Allow to provide 'grep' regex for print to narrow output.

README.md

@@ -60,6 +60,13 @@ of these and print the final form:
 bant print -b @googletest//:gtest
 ```
 
+The `-g` option allows to 'grep' for targets where any of the strings in the
+AST of a rule matches a pattern
+
+```
+bant print ... -g "arena.*test"
+```
+
 #### Workspace
 
 **`workspace`** prints all the external projects found in the workspace.
@@ -229,10 +236,13 @@ Options
 Commands (unique prefix sufficient):
     == Parsing ==
     print          : Print AST matching pattern. -e : only files w/ parse errors
-                     -b : elaBorate; light eval: expand variables, concat etc.
+                      -b : elaBorate; light eval: expand variables, concat etc.
+                      -g <regex> : 'grep' - only print targets where any string
+                                    matches regex.
+                      -i If '-g' is given: case insensitive
     parse          : Parse all BUILD files from pattern. Follow deps with -r
                      Emit parse errors. Silent otherwise: No news are good news.
-                     -v : some stats.
+                      -v : some stats.
 
     == Extract facts == (Use -f to choose output format) ==
     workspace      : Print external projects found in WORKSPACE.
@@ -258,7 +268,7 @@ Commands (unique prefix sufficient):
 
     == Tools ==
     dwyu           : DWYU: Depend on What You Use (emit buildozer edit script)
-                     -k strict: emit remove even if # keep comment in line.
+                      -k strict: emit remove even if # keep comment in line.
     canonicalize   : Emit rename edits to canonicalize targets.
 ```
 

bant/bant.cc

@@ -81,10 +81,13 @@ static int usage(const char *prog, const char *message, int exit_code) {
 Commands (unique prefix sufficient):
     %s== Parsing ==%s
     print          : Print AST matching pattern. -e : only files w/ parse errors
-                     -b : elaBorate; light eval: expand variables, concat etc.
+                      -b : elaBorate; light eval: expand variables, concat etc.
+                      -g <regex> : 'grep' - only print targets where any string
+                                    matches regex.
+                      -i If '-g' is given: case insensitive
     parse          : Parse all BUILD files from pattern. Follow deps with -r
                      Emit parse errors. Silent otherwise: No news are good news.
-                     -v : some stats.
+                      -v : some stats.
 
     %s== Extract facts ==%s (Use -f to choose output format) ==
     workspace      : Print external projects found in WORKSPACE.
@@ -110,7 +113,7 @@ Commands (unique prefix sufficient):
 
     %s== Tools ==%s
     dwyu           : DWYU: Depend on What You Use (emit buildozer edit script)
-                     -k strict: emit remove even if # keep comment in line.
+                      -k strict: emit remove even if # keep comment in line.
     canonicalize   : Emit rename edits to canonicalize targets.
 )",
           BOLD, RESET, BOLD, RESET, BOLD, RESET);
@@ -132,14 +135,16 @@ int main(int argc, char *argv[]) {
 
   bant::CommandlineFlags flags;
 
+  bool regex_case_insesitive = false;
+
   using bant::OutputFormat;
   static const std::map<std::string_view, OutputFormat> kFormatOutNames = {
     {"native", OutputFormat::kNative}, {"s-expr", OutputFormat::kSExpr},
     {"plist", OutputFormat::kPList},   {"csv", OutputFormat::kCSV},
     {"json", OutputFormat::kJSON},     {"graphviz", OutputFormat::kGraphviz},
   };
   int opt;
-  while ((opt = getopt(argc, argv, "C:qo:vhpecbf:r::Vk")) != -1) {
+  while ((opt = getopt(argc, argv, "C:qo:vhpecbf:r::Vkg:i")) != -1) {
     switch (opt) {
     case 'C': {
       std::error_code err;
@@ -176,10 +181,13 @@ int main(int argc, char *argv[]) {
                                          : std::numeric_limits<int>::max();
       break;
 
-    case 'k':
-      flags.ignore_keep_comment = true;
-      break;
+    case 'k': flags.ignore_keep_comment = true; break;
+
+    case 'g': flags.grep_regex = optarg; break;
 
