Improve HTML table detection

docs/html-table-support.md

@@ -8,3 +8,4 @@ consistently.
 The crate `markup5ever_rcdom` provides a minimal DOM which `html5ever` populates
 and which is traversed to extract rows and cells. Only basic tables containing
 `<tr>`, `<th>` and `<td>` elements are supported.
+`mdtablefix` detects table elements regardless of attribute usage or tag case.

src/lib.rs

@@ -219,9 +219,9 @@
     out: &mut Vec<String>,
 ) {
     html_buf.push(line.trim_end().to_string());
-    *html_depth += line.matches("<table").count();
-    if line.contains("</table>") {
-        *html_depth = html_depth.saturating_sub(line.matches("</table>").count());
+    *html_depth += TABLE_START_RE.find_iter(line).count();
+    if TABLE_END_RE.is_match(line) {
+        *html_depth = html_depth.saturating_sub(TABLE_END_RE.find_iter(line).count());
         if *html_depth == 0 {
             out.extend(html_table_to_markdown(html_buf));
             html_buf.clear();
@@ -256,6 +256,10 @@
     std::sync::LazyLock::new(|| Regex::new(r"\|\s*\|\s*").unwrap());
 static SEP_RE: std::sync::LazyLock<Regex> =
     std::sync::LazyLock::new(|| Regex::new(r"^[\s|:-]+$").unwrap());
+static TABLE_START_RE: std::sync::LazyLock<Regex> =
+    std::sync::LazyLock::new(|| Regex::new(r"(?i)^<table(?:\s|>|$)").unwrap());
+static TABLE_END_RE: std::sync::LazyLock<Regex> =
+    std::sync::LazyLock::new(|| Regex::new(r"(?i)</table>").unwrap());
 
 #[must_use]
 pub fn reflow_table(lines: &[String]) -> Vec<String> {
@@ -419,7 +423,7 @@
             continue;
         }
 
-        if line.trim_start().starts_with("<table") {
+        if TABLE_START_RE.is_match(line.trim_start()) {
             if !buf.is_empty() {
                 if in_table {
                     out.extend(reflow_table(&buf));

tests/integration.rs

@@ -73,6 +73,36 @@ fn html_table() -> Vec<String> {
     ]
 }
 
+#[fixture]
+fn html_table_with_attrs() -> Vec<String> {
+    vec![
+        "<table class=\"x\">".to_string(),
+        "<tr><th>A</th><th>B</th></tr>".to_string(),
+        "<tr><td>1</td><td>2</td></tr>".to_string(),
+        "</table>".to_string(),
+    ]
+}
+
+#[fixture]
+fn html_table_uppercase() -> Vec<String> {
+    vec![
+        "<TABLE>".to_string(),
+        "<tr><th>A</th><th>B</th></tr>".to_string(),
+        "<tr><td>1</td><td>2</td></tr>".to_string(),
+        "</TABLE>".to_string(),
+    ]
+}
+
+#[fixture]
+fn html_table_mixed_case() -> Vec<String> {
+    vec![
+        "<TaBlE>".to_string(),
+        "<tr><th>A</th><th>B</th></tr>".to_string(),
+        "<tr><td>1</td><td>2</td></tr>".to_string(),
+        "</TaBlE>".to_string(),
+    ]
+}
+
 #[fixture]
 fn multiple_tables() -> Vec<String> {
     vec![
@@ -135,6 +165,24 @@ fn test_process_stream_html_table(html_table: Vec<String>) {
     assert_eq!(process_stream(&html_table), expected);
 }
 
+#[rstest]
+fn test_process_stream_html_table_with_attrs(html_table_with_attrs: Vec<String>) {
+    let expected = vec!["| A | B |", "| --- | --- |", "| 1 | 2 |"];
+    assert_eq!(process_stream(&html_table_with_attrs), expected);
+}
+
+#[rstest]
+fn test_process_stream_html_table_uppercase(html_table_uppercase: Vec<String>) {
+    let expected = vec!["| A | B |", "| --- | --- |", "| 1 | 2 |"];
+    assert_eq!(process_stream(&html_table_uppercase), expected);
+}
+
+#[rstest]
+fn test_process_stream_html_table_mixed_case(html_table_mixed_case: Vec<String>) {
+    let expected = vec!["| A | B |", "| --- | --- |", "| 1 | 2 |"];
+    assert_eq!(process_stream(&html_table_mixed_case), expected);
+}
+
 #[rstest]
 fn test_process_stream_multiple_tables(multiple_tables: Vec<String>) {
     let expected = vec![