From 2062a11e7aae29aeef3e4ef27742bb4ec0ad02a3 Mon Sep 17 00:00:00 2001
From: Oleh Prypin <oleh@pryp.in>
Date: Sun, 18 Jun 2023 23:24:01 +0200
Subject: [PATCH] Implement cache busting: append a suffix to CSS and JS URLs

---
 docs/user-guide/configuration.md           | 44 +++++++++++
 mkdocs/commands/build.py                   |  6 +-
 mkdocs/config/config_options.py            |  2 +-
 mkdocs/config/defaults.py                  |  5 ++
 mkdocs/structure/files.py                  | 85 +++++++++++++++++++---
 mkdocs/tests/build_tests.py                |  2 +-
 mkdocs/theme.py                            | 25 +++++--
 mkdocs/themes/mkdocs/mkdocs_theme.yml      |  6 ++
 mkdocs/themes/readthedocs/mkdocs_theme.yml |  4 +
 mkdocs/utils/__init__.py                   |  9 ++-
 mkdocs/utils/templates.py                  |  7 +-
 11 files changed, 171 insertions(+), 24 deletions(-)

diff --git a/docs/user-guide/configuration.md b/docs/user-guide/configuration.md
index 401caa4a..75481097 100644
--- a/docs/user-guide/configuration.md
+++ b/docs/user-guide/configuration.md
@@ -497,6 +497,50 @@ Only the plain string variant detects the `.mjs` extension and adds `type="modul
 
 NOTE: `*.js` and `*.css` files, just like any other type of file, are always copied from `docs_dir` into the site's deployed copy, regardless if they're linked to the pages via the above configs or not.
 
+### hash_rename_assets
+
+NEW: **New in version 1.5.**
+
+Set patterns of files to rename (on the fly, upon copying to the built site) by inserting a hash of the content. This is done for purposes of *cache busting*.
+
+It is recommended to use this setting whenever possible.
+
+The patterns follow the [.gitignore pattern format](https://git-scm.com/docs/gitignore#_pattern_format). But in this case these are positive matches, not "ignore" matches.
+
+For example:
+
+```yaml
+extra_javascript:
+    - js/foo.js
+    - vendored/jquery-1.2.3.js
+hash_rename_assets: |
+    *.css
+    *.js
+    !/vendored/*.js
+```
+
+Then the matched file is copied with a modified name and references to it are modified accordingly, e.g.:
+
+```html
+<script src="../js/foo.e3b0c442.js">
+```
+
+but one doesn't need to remember to update this hash, you can just keep modifying `foo.js` normally and referring to it normally.
+
+Note how in this example we chose not to hash the file that already has its own version. But we could rename it, too.
+
+### hash_append_assets
+
+NEW: **New in version 1.5.**
+
+Same as [hash_rename_assets](#hash_rename_assets) but the file doesn't get renamed, instead whenever it is referred to, a URL parameter is appended to it.
+
+E.g. a script might get linked to as:
+
+```html
+<script src="../js/foo.js?h=e3b0c442">
+```
+
 ### extra_templates
 
 Set a list of templates in your `docs_dir` to be built by MkDocs. To see more
diff --git a/mkdocs/commands/build.py b/mkdocs/commands/build.py
index efe210a7..9f43da88 100644
--- a/mkdocs/commands/build.py
+++ b/mkdocs/commands/build.py
@@ -42,9 +42,10 @@ def get_context(
         base_url = utils.get_relative_url('.', page.url)
 
     extra_javascript = [
-        utils.normalize_url(str(script), page, base_url) for script in config.extra_javascript
+        utils.normalize_url(str(script), page, base_url, files)
+        for script in config.extra_javascript
     ]
-    extra_css = [utils.normalize_url(path, page, base_url) for path in config.extra_css]
+    extra_css = [utils.normalize_url(path, page, base_url, files) for path in config.extra_css]
 
     return templates.TemplateContext(
         nav=nav,
@@ -56,6 +57,7 @@ def get_context(
         build_date_utc=utils.get_build_datetime(),
         config=config,
         page=page,
+        _files=files,
     )
 
 
diff --git a/mkdocs/config/config_options.py b/mkdocs/config/config_options.py
index 73ff8f2e..b7b9b7df 100644
--- a/mkdocs/config/config_options.py
+++ b/mkdocs/config/config_options.py
@@ -1164,6 +1164,6 @@ class PathSpec(BaseConfigOption[pathspec.gitignore.GitIgnoreSpec]):
         if not isinstance(value, str):
             raise ValidationError(f'Expected a multiline string, but a {type(value)} was given.')
         try:
-            return pathspec.gitignore.GitIgnoreSpec.from_lines(lines=value.splitlines())
+            return pathspec.gitignore.GitIgnoreSpec.from_lines(value.splitlines())
         except ValueError as e:
             raise ValidationError(str(e))
diff --git a/mkdocs/config/defaults.py b/mkdocs/config/defaults.py
index a75ea8f3..9f5ca109 100644
--- a/mkdocs/config/defaults.py
+++ b/mkdocs/config/defaults.py
@@ -96,6 +96,11 @@ class MkDocsConfig(base.Config):
     """Specify which css or javascript files from the docs directory should be
     additionally included in the site."""
 
