chore(linux): own ignore, editorconfig and generated-file attributes

.gitattributes

@@ -43,7 +43,21 @@
 *.yaml text eol=lf
 *.plist text eol=lf
 
-# Vendored C headers (database driver bridges, tree-sitter grammars)
+# Vendored C headers (database driver bridges, SQLite, tree-sitter grammars)
 Plugins/*/C*/include/** linguist-vendored
+Packages/TableProCore/Sources/CSQLite/include/sqlite3*.h linguist-vendored
 TablePro/Core/SSH/CLibSSH2/** linguist-vendored
 Packages/TableProGrammars/Sources/TreeSitterGrammars/** linguist-vendored
+
+# Third-party license texts, copied verbatim
+Native/DamengBridge/LICENSES/** linguist-vendored
+Packages/TableProEditor/LICENSE-*.md linguist-vendored
+TablePro/Resources/ThirdPartyLicenses/texts/** linguist-vendored
+# Joins the SQLite copyright page and the source blessing with a sentence of its own.
+TablePro/Resources/ThirdPartyLicenses/texts/sqlite.txt -linguist-vendored
+
+# Generated Linux files (Cargo, xgettext, cargo-about, license_metadata.py)
+linux/Cargo.lock linguist-generated
+linux/po/tablepro.pot linguist-generated
+linux/data/licenses/** linguist-generated
+linux/build-aux/debian/copyright linguist-generated

.gitignore

@@ -173,9 +173,6 @@ Native/HanaBridge/bin
 .docs/
 /plans/reports
 
-# Linux meson build directory
-linux/_build/
-
 # Translation exports from `xcodebuild -exportLocalizations`; the catalog is the source of truth.
 # Anchored: unanchored, it also matched TableProTests/Localization/, so the string-catalog
 # guard suite existed on disk and was invisible to git and to CI.

linux/.editorconfig

@@ -0,0 +1,25 @@
+# No root = true: the repository's .editorconfig sets the defaults, and this file overrides them.
+
+[{meson.build,meson.options,meson_options.txt}]
+indent_size = 2
+
+[*.{ui,xml,xml.in,xml.in.in,svg}]
+indent_size = 2
+
+[*.css]
+indent_size = 2
+
+[*.hbs]
+indent_size = 2
+
+[build-aux/flatpak/*.json]
+indent_size = 4
+
+[*.rs]
+max_line_length = 120
+
+[build-aux/debian/rules]
+indent_style = tab
+
+[build-aux/debian/{control,copyright}]
+indent_size = 1

linux/.gitignore

@@ -1,6 +1,5 @@
 target/
+_build/
 _packages/
-*.swp
-.DS_Store
-
-docs/audit-*.md
+_flatpak/
+.flatpak-builder/

linux/build-aux/packaging/add-rust-toolchain-ppa.sh

@@ -20,6 +20,7 @@ fi
 # The key is committed, so no run trusts a key it fetched. apt checks
 # InRelease against it, which makes plain HTTP as safe as the archive's
 # own sources, and the base image has no CA bundle for HTTPS.
+# shellcheck source=/dev/null
 . /etc/os-release
 mkdir -p /etc/apt/keyrings
 cp "$here/rust-toolchain-staging.asc" /etc/apt/keyrings/rust-toolchain-staging.asc

linux/build-aux/packaging/build-deb.sh

@@ -34,6 +34,7 @@ build() {
     # login that is not root, 1024 open files among them. Launchpad passes
     # the builder's CPU count the same way as here.
     status=0
+    # shellcheck disable=SC2016 # the inner sh expands $1
     runuser -u builder -- env HOME=/home/builder DEB_BUILD_OPTIONS="parallel=$(nproc)" \
         sh -euc 'cd /build/work && dpkg-source -x "$1" src && cd src && dpkg-buildpackage -b -us -uc' sh "$1" ||
         status=$?

