Don't generate SemanticDB for generated or external sources

Previously, we only checked if a source ended with the "scala" extension
before adding it to `SemanticDbInfo`. However, we don't want to generate
SemanticDB files for generated files or external files, because they'll
have the output directory (beginning with `bazel-out`) in their paths,
which isn't guaranteed to be consistent.

Also, tell the SemanticDB compiler plugin to generate SemanticDB outside
of the output JAR, even if we don't want SemanticDB generated for any of
the sources. That way, the output JAR won't include SemanticDB files we
don't want.

rules/private/phases/phase_semanticdb.bzl

@@ -5,17 +5,15 @@ load(
     _SemanticDbInfo = "SemanticDbInfo",
 )
 
-def _semanticdb_directory_from_file(file):
+def _semanticdb_directory_from_output_jar(file):
     """
-    This is janky, but we're limited in what we can do in this function. From the
-    [documentation](https://bazel.build/rules/lib/builtins/Args#add_all) on `Args#add_all`:
-
-    To avoid unintended retention of large analysis-phase data structures into the execution phase,
-    the `map_each` function must be declared by a top-level `def` statement; it may not be a
-    nested function closure by default.
+    Using the path of the output JAR to determine the SemanticDB target root is janky, but the
+    output directory won't be known at build-time, so we have to use the output JAR path as a proxy.
+    It should be built under the same configuration as the SemanticDB files, since both are produced
+    by the same compilation action.
     """
 
-    return file.path[:file.path.find("META-INF") - 1]
+    return "{}/semanticdb".format(file.dirname)
 
 #
 # PHASE: semanticdb
@@ -33,7 +31,13 @@ def phase_semanticdb(ctx, g):
     outputs = []
 
     for source in ctx.files.srcs:
-        if source.extension == "scala":
+        # Generated or external files will have the output directory (beginning with `bazel-out`) in
+        # their paths, which we don't want because it isn't guaranteed to be consistent
+        if (
+            source.extension == "scala" and \
+            source.is_source and \
+            source.owner.repo_name == ctx.label.repo_name
+        ):
             path = paths.join(
                 directory_name,
                 "META-INF",
@@ -44,22 +48,21 @@ def phase_semanticdb(ctx, g):
             outputs.append(ctx.actions.declare_file(path))
 
     def add_scalacopts(arguments):
-        if len(outputs) == 0:
-            return
+        output_jar = g.classpaths.jar
 
         if toolchain.scala_configuration.version.startswith("2"):
             arguments.add("--compiler_option=-P:semanticdb:failures:error")
             arguments.add("--compiler_option_referencing_path=-P:semanticdb:sourceroot:${workDir}")
             arguments.add_all(
-                [outputs[0]],
+                [output_jar],
                 format_each = "--compiler_option_referencing_path=-P:semanticdb:targetroot:${path} %s",
-                map_each = _semanticdb_directory_from_file,
+                map_each = _semanticdb_directory_from_output_jar,
             )
         else:
             arguments.add_all(
-                [outputs[0]],
+                [output_jar],
                 format_each = "--compiler_option_referencing_path=-semanticdb-target:${path} %s",
-                map_each = _semanticdb_directory_from_file,
+                map_each = _semanticdb_directory_from_output_jar,
             )
 
             arguments.add("--compiler_option_referencing_path=-sourceroot:${workDir}")

tests/plugins/semanticdb/BUILD

@@ -1,3 +1,5 @@
+load("@bazel_skylib//rules:copy_file.bzl", "copy_file")
+load("@rules_java//java:java_library.bzl", "java_library")
 load("@rules_scala_annex//rules:register_toolchain.bzl", "register_zinc_toolchain")
 load("@rules_scala_annex//rules:scala.bzl", "scala_library")
 load("@rules_scala_annex//rules/scala:versions.bzl", "scala_2_13_version", "scala_3_version")
@@ -38,7 +40,6 @@ scala_library(
     name = "semanticdb-2_13",
     srcs = glob(["*.scala"]),
     scala_toolchain_name = "scala_2_13_with_semanticdb",
-    tags = ["manual"],
 )
 
 read_semanticdb_info(
@@ -50,7 +51,6 @@ scala_library(
     name = "semanticdb-3",
     srcs = glob(["*.scala"]),
     scala_toolchain_name = "scala_3_with_semanticdb",
-    tags = ["manual"],
 )
 
