From 38add74bd6bdcca7f720b669ba8aef33e1a8f5a5 Mon Sep 17 00:00:00 2001
From: Taiki Endo <te316e89@gmail.com>
Date: Mon, 3 Feb 2025 03:04:04 +0900
Subject: [PATCH] Update cspell config & dictionary

---
 .cspell.json                                | 15 ++++----
 .github/.cspell/organization-dictionary.txt | 39 ++------------------
 .github/workflows/ci.yml                    |  8 ++++
 tools/cspell-dict.sh                        | 41 +++++++++++++++++++++
 4 files changed, 61 insertions(+), 42 deletions(-)
 create mode 100755 tools/cspell-dict.sh

diff --git a/.cspell.json b/.cspell.json
index 33996c3..e80166a 100644
--- a/.cspell.json
+++ b/.cspell.json
@@ -27,24 +27,25 @@
   "ignoreRegExpList": [
     // Copyright notice
     "Copyright .*",
+    "SPDX-(File|Snippet)CopyrightText: .*",
     // GHA actions/workflows
-    "uses: .+@",
+    "uses: .+@[\\w_.-]+",
     // GHA context (repo name, owner name, etc.)
-    "github.\\w+ (=|!)= '.+'",
+    "github.[\\w_.-]+ (=|!)= '[^']+'",
     // GH username
     "( |\\[)@[\\w_-]+",
     // Git config username
-    "git config user.name .*",
-    // Username in todo comment
+    "git config( --[^ ]+)? user.name .*",
+    // Username in TODO|FIXME comment
     "(TODO|FIXME)\\([\\w_., -]+\\)",
     // Cargo.toml authors
-    "authors *= *\\[.*\\]",
-    "\".* <[\\w_.+-]+@[\\w.-]+>\""
+    "authors *= *\\[[^\\]]*\\]",
+    "\"[^\"]* <[\\w_.+-]+@[\\w.-]+>\""
   ],
   "languageSettings": [
     {
       "languageId": ["*"],
-      "dictionaries": ["bash", "rust"]
+      "dictionaries": ["bash", "cpp-refined", "rust"]
     }
   ],
   "ignorePaths": []
diff --git a/.github/.cspell/organization-dictionary.txt b/.github/.cspell/organization-dictionary.txt
index 4bb99aa..5c83e4e 100644
--- a/.github/.cspell/organization-dictionary.txt
+++ b/.github/.cspell/organization-dictionary.txt
@@ -8,7 +8,6 @@ asan
 cdylib
 cflags
 clif
-clippy
 codegen
 cranelift
 cxxflags
@@ -19,11 +18,9 @@ miriflags
 msan
 rlib
 rustc
-rustdoc
 rustdocflags
 rustflags
 rustsec
-rustup
 staticlib
 tsan
 xcompile
@@ -31,7 +28,6 @@ Zmiri
 
 // Rust target triple/spec
 aarch
-amdgpu
 androideabi
 armeb
 armebv
@@ -40,10 +36,8 @@ atmega
 bpfeb
 bpfel
 csky
-cuda
 eabi
 eabihf
-emscripten
 espidf
 fortanix
 gnuabi
@@ -62,7 +56,6 @@ loongarch
 macabi
 mipsel
 mipsisa
-msvc
 muslabi
 musleabi
 musleabihf
@@ -104,83 +97,62 @@ bufs
 bytecount
 cfgs
 ctypes
-dealloc
-deque
 docsrs
-doctest
 doctests
+elems
 idents
 ilog
 impls
 incompat
 inlateout
 lateout
-libdir
 mclass
-memcpy
 nand
 nanos
 nocapture
 nomem
 nonminimal
 nonoverlapping
-noreturn
 nostack
 peekable
-powf
-punct
 rchunks
 rclass
 repr
-rfind
 rfold
 rposition
 rsplit
 rustlib
 seqcst
-simd
 splitn
-structs
 subsec
 supertrait
 supertraits
-sysroot
 turbofish
 uninit
 unpark
+unparse
 upcastable
 
 // Other
 armel
 armhf
-backports
 binutils
 choco
-chsh
 connrefused
 devel
+dockerfiles
 endgroup
-endianness
-euxo
 exitcode
-ggrep
-gsed
 gsub
 gtar
 HEALTHCHECK
-libc
-markdownlint
-mmap
+libunwind
 moreutils
-msys
 noconfirm
 noninteractive
 noprofile
 norc
-nproc
 nullary
-objcopy
-objdump
 pacman
 pkgin
 powerset
@@ -189,15 +161,12 @@ readelf
 requirechecksums
 sched
 semihosting
-shellcheckrc
 SIGABRT
-SIGBUS
 SIGILL
 STOPSIGNAL
 subcmd
 taplo
 tlsv
-tmpdir
 tomlq
 valgrind
 venv
diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index b651bc0..1dde010 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -119,3 +119,11 @@ jobs:
     steps:
       - uses: taiki-e/checkout-action@v1
       - uses: ./setup-docker
+
+  cspell-dict:
+    runs-on: ubuntu-latest
+    timeout-minutes: 60
+    steps:
+      - uses: taiki-e/checkout-action@v1
+      - run: tools/cspell-dict.sh clean
+      - run: git add -N . && git diff --exit-code
diff --git a/tools/cspell-dict.sh b/tools/cspell-dict.sh
new file mode 100755
index 0000000..d184421
--- /dev/null
+++ b/tools/cspell-dict.sh
@@ -0,0 +1,41 @@
+#!/usr/bin/env bash
+# SPDX-License-Identifier: Apache-2.0 OR MIT
+# shellcheck disable=SC2013
+set -CeEuo pipefail
+IFS=$'\n\t'
+trap -- 's=$?; printf >&2 "%s\n" "${0##*/}:${LINENO}: \`${BASH_COMMAND}\` exit with ${s}"; exit ${s}' ERR
+
+bail() {
+  printf >&2 'error: %s\n' "$*"
+  exit 1
+}
+
+case "$1" in
+  clean)
+    cd -- "$(dirname -- "$0")"/..
+    dictionary=.github/.cspell/organization-dictionary.txt
+    grep_args=()
+    for word in $(grep -Ev '^//' "${dictionary}" || true); do
+      [[ -n "${word}" ]] || continue
+      if npx -y cspell trace "${word}" | grep -Fv "${dictionary}" | grep -Eq "^${word} \\* [^ ]+\*"; then
+        grep_args+=(-e "^${word}$")
+      fi
+    done
+    if [[ ${#grep_args[@]} -gt 0 ]]; then
+      printf 'info: %s\n' "removing needless words from ${dictionary}"
+      res=$(grep -Ev "${grep_args[@]}" "${dictionary}")
+      printf '%s\n' "${res}" >|"${dictionary}"
+    fi
+    ;;
+  trace)
+    for dictionary in .github/.cspell/*; do
+      if [[ -f "${dictionary}" ]]; then
+        for word in $(grep -Ev '^//' "${dictionary}" || true); do
+          [[ -n "${word}" ]] || continue
+          npx -y cspell trace "${word}" | grep -E "^${word} \\*" | { grep -Fv "${dictionary}" || true; }
+        done
+      fi
+    done
+    ;;
+  *) bail "unrecognized subcommand '$1'" ;;
+esac