scripts/check-banned-words.sh

@@ -10,15 +10,14 @@
 # A term matches case-insensitively at the start of a word, so a capitalized or suffixed form is a
 # hit and a word that only ends in a term is not. A term that starts with punctuation, the em dash,
 # matches anywhere. Each hit prints as file:line: term. Exits 1 on a hit, 0 when clean and 2 on a
-# usage error. A diff skips linux/data/licenses/ and linux/build-aux/debian/copyright, which reproduce
-# third-party license text verbatim.
+# usage error. A diff skips a file .gitattributes marks linguist-generated or linguist-vendored,
+# since a generator or an upstream project wrote its text.
 
 set -uo pipefail
 
 SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd -P)"
 LIST="$SCRIPT_DIR/banned-words.txt"
 LIST_PATH_IN_REPO="scripts/banned-words.txt"
-VERBATIM_IN_REPO="linux/data/licenses/ linux/build-aux/debian/copyright"
 
 usage() {
     sed -n "5,8s/^#   //p" "${BASH_SOURCE[0]}"
@@ -72,23 +71,19 @@ function check(text, location,    lowered, i) {
 '
 
 DIFF_READER='
-BEGIN { verbatim_count = split(verbatim, verbatim_paths, " ") }
-
-function is_verbatim(path,    i) {
-    for (i = 1; i <= verbatim_count; i++) {
-        if (index(path, verbatim_paths[i]) == 1) {
-            return 1
-        }
+BEGIN {
+    while ((getline path < exempt_list) > 0) {
+        exempt[path] = 1
     }
-    return 0
+    close(exempt_list)
 }
 
 /^diff --git / { in_header = 1; file = ""; next }
 in_header && /^\+\+\+ / {
     file = substr($0, 5)
     sub(/^b\//, "", file)
     sub(/\t$/, "", file)
-    if (file == "/dev/null" || file == skip || is_verbatim(file)) {
+    if (file == "/dev/null" || file == skip || file in exempt) {
         file = ""
     }
     next
@@ -117,13 +112,22 @@ END { exit(hits > 0 ? 1 : 0) }
 '
 
 check_diff() {
-    git rev-parse --show-toplevel > /dev/null 2>&1 || {
+    local top
+    top="$(git rev-parse --show-toplevel 2> /dev/null)" || {
         echo "check-banned-words.sh: not inside a git repository" >&2
         exit 2
     }
+    EXEMPT_LIST="$(mktemp)" || exit 2
+    trap 'rm -f "$EXEMPT_LIST"' EXIT
+    # A renamed file still lists under its new path, and the diff below already prints any rename
+    # limit warning. The index holds the staged .gitattributes, and HEAD's when nothing is staged.
+    git -c core.quotePath=false diff --name-only --no-renames --diff-filter=ACMR "$@" |
+        git -C "$top" -c core.quotePath=false check-attr --cached --stdin \
+            linguist-generated linguist-vendored |
+        sed -n -E 's/^(.*): linguist-(generated|vendored): (set|true)$/\1/p' > "$EXEMPT_LIST" || exit 2
     git -c core.quotePath=false diff -U0 --no-color --no-ext-diff --diff-filter=ACMR \
         --src-prefix=a/ --dst-prefix=b/ "$@" |
-        LC_ALL=C awk -v skip="$LIST_PATH_IN_REPO" -v verbatim="$VERBATIM_IN_REPO" "$MATCHER$DIFF_READER"
+        LC_ALL=C awk -v skip="$LIST_PATH_IN_REPO" -v exempt_list="$EXEMPT_LIST" "$MATCHER$DIFF_READER"
 }
 
