From 12e99afb5e0deb32f0c6b7357c36d24da2470b46 Mon Sep 17 00:00:00 2001
From: Simon Holthausen <simon.holthausen@accso.de>
Date: Sat, 26 Sep 2020 16:25:23 +0200
Subject: [PATCH] (fix) try to be smarter about unclosed tags

Tries to parse the HTML document and work around the limitations of the underlying parser which will trip up when there's a non-closed tag and treat everything afterwards as its child.
#547
---
 .../src/lib/documents/utils.ts                | 49 +++++++++++++++++++
 .../test/lib/documents/utils.test.ts          | 15 ++++++
 2 files changed, 64 insertions(+)

diff --git a/packages/language-server/src/lib/documents/utils.ts b/packages/language-server/src/lib/documents/utils.ts
index c3ea5f004..45440f249 100644
--- a/packages/language-server/src/lib/documents/utils.ts
+++ b/packages/language-server/src/lib/documents/utils.ts
@@ -44,6 +44,46 @@ const parser = getLanguageService();
  * Parses text as HTML
  */
 export function parseHtml(text: string): HTMLDocument {
+    const html = parseHTMLDocument(text);
+    return tryParse(text, html, 0) || html;
+}
+
+/**
+ * Tries to parse the HTML document and work around the limitations of the underlying
+ * parsers which will trip up when there's a non-closed tag and treat everything afterwards as its child.
+ */
+function tryParse(text: string, html: HTMLDocument, run: number): HTMLDocument | null {
+    if (run > 5) {
+        // Stop before doing too much work, possibly having an endless loop
+        return null;
+    }
+
+    let sanitizedText = text;
+    html.roots.forEach((node) => {
+        if (
+            !(<any>node).closed &&
+            // This indicates a child tag falsely identified as an attribute
+            Object.keys(node.attributes || {}).some((key) => key.startsWith('<')) &&
+            node.tag
+        ) {
+            // Replace <some-unclosed-tag with whitespaces of same length
+            sanitizedText = replaceRangeWithStr(
+                sanitizedText,
+                node.start,
+                node.start + node.tag.length + 1,
+                ' '.repeat(node.tag.length + 1),
+            );
+        }
+    });
+
+    if (sanitizedText !== text) {
+        return tryParse(sanitizedText, parseHTMLDocument(sanitizedText), run + 1);
+    } else {
+        return html;
+    }
+}
+
+function parseHTMLDocument(text: string) {
     // We can safely only set getText because only this is used for parsing
     return parser.parseHTMLDocument(<any>{ getText: () => text });
 }
@@ -332,3 +372,12 @@ export function getWordAt(
 
     return str.slice(left, right + pos);
 }
+
+export function replaceRangeWithStr(
+    text: string,
+    start: number,
+    end: number,
+    replacement: string,
+): string {
+    return text.substring(0, start) + replacement + text.substring(end);
+}
diff --git a/packages/language-server/test/lib/documents/utils.test.ts b/packages/language-server/test/lib/documents/utils.test.ts
index 013fc9fd0..77d09b81d 100644
--- a/packages/language-server/test/lib/documents/utils.test.ts
+++ b/packages/language-server/test/lib/documents/utils.test.ts
@@ -286,6 +286,21 @@ describe('document/utils', () => {
                 container: { start: 151, end: 181 },
             });
         });
+
+        it('extract script when unclosed tag preceeding', () => {
+            const text = `
+          <Unclosed-Tag
+          <script>let value = 2</script>`;
+            assert.deepStrictEqual(extractScriptTags(text)?.script, {
+                content: 'let value = 2',
+                attributes: {},
+                start: 43,
+                end: 56,
+                startPos: Position.create(2, 18),
+                endPos: Position.create(2, 31),
+                container: { start: 35, end: 65 },
+            });
+        });
     });
 
     describe('#getLineAtPosition', () => {