From 6005c9dc75a8d17761d775b6a86d79892aef1a56 Mon Sep 17 00:00:00 2001 From: Michael Williamson Date: Sat, 26 Sep 2015 18:11:53 +0100 Subject: Add conversion of single paragraph documents --- src/main/java/org/zwobble/mammoth/Mammoth.java | 50 +++++++++++++++++++++- .../java/org/zwobble/mammoth/xml/XmlElement.java | 9 ++++ src/main/java/org/zwobble/mammoth/xml/XmlNode.java | 2 +- .../org/zwobble/mammoth/xml/XmlNodeVisitor.java | 6 +++ .../java/org/zwobble/mammoth/xml/XmlTextNode.java | 5 +++ .../org/zwobble/mammoth/tests/MammothTests.java | 6 +-- .../java/org/zwobble/mammoth/tests/TestData.java | 13 +++--- 7 files changed, 79 insertions(+), 12 deletions(-) create mode 100644 src/main/java/org/zwobble/mammoth/xml/XmlNodeVisitor.java (limited to 'src') diff --git a/src/main/java/org/zwobble/mammoth/Mammoth.java b/src/main/java/org/zwobble/mammoth/Mammoth.java index 5a2ed17..aa3514f 100644 --- a/src/main/java/org/zwobble/mammoth/Mammoth.java +++ b/src/main/java/org/zwobble/mammoth/Mammoth.java @@ -1,11 +1,57 @@ package org.zwobble.mammoth; +import java.io.File; +import java.io.IOException; import java.io.InputStream; +import java.util.zip.ZipFile; + +import org.zwobble.mammoth.xml.XmlElement; +import org.zwobble.mammoth.xml.XmlNodeVisitor; +import org.zwobble.mammoth.xml.XmlParser; +import org.zwobble.mammoth.xml.XmlTextNode; + +import com.google.common.base.Joiner; +import com.google.common.base.Strings; +import com.google.common.collect.ImmutableBiMap; + +import static com.google.common.collect.Iterables.transform; + +import lombok.val; public class Mammoth { - public static String convertToHtml(InputStream stream) { - return ""; + public static String convertToHtml(File file) { + try (val zipFile = new ZipFile(file)) { + val entry = zipFile.getEntry("word/document.xml"); + val documentXml = parseXml(zipFile.getInputStream(entry)); + val text = extractText(documentXml); + if (Strings.isNullOrEmpty(text)) { + return ""; + } else { + return "

" + text + "

"; + } + } catch (IOException e) { + throw new UnsupportedOperationException("Should return a result of failure"); + } + } + + private static XmlElement parseXml(InputStream inputStream) { + return new XmlParser(ImmutableBiMap.of()).parseStream(inputStream); + } + + private static String extractText(XmlElement documentXml) { + return documentXml.accept(new XmlNodeVisitor() { + @Override + public String visit(XmlElement element) { + return Joiner.on("").join( + transform(element.children(), child -> child.accept(this))); + } + + @Override + public String Visit(XmlTextNode textNode) { + return textNode.getValue(); + } + }); } } diff --git a/src/main/java/org/zwobble/mammoth/xml/XmlElement.java b/src/main/java/org/zwobble/mammoth/xml/XmlElement.java index a49f5f8..b5e8f1b 100644 --- a/src/main/java/org/zwobble/mammoth/xml/XmlElement.java +++ b/src/main/java/org/zwobble/mammoth/xml/XmlElement.java @@ -27,4 +27,13 @@ public class XmlElement implements XmlNode { private final String name; private final Map attributes; private final List children; + + @Override + public T accept(XmlNodeVisitor visitor) { + return visitor.visit(this); + } + + public Iterable children() { + return children; + } } diff --git a/src/main/java/org/zwobble/mammoth/xml/XmlNode.java b/src/main/java/org/zwobble/mammoth/xml/XmlNode.java index c91e100..37ce2cf 100644 --- a/src/main/java/org/zwobble/mammoth/xml/XmlNode.java +++ b/src/main/java/org/zwobble/mammoth/xml/XmlNode.java @@ -1,5 +1,5 @@ package org.zwobble.mammoth.xml; public interface XmlNode { - + T accept(XmlNodeVisitor visitor); } diff --git a/src/main/java/org/zwobble/mammoth/xml/XmlNodeVisitor.java b/src/main/java/org/zwobble/mammoth/xml/XmlNodeVisitor.java new file mode 100644 index 0000000..717e3db --- /dev/null +++ b/src/main/java/org/zwobble/mammoth/xml/XmlNodeVisitor.java @@ -0,0 +1,6 @@ +package org.zwobble.mammoth.xml; + +public interface XmlNodeVisitor { + T visit(XmlElement element); + T Visit(XmlTextNode textNode); +} diff --git a/src/main/java/org/zwobble/mammoth/xml/XmlTextNode.java b/src/main/java/org/zwobble/mammoth/xml/XmlTextNode.java index 0d63273..4d47626 100644 --- a/src/main/java/org/zwobble/mammoth/xml/XmlTextNode.java +++ b/src/main/java/org/zwobble/mammoth/xml/XmlTextNode.java @@ -5,4 +5,9 @@ import lombok.Value; @Value public class XmlTextNode implements XmlNode { private String value; + + @Override + public T accept(XmlNodeVisitor visitor) { + return visitor.Visit(this); + } } diff --git a/src/test/java/org/zwobble/mammoth/tests/MammothTests.java b/src/test/java/org/zwobble/mammoth/tests/MammothTests.java index da06d55..013deae 100644 --- a/src/test/java/org/zwobble/mammoth/tests/MammothTests.java +++ b/src/test/java/org/zwobble/mammoth/tests/MammothTests.java @@ -1,6 +1,5 @@ package org.zwobble.mammoth.tests; -import org.junit.Ignore; import org.junit.Test; import org.zwobble.mammoth.Mammoth; @@ -16,7 +15,6 @@ public class MammothTests { } @Test - @Ignore("WIP") public void docxContainingOneParagraphIsConvertedToSingleParagraphElement() { assertThat( convertToHtml("single-paragraph.docx"), @@ -24,7 +22,7 @@ public class MammothTests { } private String convertToHtml(String name) { - val stream = TestData.stream(name); - return Mammoth.convertToHtml(stream); + val file = TestData.file(name); + return Mammoth.convertToHtml(file); } } diff --git a/src/test/java/org/zwobble/mammoth/tests/TestData.java b/src/test/java/org/zwobble/mammoth/tests/TestData.java index 45968d8..e84caa4 100644 --- a/src/test/java/org/zwobble/mammoth/tests/TestData.java +++ b/src/test/java/org/zwobble/mammoth/tests/TestData.java @@ -1,13 +1,16 @@ package org.zwobble.mammoth.tests; -import java.io.IOException; -import java.io.InputStream; +import java.io.File; +import java.net.URISyntaxException; + +import lombok.val; public class TestData { - public static InputStream stream(String name) { + public static File file(String name) { try { - return TestData.class.getResource("/test-data/" + name).openStream(); - } catch (IOException exception) { + val uri = TestData.class.getResource("/test-data/" + name).toURI(); + return new File(uri); + } catch (URISyntaxException exception) { throw new RuntimeException(exception); } } -- cgit v1.2.3