+    hash_rename_assets = c.Optional(c.PathSpec())
+    hash_append_assets = c.Optional(c.PathSpec())
+    """Specify which css or javascript files from the docs directory should be
+    renamed to contain a hash suffix, for cache busting."""
+
     extra_templates = c.Type(list, default=[])
     """Similar to the above, but each template (HTML or XML) will be build with
     Jinja2 and the global context."""
diff --git a/mkdocs/structure/files.py b/mkdocs/structure/files.py
index bf9bde00..3cde8cb8 100644
--- a/mkdocs/structure/files.py
+++ b/mkdocs/structure/files.py
@@ -2,6 +2,7 @@ from __future__ import annotations
 
 import enum
 import fnmatch
+import hashlib
 import logging
 import os
 import posixpath
@@ -21,6 +22,7 @@ from mkdocs import utils
 if TYPE_CHECKING:
     from mkdocs.config.defaults import MkDocsConfig
     from mkdocs.structure.pages import Page
+    from mkdocs.theme import Theme
 
 
 log = logging.getLogger(__name__)
@@ -52,6 +54,15 @@ class InclusionLevel(enum.Enum):
         return self.value <= self.NOT_IN_NAV.value
 
 
+class AssetVersioning(enum.Enum):
+    NONE = 'NONE'
+    """The asset file is copied as is."""
+    HASH_RENAME = 'HASH_RENAME'
+    """The file (such as 'main.js') gets renamed to e.g. 'main.e3b0c442.js'."""
+    HASH_SUFFIX = 'HASH_SUFFIX'
+    """The file (such as 'main.js') always gets a hash appended when linking to it, e.g. 'main.js?h=e3b0c442'."""
+
+
 class Files:
     """A collection of [File][mkdocs.structure.files.File] objects."""
 
@@ -164,7 +175,15 @@ class Files:
                 for dir in config.theme.dirs:
                     # Find the first theme dir which contains path
                     if os.path.isfile(os.path.join(dir, path)):
-                        self.append(File(path, dir, config.site_dir, config.use_directory_urls))
+                        self.append(
+                            File(
+                                path,
+                                dir,
+                                config.site_dir,
+                                config.use_directory_urls,
+                                asset_versioning=_asset_versioning(path, config.theme),
+                            )
+                        )
                         break
 
 
@@ -233,15 +252,17 @@ class File:
         *,
         dest_uri: str | None = None,
         inclusion: InclusionLevel = InclusionLevel.UNDEFINED,
+        asset_versioning: AssetVersioning = AssetVersioning.NONE,
     ) -> None:
         self.page = None
         self.src_path = path
         self.name = self._get_stem()
+        self.abs_src_path = os.path.normpath(os.path.join(src_dir, self.src_uri))
+        self.asset_versioning = asset_versioning
         if dest_uri is None:
             dest_uri = self._get_dest_path(use_directory_urls)
         self.dest_uri = dest_uri
         self.url = self._get_url(use_directory_urls)
-        self.abs_src_path = os.path.normpath(os.path.join(src_dir, self.src_uri))
         self.abs_dest_path = os.path.normpath(os.path.join(dest_dir, self.dest_uri))
         self.inclusion = inclusion
 
@@ -250,7 +271,7 @@ class File:
             isinstance(other, self.__class__)
             and self.src_uri == other.src_uri
             and self.abs_src_path == other.abs_src_path
-            and self.url == other.url
+            and self.dest_uri == other.dest_uri
         )
 
     def __repr__(self):
@@ -276,15 +297,31 @@ class File:
             else:
                 # foo.md => foo/index.html
                 return posixpath.join(parent, self.name, 'index.html')
+
+        if self.asset_versioning is AssetVersioning.HASH_RENAME:
+            try:
+                suf = _hash_suffix(self.abs_src_path)
+            except FileNotFoundError:
+                pass
+            else:
+                name, ext = posixpath.splitext(self.src_uri)
+                return f'{name}.{suf}{ext}'
+
         return self.src_uri
 
     def _get_url(self, use_directory_urls: bool) -> str:
-        """Return url based in destination path."""
+        """Return url based on destination path."""
         url = self.dest_uri
-        dirname, filename = posixpath.split(url)
-        if use_directory_urls and filename == 'index.html':
-            url = (dirname or '.') + '/'
-        return urlquote(url)
+        if use_directory_urls:
+            if url == 'index.html' or url.endswith('/index.html'):
+                url = (posixpath.dirname(url) or '.') + '/'
+        url = urlquote(url)
+        if self.asset_versioning is AssetVersioning.HASH_SUFFIX:
+            try:
+                url += '?h=' + _hash_suffix(self.abs_src_path)
+            except FileNotFoundError:
+                pass
+        return url
 
     def url_relative_to(self, other: File | str) -> str:
         """Return url for file relative to other file."""
@@ -327,6 +364,30 @@ class File:
         return self.src_uri.endswith('.css')
 
