Fix Vale by passing files to process instead of a glob to ignore (#3871)

* Fix Vale by passing files to process instead of a glob to ignore

* Update .vale/justfile

* ignore node_modules

* Fix term casing context; update justfile recipe; fix Dockerfile order

.vale/Dockerfile

@@ -2,8 +2,9 @@ FROM jdkato/vale:v2.30.0
 
 WORKDIR /vale
 COPY .vale.ini .
-COPY styles styles
 
 RUN vale sync
 
+COPY styles styles
+
 ENTRYPOINT ["vale", "--config=/vale/.vale.ini"]

.vale/justfile

@@ -1,64 +1,41 @@
 COLOR := "\\033[0;34m"
 NO_COLOR := "\\033[0m"
 
-
 # Show all available recipes
 @_default:
     printf "\n{{ COLOR }}# Vale (path: \`.vale/\`)\n"
     printf "===================={{ NO_COLOR }}\n"
     just --list --unsorted
 
-
 # Build a local version of `openverse-vale:local` for testing
 build:
     docker build . -t openverse-vale:local
 
-
-# The --glob setting excludes things that should not be linted, including
-# build artefacts, dependencies, and automatically generated files like
-# the changelogs (without excluding `changelogs/index.md`). Project proposals
-# are also excluded for want of a good way to ignore existing proposals and only
-# lint _new_ proposals. Project proposals aren't "living documents" in the way
-# the rest of our documentation is, so it doesn't seem right to retroactively
-# edit them for mere editorial purposes (rather than, for example, to correct
-# some grave inacurracy).
-# Leading `,` compensates for lack of trailing comma support
-# Note the lack of space before the trailing \ on each line, this is to prevent
-# the addition of a space between each pattern, which isn't supported by Vale's glob
-# If you change this and Vale takes a very long time to run (more than 30 seconds)
-# then chances are the change is breaking the glob pattern. Unfortunately the only
-# feedback you get when the glob pattern is not working as intended is a very long
-# run time for Vale as it checks everything that should have otherwise been ignored,
-# but wasn't due to some minor issue in the pattern syntax.
-# This is fiddly, but I can't find a good way around it.
-_IGNORE_PATTERNS := """
-    _build\
-    ,_static\
-    ,venv\
-    ,.venv\
-    ,.nuxt\
-    ,.pnpm-store\
-    ,node_modules\
-    ,test-results\
-    ,storebook-static\
-    ,.ruff-cache\
-    ,projects/proposals\
-    ,changelogs/api\
-    ,changelogs/frontend\
-    ,changelogs/catalog\
-    ,changelogs/ingestion_server\
-"""
-
-VALE_GLOB := "--glob='!*{" + _IGNORE_PATTERNS + "}*'"
-
+IGNORE_FILES_PATTERN := ".pnpm|changelogs|projects/proposals|node_modules|.?venv"
+
+# This recipe generates a list of Markdown and MDX files to be processed, excluding those
+# matched by the IGNORE_FILES_PATTERN. Files are listed relative to the parent directory
+# of the current working directory. If a custom separator is specified, newlines are
+# replaced with this separator before outputting the file list.
+@files separator="\n":
+    #! /usr/bin/env sh
+    files=$(
+        find "$PWD/.." -type f \( -name "*.md" -o -name "*.mdx" \) \
+            | grep -vE "{{ IGNORE_FILES_PATTERN }}" \
+            | sed "s|$PWD/../||"
+    )
+    if [ '{{ separator }}' != '\n' ]; then
+        echo "$files" | tr '\n' '{{ separator }}'
+        exit 0
+    fi
+    echo "$files"
 
 # Run Vale configured for Openverse in Docker.
 # Using Docker avoids the need for contributors to install the Vale binary.
 #Configuration defaults to what is used for CI.
-run +files=".": build
+run: build
     docker run --rm \
         -v $PWD/..:/src:rw,Z \
         --workdir=/src \
         openverse-vale:local \
-        {{ files }} \
-        {{ VALE_GLOB }}
+        {{ "`just files`" }}

.vale/styles/Openverse/TermCasing.yml

@@ -5,7 +5,7 @@ message: "Incorrect casing. Use '%s' instead of '%s'."
 level: error
 ignorecase: false
 scope:
-  - paragraph
+  - text
 swap:
   # [^/\.] prevents matching things that look like URLs, file paths, or GitHub team mentions
   # For example: @WordPress/openverse-maintainers