Add `StringRef::trim()` functions

Add three functions that trim characters from the front & end of a
`StringRef`. All functions return a new `StringRef` that references a
sub-string of the original `StringRef`.

- `trim(chars_to_remove)`: strips all characters from the start and end
  that occur in `chars_to_remove`.
- `trim(char_to_remove)`: same, but with a single character to remove.
- `trim()`: remove leading & trailing whitespace, so same as
  `trim(" \r\n\t")`

Reviewed By: JacquesLucke

Differential Revision: https://developer.blender.org/D12031

2 files changed, 80 insertions, 0 deletions

diff --git a/source/blender/blenlib/BLI_string_ref.hh b/source/blender/blenlib/BLI_string_ref.hh
index 0cff8fa7fb4..257a0ba143e 100644
--- a/source/blender/blenlib/BLI_string_ref.hh
+++ b/source/blender/blenlib/BLI_string_ref.hh
@@ -208,6 +208,18 @@ class StringRefBase {
   constexpr int64_t find_first_not_of(char c, int64_t pos = 0) const;
   constexpr int64_t find_last_not_of(StringRef chars, int64_t pos = INT64_MAX) const;
   constexpr int64_t find_last_not_of(char c, int64_t pos = INT64_MAX) const;
+
+  /**
+   * Return a new StringRef that does not contain leading and trailing whitespace.
+   */
+  constexpr StringRef trim() const;
+
+  /**
+   * Return a new StringRef that removes all the leading and trailing characters
+   * that occur in `characters_to_remove`.
+   */
+  constexpr StringRef trim(StringRef characters_to_remove) const;
+  constexpr StringRef trim(char character_to_remove) const;
 };
 
 /**
@@ -540,4 +552,34 @@ constexpr inline int64_t StringRefBase::find_last_not_of(char c, int64_t pos) co
   return this->find_last_not_of(StringRef(&c, 1), pos);
 }
 
+constexpr StringRef StringRefBase::trim() const
+{
+  return this->trim(" \t\r\n");
+}
+
+constexpr StringRef StringRefBase::trim(const char character_to_remove) const
+{
+  return this->trim(StringRef(&character_to_remove, 1));
+}
+
+/**
+ * Return a new StringRef that removes all the leading and trailing characters
+ * that occur in `characters_to_remove`.
+ */
+constexpr StringRef StringRefBase::trim(StringRef characters_to_remove) const
+{
+  const int64_t find_front = this->find_first_not_of(characters_to_remove);
+  if (find_front == not_found) {
+    return StringRef();
+  }
+  const int64_t find_end = this->find_last_not_of(characters_to_remove);
+  /* `find_end` cannot be `not_found`, because that means the string is only
+   * `characters_to_remove`, in which case `find_front` would already have
+   * been `not_found`. */
+  BLI_assert_msg(find_end != not_found,
+                 "forward search found characters-to-not-remove, but backward search did not");
+  const int64_t substr_len = find_end - find_front + 1;
+  return this->substr(find_front, substr_len);
+}
+
 }  // namespace blender
diff --git a/source/blender/blenlib/tests/BLI_string_ref_test.cc b/source/blender/blenlib/tests/BLI_string_ref_test.cc
index fb8b894bfd5..4ecea8031ca 100644
--- a/source/blender/blenlib/tests/BLI_string_ref_test.cc
+++ b/source/blender/blenlib/tests/BLI_string_ref_test.cc
@@ -278,6 +278,44 @@ TEST(string_ref, DropSuffixLargeN)
   EXPECT_EQ(ref2, "");
 }
 
+TEST(string_ref, TrimArbitrary)
+{
+  StringRef ref1("test");
+  StringRef ref2("   test ");
+  StringRef ref3(" \t  Urož with spaces ");
+  StringRef ref4("žžžžleepyžžž");
+  EXPECT_EQ(ref1.trim("t"), "es");
+  EXPECT_EQ(ref1.trim("te"), "s");
+  EXPECT_EQ(ref1.trim("test"), "");
+  EXPECT_EQ(ref2.trim("t"), "   test ");
+  EXPECT_EQ(ref2.trim(""), "   test ");
+  EXPECT_EQ(ref3.trim(" "), "\t  Urož with spaces"); /* TAB should be kept. */
+  EXPECT_EQ(ref4.trim("ž"), "leepy");
+}
+
+TEST(string_ref, TrimWhitespace)
+{
+  StringRef ref1("test");
+  StringRef ref2("   test ");
+  StringRef ref3(" \t  Urož with spaces ");
+  StringRef ref4(" \t \n\r  \t ");
+  EXPECT_EQ(ref1.trim(), "test");
+  EXPECT_EQ(ref2.trim(), "test");
+  EXPECT_EQ(ref3.trim(), "Urož with spaces");
+  EXPECT_EQ(ref4.trim(), "");
+}
+
+TEST(string_ref, TrimCharacter)
+{
+  StringRef ref1("test");
+  StringRef ref2("   test ");
+  StringRef ref3("does this work?");
+  EXPECT_EQ(ref1.trim('t'), "es");
+  EXPECT_EQ(ref1.trim('p'), "test");
+  EXPECT_EQ(ref2.trim(' '), "test");
+  EXPECT_EQ(ref3.trim('\000'), "does this work?");
+}
+
 TEST(string_ref, Substr)
 {
   StringRef ref("hello world");