+    case 'i':
+      regex_case_insesitive = true;
+      break;
       // "print" options
     case 'p': flags.print_ast = true; break;
     case 'e': flags.print_only_errors = true; break;
@@ -197,6 +205,10 @@ int main(int argc, char *argv[]) {
     }
   }
 
+  if (regex_case_insesitive && !flags.grep_regex.empty()) {
+    flags.grep_regex.insert(0, "(?i)");
+  }
+
   bant::FilesystemPrewarmCacheInit(argc, argv);
 
   bant::Session session(primary_out, info_out, flags);

bant/cli-commands.cc

@@ -180,10 +180,12 @@ CliStatus RunCommand(Session &session, Command cmd,
   switch (cmd) {
   case Command::kPrint: flags.print_ast = true; [[fallthrough]];
   case Command::kParse:
-    // Parsing has already be done by now by building the dependency graph
+    // Parsing has already be done by now by building the dependency graph,
+    // so it would already have emitted parse errors. Here we only have to
+    // decide if we print anything.
     if (flags.print_ast || flags.print_only_errors) {
       bant::PrintProject(pattern, session.out(), session.info(), project,
-                         flags.print_only_errors);
+                         flags.print_only_errors, flags.grep_regex);
     }
     break;
 

bant/frontend/BUILD

@@ -73,6 +73,7 @@ cc_library(
         "//bant/util:memory",
         "@abseil-cpp//absl/log:check",
         "@abseil-cpp//absl/strings",
+        "@re2",
     ],
 )
 
@@ -120,6 +121,7 @@ cc_library(
         "//bant/util:memory",
         "//bant/util:stat",
         "@abseil-cpp//absl/log:check",
+        "@re2",
     ],
 )
 

bant/frontend/parsed-project.cc

@@ -19,6 +19,7 @@
 
 #include <cstdlib>
 #include <iostream>
+#include <memory>
 #include <optional>
 #include <sstream>
 #include <string>
@@ -38,6 +39,7 @@
 #include "bant/util/file-utils.h"
 #include "bant/util/stat.h"
 #include "bant/workspace.h"
+#include "re2/re2.h"
 
 namespace bant {
 namespace {
@@ -236,7 +238,19 @@ const ParsedBuildFile *ParsedProject::FindParsedOrNull(
 
 void PrintProject(const BazelPattern &pattern, std::ostream &out,
                   std::ostream &info_out, const ParsedProject &project,
-                  bool only_files_with_errors) {
+                  bool only_files_with_errors, std::string_view grep_regex) {
+  std::unique_ptr<RE2> regex;
+  if (!grep_regex.empty()) {
+    // TODO: pass options to not log error
+    regex = std::make_unique<RE2>(grep_regex);
+    if (!regex->ok()) {
+      // This really needs the session passed in so that we can reach the
+      // correct error stream.
+      std::cerr << "Grep pattern: " << regex->error() << "\n";
+      return;
+    }
+  }
+
   for (const auto &[package, file_content] : project.ParsedFiles()) {
     if (only_files_with_errors && file_content->errors.empty()) {
       continue;
@@ -245,26 +259,28 @@ void PrintProject(const BazelPattern &pattern, std::ostream &out,
       continue;
     }
 
