diff options
| author | Michael Williamson <mike@zwobble.org> | 2016-04-17 18:57:29 +0100 |
|---|---|---|
| committer | Michael Williamson <mike@zwobble.org> | 2016-04-17 18:58:12 +0100 |
| commit | 52140d52210eda0f25ecab9f89b5c73473e9d8e0 (patch) | |
| tree | cb383edb71e44b8ae7b2b0d0890f13574d1922f0 /src/main/java/org/zwobble/mammoth/internal/InternalDocumentConverter.java | |
| parent | 0f9ef005f063198ea2e16ca36beaead4bd5e3824 (diff) | |
Add CLI for .NET version
Diffstat (limited to 'src/main/java/org/zwobble/mammoth/internal/InternalDocumentConverter.java')
| -rw-r--r-- | src/main/java/org/zwobble/mammoth/internal/InternalDocumentConverter.java | 110 |
1 files changed, 110 insertions, 0 deletions
diff --git a/src/main/java/org/zwobble/mammoth/internal/InternalDocumentConverter.java b/src/main/java/org/zwobble/mammoth/internal/InternalDocumentConverter.java new file mode 100644 index 0000000..4cc0e28 --- /dev/null +++ b/src/main/java/org/zwobble/mammoth/internal/InternalDocumentConverter.java @@ -0,0 +1,110 @@ +package org.zwobble.mammoth.internal; + +import org.zwobble.mammoth.internal.conversion.DocumentToHtml; +import org.zwobble.mammoth.internal.conversion.DocumentToHtmlOptions; +import org.zwobble.mammoth.internal.documents.DocumentElement; +import org.zwobble.mammoth.internal.documents.HasChildren; +import org.zwobble.mammoth.internal.documents.Paragraph; +import org.zwobble.mammoth.internal.documents.Text; +import org.zwobble.mammoth.internal.docx.DocxFile; +import org.zwobble.mammoth.internal.docx.InMemoryDocxFile; +import org.zwobble.mammoth.internal.docx.ZippedDocxFile; +import org.zwobble.mammoth.internal.html.Html; +import org.zwobble.mammoth.internal.results.InternalResult; +import org.zwobble.mammoth.internal.util.PassThroughException; + +import java.io.File; +import java.io.IOException; +import java.io.InputStream; +import java.nio.file.Path; +import java.util.List; +import java.util.Optional; +import java.util.function.Function; + +import static org.zwobble.mammoth.internal.docx.DocumentReader.readDocument; +import static org.zwobble.mammoth.internal.util.Casts.tryCast; +import static org.zwobble.mammoth.internal.util.Iterables.lazyMap; +import static org.zwobble.mammoth.internal.util.Lists.list; + +public class InternalDocumentConverter { + private final DocumentToHtmlOptions options; + + public InternalDocumentConverter(DocumentToHtmlOptions options) { + this.options = options; + } + + public InternalResult<String> convertToHtml(InputStream stream) throws IOException { + return PassThroughException.unwrap(() -> + withDocxFile(stream, zipFile -> + convertToHtml(Optional.empty(), zipFile))); + } + + public InternalResult<String> convertToHtml(File file) throws IOException { + return PassThroughException.unwrap(() -> + withDocxFile(file, zipFile -> + convertToHtml(Optional.of(file.toPath()), zipFile))); + } + + private InternalResult<String> convertToHtml(Optional<Path> path, DocxFile zipFile) { + return readDocument(path, zipFile) + .flatMap(nodes -> DocumentToHtml.convertToHtml(nodes, options)) + .map(Html::stripEmpty) + .map(Html::collapse) + .map(Html::write); + } + + public InternalResult<String> extractRawText(InputStream stream) throws IOException { + return PassThroughException.unwrap(() -> + withDocxFile(stream, zipFile -> + extractRawText(Optional.empty(), zipFile))); + } + + public InternalResult<String> extractRawText(File file) throws IOException { + return PassThroughException.unwrap(() -> + withDocxFile(file, zipFile -> + extractRawText(Optional.of(file.toPath()), zipFile))); + } + + private InternalResult<String> extractRawText(Optional<Path> path, DocxFile zipFile) { + return readDocument(path, zipFile) + .map(InternalDocumentConverter::extractRawTextOfChildren); + } + + private static <T> T withDocxFile(File file, Function<DocxFile, T> function) throws IOException { + DocxFile zipFile = new ZippedDocxFile(file); + try { + return function.apply(zipFile); + } finally { + zipFile.close(); + } + } + + private static <T> T withDocxFile(InputStream stream, Function<DocxFile, T> function) throws IOException { + DocxFile zipFile = InMemoryDocxFile.fromStream(stream); + try { + return function.apply(zipFile); + } finally { + zipFile.close(); + } + } + + private static String extractRawTextOfChildren(HasChildren parent) { + return extractRawText(parent.getChildren()); + } + + private static String extractRawText(List<DocumentElement> nodes) { + return String.join("", lazyMap(nodes, node -> extractRawText(node))); + } + + private static String extractRawText(DocumentElement node) { + return tryCast(Text.class, node) + .map(Text::getValue) + .orElseGet(() -> { + List<DocumentElement> children = tryCast(HasChildren.class, node) + .map(HasChildren::getChildren) + .orElse(list()); + String suffix = tryCast(Paragraph.class, node).map(paragraph -> "\n\n").orElse(""); + return extractRawText(children) + suffix; + }); + } +} |