 check_files() {

scripts/ci/test_check_banned_words.py

@@ -0,0 +1,92 @@
+#!/usr/bin/env python3
+"""Tests check-banned-words.sh against throwaway git repositories."""
+
+import os
+import shutil
+import subprocess
+import tempfile
+import unittest
+from pathlib import Path
+
+SCRIPT = Path(__file__).resolve().parent.parent / "check-banned-words.sh"
+TERM = "frobnicate"
+SEED_TEXT = "".join(f"seed line {number}\n" for number in range(20))
+
+
+class CheckBannedWordsTests(unittest.TestCase):
+    def setUp(self):
+        self.temp = tempfile.TemporaryDirectory()
+        self.repo = Path(self.temp.name)
+        self.env = {key: value for key, value in os.environ.items() if not key.startswith("GIT_")}
+        self.git("-c", "init.defaultBranch=main", "init", "-q")
+        (self.repo / "scripts").mkdir()
+        shutil.copy(SCRIPT, self.repo / "scripts/check-banned-words.sh")
+        self.write("scripts/banned-words.txt", f"{TERM}\n")
+        self.write(".gitattributes", "gen/** linguist-generated\nvendor/** linguist-vendored\n")
+        self.write("README.md", SEED_TEXT)
+        self.write("gen/data.txt", SEED_TEXT)
+        self.git("add", "-A")
+        self.git("commit", "-q", "-m", "seed")
+        self.seed = self.git("rev-parse", "HEAD")
+
+    def tearDown(self):
+        self.temp.cleanup()
+
+    def git(self, *args):
+        result = subprocess.run(
+            ["git", "-c", "user.name=test", "-c", "user.email=test@example.com", "-c", "commit.gpgsign=false", *args],
+            cwd=self.repo, env=self.env, check=True, capture_output=True, text=True,
+        )
+        return result.stdout.strip()
+
+    def write(self, path, text):
+        file = self.repo / path
+        file.parent.mkdir(parents=True, exist_ok=True)
+        with open(file, "a", encoding="utf-8") as handle:
+            handle.write(text)
+
+    def stage_term(self, *paths):
+        for path in paths:
+            self.write(path, f"a {TERM} line\n")
+        self.git("add", "-A")
+
+    def check(self, *args, cwd=None):
+        result = subprocess.run(
+            ["bash", str(self.repo / "scripts/check-banned-words.sh"), *args],
+            cwd=cwd or self.repo, env=self.env, capture_output=True, text=True,
+        )
+        return result.returncode, sorted(line.split(":")[0] for line in result.stdout.splitlines())
+
+    def commit_term(self, *paths):
+        self.stage_term(*paths)
+        self.git("commit", "-q", "-m", "change")
+        return self.check("--base", self.seed)
+
+    def test_a_generated_or_vendored_file_is_skipped(self):
+        self.assertEqual(self.commit_term("gen/notice.md", "vendor/lib.h", "src/app.txt"), (1, ["src/app.txt"]))
+
+    def test_a_diff_of_only_generated_and_vendored_files_passes(self):
+        self.assertEqual(self.commit_term("gen/notice.md", "vendor/lib.h"), (0, []))
+
+    def test_an_unset_or_false_attribute_is_checked(self):
+        self.write(".gitattributes", "vendor/own.txt -linguist-vendored\ngen/kept.txt linguist-generated=false\n")
+        self.assertEqual(self.commit_term("vendor/own.txt", "gen/kept.txt"), (1, ["gen/kept.txt", "vendor/own.txt"]))
+
+    def test_staged_reads_the_staged_gitattributes(self):
+        self.write(".gitattributes", "notes/** linguist-generated\n")
+        self.stage_term("notes/new.txt", "src/app.txt")
+        self.assertEqual(self.check("--staged"), (1, ["src/app.txt"]))
+
+    def test_a_renamed_file_follows_its_new_path(self):
+        (self.repo / "src").mkdir()
+        self.git("mv", "README.md", "gen/README.md")
+        self.git("mv", "gen/data.txt", "src/data.txt")
+        self.assertEqual(self.commit_term("gen/README.md", "src/data.txt"), (1, ["src/data.txt"]))
+
+    def test_a_subdirectory_still_resolves_attributes_from_the_top_level(self):
+        self.commit_term("gen/notice.md", "src/app.txt")
+        self.assertEqual(self.check("--base", self.seed, cwd=self.repo / "src"), (1, ["src/app.txt"]))
+
+
+if __name__ == "__main__":
+    unittest.main()