From 577db37b11dfd0243d22986a6d224700f0b49693 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Dominique=20Eifl=C3=A4nder?=
 <deiflaender@dominiques-mbp.speedport.ip>
Date: Fri, 5 Feb 2021 13:05:12 +0100
Subject: [PATCH] RED-1045: Enabled to redact in headers and footers

---
 .../server/classification/model/Document.java |  2 ++
 .../server/classification/model/Footer.java   | 24 +++++++++++++
 .../server/classification/model/Header.java   | 24 +++++++++++++
 .../service/EntityRedactionService.java       | 26 ++++++++++----
 .../segmentation/SectionsBuilderService.java  | 36 ++++++++++++++-----
 5 files changed, 97 insertions(+), 15 deletions(-)
 create mode 100644 redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/classification/model/Footer.java
 create mode 100644 redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/classification/model/Header.java

diff --git a/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/classification/model/Document.java b/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/classification/model/Document.java
index e1cb08ad..9aae9032 100644
--- a/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/classification/model/Document.java
+++ b/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/classification/model/Document.java
@@ -18,6 +18,8 @@ public class Document {
 
     private List<Page> pages = new ArrayList<>();
     private List<Paragraph> paragraphs = new ArrayList<>();
+    private List<Header> headers = new ArrayList<>();
+    private List<Footer> footers = new ArrayList<>();
     private Map<Integer, List<Entity>> entities = new HashMap<>();
     private FloatFrequencyCounter textHeightCounter = new FloatFrequencyCounter();
     private FloatFrequencyCounter fontSizeCounter= new FloatFrequencyCounter();
diff --git a/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/classification/model/Footer.java b/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/classification/model/Footer.java
new file mode 100644
index 00000000..37b691e6
--- /dev/null
+++ b/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/classification/model/Footer.java
@@ -0,0 +1,24 @@
+package com.iqser.red.service.redaction.v1.server.classification.model;
+
+import java.util.List;
+
+import com.iqser.red.service.redaction.v1.server.redaction.model.SearchableText;
+
+import lombok.AllArgsConstructor;
+import lombok.Data;
+
+@Data
+@AllArgsConstructor
+public class Footer {
+
+    private List<TextBlock> textBlocks;
+
+
+    public SearchableText getSearchableText() {
+
+        SearchableText searchableText = new SearchableText();
+        textBlocks.forEach(block -> searchableText.addAll(block.getSequences()));
+        return searchableText;
+    }
+
+}
\ No newline at end of file
diff --git a/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/classification/model/Header.java b/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/classification/model/Header.java
new file mode 100644
index 00000000..8a4b67ae
--- /dev/null
+++ b/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/classification/model/Header.java
@@ -0,0 +1,24 @@
+package com.iqser.red.service.redaction.v1.server.classification.model;
+
+import java.util.List;
+
+import com.iqser.red.service.redaction.v1.server.redaction.model.SearchableText;
+
+import lombok.AllArgsConstructor;
+import lombok.Data;
+
+@Data
+@AllArgsConstructor
+public class Header {
+
+    private List<TextBlock> textBlocks;
+
+
+    public SearchableText getSearchableText() {
+
+        SearchableText searchableText = new SearchableText();
+        textBlocks.forEach(block -> searchableText.addAll(block.getSequences()));
+        return searchableText;
+    }
+
+}
\ No newline at end of file
diff --git a/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/redaction/service/EntityRedactionService.java b/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/redaction/service/EntityRedactionService.java
index a649c356..7be4069b 100644
--- a/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/redaction/service/EntityRedactionService.java
+++ b/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/redaction/service/EntityRedactionService.java
@@ -19,6 +19,8 @@ import com.iqser.red.service.redaction.v1.model.ManualRedactionEntry;
 import com.iqser.red.service.redaction.v1.model.ManualRedactions;
 import com.iqser.red.service.redaction.v1.model.Rectangle;
 import com.iqser.red.service.redaction.v1.server.classification.model.Document;
+import com.iqser.red.service.redaction.v1.server.classification.model.Footer;
+import com.iqser.red.service.redaction.v1.server.classification.model.Header;
 import com.iqser.red.service.redaction.v1.server.classification.model.Paragraph;
 import com.iqser.red.service.redaction.v1.server.classification.model.TextBlock;
 import com.iqser.red.service.redaction.v1.server.redaction.model.CellValue;
@@ -113,7 +115,18 @@ public class EntityRedactionService {
                 }
                 sectionNumber.incrementAndGet();
             }
-            sectionSearchableTextPairs.add(processText(paragraph, manualRedactions, sectionNumber, dictionary, local, hintsPerSectionNumber));
+            sectionSearchableTextPairs.add(processText(paragraph.getSearchableText(), paragraph.getTextBlocks(), paragraph
+                    .getHeadline(), manualRedactions, sectionNumber, dictionary, local, hintsPerSectionNumber));
+            sectionNumber.incrementAndGet();
+        }
+
+        for (Header header : classifiedDoc.getHeaders()) {
+            sectionSearchableTextPairs.add(processText(header.getSearchableText(), header.getTextBlocks(), "Header", manualRedactions, sectionNumber, dictionary, local, hintsPerSectionNumber));
+            sectionNumber.incrementAndGet();
+        }
+
+        for (Footer footer : classifiedDoc.getFooters()) {
+            sectionSearchableTextPairs.add(processText(footer.getSearchableText(), footer.getTextBlocks(), "Footer", manualRedactions, sectionNumber, dictionary, local, hintsPerSectionNumber));
             sectionNumber.incrementAndGet();
         }
 