 read_semanticdb_info(
@@ -60,7 +60,27 @@ read_semanticdb_info(
 
 scala_library(
     name = "semanticdb-empty",
-    srcs = [],
     scala_toolchain_name = "scala_2_13_with_semanticdb",
-    tags = ["manual"],
+)
+
+java_library(
+    name = "sourcejar-library",
+    resources = glob(["*.scala"]),
+)
+
+copy_file(
+    name = "sourcejar",
+    src = ":sourcejar-library",
+    out = "sourcejar.srcjar",
+)
+
+scala_library(
+    name = "semanticdb-sourcejar",
+    srcs = [":sourcejar"],
+    scala_toolchain_name = "scala_2_13_with_semanticdb",
+)
+
+read_semanticdb_info(
+    name = "semanticdb-sourcejar-semanticdb-info",
+    scala_target = ":semanticdb-sourcejar",
 )

tests/plugins/semanticdb/test

@@ -5,7 +5,7 @@ bazel_bin="$(bazel info bazel-bin)"
 
 check_for_semanticdb_files() {
 	for filename in "A.scala.semanticdb" "B.scala.semanticdb"; do
-		path="../../bazel-bin/plugins/semanticdb/semanticdb-$1/semanticdb/META-INF/semanticdb/plugins/semanticdb/$filename"
+		path="../../bazel-bin/plugins/semanticdb/$1/semanticdb/META-INF/semanticdb/plugins/semanticdb/$filename"
 
 		if [ ! -f "$path" ]; then
 			echo "Error: $path doesn't exist"
@@ -15,22 +15,42 @@ check_for_semanticdb_files() {
 }
 
 check_semanticdb_info() {
-	bazel build ":semanticdb-$1-semanticdb-info"
+	bazel build ":$1-semanticdb-info"
 
-	output_path="$bazel_bin/plugins/semanticdb/semanticdb-$1-semanticdb-info.txt"
+	output_path="$bazel_bin/plugins/semanticdb/$1-semanticdb-info.txt"
 	reported_semanticdb_files="$(jq -r '.semanticDbFiles[]' "$output_path" | sed -E 's_^bazel-out/[^\n/]+/bin/__g')"
-	semanticdb_file_directory="plugins/semanticdb/semanticdb-$1/semanticdb/META-INF/semanticdb/plugins/semanticdb"
+	semanticdb_file_directory="plugins/semanticdb/$1/semanticdb/META-INF/semanticdb/plugins/semanticdb"
 
-	[ "$(jq '.targetRoot' "$output_path")" = "\"plugins/semanticdb/semanticdb-$1/semanticdb\"" ]
+	[ "$(jq '.targetRoot' "$output_path")" = "\"plugins/semanticdb/$1/semanticdb\"" ]
 	[ "$reported_semanticdb_files" = "$semanticdb_file_directory/A.scala.semanticdb"$'\n'"$semanticdb_file_directory/B.scala.semanticdb" ]
 }
 
-bazel build :semanticdb-2_13
-check_for_semanticdb_files 2_13
-check_semanticdb_info 2_13
+works_with_scala_2_13() {
+	bazel build :semanticdb-2_13
+	check_for_semanticdb_files semanticdb-2_13
+	check_semanticdb_info semanticdb-2_13
+}
+
+works_with_scala_3() {
+	bazel build :semanticdb-3
+	check_for_semanticdb_files semanticdb-3
+	check_semanticdb_info semanticdb-3
+}
+
+works_with_no_sources() {
+	bazel build :semanticdb-empty
+}
+
+does_not_produce_semanticdb_for_source_jars() {
+	bazel build :sourcejar-library
 
-bazel build :semanticdb-3
-check_for_semanticdb_files '3'
-check_semanticdb_info '3'
+	semanticdb_info_path="$bazel_bin/plugins/semanticdb/semanticdb-sourcejar-semanticdb-info.txt"
+
+	[ "$(jq '.semanticDbFiles' "$semanticdb_info_path")" = '[]' ]
+	[ "$(jq '.targetRoot' "$semanticdb_info_path")" = 'plugins/semanticdb/semanticdb-sourcejar/semanticdb' ]
+}
 
-bazel build :semanticdb-empty
+works_with_scala_2_13
+works_with_scala_3
+works_with_no_sources
+does_not_produce_semanticdb_for_source_jars