lib-hdrs, lib-srcs: also expand to data. And rename target-...

Also with `-P` it only emits elements that are physical files.

Author: hzeller

bant/bant.cc

@@ -131,13 +131,16 @@ Commands (unique prefix sufficient):
                      → 2 column table: (target, dependency*)
     has-dependent  : List cc library targets and the libraries that depend on it
                      → 2 column table: (target, dependent*)
-    lib-hdrs       : Print headers provided by cc_library()s matching pattern.
-                     → 2 column table: (header-filename, cc-library-target)
-    lib-srcs       : Print sources provided by cc_library()s matching pattern.
-                     → 2 column table: (source-filename, cc-library-target)
-    lib-headers     : like lib-hdrs, but expands all reachable paths due to
-                      includes = [] expansion; so same header can show up
-                      multiple times.
+    target-hdrs    : Print hdrs mentioned in targets (-P: only physical files)
+                     → 2 column table: (header-filename, target)
+    target-srcs    : Print srcs mentioned in targets (-P: only physical files)
+                     → 2 column table: (srcs-filename, target)
+    target-data    : Print data mentioned in targets (-P: only physical files)
+                     → 2 column table: (data-filename, target)
+    lib-headers    : Like target-hdrs, but all reachable paths expanded with all
+                     combinations of includes = [], include_prefix, etc.
+                     So same header can show up multiple times with different
+                     paths. This is the relevant list used by dwyu.
                      → 2 column table: (header-filename, cc-library-target)
     genrule-outputs: Print generated files by genrule()s matching pattern.
                      → 2 column table: (filename, genrule-target)
@@ -211,7 +214,7 @@ int main(int argc, char *argv[]) {
     {"json", OutputFormat::kJSON},     {"graphviz", OutputFormat::kGraphviz},
   };
   int opt;
-  while ((opt = getopt(argc, argv, "C:qo:vhaEF:ecbf:r::Vkg:iT:")) != -1) {
+  while ((opt = getopt(argc, argv, "C:qo:vhaEF:ecbf:r::Vkg:iT:P")) != -1) {
     switch (opt) {
     case 'C': {
       std::error_code err;
@@ -277,6 +280,7 @@ int main(int argc, char *argv[]) {
         return usage(argv[0], "-T needs a numeric parameter", EXIT_FAILURE);
       }
       break;
+    case 'P': flags.only_physical_files = true; break;
     case 'v': flags.verbose++; break;  // More -v, more detail.
     case 'V': return print_version();
     default: return usage(argv[0], nullptr, EXIT_SUCCESS);

bant/cli-commands.cc

@@ -67,8 +67,9 @@ enum class Command {
   kListTargets,
   kListLeafs,
   kListWorkkspace,
-  kLibraryHdrs,
-  kLibrarySrcs,
+  kTargetHdrs,
+  kTargetSrcs,
+  kTargetData,
   kExpandedLibraryHeaders,
   kAliasedBy,
   kGenruleOutputs,
@@ -212,9 +213,10 @@ CliStatus RunCommand(Session &session, Command cmd,
   switch (cmd) {
   case Command::kDWYU:
   case Command::kParse:
-  case Command::kLibraryHdrs:
+  case Command::kTargetHdrs:
+  case Command::kTargetData:
   case Command::kExpandedLibraryHeaders:
-  case Command::kLibrarySrcs:
+  case Command::kTargetSrcs:
   case Command::kGenruleOutputs:
   case Command::kListTargets:
   case Command::kListLeafs:
@@ -279,18 +281,29 @@ CliStatus RunCommand(Session &session, Command cmd,
       ExtractExpandedHeaderToLibMapping(project, session.info()));
     break;
 
-  case Command::kLibrarySrcs:  //
+    // TODO: these target srcs/hdrs/data should include target type.
+  case Command::kTargetSrcs:  //
     bant::PrintProvidedSources(
-      session, "source", print_pattern,
-      ExtractComponentToLibMapping(project, ExtractComponent::kSrcs,
-                                   session.info()));
+      session, "srcs", print_pattern,
+      ExtractComponentToTargetMapping(project, ExtractComponent::kSrcs,
+                                      session.flags().only_physical_files,
+                                      session.info()));
     break;
 