@@ -253,14 +266,13 @@ public class EntityRedactionService {
     }
 
 
-    private SectionSearchableTextPair processText(Paragraph paragraph, ManualRedactions manualRedactions,
+    private SectionSearchableTextPair processText(SearchableText searchableText, List<TextBlock> paragraphTextBlocks,
+                                                  String headline, ManualRedactions manualRedactions,
                                                   AtomicInteger sectionNumber, Dictionary dictionary, boolean local,
                                                   Map<Integer, Set<Entity>> hintsPerSectionNumber) {
 
-        SearchableText searchableText = paragraph.getSearchableText();
-        addSectionToManualRedactions(paragraph.getTextBlocks(), manualRedactions, paragraph.getHeadline(), sectionNumber
-                .intValue());
-        Set<Entity> entities = findEntities(searchableText, paragraph.getHeadline(), sectionNumber.intValue(), dictionary, local);
+        addSectionToManualRedactions(paragraphTextBlocks, manualRedactions, headline, sectionNumber.intValue());
+        Set<Entity> entities = findEntities(searchableText, headline, sectionNumber.intValue(), dictionary, local);
         surroundingWordsService.addSurroundingText(entities, searchableText, dictionary);
 
         return new SectionSearchableTextPair(Section.builder()
@@ -271,7 +283,7 @@ public class EntityRedactionService {
                         .collect(Collectors.toSet()) : entities)
                 .text(searchableText.getAsStringWithLinebreaks())
                 .searchText(searchableText.toString())
-                .headline(paragraph.getHeadline())
+                .headline(headline)
                 .sectionNumber(sectionNumber.intValue())
                 .searchableText(searchableText)
                 .dictionary(dictionary)
diff --git a/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/segmentation/SectionsBuilderService.java b/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/segmentation/SectionsBuilderService.java
index 1b824ea9..e9e3c6e5 100644
--- a/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/segmentation/SectionsBuilderService.java
+++ b/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/segmentation/SectionsBuilderService.java
@@ -10,6 +10,8 @@ import org.apache.commons.collections4.CollectionUtils;
 import org.springframework.stereotype.Service;
 
 import com.iqser.red.service.redaction.v1.server.classification.model.Document;
+import com.iqser.red.service.redaction.v1.server.classification.model.Footer;
+import com.iqser.red.service.redaction.v1.server.classification.model.Header;
 import com.iqser.red.service.redaction.v1.server.classification.model.Page;
 import com.iqser.red.service.redaction.v1.server.classification.model.Paragraph;
 import com.iqser.red.service.redaction.v1.server.classification.model.TextBlock;
@@ -24,25 +26,39 @@ public class SectionsBuilderService {
 
         List<AbstractTextContainer> chunkWords = new ArrayList<>();
         List<Paragraph> chunkBlockList = new ArrayList<>();
+        List<Header> headers = new ArrayList<>();
+        List<Footer> footers = new ArrayList<>();
 
         AbstractTextContainer prev = null;
 
         String lastHeadline = "";
         Table previousTable = null;
         for (Page page : document.getPages()) {
+            List<TextBlock> header = new ArrayList<>();
+            List<TextBlock> footer = new ArrayList<>();
             for (AbstractTextContainer current : page.getTextBlocks()) {
 
-                if (current.getClassification() == null || current.getClassification()
-                        .equals("Header") || current.getClassification().equals("Footer")) {
+                if (current.getClassification() == null) {
                     continue;
                 }
 
                 current.setPage(page.getPageNumber());
 
-                if (prev != null && current.getClassification().startsWith("H ") && !prev.getClassification().startsWith("H ") || !document.isHeadlines()) {
+                if (current.getClassification().equals("Header")) {
+                    header.add((TextBlock) current);
+                    continue;
+                }
+
+                if (current.getClassification().equals("Footer")) {
+                    footer.add((TextBlock) current);
+                    continue;
+                }
+
+                if (prev != null && current.getClassification().startsWith("H ") && !prev.getClassification()
+                        .startsWith("H ") || !document.isHeadlines()) {
                     Paragraph chunkBlock = buildTextBlock(chunkWords, lastHeadline);
                     chunkBlock.setHeadline(lastHeadline);
-                    if(document.isHeadlines()) {
+                    if (document.isHeadlines()) {
                         lastHeadline = current.getText();
                     }
                     chunkBlockList.add(chunkBlock);
@@ -60,6 +76,8 @@ public class SectionsBuilderService {
                 chunkWords.add(current);
                 prev = current;
             }
+            headers.add(new Header(header));
+            footers.add(new Footer(footer));
         }
 
         Paragraph chunkBlock = buildTextBlock(chunkWords, lastHeadline);
@@ -67,6 +85,8 @@ public class SectionsBuilderService {
         chunkBlockList.add(chunkBlock);
 
         document.setParagraphs(chunkBlockList);
+        document.setHeaders(headers);
+        document.setFooters(footers);
     }
 
 
@@ -175,9 +195,9 @@ public class SectionsBuilderService {
 
     private boolean hasInvalidHeaderInformation(Table table) {
 
-        return table.getRows().stream()
-                .flatMap(row -> row.stream()
-                        .filter(cell -> CollectionUtils.isNotEmpty(cell.getHeaderCells())))
+        return table.getRows()
+                .stream()
+                .flatMap(row -> row.stream().filter(cell -> CollectionUtils.isNotEmpty(cell.getHeaderCells())))
                 .findAny()
                 .isEmpty();
 
@@ -188,7 +208,7 @@ public class SectionsBuilderService {
 
         for (int i = table.getRowCount() - 1; i >= 0; i--) { // Non header rows are most likely at bottom of table
             List<Cell> row = table.getRows().get(i);
-            if(row.size() == 1){
+            if (row.size() == 1) {
                 continue;
             }
             boolean allNonHeader = true;