Keep escaped characters in CSS strings

src/zimscraperlib/rewriting/css.py

@@ -10,6 +10,8 @@
 """
 
 import re
+import string
+import unicodedata
 from collections.abc import Iterable, Sequence
 from functools import partial
 from typing import Any
@@ -29,6 +31,42 @@
 from zimscraperlib.rewriting.rx_replacer import RxRewriter, TransformationRule
 from zimscraperlib.rewriting.url_rewriting import ArticleUrlRewriter
 
+_HEX = set(string.hexdigits)
+_WS = set(" \t\n\r\f")
+
+
+def _escape_string_value(value: str) -> str:
+    """Serialize a CSS string value while preserving non-printable characters."""
+    out: list[str] = []
+    for idx, char in enumerate(value):
+        if char == '"':
+            out.append('\\"')
+        elif char == "\\":
+            out.append("\\\\")
+        elif char == "\n":
+            out.append("\\A ")
+        elif char == "\r":
+            out.append("\\D ")
+        elif char == "\f":
+            out.append("\\C ")
+        elif unicodedata.category(char) in ("Cc", "Co", "Cs", "Cn"):
+            out.append(f"\\{ord(char):x}")
+            nxt = value[idx + 1 : idx + 2]
+            if nxt and (nxt in _HEX or nxt in _WS):
+                out.append(" ")
+        else:
+            out.append(char)
+    return "".join(out)
+
+
+def _is_terminated_string(token: ast.StringToken) -> bool:
+    """Check whether a StringToken has a valid (unescaped) closing quote."""
+    rep = token.representation
+    if len(rep) <= 1 or rep[-1] != rep[0]:
+        return False
+    num_backslashes = len(rep[:-1]) - len(rep[:-1].rstrip("\\"))
+    return num_backslashes % 2 == 0
+
 
 class FallbackRegexCssRewriter(RxRewriter):
     """
@@ -211,6 +249,9 @@ def _process_node(self, node: ast.Node):
             self._process_list(
                 node.value  # pyright: ignore[reportUnknownMemberType, reportUnknownArgumentType]
             )
+        elif isinstance(node, ast.StringToken):
+            if _is_terminated_string(node):
+                node.representation = f'"{_escape_string_value(node.value)}"'
         elif isinstance(node, ast.URLToken):
             new_url = self.url_rewriter(
                 node.value,  # pyright: ignore[reportUnknownMemberType, reportUnknownArgumentType]

tests/rewriting/test_css_rewriting.py

@@ -216,3 +216,67 @@ def test_rewrite():
         ).rewrite(content)
         == expected
     )
+
+
+@pytest.mark.parametrize(
+    "css",
+    [
+        r'.test { content="\e123" }',
+        r'.test { content: "\e123" }',
+        r'.test::before { content: "\f101" }',
+        r'.test::before { content: "\e123\e124" }',
+        r'.test::before { content: "a\"b\\c" }',
+        r'.test { content: "\A " }',
+        r'.test { content: "\D " }',
+        r'.test { content: "\C " }',
+        r'.test { content: "\e123a" }',
+        r'.test { content: "a\\" }',
+        '"',
+    ],
+)
+def test_escaped_chars_are_kept(css: str):
+    rewriter = CssRewriter(
+        ArticleUrlRewriter(article_url=HttpUrl("http://kiwix.org/")),
+        base_href=None,
+    )
+    assert rewriter.rewrite(css) == css
+
+
+@pytest.mark.parametrize(
+    "css, expected",
+    [
+        (r'.test { content: "\e123 a" }', r'.test { content: "\e123 a" }'),
+        (r'.test { content: "\e123 x" }', r'.test { content: "\e123x" }'),
+        (r'.test { content: "\e123  x" }', r'.test { content: "\e123  x" }'),
+    ],
+)
+def test_escaped_chars_separator(css: str, expected: str):
+    rewriter = CssRewriter(
+        ArticleUrlRewriter(article_url=HttpUrl("http://kiwix.org/")),
+        base_href=None,
+    )
+    assert rewriter.rewrite(css) == expected
+
+
+def test_escaped_chars_and_url_rewriting():
+    rewriter = CssRewriter(
+        ArticleUrlRewriter(article_url=HttpUrl("http://kiwix.org/")),
+        base_href=None,
+    )
+    assert (
+        rewriter.rewrite(
+            r'.a { content: "\e123"; background: url(http://exemple.com/img.png) }'
+        )
+        == r'.a { content: "\e123"; background: url(../exemple.com/img.png) }'
+    )
+
+
+def test_unterminated_string_ending_in_escaped_quote():
+    rewriter = CssRewriter(
+        ArticleUrlRewriter(article_url=HttpUrl("http://kiwix.org/")),
+        base_href=None,
+    )
+    assert (
+        rewriter.rewrite(r'.test { content: "unterminated\" }')
+        == r'.test { content: "unterminated\" }}'
+    )