fixed missing spaces inside some tags

Signed-off-by: Jonas Kalderstam <[email protected]>
spacecowboy · Dec 18, 2024 · 32cb38f · 32cb38f
1 parent 235228e
commit 32cb38f
Show file tree

Hide file tree

Showing 2 changed files with 21 additions and 18 deletions.
diff --git a/app/src/main/java/com/nononsenseapps/feeder/model/html/HtmlLinearizer.kt b/app/src/main/java/com/nononsenseapps/feeder/model/html/HtmlLinearizer.kt
@@ -69,7 +69,6 @@ class HtmlLinearizer {
                     if (blockStyle.shouldSoftWrap) {
                         node.appendCorrectlyNormalizedWhiteSpace(
                             linearTextBuilder,
-                            stripLeading = linearTextBuilder.endsWithWhitespace,
                         )
                     } else {
                         append(node.wholeText)
@@ -113,7 +112,6 @@ class HtmlLinearizer {
                                 withLinearTextAnnotation(LinearTextAnnotationH1) {
                                     element.appendCorrectlyNormalizedWhiteSpaceRecursively(
                                         linearTextBuilder,
-                                        stripLeading = linearTextBuilder.endsWithWhitespace,
                                     )
                                 }
                             }
@@ -124,7 +122,6 @@ class HtmlLinearizer {
                                 withLinearTextAnnotation(LinearTextAnnotationH2) {
                                     element.appendCorrectlyNormalizedWhiteSpaceRecursively(
                                         linearTextBuilder,
-                                        stripLeading = linearTextBuilder.endsWithWhitespace,
                                     )
                                 }
                             }
@@ -135,7 +132,6 @@ class HtmlLinearizer {
                                 withLinearTextAnnotation(LinearTextAnnotationH3) {
                                     element.appendCorrectlyNormalizedWhiteSpaceRecursively(
                                         linearTextBuilder,
-                                        stripLeading = linearTextBuilder.endsWithWhitespace,
                                     )
                                 }
                             }
@@ -146,7 +142,6 @@ class HtmlLinearizer {
                                 withLinearTextAnnotation(LinearTextAnnotationH4) {
                                     element.appendCorrectlyNormalizedWhiteSpaceRecursively(
                                         linearTextBuilder,
-                                        stripLeading = linearTextBuilder.endsWithWhitespace,
                                     )
                                 }
                             }
@@ -157,7 +152,6 @@ class HtmlLinearizer {
                                 withLinearTextAnnotation(LinearTextAnnotationH5) {
                                     element.appendCorrectlyNormalizedWhiteSpaceRecursively(
                                         linearTextBuilder,
-                                        stripLeading = linearTextBuilder.endsWithWhitespace,
                                     )
                                 }
                             }
@@ -168,7 +162,6 @@ class HtmlLinearizer {
                                 withLinearTextAnnotation(LinearTextAnnotationH6) {
                                     element.appendCorrectlyNormalizedWhiteSpaceRecursively(
                                         linearTextBuilder,
-                                        stripLeading = linearTextBuilder.endsWithWhitespace,
                                     )
                                 }
                             }
@@ -925,13 +918,10 @@ class HtmlLinearizer {
  * Can't use JSoup's text() method because that strips invisible characters
  * such as ZWNJ which are crucial for several languages.
  */
-fun TextNode.appendCorrectlyNormalizedWhiteSpace(
-    builder: LinearTextBuilder,
-    stripLeading: Boolean,
-) {
+fun TextNode.appendCorrectlyNormalizedWhiteSpace(builder: LinearTextBuilder) {
     wholeText.asUTF8Sequence()
         .dropWhile {
-            stripLeading && isCollapsableWhiteSpace(it)
+            builder.endsWithWhitespace && isCollapsableWhiteSpace(it)
         }
         .fold(false) { lastWasWhite, char ->
             if (isCollapsableWhiteSpace(char)) {
@@ -946,17 +936,13 @@ fun TextNode.appendCorrectlyNormalizedWhiteSpace(
         }
 }
 
-fun Element.appendCorrectlyNormalizedWhiteSpaceRecursively(
-    builder: LinearTextBuilder,
-    stripLeading: Boolean,
-) {
+fun Element.appendCorrectlyNormalizedWhiteSpaceRecursively(builder: LinearTextBuilder) {
     for (child in childNodes()) {
         when (child) {
-            is TextNode -> child.appendCorrectlyNormalizedWhiteSpace(builder, stripLeading)
+            is TextNode -> child.appendCorrectlyNormalizedWhiteSpace(builder)
             is Element ->
                 child.appendCorrectlyNormalizedWhiteSpaceRecursively(
                     builder,
-                    stripLeading,
                 )
         }
     }

diff --git a/app/src/test/java/com/nononsenseapps/feeder/model/html/HtmlLinearizerTest.kt b/app/src/test/java/com/nononsenseapps/feeder/model/html/HtmlLinearizerTest.kt
@@ -36,6 +36,23 @@ class HtmlLinearizerTest {
         assertEquals(LinearText("Hello, world!", LinearTextBlockStyle.TEXT), result[0])
     }
 
+    @Test
+    fun `spaces inside headers are kept`() {
+        val html =
+            """
+            <html><body><h2><a href="http://example.com">Link</a> <small>small</small></h2></body></html>
+            """.trimIndent()
+        val baseUrl = "https://example.com"
+
+        val result = linearizer.linearize(html, baseUrl).elements
+
+        assertEquals(1, result.size)
+        assertEquals(
+            LinearText("Link small", LinearTextBlockStyle.TEXT, LinearTextAnnotation(LinearTextAnnotationH2, 0, 9)),
+            result[0],
+        )
+    }
+
     @Test
     fun `should return annotations with bold, italic, and underline`() {
         val html = "<html><body><b><i><u>Hello, world!</u></i></b></body></html>"