innertext 0.2.3__tar.gz → 0.2.4__tar.gz

{innertext-0.2.3 → innertext-0.2.4}/Cargo.lock

@@ -115,7 +115,7 @@ dependencies = [
 
 [[package]]
 name = "innertext-core"
-version = "0.2.3"
+version = "0.2.4"
 dependencies = [
  "html5ever",
  "markup5ever",
@@ -128,7 +128,7 @@ dependencies = [
 
 [[package]]
 name = "innertext-java"
-version = "0.2.3"
+version = "0.2.4"
 dependencies = [
  "innertext-core",
  "jni",
@@ -137,7 +137,7 @@ dependencies = [
 
 [[package]]
 name = "innertext-node"
-version = "0.2.3"
+version = "0.2.4"
 dependencies = [
  "innertext-core",
  "napi",
@@ -146,7 +146,7 @@ dependencies = [
 
 [[package]]
 name = "innertext-python"
-version = "0.2.3"
+version = "0.2.4"
 dependencies = [
  "innertext-core",
  "pyo3",

{innertext-0.2.3 → innertext-0.2.4}/Cargo.toml

@@ -6,7 +6,7 @@ members = ["crates/innertext-core", "bindings/innertext-python"]
 resolver = "2"
 
 [workspace.package]
-version = "0.2.3"
+version = "0.2.4"
 edition = "2021"
 license = "MIT"
 

{innertext-0.2.3 → innertext-0.2.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: innertext
-Version: 0.2.3
+Version: 0.2.4
 Classifier: Development Status :: 3 - Alpha
 Classifier: Environment :: Console
 Classifier: Intended Audience :: Developers

{innertext-0.2.3 → innertext-0.2.4}/crates/innertext-core/src/inner_text.rs

@@ -22,6 +22,8 @@ pub(crate) fn extract(dom: &RcDom) -> String {
     }
 
     let normalized = normalize_tokens(tokens);
+    let normalized = drop_inter_block_separator_spaces(normalized);
+    let normalized = drop_boundaries_between_required_breaks(normalized);
     let folded = fold_required_break_runs(normalized);
     serialize_tokens(folded)
 }
@@ -133,6 +135,48 @@ fn normalize_tokens(mut tokens: Vec<RenderToken>) -> Vec<RenderToken> {
     tokens
 }
 
+fn drop_inter_block_separator_spaces(tokens: Vec<RenderToken>) -> Vec<RenderToken> {
+    let mut out = Vec::with_capacity(tokens.len());
+
+    for i in 0..tokens.len() {
+        if let RenderToken::Text(text) = &tokens[i] {
+            let prev_is_marker = i > 0
+                && matches!(
+                    tokens.get(i - 1),
+                    Some(RenderToken::RequiredBreak(_) | RenderToken::CollapseBoundary)
+                );
+            let next_is_marker = matches!(
+                tokens.get(i + 1),
+                Some(RenderToken::RequiredBreak(_) | RenderToken::CollapseBoundary)
+            );
+            if text.chars().all(|ch| ch == ' ') && prev_is_marker && next_is_marker {
+                continue;
+            }
+        }
+        out.push(tokens[i].clone());
+    }
+
+    out
+}
+
+fn drop_boundaries_between_required_breaks(tokens: Vec<RenderToken>) -> Vec<RenderToken> {
+    let mut out = Vec::with_capacity(tokens.len());
+
+    for i in 0..tokens.len() {
+        if matches!(tokens[i], RenderToken::CollapseBoundary) {
+            let prev_is_break =
+                i > 0 && matches!(tokens.get(i - 1), Some(RenderToken::RequiredBreak(_)));
+            let next_is_break = matches!(tokens.get(i + 1), Some(RenderToken::RequiredBreak(_)));
+            if prev_is_break && next_is_break {
+                continue;
+            }
+        }
+        out.push(tokens[i].clone());
+    }
+
+    out
+}
+
 fn fold_required_break_runs(tokens: Vec<RenderToken>) -> Vec<RenderToken> {
     let mut out = Vec::new();
     let mut i = 0;

{innertext-0.2.3 → innertext-0.2.4}/crates/innertext-core/tests/spec_behavior.rs

@@ -51,6 +51,48 @@ fn sibling_blocks_have_two_newlines_between_them() {
     assert_eq!(got, "Block A\nBlock B");
 }
 
+#[test]
+fn formatting_whitespace_between_p_tags_does_not_change_inner_text() {
+    let compact = "<p>Some text</p><p>Some more text which <span>is inside a span</span></p><p style='display: none'>This text is hidden</p>";
+    let spaced = "<p>Some text</p>\n\n<p>Some more text which <span>is inside a span</span></p><p style='display: none'>This text is hidden</p>";
+    let heavily_spaced = "<p>Some text</p>\n\n\n\n   <p>Some more text which <span>is inside a span</span></p><p style='display: none'>This text is hidden</p>";
+
+    let expected = inner_text_from_html(compact).expect("must extract compact text");
+    let spaced_actual = inner_text_from_html(spaced).expect("must extract spaced text");
+    let heavily_spaced_actual =
+        inner_text_from_html(heavily_spaced).expect("must extract heavily spaced text");
+
+    assert_eq!(spaced_actual, expected);
+    assert_eq!(heavily_spaced_actual, expected);
+    assert_eq!(
+        expected,
+        "Some text\n\nSome more text which is inside a span"
+    );
+}
+
+#[test]
+fn formatting_whitespace_between_div_tags_does_not_change_inner_text() {
+    let compact = "<div>Block A</div><div>Block B</div>";
+    let spaced = "<div>Block A</div>\n\n\n    <div>Block B</div>";
+
+    let expected = inner_text_from_html(compact).expect("must extract compact text");
+    let actual = inner_text_from_html(spaced).expect("must extract spaced text");
+
+    assert_eq!(actual, expected);
+    assert_eq!(expected, "Block A\nBlock B");
+}
+
+#[test]
+fn hidden_block_with_separator_whitespace_does_not_add_extra_gaps() {
+    let compact = "<p>One</p><p style='display:none'>Hidden</p><p>Two</p>";
+    let spaced = "<p>One</p>\n\n\n<p style='display:none'>Hidden</p>\n\n\n<p>Two</p>";
+
+    let expected = inner_text_from_html(compact).expect("must extract compact text");
+    let actual = inner_text_from_html(spaced).expect("must extract spaced text");
+
+    assert_eq!(actual, expected);
+}
+
 // ── outerText getter ───────────────────────────────────────────────────────
 
 #[test]

{innertext-0.2.3 → innertext-0.2.4}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "maturin"
 
 [project]
 name = "innertext"
-version = "0.2.3"
+version = "0.2.4"
 description = "Pure Rust WHATWG-compliant innerText, outerText, and textContent extraction"
 readme = "README.md"
 license = { text = "MIT" }