krino: foundation — sexp reader, config language, init/new/check

2 files changed, 74 insertions, 0 deletions

diff --git a/scripts/hooks/pre-commit b/scripts/hooks/pre-commit
new file mode 100755
index 0000000..ec9f507
--- /dev/null
+++ b/scripts/hooks/pre-commit
@@ -0,0 +1,4 @@
+#!/bin/sh
+# SPDX-License-Identifier: GPL-3.0-or-later
+# Installed by `make install-hooks`: refuse commits that carry personal data.
+exec "$(git rev-parse --show-toplevel)/scripts/leak-check"
diff --git a/scripts/leak-check b/scripts/leak-check
new file mode 100755
index 0000000..fbbd5d5
--- /dev/null
+++ b/scripts/leak-check
@@ -0,0 +1,70 @@
+#!/bin/sh
+# SPDX-License-Identifier: GPL-3.0-or-later
+#
+# leak-check: refuse personal data in the files git would commit.
+#
+# Scans every staged file (the index, not the working tree), and the names
+# of staged files against the private list, for:
+#   - the current user's home directory, e.g. /home/alice;
+#   - email addresses (LICENSE is exempt);
+#   - a private pattern list: one extended regular expression per line,
+#     blank lines and lines starting with # ignored, matched ignoring case.
+#     Its path is set per clone with:  git config krino.leakpatterns FILE
+#     Keep that file outside the repository so the list is never published.
+#
+# A line matching the regex in `git config krino.leakallow` is never
+# reported (for example a public contact address in the README).
+#
+# Prints file:line for each hit, never the matched text, so a hit on a
+# private pattern does not echo the secret. Exits 0 when clean, 1 when
+# something matched, 2 when the private list is configured but missing.
+
+set -eu
+
+cd "$(git rev-parse --show-toplevel)"
+
+allow=$(git config --get krino.leakallow || true)
+hits=0
+
+# check LABEL GREP-ARGS...: report staged lines that match, as file:line.
+check() {
+	label=$1
+	shift
+	out=$(git grep --cached -n -I "$@" | { if [ -n "$allow" ]; then grep -v -E -i -e "$allow"; else cat; fi; } | cut -d: -f1,2 || true)
+	if [ -n "$out" ]; then
+		printf 'leak-check: %s:\n%s\n' "$label" "$out" | sed '2,$s/^/  /' >&2
+		hits=1
+	fi
+}
+
+home=${HOME%/} # a HOME of / becomes empty and is skipped
+if [ -n "$home" ]; then
+	check "your home directory ($home)" -F -e "$home" -- .
+fi
+
+check "an email address" -E -e '[[:alnum:]._%+-]+@[[:alnum:].-]+\.[[:alpha:]]{2,}' -- . ':(exclude)LICENSE'
+
+list=$(git config --type=path --get krino.leakpatterns || true)
+if [ -z "$list" ]; then
+	echo "leak-check: no private pattern list (git config krino.leakpatterns FILE); built-in checks only" >&2
+elif [ ! -r "$list" ]; then
+	echo "leak-check: private pattern list $list is configured but cannot be read" >&2
+	exit 2
+else
+	patterns=$(mktemp)
+	trap 'rm -f "$patterns"' EXIT
+	grep -v -e '^[[:space:]]*#' -e '^[[:space:]]*$' "$list" > "$patterns" || true
+	if [ -s "$patterns" ]; then
+		check "a private pattern from $list" -i -E -f "$patterns" -- .
+		named=$(git ls-files --cached | grep -i -E -f "$patterns" || true)
+		if [ -n "$named" ]; then
+			printf 'leak-check: a file name matching a private pattern:\n%s\n' "$named" | sed '2,$s/^/  /' >&2
+			hits=1
+		fi
+	fi
+fi
+
+if [ "$hits" -ne 0 ]; then
+	echo "leak-check: remove the data above before committing; for a false positive, narrow the pattern or set krino.leakallow" >&2
+	exit 1
+fi