aboutsummaryrefslogtreecommitdiff
path: root/src/main/java/org/zwobble/mammoth/internal/conversion/RawText.java
diff options
context:
space:
mode:
authorMichael Williamson <mike@zwobble.org>2021-08-01 11:17:17 +0100
committerMichael Williamson <mike@zwobble.org>2021-08-01 11:17:17 +0100
commit3287e71ce896193eaf2bba8e6da9bf1827587ab3 (patch)
tree107ee1e381ade725cc517532778c08a229597c53 /src/main/java/org/zwobble/mammoth/internal/conversion/RawText.java
parent557ea3126459943147e17267b7af23680dd128f8 (diff)
Extract RawText class
Diffstat (limited to 'src/main/java/org/zwobble/mammoth/internal/conversion/RawText.java')
-rw-r--r--src/main/java/org/zwobble/mammoth/internal/conversion/RawText.java35
1 files changed, 35 insertions, 0 deletions
diff --git a/src/main/java/org/zwobble/mammoth/internal/conversion/RawText.java b/src/main/java/org/zwobble/mammoth/internal/conversion/RawText.java
new file mode 100644
index 0000000..0834874
--- /dev/null
+++ b/src/main/java/org/zwobble/mammoth/internal/conversion/RawText.java
@@ -0,0 +1,35 @@
+package org.zwobble.mammoth.internal.conversion;
+
+import org.zwobble.mammoth.internal.documents.*;
+
+import java.util.List;
+
+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 RawText {
+ public static String extractRawText(Document document) {
+ return extractRawTextOfChildren(document);
+ }
+
+ 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)));
+ }
+
+ public 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;
+ });
+ }
+}