-    if (pattern.is_recursive()) {
-      out << "# " << file_content->name() << ": "
-          << file_content->package.ToString() << "\n";
-      info_out << file_content->errors;
-      PrintVisitor(out).WalkNonNull(file_content->ast);
-      out << "\n";
-    } else {
-      query::FindTargets(
-        file_content->ast, {}, [&](const query::Result &result) {
+    query::FindTargetsAllowEmptyName(
+      file_content->ast, {}, [&](const query::Result &result) {
+        if (!pattern.is_recursive()) {
+          // Need more specific check if matches.
           auto self = BazelTarget::ParseFrom(result.name, package);
           if (!self.has_value() || !pattern.Match(*self)) {
             return;
           }
-          // TODO: instead of just marking the range of the function name,
-          // show the range the whole function covers until closed parenthesis.
-          out << "# " << project.Loc(result.node->identifier()->id()) << "\n";
-          PrintVisitor(out).WalkNonNull(result.node);
-          out << "\n";
-        });
-    }
+        }
+
+        // TODO: instead of just marking the range of the function name,
+        // show the range the whole function covers until closed parenthesis.
+        // TODO: if isatty(out), color filename gray
+        std::stringstream tmp_out;
+        tmp_out << "# " << project.Loc(result.node->identifier()->id()) << "\n";
+        PrintVisitor printer(tmp_out, regex.get());
+        printer.WalkNonNull(result.node);
+        tmp_out << "\n";
+        if (!regex || printer.any_highlight()) {  // w/o regex: always print.
+          out << tmp_out.str();
+        }
+      });
   }
 }
 }  // namespace bant

bant/frontend/parsed-project.h

@@ -133,8 +133,11 @@ class ParsedProject : public SourceLocator {
 // "out" is the destination of the acutal parse tree, "info_out" will
 // print error message and filenames.
 // If "only_files_with_errors" is set, prints only the files that had issues.
+// With "grep_regex", only targets are printed that have any substring match
+// expression.
+// TODO: this function has too many parameters.
 void PrintProject(const BazelPattern &pattern, std::ostream &out,
                   std::ostream &info_out, const ParsedProject &project,
-                  bool only_files_with_errors);
+                  bool only_files_with_errors, std::string_view grep_regex);
 }  // namespace bant
 #endif  // BANT_PROJECT_PARDER_

bant/frontend/print-visitor.cc

@@ -22,6 +22,7 @@
 
 #include "bant/frontend/ast.h"
 #include "bant/frontend/scanner.h"
+#include "re2/re2.h"
 
 namespace bant {
 void PrintVisitor::VisitFunCall(FunCall *f) {
@@ -122,6 +123,11 @@ void PrintVisitor::VisitScalar(Scalar *s) {
       str->AsString().find_first_of('"') != std::string_view::npos;
     const char quote_char = has_any_double_quote ? '\'' : '"';
     if (str->is_triple_quoted()) out_ << quote_char << quote_char;
+    if (optional_highlight_) {
+      // TODO: actually highlight.
+      any_highlight_ |=
+        RE2::PartialMatch(str->AsString(), *optional_highlight_);
+    }
     out_ << quote_char << str->AsString() << quote_char;
     if (str->is_triple_quoted()) out_ << quote_char << quote_char;
   }

bant/frontend/print-visitor.h

@@ -18,14 +18,18 @@
 #ifndef BANT_PRINT_VISITOR_H
 #define BANT_PRINT_VISITOR_H
 
+#include <memory>
 #include <ostream>
 
 #include "bant/frontend/ast.h"
+#include "re2/re2.h"
 
 namespace bant {
 class PrintVisitor : public BaseVoidVisitor {
  public:
-  explicit PrintVisitor(std::ostream &out) : out_(out) {}
+  explicit PrintVisitor(std::ostream &out,
+                        const RE2 *optional_highlight = nullptr)
+      : out_(out), optional_highlight_(optional_highlight) {}
   // Using default impl. for Assignment.
   void VisitFunCall(FunCall *f) final;
   void VisitList(List *l) final;
@@ -38,9 +42,13 @@ class PrintVisitor : public BaseVoidVisitor {
   void VisitScalar(Scalar *s) final;
   void VisitIdentifier(Identifier *i) final;
 
+  bool any_highlight() const { return any_highlight_; }
+
  private:
   int indent_ = 0;
   std::ostream &out_;
+  const RE2 *const optional_highlight_;
+  bool any_highlight_ = false;
 };
 }  // namespace bant
 

bant/session.h

@@ -53,6 +53,7 @@ struct CommandlineFlags {
   bool ignore_keep_comment = false;
   int recurse_dependency_depth = 0;
   OutputFormat output_format = OutputFormat::kNative;
+  std::string grep_regex;
 };
 
 // A session contains some settings such as output/verbose requests