 
+def _asset_versioning(
+    src_uri: str,
+    config: MkDocsConfig | Theme,
+) -> AssetVersioning:
+    hash_rename_assets = getattr(config, 'hash_rename_assets', None)
+    hash_append_assets = getattr(config, 'hash_append_assets', None)
+    if hash_rename_assets and hash_rename_assets.match_file(src_uri):
+        return AssetVersioning.HASH_RENAME
+    elif hash_append_assets and hash_append_assets.match_file(src_uri):
+        return AssetVersioning.HASH_SUFFIX
+    return AssetVersioning.NONE
+
+
+def _hash_suffix(abs_src_path: str) -> str:
+    digest = hashlib.sha256()
+    with open(abs_src_path, 'rb') as f:
+        while True:
+            data = f.read(65536)
+            if not data:
+                break
+            digest.update(data)
+    return digest.hexdigest()[:8]
+
+
 _default_exclude = pathspec.gitignore.GitIgnoreSpec.from_lines(['.*', '/templates/'])
 
 
@@ -346,22 +407,24 @@ def _set_exclusions(files: Iterable[File], config: MkDocsConfig | Mapping[str, A
                 file.inclusion = InclusionLevel.INCLUDED
 
 
-def get_files(config: MkDocsConfig | Mapping[str, Any]) -> Files:
+def get_files(config: MkDocsConfig) -> Files:
     """Walk the `docs_dir` and return a Files collection."""
     files: list[File] = []
     conflicting_files: list[tuple[File, File]] = []
     for source_dir, dirnames, filenames in os.walk(config['docs_dir'], followlinks=True):
-        relative_dir = os.path.relpath(source_dir, config['docs_dir'])
+        relative_dir = PurePath(os.path.relpath(source_dir, config['docs_dir'])).as_posix()
         dirnames.sort()
         filenames.sort(key=_file_sort_key)
 
         files_by_dest: dict[str, File] = {}
         for filename in filenames:
+            src_uri = posixpath.join(relative_dir, filename)
             file = File(
-                os.path.join(relative_dir, filename),
+                src_uri,
                 config['docs_dir'],
                 config['site_dir'],
                 config['use_directory_urls'],
+                asset_versioning=_asset_versioning(src_uri, config),
             )
             # Skip README.md if an index file also exists in dir (part 1)
             prev_file = files_by_dest.setdefault(file.dest_uri, file)
diff --git a/mkdocs/tests/build_tests.py b/mkdocs/tests/build_tests.py
index bc161fce..b59b1d2f 100644
--- a/mkdocs/tests/build_tests.py
+++ b/mkdocs/tests/build_tests.py
@@ -220,7 +220,7 @@ class BuildTests(PathAssertionMixin, unittest.TestCase):
 
     def test_context_extra_css_js_no_page(self):
         cfg = load_config(extra_css=['style.css'], extra_javascript=['script.js'])
-        context = build.get_context(mock.Mock(), mock.Mock(), cfg, base_url='..')
+        context = build.get_context(mock.Mock(), Files([]), cfg, base_url='..')
         self.assertEqual(context['extra_css'], ['../style.css'])
         self.assertEqual(context['extra_javascript'], ['../script.js'])
 
diff --git a/mkdocs/theme.py b/mkdocs/theme.py
index 467d26ea..2d3990f4 100644
--- a/mkdocs/theme.py
+++ b/mkdocs/theme.py
@@ -5,6 +5,7 @@ import os
 from typing import Any, Collection, MutableMapping
 
 import jinja2
+import pathspec.gitignore
 
 from mkdocs import localization, utils
 from mkdocs.config.base import ValidationError
@@ -66,6 +67,15 @@ class Theme(MutableMapping[str, Any]):
             locale if locale is not None else _vars['locale']
         )
 
+        if 'hash_rename_assets' in _vars:
+            self.hash_rename_assets = pathspec.gitignore.GitIgnoreSpec.from_lines(
+                _vars.pop('hash_rename_assets').splitlines()
+            )
+        if 'hash_append_assets' in _vars:
+            self.hash_append_assets = pathspec.gitignore.GitIgnoreSpec.from_lines(
+                _vars.pop('hash_append_assets').splitlines()
+            )
+
     name: str | None
 
