From 0caa67864e1ece83b516a3d29bd560df7599896e Mon Sep 17 00:00:00 2001 From: Michael Williamson Date: Sat, 20 May 2023 20:47:25 +0100 Subject: Support merged paragraphs when revisions are tracked --- .../mammoth/internal/docx/BodyXmlReader.java | 2 +- .../internal/docx/StatefulBodyXmlReader.java | 31 +++++++++++++++++----- 2 files changed, 26 insertions(+), 7 deletions(-) (limited to 'src/main/java/org') diff --git a/src/main/java/org/zwobble/mammoth/internal/docx/BodyXmlReader.java b/src/main/java/org/zwobble/mammoth/internal/docx/BodyXmlReader.java index ace6628..4282bf6 100644 --- a/src/main/java/org/zwobble/mammoth/internal/docx/BodyXmlReader.java +++ b/src/main/java/org/zwobble/mammoth/internal/docx/BodyXmlReader.java @@ -29,7 +29,7 @@ public class BodyXmlReader { this.fileReader = fileReader; } - ReadResult readElements(Iterable nodes) { + public ReadResult readElements(Iterable nodes) { return new StatefulBodyXmlReader( styles, numbering, diff --git a/src/main/java/org/zwobble/mammoth/internal/docx/StatefulBodyXmlReader.java b/src/main/java/org/zwobble/mammoth/internal/docx/StatefulBodyXmlReader.java index b549387..3be34ad 100644 --- a/src/main/java/org/zwobble/mammoth/internal/docx/StatefulBodyXmlReader.java +++ b/src/main/java/org/zwobble/mammoth/internal/docx/StatefulBodyXmlReader.java @@ -19,6 +19,7 @@ import static org.zwobble.mammoth.internal.docx.ReadResult.*; import static org.zwobble.mammoth.internal.docx.Uris.uriToZipEntryName; import static org.zwobble.mammoth.internal.util.Iterables.lazyFilter; import static org.zwobble.mammoth.internal.util.Iterables.tryGetLast; +import static org.zwobble.mammoth.internal.util.Lists.eagerConcat; import static org.zwobble.mammoth.internal.util.Lists.list; import static org.zwobble.mammoth.internal.util.Maps.entry; import static org.zwobble.mammoth.internal.util.Maps.lookup; @@ -36,6 +37,7 @@ class StatefulBodyXmlReader { private final FileReader fileReader; private final StringBuilder currentInstrText; private final Queue complexFieldStack; + private List deletedParagraphContents; private interface ComplexField { ComplexField UNKNOWN = new ComplexField() {}; @@ -70,6 +72,7 @@ class StatefulBodyXmlReader { this.fileReader = fileReader; this.currentInstrText = new StringBuilder(); this.complexFieldStack = Queues.stack(); + this.deletedParagraphContents = new ArrayList<>(); } ReadResult readElement(XmlElement element) { @@ -249,17 +252,33 @@ class StatefulBodyXmlReader { return readStyle(properties, "w:rStyle", "Run", styles::findCharacterStyleById); } - ReadResult readElements(Iterable nodes) { + ReadResult readElements(Iterable nodes) { return ReadResult.flatMap(lazyFilter(nodes, XmlElement.class), this::readElement); } private ReadResult readParagraph(XmlElement element) { XmlElementLike properties = element.findChildOrEmpty("w:pPr"); - ParagraphIndent indent = readParagraphIndent(properties); - return ReadResult.map( - readParagraphStyle(properties), - readElements(element.getChildren()), - (style, children) -> new Paragraph(style, readNumbering(style, properties), indent, children)).appendExtra(); + + boolean isDeleted = properties + .findChildOrEmpty("w:rPr") + .findChild("w:del") + .isPresent(); + + if (isDeleted) { + deletedParagraphContents.addAll(element.getChildren()); + return ReadResult.success(list()); + } else { + ParagraphIndent indent = readParagraphIndent(properties); + List childrenXml = element.getChildren(); + if (!deletedParagraphContents.isEmpty()) { + childrenXml = eagerConcat(deletedParagraphContents, childrenXml); + deletedParagraphContents = new ArrayList<>(); + } + return ReadResult.map( + readParagraphStyle(properties), + readElements(childrenXml), + (style, children) -> new Paragraph(style, readNumbering(style, properties), indent, children)).appendExtra(); + } } private ReadResult readFieldChar(XmlElement element) { -- cgit v1.2.3