regress/plus_build_version: RCS Id extraction for +BUILD_VERSION

In pkgtools/pkglint, there are several lines that look almost like RCS
Ids. Some parts of the pkgsrc infrastructure expand them and some others
don't. This needs to be fixed so that all parts of pkgsrc agree what is a
complete RCS Id and what isn't.

As long as that is not the case, pbulk unnecessarily builds pkglint over
and over again, even if nothing changed. There are probably other
unintended side effects as well that just haven't been discovered or
considered grave enough.

8 files changed, 84 insertions, 0 deletions

diff --git a/regress/plus_build_version/files/010-only-first-line b/regress/plus_build_version/files/010-only-first-line
new file mode 100644
index 00000000000..2575fc19910
--- /dev/null
+++ b/regress/plus_build_version/files/010-only-first-line
@@ -0,0 +1,6 @@
+From each file, only the first RCS Id is extracted.
+
+first	$NetBSD: 010-only-first-line,v 1.1 2020/01/11 12:24:32 rillig Exp $
+second	$NetBSD: 010-only-first-line,v 1.1 2020/01/11 12:24:32 rillig Exp $
+
+For this reason, each of these test cases is in its own file.
diff --git a/regress/plus_build_version/files/020-short-name b/regress/plus_build_version/files/020-short-name
new file mode 100644
index 00000000000..4bc1d5206f6
--- /dev/null
+++ b/regress/plus_build_version/files/020-short-name
@@ -0,0 +1,6 @@
+The D at the end of the name is missing intentionally.
+
+Because of this missing letter, this file does not get recorded in
++BUILD_VERSION.
+
+$NetBS
diff --git a/regress/plus_build_version/files/021-long-name b/regress/plus_build_version/files/021-long-name
new file mode 100644
index 00000000000..0c969c1657c
--- /dev/null
+++ b/regress/plus_build_version/files/021-long-name
@@ -0,0 +1,7 @@
+The RCS Id must be exactly NetBSD, not any longer name that happens to
+start with NetBSD. Only the exact name is expanded by CVS, therefore
+recording the other occurrences doesn't make sense.
+
+As of January 2020, this line gets recorded nevertheless.
+
+$NetBSDFoundation$
diff --git a/regress/plus_build_version/files/030-missing-colon-or-dollar b/regress/plus_build_version/files/030-missing-colon-or-dollar
new file mode 100644
index 00000000000..b095a93ff27
--- /dev/null
+++ b/regress/plus_build_version/files/030-missing-colon-or-dollar
@@ -0,0 +1,7 @@
+A complete RCS Id needs a colon or a dollar, otherwise it is not
+expanded.
+
+As of January 2020, this line gets unnecessarily recorded in
++BUILD_VERSION.
+
+$NetBSD
diff --git a/regress/plus_build_version/files/040-missing-dollar-after-colon b/regress/plus_build_version/files/040-missing-dollar-after-colon
new file mode 100644
index 00000000000..535cc701fb7
--- /dev/null
+++ b/regress/plus_build_version/files/040-missing-dollar-after-colon
@@ -0,0 +1,7 @@
+Each RCS Id must end with a second dollar in the same line. Just having
+the colon is not enough.
+
+As of January 2020, this line gets unnecessarily recorded in
++BUILD_VERSION.
+
+$NetBSD:
diff --git a/regress/plus_build_version/files/050-cvs b/regress/plus_build_version/files/050-cvs
new file mode 100644
index 00000000000..07a3a2c3dff
--- /dev/null
+++ b/regress/plus_build_version/files/050-cvs
@@ -0,0 +1,4 @@
+The default CVS Id is not taken into account for the package build
+versions, even though it is expanded(?).
+
+$Id: 050-cvs,v 1.1 2020/01/11 12:24:32 rillig Exp $
diff --git a/regress/plus_build_version/files/expected b/regress/plus_build_version/files/expected
new file mode 100644
index 00000000000..92f1dba2f75
--- /dev/null
+++ b/regress/plus_build_version/files/expected
@@ -0,0 +1,5 @@
+regress/plus_build_version/files/010-only-first-line:	*NetBSD*
+regress/plus_build_version/files/021-long-name:	*NetBSDFoundation*
+regress/plus_build_version/files/030-missing-colon-or-dollar:	*
+regress/plus_build_version/files/040-missing-dollar-after-colon:	*
+regress/plus_build_version/spec:	*NetBSD*
diff --git a/regress/plus_build_version/spec b/regress/plus_build_version/spec
new file mode 100644
index 00000000000..d99fb982da2
--- /dev/null
+++ b/regress/plus_build_version/spec
@@ -0,0 +1,42 @@
+#! /bin/sh
+# $NetBSD: spec,v 1.1 2020/01/11 12:24:32 rillig Exp $
+
+# This test demonstrates how the RCS Ids from the package files end up in
+# the +BUILD_VERSION file of the binary package.
+#
+# There are several places in pkgsrc that make use of these RCS Ids:
+#
+# * The SHA1 hash for a patch file ignores RCS Id lines.
+# * The RCS Ids end up in the +BUILD_VERSION of the binary package.
+# * The pbulk checks the RCS Ids to see whether the package needs to be
+#   built again.
+#
+# All these places must use the same patterns for extracting the RCS Ids.
+# As of January 2020, they don't.
+
+set -u
+
+tmpdir="${TMPDIR-/tmp}/plus-build-version"
+actual="$tmpdir/.pkgdb/+BUILD_VERSION"
+
+do_test() {
+	rm -rf "$tmpdir"
+	mkdir -p "$tmpdir/.pkgdb"
+	$TEST_MAKE \
+		PKGNAME="package-1.0" \
+		WRKDIR="$tmpdir" \
+		-f "../../mk/bsd.pkg.mk" \
+		"$actual"
+}
+
+check_result() {
+	# In files/expected, the $ characters are replaced with * to
+	# prevent them from being expanded by CVS.
+	tr '*' '$' < "files/expected" > "$tmpdir/expected"
+
+	diff -u "$tmpdir/expected" "$actual" || regress_fail "differ"
+}
+
+do_cleanup() {
+	rm -rf "$tmpdir"
+}