| 1 | package pro.verron.officestamper.core; | |
| 2 | ||
| 3 | import jakarta.xml.bind.JAXBElement; | |
| 4 | import org.docx4j.dml.Graphic; | |
| 5 | import org.docx4j.dml.picture.Pic; | |
| 6 | import org.docx4j.dml.wordprocessingDrawing.Inline; | |
| 7 | import org.docx4j.openpackaging.exceptions.Docx4JException; | |
| 8 | import org.docx4j.openpackaging.packages.WordprocessingMLPackage; | |
| 9 | import org.docx4j.wml.Drawing; | |
| 10 | import org.docx4j.wml.R; | |
| 11 | import pro.verron.officestamper.api.OfficeStamperException; | |
| 12 | ||
| 13 | import java.io.IOException; | |
| 14 | import java.io.InputStream; | |
| 15 | import java.util.Arrays; | |
| 16 | ||
| 17 | /** | |
| 18 | * Extracts images from a docx document. | |
| 19 | * | |
| 20 | * @author Joseph Verron | |
| 21 | * @version ${version} | |
| 22 | * @since 1.4.7 | |
| 23 | */ | |
| 24 | public class DocxImageExtractor { | |
| 25 | ||
| 26 | private final WordprocessingMLPackage wordprocessingMLPackage; | |
| 27 | ||
| 28 | /** | |
| 29 | * Creates a new image extractor for the given docx document. | |
| 30 | * | |
| 31 | * @param wordprocessingMLPackage the docx document to extract images from. | |
| 32 | */ | |
| 33 | public DocxImageExtractor(WordprocessingMLPackage wordprocessingMLPackage) { | |
| 34 | this.wordprocessingMLPackage = wordprocessingMLPackage; | |
| 35 | } | |
| 36 | ||
| 37 | /** | |
| 38 | * Extract an image bytes from an embedded image run. | |
| 39 | * | |
| 40 | * @param run run containing the embedded drawing. | |
| 41 | */ | |
| 42 | byte[] getRunDrawingData(R run) { | |
| 43 | String imageRelId = getPic(run).getBlipFill().getBlip().getEmbed(); | |
| 44 | String imageRelPartName = getImageRelPartName(imageRelId); | |
| 45 | long size = getImageSize(imageRelPartName); | |
| 46 | InputStream stream = getImageStream(imageRelPartName); | |
| 47 |
1
1. getRunDrawingData : replaced return value with null for pro/verron/officestamper/core/DocxImageExtractor::getRunDrawingData → KILLED |
return streamToByteArray(size, stream); |
| 48 | } | |
| 49 | ||
| 50 | private static Pic getPic(R run) { | |
| 51 |
1
1. getPic : negated conditional → KILLED |
for (Object runContent : run.getContent()) { |
| 52 |
1
1. getPic : negated conditional → KILLED |
if (!(runContent instanceof JAXBElement<?> runElement)) break; |
| 53 |
1
1. getPic : negated conditional → KILLED |
if (!(runElement.getValue() instanceof Drawing drawing)) break; |
| 54 | Graphic graphic = getInlineGraphic(drawing); | |
| 55 |
1
1. getPic : replaced return value with null for pro/verron/officestamper/core/DocxImageExtractor::getPic → KILLED |
return graphic.getGraphicData().getPic(); |
| 56 | } | |
| 57 | throw new OfficeStamperException("Run drawing not found !"); | |
| 58 | } | |
| 59 | ||
| 60 | private String getImageRelPartName(String imageRelId) { | |
| 61 | // TODO_LATER: find a better way to find image rel part name in source part store | |
| 62 |
1
1. getImageRelPartName : replaced return value with "" for pro/verron/officestamper/core/DocxImageExtractor::getImageRelPartName → KILLED |
return wordprocessingMLPackage |
| 63 | .getMainDocumentPart() | |
| 64 | .getRelationshipsPart() | |
| 65 | .getPart(imageRelId) | |
| 66 | .getPartName() | |
| 67 | .getName() | |
| 68 | .substring(1); | |
| 69 | } | |
| 70 | ||
| 71 | /** | |
| 72 | * Extract an inline graphic from a drawing. | |
| 73 | * | |
| 74 | * @param drawing the drawing containing the graphic. | |
| 75 | */ | |
| 76 | private static Graphic getInlineGraphic(Drawing drawing) { | |
| 77 | if (drawing.getAnchorOrInline() | |
| 78 |
1
1. getInlineGraphic : negated conditional → KILLED |
.isEmpty()) { |
| 79 | throw new OfficeStamperException("Anchor or Inline is empty !"); | |
| 80 | } | |
| 81 | Object anchorOrInline = drawing.getAnchorOrInline() | |
| 82 | .get(0); | |
| 83 |
1
1. getInlineGraphic : negated conditional → KILLED |
if (anchorOrInline instanceof Inline inline) { |
| 84 |
1
1. getInlineGraphic : replaced return value with null for pro/verron/officestamper/core/DocxImageExtractor::getInlineGraphic → KILLED |
return inline.getGraphic(); |
| 85 | } | |
| 86 | else { | |