     @property
@@ -80,13 +90,16 @@ class Theme(MutableMapping[str, Any]):
 
     static_templates: set[str]
 
+    hash_rename_assets: pathspec.gitignore.GitIgnoreSpec | None = None
+    hash_append_assets: pathspec.gitignore.GitIgnoreSpec | None = None
+
     def __repr__(self) -> str:
-        return "{}(name={!r}, dirs={!r}, static_templates={!r}, {})".format(
-            self.__class__.__name__,
-            self.name,
-            self.dirs,
-            self.static_templates,
-            ', '.join(f'{k}={v!r}' for k, v in self.items()),
+        return (
+            f'{self.__class__.__name__}('
+            f'name={self.name!r}, dirs={self.dirs!r}, static_templates={self.static_templates!r}, '
+            f'hash_rename_assets={self.hash_rename_assets!r}, hash_append_assets={self.hash_append_assets!r}, '
+            + ', '.join(f'{k}={v!r}' for k, v in self.items())
+            + ')'
         )
 
     def __getitem__(self, key: str) -> Any:
diff --git a/mkdocs/themes/mkdocs/mkdocs_theme.yml b/mkdocs/themes/mkdocs/mkdocs_theme.yml
index 2be349c2..eb942b02 100644
--- a/mkdocs/themes/mkdocs/mkdocs_theme.yml
+++ b/mkdocs/themes/mkdocs/mkdocs_theme.yml
@@ -3,6 +3,12 @@
 static_templates:
     - 404.html
 
+hash_rename_assets: |
+    *.js
+    *.css
+    !*-*
+    !*worker*
+
 locale: en
 
 include_search_page: false
diff --git a/mkdocs/themes/readthedocs/mkdocs_theme.yml b/mkdocs/themes/readthedocs/mkdocs_theme.yml
index d9553ae8..9c943119 100644
--- a/mkdocs/themes/readthedocs/mkdocs_theme.yml
+++ b/mkdocs/themes/readthedocs/mkdocs_theme.yml
@@ -3,6 +3,10 @@
 static_templates:
     - 404.html
 
+hash_rename_assets: |
+    theme*.js
+    theme*.css
+
 locale: en
 
 include_search_page: true
diff --git a/mkdocs/utils/__init__.py b/mkdocs/utils/__init__.py
index 23ec3a0b..a7130ee7 100644
--- a/mkdocs/utils/__init__.py
+++ b/mkdocs/utils/__init__.py
@@ -32,6 +32,7 @@ from yaml_env_tag import construct_env_tag
 from mkdocs import exceptions
 
 if TYPE_CHECKING:
+    from mkdocs.structure.files import Files
     from mkdocs.structure.pages import Page
 
 T = TypeVar('T')
@@ -281,11 +282,17 @@ def get_relative_url(url: str, other: str) -> str:
     return relurl + '/' if url.endswith('/') else relurl
 
 
-def normalize_url(path: str, page: Page | None = None, base: str = '') -> str:
+def normalize_url(
+    path: str, page: Page | None = None, base: str = '', files: Files | None = None
+) -> str:
     """Return a URL relative to the given page or using the base."""
     path, relative_level = _get_norm_url(path)
     if relative_level == -1:
         return path
+    if files is not None:
+        file = files.get_file_from_path(path)
+        if file is not None:
+            path = file.url
     if page is not None:
         result = get_relative_url(path, page.url)
         if relative_level > 0:
diff --git a/mkdocs/utils/templates.py b/mkdocs/utils/templates.py
index 9c7ccdf8..eefc6fc1 100644
--- a/mkdocs/utils/templates.py
+++ b/mkdocs/utils/templates.py
@@ -23,7 +23,7 @@ from mkdocs.utils import normalize_url
 if TYPE_CHECKING:
     from mkdocs.config.config_options import ExtraScriptValue
     from mkdocs.config.defaults import MkDocsConfig
-    from mkdocs.structure.files import File
+    from mkdocs.structure.files import File, Files
     from mkdocs.structure.nav import Navigation
     from mkdocs.structure.pages import Page
 
@@ -38,12 +38,15 @@ class TemplateContext(TypedDict):
     build_date_utc: datetime.datetime
     config: MkDocsConfig
     page: Page | None
+    _files: Files
 
 
 @contextfilter
 def url_filter(context: TemplateContext, value: str) -> str:
     """A Template filter to normalize URLs."""
-    return normalize_url(str(value), page=context['page'], base=context['base_url'])
+    return normalize_url(
+        value, page=context['page'], base=context['base_url'], files=context.get('_files')
+    )
 
 
 @contextfilter