Move _file_checksum() to sphinx.builders.html._assets

Author: AA-Turner

sphinx/builders/html/__init__.py

@@ -9,7 +9,6 @@
 import re
 import sys
 import warnings
-import zlib
 from datetime import datetime, timezone
 from os import path
 from typing import IO, TYPE_CHECKING, Any
@@ -27,6 +26,7 @@
 from sphinx import version_info as sphinx_version
 from sphinx.application import Sphinx
 from sphinx.builders import Builder
+from sphinx.builders.html._assets import _file_checksum
 from sphinx.config import ENUM, Config
 from sphinx.domains import Domain, Index, IndexEntry
 from sphinx.environment import BuildEnvironment
@@ -1254,29 +1254,6 @@ def js_tag(js: JavaScript) -> str:
     context['js_tag'] = js_tag
 
 
-def _file_checksum(outdir: str | os.PathLike[str], filename: str | os.PathLike[str]) -> str:
-    filename = os.fspath(filename)
-    # Don't generate checksums for HTTP URIs
-    if '://' in filename:
-        return ''
-    # Some themes and extensions have used query strings
-    # for a similar asset checksum feature.
-    # As we cannot safely strip the query string,
-    # raise an error to the user.
-    if '?' in filename:
-        msg = f'Local asset file paths must not contain query strings: {filename!r}'
-        raise ThemeError(msg)
-    try:
-        # Ensure universal newline mode is used to avoid checksum differences
-        with open(path.join(outdir, filename), encoding='utf-8') as f:
-            content = f.read().encode(encoding='utf-8')
-    except FileNotFoundError:
-        return ''
-    if not content:
-        return ''
-    return f'{zlib.crc32(content):08x}'
-
-
 def setup_resource_paths(app: Sphinx, pagename: str, templatename: str,
                          context: dict, doctree: Node) -> None:
     """Set up relative resource paths."""

sphinx/builders/html/_assets.py

@@ -0,0 +1,32 @@
+from __future__ import annotations
+
+import os
+import zlib
+from typing import TYPE_CHECKING
+
+from sphinx.errors import ThemeError
+
+if TYPE_CHECKING:
+    from pathlib import Path
+
+
+def _file_checksum(outdir: Path, filename: str | os.PathLike[str]) -> str:
+    filename = os.fspath(filename)
+    # Don't generate checksums for HTTP URIs
+    if '://' in filename:
+        return ''
+    # Some themes and extensions have used query strings
+    # for a similar asset checksum feature.
+    # As we cannot safely strip the query string,
+    # raise an error to the user.
+    if '?' in filename:
+        msg = f'Local asset file paths must not contain query strings: {filename!r}'
+        raise ThemeError(msg)
+    try:
+        # Remove all carriage returns to avoid checksum differences
+        content = outdir.joinpath(filename).read_bytes().translate(None, b'\r')
+    except FileNotFoundError:
+        return ''
+    if not content:
+        return ''
+    return f'{zlib.crc32(content):08x}'

tests/test_build_html.py

@@ -12,10 +12,10 @@
 
 import sphinx.builders.html
 from sphinx.builders.html import (
-    _file_checksum,
     validate_html_extra_path,
     validate_html_static_path,
 )
+from sphinx.builders.html._assets import _file_checksum
 from sphinx.errors import ConfigError, ThemeError
 from sphinx.testing.util import strip_escseq
 from sphinx.util.inventory import InventoryFile
@@ -1248,10 +1248,10 @@ def test_file_checksum(app):
 
 def test_file_checksum_query_string():
     with pytest.raises(ThemeError, match='Local asset file paths must not contain query strings'):
-        _file_checksum('', 'with_query_string.css?dead_parrots=1')
+        _file_checksum(Path(), 'with_query_string.css?dead_parrots=1')
 
     with pytest.raises(ThemeError, match='Local asset file paths must not contain query strings'):
-        _file_checksum('', 'with_query_string.js?dead_parrots=1')
+        _file_checksum(Path(), 'with_query_string.js?dead_parrots=1')
 
     with pytest.raises(ThemeError, match='Local asset file paths must not contain query strings'):
         _file_checksum(Path.cwd(), '_static/with_query_string.css?dead_parrots=1')