICTIONARY"],"reference":[],"importedRedactionIntersections":[],"dossierDictionaryEntry":false,"recommendation":false,"hint":false,"dictionaryEntry":true,"falsePositive":false,"image":false,"localManualRedaction":false,"manuallyRemoved":false}],"legalBasis":[],"dictionaryVersion":0,"dossierDictionaryVersion":0,"rulesVersion":0,"legalBasisVersion":0}

redaction-service-v1/redaction-service-server-v1/src/test/resources/RedactionLog/files/Minimal Examples/merge_images.json

@@ -0,0 +1 @@
+{"analysisVersion":1,"analysisNumber":0,"redactionLogEntry":[],"legalBasis":[],"dictionaryVersion":0,"dossierDictionaryVersion":0,"rulesVersion":0,"legalBasisVersion":0}