| 87 | throw new OfficeStamperException("Don't know how to process anchor !"); | |
| 88 | } | |
| 89 | } | |
| 90 | ||
| 91 | /** | |
| 92 | * Converts an InputStream to a byte array. | |
| 93 | * | |
| 94 | * @param size expected size of the byte array. | |
| 95 | * @param is input stream to read data from. | |
| 96 | * @return the data from the input stream. | |
| 97 | */ | |
| 98 | private static byte[] streamToByteArray(long size, InputStream is) { | |
| 99 |
2
1. streamToByteArray : changed conditional boundary → SURVIVED 2. streamToByteArray : negated conditional → KILLED |
if (size > Integer.MAX_VALUE) |
| 100 | throw new OfficeStamperException("Image size exceeds maximum allowed (2GB)"); | |
| 101 | ||
| 102 | int intSize = (int) size; | |
| 103 | byte[] data = new byte[intSize]; | |
| 104 | int numRead = tryRead(is, data); | |
| 105 |
1
1. streamToByteArray : replaced return value with null for pro/verron/officestamper/core/DocxImageExtractor::streamToByteArray → KILLED |
return Arrays.copyOfRange(data, 0, numRead); |
| 106 | } | |
| 107 | ||
| 108 | private static int tryRead(InputStream is, byte[] data) { | |
| 109 | try { | |
| 110 |
1
1. tryRead : replaced int return with 0 for pro/verron/officestamper/core/DocxImageExtractor::tryRead → KILLED |
return is.read(data); |
| 111 | } catch (IOException e) { | |
| 112 | throw new OfficeStamperException(e); | |
| 113 | } | |
| 114 | } | |
| 115 | ||
| 116 | private long getImageSize(String imageRelPartName) { | |
| 117 | try { | |
| 118 |
1
1. getImageSize : replaced long return with 0 for pro/verron/officestamper/core/DocxImageExtractor::getImageSize → KILLED |
return wordprocessingMLPackage |
| 119 | .getSourcePartStore() | |
| 120 | .getPartSize(imageRelPartName); | |
| 121 | } catch (Docx4JException e) { | |
| 122 | throw new OfficeStamperException(e); | |
| 123 | } | |
| 124 | } | |
| 125 | ||
| 126 | private InputStream getImageStream(String imageRelPartName) { | |
| 127 | try { | |
| 128 |
1
1. getImageStream : replaced return value with null for pro/verron/officestamper/core/DocxImageExtractor::getImageStream → KILLED |
return wordprocessingMLPackage |
| 129 | .getSourcePartStore() | |
| 130 | .loadPart(imageRelPartName); | |
| 131 | } catch (Docx4JException e) { | |
| 132 | throw new OfficeStamperException(e); | |
| 133 | } | |
| 134 | } | |
| 135 | ||
| 136 | /** | |
| 137 | * Extract the name of the image from an embedded image run. | |
| 138 | * | |
| 139 | * @param run run containing the embedded drawing. | |
| 140 | * @return a {@link String} object | |
| 141 | */ | |
| 142 | public String getRunDrawingFilename(R run) { | |
| 143 |
1
1. getRunDrawingFilename : replaced return value with "" for pro/verron/officestamper/core/DocxImageExtractor::getRunDrawingFilename → NO_COVERAGE |
return getPic(run).getNvPicPr().getCNvPr().getName(); |
| 144 | } | |
| 145 | ||
| 146 | /** | |
| 147 | * Extract the content type of the image from an embedded image run. | |
| 148 | * | |
| 149 | * @param run run containing the embedded drawing. | |
| 150 | * @return a {@link String} object | |
| 151 | */ | |
| 152 | public String getRunDrawingAltText(R run) { | |
| 153 |
1
1. getRunDrawingAltText : replaced return value with "" for pro/verron/officestamper/core/DocxImageExtractor::getRunDrawingAltText → NO_COVERAGE |
return getPic(run).getNvPicPr().getCNvPr().getDescr(); |
| 154 | } | |
| 155 | ||
| 156 | /** | |
| 157 | * Extract the width of the image from an embedded image run. | |
| 158 | * | |
| 159 | * @param run run containing the embedded drawing. | |
| 160 | * @return a {@link Integer} object | |
| 161 | */ | |
| 162 | public Integer getRunDrawingMaxWidth(R run) { | |
| 163 |
1
1. getRunDrawingMaxWidth : replaced Integer return value with 0 for pro/verron/officestamper/core/DocxImageExtractor::getRunDrawingMaxWidth → SURVIVED |
return (int) getPic(run).getSpPr().getXfrm().getExt().getCx(); |
| 164 | } | |
| 165 | } | |
Mutations | ||
| 47 |
1.1 |
|
| 51 |
1.1 |
|
| 52 |
1.1 |
|
| 53 |
1.1 |
|
| 55 |
1.1 |
|
| 62 |
1.1 |
|
| 78 |
1.1 |
|
| 83 |
1.1 |
|
| 84 |
1.1 |
|
| 99 |
1.1 2.2 |
|
| 105 |
1.1 |
|
| 110 |
1.1 |
|
| 118 |
1.1 |
|
| 128 |
1.1 |
|
| 143 |
1.1 |
|
| 153 |
1.1 |
|
| 163 |
1.1 |