-  case Command::kLibraryHdrs:  //
+  case Command::kTargetHdrs:  //
     bant::PrintProvidedSources(
-      session, "header", print_pattern,
-      ExtractComponentToLibMapping(project, ExtractComponent::kHdrs,
-                                   session.info()));
+      session, "hdrs", print_pattern,
+      ExtractComponentToTargetMapping(project, ExtractComponent::kHdrs,
+                                      session.flags().only_physical_files,
+                                      session.info()));
+    break;
+
+  case Command::kTargetData:  //
+    bant::PrintProvidedSources(
+      session, "data", print_pattern,
+      ExtractComponentToTargetMapping(project, ExtractComponent::kData,
+                                      session.flags().only_physical_files,
+                                      session.info()));
     break;
 
   case Command::kGenruleOutputs:
@@ -396,8 +409,9 @@ CliStatus RunCliCommand(Session &session, std::span<std::string_view> args) {
     {"list-targets", Command::kListTargets},
     {"list-leafs", Command::kListLeafs},
     {"workspace", Command::kListWorkkspace},
-    {"lib-hdrs", Command::kLibraryHdrs},
-    {"lib-srcs", Command::kLibrarySrcs},
+    {"target-hdrs", Command::kTargetHdrs},
+    {"target-data", Command::kTargetData},
+    {"target-srcs", Command::kTargetSrcs},
     {"lib-headers", Command::kExpandedLibraryHeaders},
     {"aliased-by", Command::kAliasedBy},
     {"depends-on", Command::kDependsOn},

bant/explore/BUILD

@@ -67,6 +67,8 @@ cc_library(
         "//bant:types",
         "//bant:types-bazel",
         "//bant/frontend:parsed-project",
+        "//bant/frontend:parser",
+        "//bant/util:file-utils",
         "//bant/util:table-printer",
         "@abseil-cpp//absl/container:btree",
         "@abseil-cpp//absl/strings",

bant/explore/header-providers.cc

@@ -32,10 +32,12 @@
 #include "absl/strings/str_cat.h"
 #include "bant/explore/aliased-by.h"
 #include "bant/explore/query-utils.h"
+#include "bant/frontend/ast.h"
 #include "bant/frontend/parsed-project.h"
 #include "bant/session.h"
 #include "bant/types-bazel.h"
 #include "bant/types.h"
+#include "bant/util/filesystem.h"
 #include "bant/util/table-printer.h"
 
 // The header providers maps header filenames to all the libraries that
@@ -336,27 +338,33 @@ ProvidedFromTargetSet ExtractExpandedHeaderToLibMapping(
   return result;
 }
 
-ProvidedFromTargetSet ExtractComponentToLibMapping(const ParsedProject &project,
-                                                   ExtractComponent which,
-                                                   std::ostream &info_out) {
+ProvidedFromTargetSet ExtractComponentToTargetMapping(
+  const ParsedProject &project, ExtractComponent which,
+  bool only_physical_files, std::ostream &info_out) {
+  auto &fs = Filesystem::instance();
   ProvidedFromTargetSet result;
 
   for (const auto &[_, build_file] : project.ParsedFiles()) {
     if (!build_file->ast) continue;
 
