[utils] js_to_json: Implement template strings (yt-dlp#6623)

Authored by: Grub4K

test/test_utils.py

@@ -1190,6 +1190,13 @@ def test_js_to_json_malformed(self):
         self.assertEqual(js_to_json('42a1'), '42"a1"')
         self.assertEqual(js_to_json('42a-1'), '42"a"-1')
 
+    def test_js_to_json_template_literal(self):
+        self.assertEqual(js_to_json('`Hello ${name}`', {'name': '"world"'}), '"Hello world"')
+        self.assertEqual(js_to_json('`${name}${name}`', {'name': '"X"'}), '"XX"')
+        self.assertEqual(js_to_json('`${name}${name}`', {'name': '5'}), '"55"')
+        self.assertEqual(js_to_json('`${name}"${name}"`', {'name': '5'}), '"5\\"5\\""')
+        self.assertEqual(js_to_json('`${name}`', {}), '"name"')
+
     def test_extract_attributes(self):
         self.assertEqual(extract_attributes('<e x="y">'), {'x': 'y'})
         self.assertEqual(extract_attributes("<e x='y'>"), {'x': 'y'})

yt_dlp/utils.py

@@ -3366,7 +3366,7 @@ def strip_jsonp(code):
 
 def js_to_json(code, vars={}, *, strict=False):
     # vars is a dict of var, val pairs to substitute
-    STRING_QUOTES = '\'"'
+    STRING_QUOTES = '\'"`'
     STRING_RE = '|'.join(rf'{q}(?:\\.|[^\\{q}])*{q}' for q in STRING_QUOTES)
     COMMENT_RE = r'/\*(?:(?!\*/).)*?\*/|//[^\n]*\n'
     SKIP_RE = fr'\s*(?:{COMMENT_RE})?\s*'
@@ -3384,6 +3384,12 @@ def process_escape(match):
                 else '' if escape == '\n'
                 else escape)
 
+    def template_substitute(match):
+        evaluated = js_to_json(match.group(1), vars, strict=strict)
+        if evaluated[0] == '"':
+            return json.loads(evaluated)
+        return evaluated
+
     def fix_kv(m):
         v = m.group(0)
         if v in ('true', 'false', 'null'):
@@ -3394,7 +3400,8 @@ def fix_kv(m):
             return ''
 
         if v[0] in STRING_QUOTES:
-            escaped = re.sub(r'(?s)(")|\\(.)', process_escape, v[1:-1])
+            v = re.sub(r'(?s)\${([^}]+)}', template_substitute, v[1:-1]) if v[0] == '`' else v[1:-1]
+            escaped = re.sub(r'(?s)(")|\\(.)', process_escape, v)
             return f'"{escaped}"'
 
         for regex, base in INTEGER_TABLE: