Reland "[libc] Implement basename and dirname in libgen.h #204554" (#205352)

Added the POSIX standard functions basename and dirname under a new libgen.h header. The implementations modify the input path in-place using cpp::string_view to determine boundaries safely.

Added find_last_not_of to cpp::string_view to support trailing slash removal.

Implemented:

libc/include/libgen.yaml, libgen.h.def: Public API definitions.
libc/src/libgen/basename.cpp, dirname.cpp: Generic implementations.
libc/test/src/libgen/: Unit and hermetic tests.
Registered the new entrypoints for all active Linux targets (x86_64, aarch64, arm, riscv) and added docgen configuration.

The tests are skipped when using ASan because death tests do not currently work with them.

Assisted-by: Automated tooling, human reviewed.

libc/config/linux/aarch64/entrypoints.txt

@@ -109,6 +109,10 @@ set(TARGET_LIBC_ENTRYPOINTS
     libc.src.inttypes.strtoimax
     libc.src.inttypes.strtoumax
 
+    # libgen.h entrypoints
+    libc.src.libgen.basename
+    libc.src.libgen.dirname
+
     # stdbit.h entrypoints
     libc.src.stdbit.stdc_bit_ceil_uc
     libc.src.stdbit.stdc_bit_ceil_ui

libc/config/linux/arm/entrypoints.txt

@@ -73,6 +73,10 @@ set(TARGET_LIBC_ENTRYPOINTS
     libc.src.inttypes.strtoimax
     libc.src.inttypes.strtoumax
 
+    # libgen.h entrypoints
+    libc.src.libgen.basename
+    libc.src.libgen.dirname
+
     # stdbit.h entrypoints
     libc.src.stdbit.stdc_bit_ceil_uc
     libc.src.stdbit.stdc_bit_ceil_ui

libc/config/linux/riscv/entrypoints.txt

@@ -109,6 +109,10 @@ set(TARGET_LIBC_ENTRYPOINTS
     libc.src.inttypes.strtoimax
     libc.src.inttypes.strtoumax
 
+    # libgen.h entrypoints
+    libc.src.libgen.basename
+    libc.src.libgen.dirname
+
     # stdbit.h entrypoints
     libc.src.stdbit.stdc_bit_ceil_uc
     libc.src.stdbit.stdc_bit_ceil_ui

libc/config/linux/x86_64/entrypoints.txt

@@ -125,6 +125,10 @@ set(TARGET_LIBC_ENTRYPOINTS
     libc.src.inttypes.wcstoimax
     libc.src.inttypes.wcstoumax
 
+    # libgen.h entrypoints
+    libc.src.libgen.basename
+    libc.src.libgen.dirname
+
     # stdbit.h entrypoints
     libc.src.stdbit.stdc_bit_ceil_uc
     libc.src.stdbit.stdc_bit_ceil_ui

libc/docs/CMakeLists.txt

@@ -54,6 +54,7 @@ if (SPHINX_FOUND)
       float
       glob
       inttypes
+      libgen
       locale
       nl_types
       net/if

libc/docs/headers/index.rst

@@ -20,6 +20,7 @@ Implementation Status
    float
    glob
    inttypes
+   libgen
    locale
    math/index.rst
    net/if

libc/include/CMakeLists.txt

@@ -130,6 +130,14 @@ add_header_macro(
     .llvm-libc-macros.float_macros
 )
 
+add_header_macro(
+  libgen
+  ../libc/include/libgen.yaml
+  libgen.h
+  DEPENDS
+    .llvm_libc_common_h
+)
+
 add_header_macro(
   limits
   ../libc/include/limits.yaml

libc/include/libgen.yaml

@@ -0,0 +1,20 @@
+header: libgen.h
+standards:
+  - posix
+macros: []
+types: []
+enums: []
+objects: []
+functions:
+  - name: basename
+    standards:
+      - posix
+    return_type: char *
+    arguments:
+      - type: char *
+  - name: dirname
+    standards:
+      - posix
+    return_type: char *
+    arguments:
+      - type: char *

libc/src/CMakeLists.txt

@@ -7,6 +7,7 @@ add_subdirectory(dlfcn)
 add_subdirectory(errno)
 add_subdirectory(fenv)
 add_subdirectory(inttypes)
+add_subdirectory(libgen)
 add_subdirectory(link)
 add_subdirectory(math)
 add_subdirectory(netinet)

libc/src/__support/CPP/string_view.h

@@ -205,6 +205,15 @@ public:
     return npos;
   }
 
+  LIBC_INLINE constexpr size_t find_last_not_of(const char c,
+                                                size_t end = npos) const {
+    end = end >= size() ? size() : end + 1;
+    for (; end > 0; --end)
+      if ((*this)[end - 1] != c)
+        return end - 1;
+    return npos;
+  }
+
   // Finds the first character not equal to c in this view, starting at
   // position From.
   LIBC_INLINE constexpr size_t find_first_not_of(const char c,

libc/src/libgen/CMakeLists.txt

@@ -0,0 +1,23 @@
+add_entrypoint_object(
+  basename
+  SRCS
+    basename.cpp
+  HDRS
+    basename.h
+  DEPENDS
+    libc.src.__support.CPP.string_view
+    libc.src.__support.common
+    libc.src.__support.macros.config
+)
+
+add_entrypoint_object(
+  dirname
+  SRCS
+    dirname.cpp
+  HDRS
+    dirname.h
+  DEPENDS
+    libc.src.__support.CPP.string_view
+    libc.src.__support.common
+    libc.src.__support.macros.config
+)

libc/src/libgen/basename.cpp

@@ -0,0 +1,42 @@
+//===----------------------------------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+///
+/// \file
+/// Implementation of basename.
+///
+//===----------------------------------------------------------------------===//
+
+#include "src/libgen/basename.h"
+#include "src/__support/CPP/string_view.h"
+#include "src/__support/common.h"
+#include "src/__support/macros/config.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+LLVM_LIBC_FUNCTION(char *, basename, (char *path)) {
+  if (path == nullptr || path[0] == '\0')
+    return const_cast<char *>(".");
+
+  cpp::string_view sv(path);
+  size_t last_non_slash = sv.find_last_not_of('/');
+
+  if (last_non_slash == cpp::string_view::npos)
+    return const_cast<char *>("/");
+
+  size_t last_slash = sv.substr(0, last_non_slash).find_last_of('/');
+
+  size_t start = (last_slash == cpp::string_view::npos) ? 0 : last_slash + 1;
+  size_t end = last_non_slash + 1;
+
+  if (end < sv.size())
+    path[end] = '\0';
+
+  return path + start;
+}
+
+} // namespace LIBC_NAMESPACE_DECL

libc/src/libgen/basename.h

@@ -0,0 +1,30 @@
+//===----------------------------------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+///
+/// \file
+/// Header for basename.
+///
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_LIBC_SRC_LIBGEN_BASENAME_H
+#define LLVM_LIBC_SRC_LIBGEN_BASENAME_H
+
+#include "src/__support/macros/config.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+/// Return the last component of a pathname.
+///
+/// \param path Pointer to the null-terminated pathname string.
+/// \return Pointer to the last component of path, or "." if path is null or
+/// empty, or "/" if path is all slashes.
+char *basename(char *path);
+
+} // namespace LIBC_NAMESPACE_DECL
+
+#endif // LLVM_LIBC_SRC_LIBGEN_BASENAME_H

libc/src/libgen/dirname.cpp

@@ -0,0 +1,48 @@
+//===----------------------------------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+///
+/// \file
+/// Implementation of dirname.
+///
+//===----------------------------------------------------------------------===//
+
+#include "src/libgen/dirname.h"
+#include "src/__support/CPP/string_view.h"
+#include "src/__support/common.h"
+#include "src/__support/macros/config.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+LLVM_LIBC_FUNCTION(char *, dirname, (char *path)) {
+  if (path == nullptr || path[0] == '\0')
+    return const_cast<char *>(".");
+
+  cpp::string_view sv(path);
+  size_t last_non_slash = sv.find_last_not_of('/');
+
+  if (last_non_slash == cpp::string_view::npos)
+    return const_cast<char *>("/");
+
+  size_t last_slash = sv.substr(0, last_non_slash).find_last_of('/');
+
+  if (last_slash == cpp::string_view::npos)
+    return const_cast<char *>(".");
+
+  cpp::string_view dir_sv = sv.substr(0, last_slash);
+  size_t dir_last_non_slash = dir_sv.find_last_not_of('/');
+
+  if (dir_last_non_slash == cpp::string_view::npos) {
+    path[1] = '\0';
+    return path;
+  }
+
+  path[dir_last_non_slash + 1] = '\0';
+  return path;
+}
+
+} // namespace LIBC_NAMESPACE_DECL

libc/src/libgen/dirname.h

@@ -0,0 +1,30 @@
+//===----------------------------------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+///
+/// \file
+/// Header for dirname.
+///
+//===----------------------------------------------------------------------===//
+
+#ifndef LLVM_LIBC_SRC_LIBGEN_DIRNAME_H
+#define LLVM_LIBC_SRC_LIBGEN_DIRNAME_H
+
+#include "src/__support/macros/config.h"
+
+namespace LIBC_NAMESPACE_DECL {
+
+/// Return the directory component of a pathname.
+///
+/// \param path Pointer to the null-terminated pathname string.
+/// \return Pointer to the directory component of path, or "." if path is null
+/// or empty, or "/" if path is all slashes.
+char *dirname(char *path);
+
+} // namespace LIBC_NAMESPACE_DECL
+
+#endif // LLVM_LIBC_SRC_LIBGEN_DIRNAME_H

libc/test/src/CMakeLists.txt

@@ -63,6 +63,7 @@ add_subdirectory(complex)
 add_subdirectory(ctype)
 add_subdirectory(errno)
 add_subdirectory(fenv)
+add_subdirectory(libgen)
 add_subdirectory(link)
 add_subdirectory(math)
 add_subdirectory(netinet)

libc/test/src/libgen/CMakeLists.txt

@@ -0,0 +1,45 @@
+add_custom_target(libc-libgen-tests)
+
+add_libc_test(
+  basename_test
+  SUITE
+    libc-libgen-tests
+  SRCS
+    basename_test.cpp
+  DEPENDS
+    libc.src.libgen.basename
+)
+
+add_libc_test(
+  dirname_test
+  SUITE
+    libc-libgen-tests
+  SRCS
+    dirname_test.cpp
+  DEPENDS
+    libc.src.libgen.dirname
+)
+
+if (NOT LLVM_USE_SANITIZER)
+  add_libc_test(
+    basename_death_test
+    UNIT_TEST_ONLY
+    SUITE
+      libc-libgen-tests
+    SRCS
+      basename_death_test.cpp
+    DEPENDS
+      libc.src.libgen.basename
+  )
+
+  add_libc_test(
+    dirname_death_test
+    UNIT_TEST_ONLY
+    SUITE
+      libc-libgen-tests
+    SRCS
+      dirname_death_test.cpp
+    DEPENDS
+      libc.src.libgen.dirname
+  )
+endif()

libc/test/src/libgen/basename_death_test.cpp

@@ -0,0 +1,22 @@
+//===----------------------------------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+///
+/// \file
+/// Death tests for basename.
+///
+//===----------------------------------------------------------------------===//
+
+#include "src/libgen/basename.h"
+#include "test/UnitTest/Test.h"
+
+#ifdef ENABLE_SUBPROCESS_TESTS
+TEST(LlvmLibcBasenameTest, ModifyReturnValue) {
+  char *r = LIBC_NAMESPACE::basename(nullptr);
+  ASSERT_DEATH([r]() { r[0] = 'a'; }, WITH_SIGNAL(-1));
+}
+#endif

libc/test/src/libgen/basename_test.cpp

@@ -0,0 +1,62 @@
+//===----------------------------------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+///
+/// \file
+/// Unittests for basename.
+///
+//===----------------------------------------------------------------------===//
+
+#include "src/libgen/basename.h"
+#include "test/UnitTest/Test.h"
+
+TEST(LlvmLibcBasenameTest, NullPointer) {
+  ASSERT_STREQ(LIBC_NAMESPACE::basename(nullptr), ".");
+}
+
+TEST(LlvmLibcBasenameTest, EmptyString) {
+  char path[] = "";
+  ASSERT_STREQ(LIBC_NAMESPACE::basename(path), ".");
+}
+
+TEST(LlvmLibcBasenameTest, RegularPath) {
+  char path[] = "/usr/lib";
+  ASSERT_STREQ(LIBC_NAMESPACE::basename(path), "lib");
+}
+
+TEST(LlvmLibcBasenameTest, TrailingSlash) {
+  char path[] = "/usr/";
+  ASSERT_STREQ(LIBC_NAMESPACE::basename(path), "usr");
+  ASSERT_STREQ(path, "/usr");
+}
+
+TEST(LlvmLibcBasenameTest, SingleSlash) {
+  char path[] = "/";
+  ASSERT_STREQ(LIBC_NAMESPACE::basename(path), "/");
+}
+
+TEST(LlvmLibcBasenameTest, MultipleSlashes) {
+  char path[] = "///";
+  ASSERT_STREQ(LIBC_NAMESPACE::basename(path), "/");
+}
+
+TEST(LlvmLibcBasenameTest, SimpleName) {
+  char path[] = "a";
+  ASSERT_STREQ(LIBC_NAMESPACE::basename(path), "a");
+}
+
+TEST(LlvmLibcBasenameTest, SimpleNameTrailingSlash) {
+  char path[] = "a/";
+  ASSERT_STREQ(LIBC_NAMESPACE::basename(path), "a");
+  ASSERT_STREQ(path, "a");
+}
+
+TEST(LlvmLibcBasenameTest, ComplexPath) {
+  char path[] = "///a///";
+  ASSERT_STREQ(LIBC_NAMESPACE::basename(path), "a");
+  ASSERT_STREQ(path, "///a");
+}

libc/test/src/libgen/dirname_death_test.cpp

@@ -0,0 +1,22 @@
+//===----------------------------------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+///
+/// \file
+/// Death tests for dirname.
+///
+//===----------------------------------------------------------------------===//
+
+#include "src/libgen/dirname.h"
+#include "test/UnitTest/Test.h"
+
+#ifdef ENABLE_SUBPROCESS_TESTS
+TEST(LlvmLibcDirnameTest, ModifyReturnValue) {
+  char *r = LIBC_NAMESPACE::dirname(nullptr);
+  ASSERT_DEATH([r]() { r[0] = 'a'; }, WITH_SIGNAL(-1));
+}
+#endif

libc/test/src/libgen/dirname_test.cpp

@@ -0,0 +1,74 @@
+//===----------------------------------------------------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+///
+/// \file
+/// Unittests for dirname.
+///
+//===----------------------------------------------------------------------===//
+
+#include "src/libgen/dirname.h"
+#include "test/UnitTest/Test.h"
+
+TEST(LlvmLibcDirnameTest, NullPointer) {
+  ASSERT_STREQ(LIBC_NAMESPACE::dirname(nullptr), ".");
+}
+
+TEST(LlvmLibcDirnameTest, EmptyString) {
+  char path[] = "";
+  ASSERT_STREQ(LIBC_NAMESPACE::dirname(path), ".");
+}
+
+TEST(LlvmLibcDirnameTest, RegularPath) {
+  char path[] = "/usr/lib";
+  ASSERT_STREQ(LIBC_NAMESPACE::dirname(path), "/usr");
+  ASSERT_STREQ(path, "/usr");
+}
+
+TEST(LlvmLibcDirnameTest, TrailingSlash) {
+  char path[] = "/usr/";
+  ASSERT_STREQ(LIBC_NAMESPACE::dirname(path), "/");
+  ASSERT_STREQ(path, "/");
+}
+
+TEST(LlvmLibcDirnameTest, SingleSlash) {
+  char path[] = "/";
+  ASSERT_STREQ(LIBC_NAMESPACE::dirname(path), "/");
+}
+
+TEST(LlvmLibcDirnameTest, MultipleSlashes) {
+  char path[] = "///";
+  ASSERT_STREQ(LIBC_NAMESPACE::dirname(path), "/");
+}
+
+TEST(LlvmLibcDirnameTest, SimpleName) {
+  char path[] = "a";
+  ASSERT_STREQ(LIBC_NAMESPACE::dirname(path), ".");
+}
+
+TEST(LlvmLibcDirnameTest, SimpleNameTrailingSlash) {
+  char path[] = "a/";
+  ASSERT_STREQ(LIBC_NAMESPACE::dirname(path), ".");
+}
+
+TEST(LlvmLibcDirnameTest, ComplexPath) {
+  char path[] = "///a///b///";
+  ASSERT_STREQ(LIBC_NAMESPACE::dirname(path), "///a");
+  ASSERT_STREQ(path, "///a");
+}
+
+TEST(LlvmLibcDirnameTest, SlashA) {
+  char path[] = "/a";
+  ASSERT_STREQ(LIBC_NAMESPACE::dirname(path), "/");
+  ASSERT_STREQ(path, "/");
+}
+
+TEST(LlvmLibcDirnameTest, MultipleSlashesA) {
+  char path[] = "///a";
+  ASSERT_STREQ(LIBC_NAMESPACE::dirname(path), "/");
+  ASSERT_STREQ(path, "/");
+}

libc/utils/docgen/libgen.yaml

@@ -0,0 +1,5 @@
+functions:
+  basename:
+    in-latest-posix: ''
+  dirname:
+    in-latest-posix: ''