-    query::FindTargets(
-      build_file->ast, {"cc_library"}, [&](const query::Result &cc_lib) {
-        auto cc_library = build_file->package.QualifiedTarget(cc_lib.name);
-        if (!cc_library.has_value()) return;
-
-        auto srcs = query::ExtractStringList((which == ExtractComponent::kSrcs)
-                                               ? cc_lib.srcs_list
-                                               : cc_lib.hdrs_list);
-        for (const std::string_view src : srcs) {
-          const std::string src_fqn = build_file->package.QualifiedFile(src);
+    query::FindTargets(build_file->ast, {}, [&](const query::Result &cc_lib) {
+      auto cc_library = build_file->package.QualifiedTarget(cc_lib.name);
+      if (!cc_library.has_value()) return;
+
+      List *search_list = nullptr;
+      switch (which) {
+      case ExtractComponent::kHdrs: search_list = cc_lib.hdrs_list; break;
+      case ExtractComponent::kSrcs: search_list = cc_lib.srcs_list; break;
+      case ExtractComponent::kData: search_list = cc_lib.data_list; break;
+      }
+      auto srcs = query::ExtractStringList(search_list);
+      for (const std::string_view src : srcs) {
+        const std::string src_fqn = build_file->package.QualifiedFile(src);
+        if (!only_physical_files || fs.Exists(src_fqn)) {
           result[src_fqn].emplace(*cc_library);
         }
-      });
+      }
+    });
   }
 
   return result;

bant/explore/header-providers.h

@@ -51,14 +51,17 @@ ProvidedFromTargetSet ExtractExpandedHeaderToLibMapping(
   bool suffix_index = false);
 
 // Unlike the Expanded*, this is much simpler, as we don't deal with multiple
-// search paths. Just extract the mentioned component.
+// search paths. Just extract the mentioned component. If "only_physical_files"
+// is on, only emit them if these are files in the filesystem (unlike
+// generated files or target names).
 enum class ExtractComponent {
   kSrcs,
   kHdrs,
+  kData,
 };
-ProvidedFromTargetSet ExtractComponentToLibMapping(const ParsedProject &project,
-                                                   ExtractComponent which,
-                                                   std::ostream &info_out);
+ProvidedFromTargetSet ExtractComponentToTargetMapping(
+  const ParsedProject &project, ExtractComponent which,
+  bool only_physical_files, std::ostream &info_out);
 
 // Find all the output generated by genrules. Should really only be 1:1.
 // If "suffix_index" is set, output is compatible with FinxBySuffix().

bant/explore/header-providers_test.cc

@@ -110,20 +110,25 @@ TEST(SourceToLibMapping, CCRuleExtraction) {
 cc_library(
   name = "foo",
   srcs = ["foo.cc", "bar.cc"],
-  hdrs = ["baz.h"]
+  hdrs = ["baz.h"],
+  data = ["data.txt"],
 )
 )");
   std::stringstream log_absorb;
-  auto srcs_map = ExtractComponentToLibMapping(
-    pp.project(), ExtractComponent::kSrcs, log_absorb);
+  auto srcs_map = ExtractComponentToTargetMapping(
+    pp.project(), ExtractComponent::kSrcs, false, log_absorb);
   EXPECT_THAT(srcs_map,
               Contains(Pair("some/path/foo.cc", Ts("//some/path:foo"))));
   EXPECT_THAT(srcs_map,
               Contains(Pair("some/path/bar.cc", Ts("//some/path:foo"))));
-  auto hdrs_map = ExtractComponentToLibMapping(
-    pp.project(), ExtractComponent::kHdrs, log_absorb);
+  auto hdrs_map = ExtractComponentToTargetMapping(
+    pp.project(), ExtractComponent::kHdrs, false, log_absorb);
   EXPECT_THAT(hdrs_map,
               Contains(Pair("some/path/baz.h", Ts("//some/path:foo"))));
+  auto data_map = ExtractComponentToTargetMapping(
+    pp.project(), ExtractComponent::kData, false, log_absorb);
+  EXPECT_THAT(data_map,
+              Contains(Pair("some/path/data.txt", Ts("//some/path:foo"))));
 }
 
 TEST(HeaderToLibMapping, InludePathsAreRelativePathCanonicalized) {

bant/session.h

@@ -55,6 +55,7 @@ struct CommandlineFlags {
   bool elaborate = false;
   bool builtin_macro_expand = false;
   bool ignore_keep_comment = false;
+  bool only_physical_files = false;  // for target-{srcs,hdrs,data}
   int recurse_dependency_depth = 0;
   OutputFormat output_format = OutputFormat::kNative;
   int io_threads = 0;  // <= 0: synchronous operation.