fix: BibleTextNode whitespace parsing

fix: BibleTextNode whitespace parsing

This makes the WMB render correctly, despite it having lots of unintended (but theoretically fine) <span> nodes.

Author: davidfedor

Sources/YouVersionPlatformCore/Bible/BibleTextNode.swift

@@ -114,29 +114,32 @@ public class BibleTextNode {
         }
 
         func parser(_ parser: XMLParser, foundCharacters string: String) {
-            let collapsed = string.replacingOccurrences(of: "\\s+", with: " ", options: .regularExpression)
-            let core = collapsed.trimmingCharacters(in: .whitespacesAndNewlines)
-            guard !core.isEmpty else {
+            guard let current = stack.last else {
                 return
             }
-            let leadingSpace = string.first?.isWhitespace == true
-            let trailingSpace = string.last?.isWhitespace == true
-            var segment = core
-            if leadingSpace {
-                segment = " " + segment
-            }
-            if trailingSpace {
-                segment += " "
-            }
-            guard let current = stack.last else {
+
+            let segment = string.replacingOccurrences(of: "\\s+", with: " ", options: .regularExpression)
+            guard !segment.isEmpty else {
                 return
             }
 
+            if segment == " " {
+                guard let previousChild = current.children.last else {
+                    return
+                }
+                guard previousChild.type == .span || previousChild.type == .text else {
+                    return
+                }
+            }
+
             if let lastChild = current.children.last, lastChild.type == .text {
-                lastChild.textSegments.append(segment)
-                let joined = lastChild.textSegments.joined()
+                let joined = (lastChild.text + segment)
+                    .replacingOccurrences(of: " {2,}", with: " ", options: .regularExpression)
+                guard !joined.isEmpty else {
+                    return
+                }
+                lastChild.textSegments = [joined]
                 lastChild.text = joined
-                lastChild.textSegments = joined.isEmpty ? [] : [joined]
             } else {
                 let textNode = BibleTextNode(name: "text")
                 textNode.textSegments = [segment]
@@ -148,7 +151,6 @@ public class BibleTextNode {
         func parser(_ parser: XMLParser, didEndElement elementName: String, namespaceURI: String?, qualifiedName qName: String?) {
             _ = stack.popLast()
         }
-
     }
 }
 

Tests/YouVersionPlatformCoreTests/BibleTextNodeTests.swift

@@ -81,3 +81,52 @@ func testParse_GenesisIntroContainsText() throws {
     let texts = collectTexts(root)
     #expect(texts.contains("In the beginning, God created the heavens and the earth."))
 }
+
+@Test
+func testParse_SpacesBetweenInlineSpansArePreserved() throws {
+    let html = "<div><span>One</span> <span>Two</span>   <span>Three three   three</span>.</div>"
+
+    let root = try #require(try BibleTextNode.parse(html))
+    let block = try #require(root.children.first)
+
+    let renderedText = collectRenderedText(from: block)
+    #expect(renderedText == "One Two Three three three.")
+}
+
+@Test
+func testParse_MixedWhitespaceAndInlineNodesCollapseToSingleSpaces() throws {
+    let html = "<div>  Start <span>middle</span>\n\t <span>end</span>   done  </div>"
+
+    let root = try #require(try BibleTextNode.parse(html))
+    let block = try #require(root.children.first)
+
+    let renderedText = collectRenderedText(from: block)
+    #expect(renderedText == "Start middle end done")
+}
+
+@Test
+func testParse_LeadingWhitespaceBeforeFirstChildIsIgnored() throws {
+    let html = "<div>   <span>One</span>   <span>Two</span></div>"
+
+    let root = try #require(try BibleTextNode.parse(html))
+    let block = try #require(root.children.first)
+
+    let renderedText = collectRenderedText(from: block)
+    #expect(renderedText == "One Two")
+}
+
+private func collectRenderedText(from node: BibleTextNode) -> String {
+    var text = ""
+    appendRenderedText(from: node, into: &text)
+    return text.trimmingCharacters(in: .whitespacesAndNewlines)
+}
+
+private func appendRenderedText(from node: BibleTextNode, into text: inout String) {
+    if node.type == .text {
+        text += node.text
+    }
+
+    for child in node.children {
+        appendRenderedText(from: child, into: &text)
+    }
+}