[clang] [clang-format] Support globstar in .clang-format-ignore (PR #121404)

[Owen Pan via cfe-commits]

https://github.com/owenca updated https://github.com/llvm/llvm-project/pull/121404

>From a1060e903d29340840cdd0a9cecf74269fc7c563 Mon Sep 17 00:00:00 2001
From: Owen Pan <owenpiano at gmail.com>
Date: Tue, 31 Dec 2024 08:13:53 -0800
Subject: [PATCH] [clang-format] Support globstar in .clang-format-ignore

Closes #114969.
---
 clang/docs/ClangFormat.rst                   |  4 +++
 clang/docs/ReleaseNotes.rst                  |  1 +
 clang/lib/Format/MatchFilePath.cpp           | 35 ++++++++++++++------
 clang/unittests/Format/MatchFilePathTest.cpp | 35 ++++++++++++++++++++
 4 files changed, 64 insertions(+), 11 deletions(-)

diff --git a/clang/docs/ClangFormat.rst b/clang/docs/ClangFormat.rst
index c8f1d7f5a77581..808ab7a0957380 100644
--- a/clang/docs/ClangFormat.rst
+++ b/clang/docs/ClangFormat.rst
@@ -150,6 +150,10 @@ names. It has the following format:
 * Patterns follow the rules specified in `POSIX 2.13.1, 2.13.2, and Rule 1 of
   2.13.3 <https://pubs.opengroup.org/onlinepubs/9699919799/utilities/
   V3_chap02.html#tag_18_13>`_.
+* Bash globstar is supported, i.e. two adjacent ``*``s used as a single pattern
+  will match all files and zero or more directories and subdirectories. If
+  followed by a ``/``, two adjacent ``*``s will match only directories and
+  subdirectories.
 * A pattern is negated if it starts with a bang (``!``).
 
 To match all files in a directory, use e.g. ``foo/bar/*``. To match all files in
diff --git a/clang/docs/ReleaseNotes.rst b/clang/docs/ReleaseNotes.rst
index b7da12bcf65818..2d59d74ddf35fc 100644
--- a/clang/docs/ReleaseNotes.rst
+++ b/clang/docs/ReleaseNotes.rst
@@ -1124,6 +1124,7 @@ clang-format
 - Adds ``RemoveEmptyLinesInUnwrappedLines`` option.
 - Adds ``KeepFormFeed`` option and set it to ``true`` for ``GNU`` style.
 - Adds ``AllowShortNamespacesOnASingleLine`` option.
+- Adds support for bash globstar in ``.clang-format-ignore``.
 
 libclang
 --------
diff --git a/clang/lib/Format/MatchFilePath.cpp b/clang/lib/Format/MatchFilePath.cpp
index 062b334dcdd8fd..aca3433dc7fa78 100644
--- a/clang/lib/Format/MatchFilePath.cpp
+++ b/clang/lib/Format/MatchFilePath.cpp
@@ -49,25 +49,38 @@ bool matchFilePath(StringRef Pattern, StringRef FilePath) {
         return false;
       break;
     case '*': {
-      while (++I < EOP && Pattern[I] == '*') { // Skip consecutive stars.
+      const bool MaybeGlobstar = I == 0 || Pattern[I - 1] == Separator;
+      int StarCount = 1;
+      for (; ++I < EOP && Pattern[I] == '*'; ++StarCount) {
+        // Skip consecutive stars.
       }
       const auto K = FilePath.find(Separator, J); // Index of next `Separator`.
       const bool NoMoreSeparatorsInFilePath = K == StringRef::npos;
+      bool Globstar = MaybeGlobstar && StarCount == 2;
       if (I == EOP) // `Pattern` ends with a star.
-        return NoMoreSeparatorsInFilePath;
-      // `Pattern` ends with a lone backslash.
-      if (Pattern[I] == '\\' && ++I == EOP)
-        return false;
+        return Globstar || NoMoreSeparatorsInFilePath;
+      if (Pattern[I] != Separator) {
+        Globstar = false;
+        // `Pattern` ends with a lone backslash.
+        if (Pattern[I] == '\\' && ++I == EOP)
+          return false;
+      }
       // The star is followed by a (possibly escaped) `Separator`.
       if (Pattern[I] == Separator) {
-        if (NoMoreSeparatorsInFilePath)
-          return false;
-        J = K; // Skip to next `Separator` in `FilePath`.
-        break;
+        if (!Globstar) {
+          if (NoMoreSeparatorsInFilePath)
+            return false;
+          J = K; // Skip to next `Separator` in `FilePath`.
+          break;
+        }
+        if (I + 1 < EOP &&
+            matchFilePath(Pattern.substr(I + 1), FilePath.substr(J))) {
+          return true;
+        }
       }
       // Recurse.
-      for (auto Pat = Pattern.substr(I); J < End && FilePath[J] != Separator;
-           ++J) {
+      for (auto Pat = Pattern.substr(I);
+           J < End && (Globstar || FilePath[J] != Separator); ++J) {
         if (matchFilePath(Pat, FilePath.substr(J)))
           return true;
       }
diff --git a/clang/unittests/Format/MatchFilePathTest.cpp b/clang/unittests/Format/MatchFilePathTest.cpp
index 28f665635718e5..9876a4583c226a 100644
--- a/clang/unittests/Format/MatchFilePathTest.cpp
+++ b/clang/unittests/Format/MatchFilePathTest.cpp
@@ -164,6 +164,41 @@ TEST_F(MatchFilePathTest, Path) {
   EXPECT_FALSE(match("foo\\", R"(foo*\)"));
 }
 
+TEST_F(MatchFilePathTest, Globstar) {
+  EXPECT_TRUE(match("/", "**"));
+  EXPECT_TRUE(match("foo", "**"));
+  EXPECT_TRUE(match("/foo", "**"));
+  EXPECT_TRUE(match("foo/", "**"));
+  EXPECT_TRUE(match("foo/bar", "**"));
+
+  EXPECT_TRUE(match("/", "**/"));
+  EXPECT_TRUE(match("foo/", "**/"));
+  EXPECT_TRUE(match("/foo/", "**/"));
+  EXPECT_TRUE(match("foo/bar/", "**/"));
+
+  EXPECT_TRUE(match("/", "/**"));
+  EXPECT_TRUE(match("/foo", "/**"));
+  EXPECT_TRUE(match("/foo/", "/**"));
+  EXPECT_TRUE(match("/foo/bar", "/**"));
+
+  EXPECT_TRUE(match("foo", "**/foo"));
+  EXPECT_TRUE(match("/foo", "**/foo"));
+  EXPECT_TRUE(match("foo/bar", "**/bar"));
+  EXPECT_TRUE(match("/foo/bar", "**/foo/bar"));
+  EXPECT_TRUE(match("foo/bar/baz", "**/bar/baz"));
+
+  EXPECT_TRUE(match("abc/foo", "abc/**"));
+  EXPECT_TRUE(match("abc/foo/", "abc/**"));
+  EXPECT_TRUE(match("abc/foo/bar", "abc/**"));
+
+  EXPECT_TRUE(match("a/b", "a/**/b"));
+  EXPECT_TRUE(match("a/x/b", "a/**/b"));
+  EXPECT_TRUE(match("a/x/y/b", "a/**/b"));
+
+  EXPECT_FALSE(match("a/x/b", "a**/b"));
+  EXPECT_FALSE(match("a/x/y/b", "a/**b"));
+}
+
 } // namespace
 } // namespace format
 } // namespace clang