diff --git a/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/classification/service/BlockificationService.java b/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/classification/service/BlockificationService.java index 2cff711a..3be9f5f0 100644 --- a/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/classification/service/BlockificationService.java +++ b/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/classification/service/BlockificationService.java @@ -54,8 +54,9 @@ public class BlockificationService { boolean newLineAfterSplit = prev != null && word.getMinYDirAdj() != prev.getMinYDirAdj() && wasSplitted && splitX1 != word.getMinXDirAdj(); boolean isSplitByRuling = isSplitByRuling(minX, minY, maxX, maxY, word, horizontalRulingLines, verticalRulingLines); boolean splitByDir = prev != null && !prev.getDir().equals(word.getDir()); + boolean splitByOtherFontAndOtherY = prev != null && prev.getMaxYDirAdj() != word.getMaxYDirAdj() && (word.getFontStyle().contains("bold") && !prev.getFontStyle().contains("bold") || prev.getFontStyle().contains("bold") && !word.getFontStyle().contains("bold")); - if (prev != null && (lineSeparation || startFromTop || splitByX || splitByDir || isSplitByRuling)) { + if (prev != null && (lineSeparation || startFromTop || splitByDir || isSplitByRuling || splitByOtherFontAndOtherY)) { Orientation prevOrientation = null; if (!chunkBlockList1.isEmpty()) { @@ -107,51 +108,51 @@ public class BlockificationService { chunkBlockList1.add(cb1); } - Iterator itty = chunkBlockList1.iterator(); - - TextBlock previousLeft = null; - TextBlock previousRight = null; - while (itty.hasNext()) { - TextBlock block = (TextBlock) itty.next(); - - if (previousLeft != null && block.getOrientation().equals(Orientation.LEFT)) { - if (previousLeft.getMinY() > block.getMinY() && block.getMaxY() + block.getMostPopularWordHeight() > previousLeft.getMinY()) { - previousLeft.add(block); - itty.remove(); - continue; - } - } - - if (previousRight != null && block.getOrientation().equals(Orientation.RIGHT)) { - if (previousRight.getMinY() > block.getMinY() && block.getMaxY() + block.getMostPopularWordHeight() > previousRight.getMinY()) { - previousRight.add(block); - itty.remove(); - continue; - } - } - - if (block.getOrientation().equals(Orientation.LEFT)) { - previousLeft = block; - } else if (block.getOrientation().equals(Orientation.RIGHT)) { - previousRight = block; - } - } - - itty = chunkBlockList1.iterator(); - TextBlock previous = null; - while (itty.hasNext()) { - TextBlock block = (TextBlock) itty.next(); - - if (previous != null && previous.getOrientation().equals(Orientation.LEFT) && block.getOrientation().equals(Orientation.LEFT) && equalsWithThreshold(block.getMaxY(), - previous.getMaxY()) || previous != null && previous.getOrientation().equals(Orientation.LEFT) && block.getOrientation() - .equals(Orientation.RIGHT) && equalsWithThreshold(block.getMaxY(), previous.getMaxY())) { - previous.add(block); - itty.remove(); - continue; - } - - previous = block; - } +// Iterator itty = chunkBlockList1.iterator(); +// +// TextBlock previousLeft = null; +// TextBlock previousRight = null; +// while (itty.hasNext()) { +// TextBlock block = (TextBlock) itty.next(); +// +// if (previousLeft != null && block.getOrientation().equals(Orientation.LEFT)) { +// if (previousLeft.getMinY() > block.getMinY() && block.getMaxY() + block.getMostPopularWordHeight() > previousLeft.getMinY()) { +// previousLeft.add(block); +// itty.remove(); +// continue; +// } +// } +// +// if (previousRight != null && block.getOrientation().equals(Orientation.RIGHT)) { +// if (previousRight.getMinY() > block.getMinY() && block.getMaxY() + block.getMostPopularWordHeight() > previousRight.getMinY()) { +// previousRight.add(block); +// itty.remove(); +// continue; +// } +// } +// +// if (block.getOrientation().equals(Orientation.LEFT)) { +// previousLeft = block; +// } else if (block.getOrientation().equals(Orientation.RIGHT)) { +// previousRight = block; +// } +// } +// +// itty = chunkBlockList1.iterator(); +// TextBlock previous = null; +// while (itty.hasNext()) { +// TextBlock block = (TextBlock) itty.next(); +// +// if (previous != null && previous.getOrientation().equals(Orientation.LEFT) && block.getOrientation().equals(Orientation.LEFT) && equalsWithThreshold(block.getMaxY(), +// previous.getMaxY()) || previous != null && previous.getOrientation().equals(Orientation.LEFT) && block.getOrientation() +// .equals(Orientation.RIGHT) && equalsWithThreshold(block.getMaxY(), previous.getMaxY())) { +// previous.add(block); +// itty.remove(); +// continue; +// } +// +// previous = block; +// } return new Page(chunkBlockList1); } diff --git a/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/classification/service/ClassificationService.java b/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/classification/service/ClassificationService.java index 6ec6f70e..e18705f4 100644 --- a/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/classification/service/ClassificationService.java +++ b/redaction-service-v1/redaction-service-server-v1/src/main/java/com/iqser/red/service/redaction/v1/server/classification/service/ClassificationService.java @@ -56,26 +56,31 @@ public class ClassificationService { var bodyTextFrame = page.getBodyTextFrame(); - var pattern = Patterns.getCompiledPattern("^(\\d{1,2}\\.){1,3}\\d{1,2}\\s[A-Za-z]{3,50}", true); + var pattern = Patterns.getCompiledPattern("^(\\d{1,1}\\.){1,3}\\d{1,2}\\.?\\s[0-9A-Za-z]{2,50}", true); var pattern2 = Patterns.getCompiledPattern(".*\\d$", true); + var pattern3 = Patterns.getCompiledPattern("^(\\d{1,1}\\.){1,3}\\d{1,2}\\.?\\s[a-z]{1,2}\\/[a-z]{1,2}.*", false); Matcher matcher = pattern.matcher(textBlock.toString()); Matcher matcher2 = pattern2.matcher(textBlock.toString()); + Matcher matcher3 = pattern3.matcher(textBlock.toString()); if (document.getFontSizeCounter().getMostPopular() == null) { textBlock.setClassification("Other"); return; } - if (textBlock.getText().length() > 5 && textBlock.getMostPopularWordHeight() > document.getTextHeightCounter().getMostPopular() && PositionUtils.getApproxLineCount(textBlock) < 5.9 + if (textBlock.getText().length() > 5 && (textBlock.getMostPopularWordHeight() > document.getTextHeightCounter().getMostPopular() || textBlock.getMostPopularWordFontSize() > document.getFontSizeCounter().getMostPopular()) && PositionUtils.getApproxLineCount(textBlock) < 5.9 - - && ((textBlock.getMostPopularWordStyle().contains("bold") && Character.isDigit(textBlock.toString().charAt(0))) - || textBlock.toString().equals(textBlock.toString().toUpperCase(Locale.ROOT)) - || textBlock.toString().startsWith("APPENDIX") || textBlock.toString().startsWith("TABLE")) && !textBlock.toString().endsWith(":")) { + && ((textBlock.getMostPopularWordStyle().contains("bold") && Character.isDigit(textBlock.toString().charAt(0)) && !matcher2.matches() && !textBlock.toString().contains(":") + || textBlock.toString().equals(textBlock.toString().toUpperCase(Locale.ROOT)) && !matcher2.matches() && !textBlock.toString().contains(":") + || textBlock.toString().startsWith("APPENDIX") || textBlock.toString().startsWith("TABLE")) && !textBlock.toString().endsWith(":"))) { textBlock.setClassification("H 1"); document.setHeadlines(true); } + else if(matcher.find() && PositionUtils.getApproxLineCount(textBlock) < 2.9 && !matcher2.matches() && !matcher3.matches()) { + textBlock.setClassification("H 2"); + document.setHeadlines(true); + } else if (PositionUtils.isOverBodyTextFrame(bodyTextFrame, textBlock, page.getRotation()) && (document.getFontSizeCounter() .getMostPopular() == null || textBlock.getHighestFontSize() <= document.getFontSizeCounter().getMostPopular())) { textBlock.setClassification("Header"); @@ -90,10 +95,7 @@ public class ClassificationService { textBlock.setClassification("Title"); } } - else if(matcher.find() && PositionUtils.getApproxLineCount(textBlock) < 2.9 && !matcher2.find()) { - textBlock.setClassification("H 2"); - document.setHeadlines(true); - } + // else if (textBlock.getMostPopularWordFontSize() > document.getFontSizeCounter() // .getMostPopular() && PositionUtils.getApproxLineCount(textBlock) < 4.9 && (textBlock.getMostPopularWordStyle().equals("bold") || !document.getFontStyleCounter() // .getCountPerValue() diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/java/com/iqser/red/service/redaction/v1/server/HeadlinesGoldStandardIntegrationTest.java b/redaction-service-v1/redaction-service-server-v1/src/test/java/com/iqser/red/service/redaction/v1/server/HeadlinesGoldStandardIntegrationTest.java index 66756dac..7dd2fd71 100644 --- a/redaction-service-v1/redaction-service-server-v1/src/test/java/com/iqser/red/service/redaction/v1/server/HeadlinesGoldStandardIntegrationTest.java +++ b/redaction-service-v1/redaction-service-server-v1/src/test/java/com/iqser/red/service/redaction/v1/server/HeadlinesGoldStandardIntegrationTest.java @@ -143,7 +143,7 @@ public class HeadlinesGoldStandardIntegrationTest { @Test public void extractHeadlines() throws IOException { - AnalyzeRequest request = prepareStorage("files/RSS/31 - CA6375 - Acute Oral Toxicity - Rats.pdf"); + AnalyzeRequest request = prepareStorage("files/RSS/26 - Sedaxane - Acute Oral Toxicity - Rat.pdf"); analyzeService.analyzeDocumentStructure(new StructureAnalyzeRequest(request.getDossierId(), request.getFileId())); analyzeService.analyze(request); @@ -165,7 +165,7 @@ public class HeadlinesGoldStandardIntegrationTest { RedactionResult result1 = redactionController.classify(redactionRequest); - try (FileOutputStream fileOutputStream = new FileOutputStream(OsUtils.getTemporaryDirectory() + "/Classified.pdf")) { + try (FileOutputStream fileOutputStream = new FileOutputStream(OsUtils.getTemporaryDirectory() + "/HeadlinesClassified.pdf")) { fileOutputStream.write(result1.getDocument()); } } diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/java/com/iqser/red/service/redaction/v1/server/RedactionIntegrationTest.java b/redaction-service-v1/redaction-service-server-v1/src/test/java/com/iqser/red/service/redaction/v1/server/RedactionIntegrationTest.java index e5cfe646..fa89ac55 100644 --- a/redaction-service-v1/redaction-service-server-v1/src/test/java/com/iqser/red/service/redaction/v1/server/RedactionIntegrationTest.java +++ b/redaction-service-v1/redaction-service-server-v1/src/test/java/com/iqser/red/service/redaction/v1/server/RedactionIntegrationTest.java @@ -1111,7 +1111,7 @@ public class RedactionIntegrationTest { System.out.println("classificationTest"); - AnalyzeRequest request = prepareStorage("files/RSS/01 - CGA100251 - Acute Oral Toxicity (Up and Down Procedure) - Rat (1).pdf"); + AnalyzeRequest request = prepareStorage("files/new/RotateTestFile.pdf"); RedactionRequest redactionRequest = RedactionRequest.builder() .dossierId(request.getDossierId()) diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21377X/A21377X - Acute Dermal Toxicity - Rats.pdf b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21377X/A21377X - Acute Dermal Toxicity - Rats.pdf new file mode 100644 index 00000000..0d916ed4 Binary files /dev/null and b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21377X/A21377X - Acute Dermal Toxicity - Rats.pdf differ diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21377X/A21377X - Acute Eye Irritation - Rabbits.pdf b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21377X/A21377X - Acute Eye Irritation - Rabbits.pdf new file mode 100644 index 00000000..809503be Binary files /dev/null and b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21377X/A21377X - Acute Eye Irritation - Rabbits.pdf differ diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21377X/A21377X - Acute Inhalation - Rats.pdf b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21377X/A21377X - Acute Inhalation - Rats.pdf new file mode 100644 index 00000000..fad8ba80 Binary files /dev/null and b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21377X/A21377X - Acute Inhalation - Rats.pdf differ diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21377X/A21377X - Acute Oral Toxicity - Rats.pdf b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21377X/A21377X - Acute Oral Toxicity - Rats.pdf new file mode 100644 index 00000000..32c90d4a Binary files /dev/null and b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21377X/A21377X - Acute Oral Toxicity - Rats.pdf differ diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21377X/A21377X - Bacterial Reverse Mutation Test.pdf b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21377X/A21377X - Bacterial Reverse Mutation Test.pdf new file mode 100644 index 00000000..f29261fa Binary files /dev/null and b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21377X/A21377X - Bacterial Reverse Mutation Test.pdf differ diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21377X/A21377X - In Vitro Eye Irritation Test - Isolated Chicken Ey.pdf b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21377X/A21377X - In Vitro Eye Irritation Test - Isolated Chicken Ey.pdf new file mode 100644 index 00000000..5bfbff1a Binary files /dev/null and b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21377X/A21377X - In Vitro Eye Irritation Test - Isolated Chicken Ey.pdf differ diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21377X/A21377X - In Vitro Skin Irritation Test - EPISKIN TM Model.pdf b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21377X/A21377X - In Vitro Skin Irritation Test - EPISKIN TM Model.pdf new file mode 100644 index 00000000..53360ecf Binary files /dev/null and b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21377X/A21377X - In Vitro Skin Irritation Test - EPISKIN TM Model.pdf differ diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21377X/A21377X - LLNA - Mice.pdf b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21377X/A21377X - LLNA - Mice.pdf new file mode 100644 index 00000000..a9fd451d Binary files /dev/null and b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21377X/A21377X - LLNA - Mice.pdf differ diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21377X/A21377X - Primary Skin Irritation - Rabbits.pdf b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21377X/A21377X - Primary Skin Irritation - Rabbits.pdf new file mode 100644 index 00000000..4558778f Binary files /dev/null and b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21377X/A21377X - Primary Skin Irritation - Rabbits.pdf differ diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21550L/A21550L - Acute Dermal Toxicity - Rats.pdf b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21550L/A21550L - Acute Dermal Toxicity - Rats.pdf new file mode 100644 index 00000000..be78a6e0 Binary files /dev/null and b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21550L/A21550L - Acute Dermal Toxicity - Rats.pdf differ diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21550L/A21550L - Acute Eye Irritation - Rabbits.pdf b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21550L/A21550L - Acute Eye Irritation - Rabbits.pdf new file mode 100644 index 00000000..651b5ec1 Binary files /dev/null and b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21550L/A21550L - Acute Eye Irritation - Rabbits.pdf differ diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21550L/A21550L - Acute Inhalation - Rats.pdf b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21550L/A21550L - Acute Inhalation - Rats.pdf new file mode 100644 index 00000000..ac17789e Binary files /dev/null and b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21550L/A21550L - Acute Inhalation - Rats.pdf differ diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21550L/A21550L - Acute Oral Toxicity - Rats.pdf b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21550L/A21550L - Acute Oral Toxicity - Rats.pdf new file mode 100644 index 00000000..366e96dc Binary files /dev/null and b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21550L/A21550L - Acute Oral Toxicity - Rats.pdf differ diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21550L/A21550L - In Vitro Eye Irritation - Isolated Chicken Eyes.pdf b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21550L/A21550L - In Vitro Eye Irritation - Isolated Chicken Eyes.pdf new file mode 100644 index 00000000..1acb5507 Binary files /dev/null and b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21550L/A21550L - In Vitro Eye Irritation - Isolated Chicken Eyes.pdf differ diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21550L/A21550L - In Vitro Skin Irritation Test - EPISKIN TM Model.pdf b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21550L/A21550L - In Vitro Skin Irritation Test - EPISKIN TM Model.pdf new file mode 100644 index 00000000..9eedffaa Binary files /dev/null and b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21550L/A21550L - In Vitro Skin Irritation Test - EPISKIN TM Model.pdf differ diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21550L/A21550L - Local Lymph Node Assay - Mouse.pdf b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21550L/A21550L - Local Lymph Node Assay - Mouse.pdf new file mode 100644 index 00000000..01355bd0 Binary files /dev/null and b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21550L/A21550L - Local Lymph Node Assay - Mouse.pdf differ diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21550L/A21550L - Primary Skin Irritation - Rabbits.pdf b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21550L/A21550L - Primary Skin Irritation - Rabbits.pdf new file mode 100644 index 00000000..a95eaf02 Binary files /dev/null and b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21550L/A21550L - Primary Skin Irritation - Rabbits.pdf differ diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21550L/A21550L - Reverse Mutation Assay - S. typhimurium & E. coli.pdf b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21550L/A21550L - Reverse Mutation Assay - S. typhimurium & E. coli.pdf new file mode 100644 index 00000000..40b41c4b Binary files /dev/null and b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/A21550L/A21550L - Reverse Mutation Assay - S. typhimurium & E. coli.pdf differ diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/TrainData/A23220A - Acute Dermal - Rat.pdf b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/TrainData/A23220A - Acute Dermal - Rat.pdf new file mode 100644 index 00000000..ccf1acb6 Binary files /dev/null and b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/TrainData/A23220A - Acute Dermal - Rat.pdf differ diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/TrainData/A23220A - Acute Inhalation - Rat.pdf b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/TrainData/A23220A - Acute Inhalation - Rat.pdf new file mode 100644 index 00000000..c6c3d6cc Binary files /dev/null and b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/TrainData/A23220A - Acute Inhalation - Rat.pdf differ diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/TrainData/A23220A - Acute Oral - Rat.pdf b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/TrainData/A23220A - Acute Oral - Rat.pdf new file mode 100644 index 00000000..1264ed66 Binary files /dev/null and b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/TrainData/A23220A - Acute Oral - Rat.pdf differ diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/TrainData/A23220A - Eye Irritation In Vivo - Final Report.pdf b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/TrainData/A23220A - Eye Irritation In Vivo - Final Report.pdf new file mode 100644 index 00000000..6c1b8988 Binary files /dev/null and b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/TrainData/A23220A - Eye Irritation In Vivo - Final Report.pdf differ diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/TrainData/A23220A - In Vitro Eye Irritation - Chicken Eyes.pdf b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/TrainData/A23220A - In Vitro Eye Irritation - Chicken Eyes.pdf new file mode 100644 index 00000000..19ff6c17 Binary files /dev/null and b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/TrainData/A23220A - In Vitro Eye Irritation - Chicken Eyes.pdf differ diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/TrainData/A23220A - In Vitro Skin Irritation - Amendment 1.pdf b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/TrainData/A23220A - In Vitro Skin Irritation - Amendment 1.pdf new file mode 100644 index 00000000..3b757080 Binary files /dev/null and b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/TrainData/A23220A - In Vitro Skin Irritation - Amendment 1.pdf differ diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/TrainData/A23220A - Salmonella Typhimurium and Escherichia Coli RMA.pdf b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/TrainData/A23220A - Salmonella Typhimurium and Escherichia Coli RMA.pdf new file mode 100644 index 00000000..8306a286 Binary files /dev/null and b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/TrainData/A23220A - Salmonella Typhimurium and Escherichia Coli RMA.pdf differ diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/TrainData/A23220A - Skin Irritation in vivo.pdf b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/TrainData/A23220A - Skin Irritation in vivo.pdf new file mode 100644 index 00000000..4b58d26c Binary files /dev/null and b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/TrainData/A23220A - Skin Irritation in vivo.pdf differ diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/TrainData/A23220A - Skin Sensitization LLNA.pdf b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/TrainData/A23220A - Skin Sensitization LLNA.pdf new file mode 100644 index 00000000..7a6cea1a Binary files /dev/null and b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/Poc2Ocr/TrainData/A23220A - Skin Sensitization LLNA.pdf differ diff --git a/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/RSS/S9.pdf b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/RSS/S9.pdf new file mode 100644 index 00000000..171186d3 Binary files /dev/null and b/redaction-service-v1/redaction-service-server-v1/src/test/resources/files/RSS/S9.pdf differ