Refactor: Use absl::ascii_* functions and improve string parsing.

majnemer · tensorflower-gardener · commit efab298fb7da · 2025-11-07T15:57:33.000-08:00
This change replaces manual character range checks (e.g., `c &gt;= '0' &amp;&amp; c &lt;= '9'`) with more idiomatic and potentially safer `absl::ascii_isdigit`, `absl::ascii_isalpha`, and `absl::ascii_isalnum` functions. Additionally, it refactors `ParseTensorName` to use `absl::string_view` methods and `tsl::str_util::ConsumeLeadingDigits`, and updates `ConsumeLeadingDigits` to use `std::from_chars`. The unused `Strnlen` function is also removed.

PiperOrigin-RevId: 829600440
diff --git a/tensorflow/core/framework/ops_util.cc b/tensorflow/core/framework/ops_util.cc
@@ -13,11 +13,13 @@ See the License for the specific language governing permissions and
 limitations under the License.
 ==============================================================================*/
 
+#include "tensorflow/core/framework/ops_util.h"
+
 #include <algorithm>
 #include <cmath>
 
+#include "absl/strings/ascii.h"
 #include "tensorflow/core/framework/attr_value.pb.h"
-#include "tensorflow/core/framework/ops_util.h"
 #include "tensorflow/core/lib/core/errors.h"
 #include "tensorflow/core/lib/strings/str_util.h"
 #include "tensorflow/core/util/padding.h"
@@ -66,8 +68,7 @@ string SanitizeThreadSuffix(string suffix) {
   string clean;
   for (int i = 0; i < suffix.size(); ++i) {
     const char ch = suffix[i];
-    if ((ch >= 'a' && ch <= 'z') || (ch >= 'A' && ch <= 'Z') ||
-        (ch >= '0' && ch <= '9') || ch == '_' || ch == '-') {
+    if (absl::ascii_isalnum(ch) || ch == '_' || ch == '-') {
       clean += ch;
     } else {
       clean += '_';
diff --git a/tensorflow/core/graph/tensor_id.cc b/tensorflow/core/graph/tensor_id.cc
@@ -15,21 +15,21 @@ limitations under the License.
 
 #include "tensorflow/core/graph/tensor_id.h"
 
+#include <cstddef>
+#include <cstdint>
 #include <string>
 
-#include "tensorflow/core/lib/core/stringpiece.h"
-#include "tensorflow/core/lib/strings/str_util.h"
+#include "absl/strings/string_view.h"
+#include "absl/strings/strip.h"
+#include "tensorflow/core/graph/graph.h"
+#include "tensorflow/core/platform/str_util.h"
 
 namespace tensorflow {
 
 TensorId::TensorId(const SafeTensorId& id) : TensorId(id.first, id.second) {}
 
 SafeTensorId::SafeTensorId(const TensorId& id)
-    : SafeTensorId(string(id.first), id.second) {}
-
-TensorId ParseTensorName(const string& name) {
-  return ParseTensorName(absl::string_view(name.data(), name.size()));
-}
+    : SafeTensorId(std::string(id.first), id.second) {}
 
 TensorId ParseTensorName(absl::string_view name) {
   // Parse either a name, ^name, or name:digits.  To do so, we go backwards from
@@ -38,28 +38,19 @@ TensorId ParseTensorName(absl::string_view name) {
   // see if the name starts with '^', indicating a control edge. If we find
   // neither ':' nor '^' characters, the output index is implicitly 0, and the
   // whole name string forms the first part of the tensor name.
-  const char* base = name.data();
-  const char* p = base + name.size() - 1;
-  unsigned int index = 0;
-  unsigned int mul = 1;
-  while (p > base && (*p >= '0' && *p <= '9')) {
-    index += ((*p - '0') * mul);
-    mul *= 10;
-    p--;
+  size_t colon_pos = name.rfind(':');
+  if (colon_pos != absl::string_view::npos) {
+    absl::string_view prefix = name.substr(0, colon_pos);
+    absl::string_view suffix = name.substr(colon_pos + 1);
+    uint64_t index;
+    if (str_util::ConsumeLeadingDigits(&suffix, &index) && suffix.empty()) {
+      return TensorId(prefix, index);
+    }
   }
-  TensorId id;
-  if (p > base && *p == ':' && mul > 1) {
-    id.first = absl::string_view(base, p - base);
-    id.second = index;
-  } else if (absl::StartsWith(name, "^")) {
-    // Control edge
-    id.first = absl::string_view(base + 1);
-    id.second = Graph::kControlSlot;
-  } else {
-    id.first = name;
-    id.second = 0;
+  if (absl::ConsumePrefix(&name, "^")) {
+    return TensorId(name, Graph::kControlSlot);
   }
-  return id;
+  return TensorId(name, 0);
 }
 
 bool IsTensorIdControl(const TensorId& tensor_id) {
diff --git a/tensorflow/core/graph/tensor_id.h b/tensorflow/core/graph/tensor_id.h
@@ -57,7 +57,6 @@ struct TensorId : public std::pair<absl::string_view, int> {
   };
 };
 
-TensorId ParseTensorName(const string& name);
 TensorId ParseTensorName(absl::string_view name);
 
 bool IsTensorIdControl(const TensorId& tensor_id);
diff --git a/tensorflow/core/lib/db/BUILD b/tensorflow/core/lib/db/BUILD
@@ -25,6 +25,7 @@ cc_library(
         "@com_google_absl//absl/log",
         "@com_google_absl//absl/log:check",
         "@com_google_absl//absl/status",
+        "@com_google_absl//absl/strings",
         "@com_google_absl//absl/strings:str_format",
         "@org_sqlite",
     ],
diff --git a/tensorflow/core/lib/db/sqlite.cc b/tensorflow/core/lib/db/sqlite.cc
@@ -19,6 +19,7 @@ limitations under the License.
 #include "absl/log/check.h"
 #include "absl/log/log.h"
 #include "absl/status/status.h"
+#include "absl/strings/ascii.h"
 #include "absl/strings/str_format.h"
 #include "tensorflow/core/lib/core/errors.h"
 #include "tensorflow/core/platform/status.h"
@@ -99,8 +100,7 @@ absl::Status SetPragma(Sqlite* db, const char* pragma,
                        const absl::string_view& value) {
   if (value.empty()) return absl::OkStatus();
   for (auto p = value.begin(); p < value.end(); ++p) {
-    if (!(('0' <= *p && *p <= '9') || ('A' <= *p && *p <= 'Z') ||
-          ('a' <= *p && *p <= 'z') || *p == '-')) {
+    if (!(absl::ascii_isalnum(*p) || *p == '-')) {
       return errors::InvalidArgument("Illegal pragma character");
     }
   }
diff --git a/tensorflow/core/platform/str_util.h b/tensorflow/core/platform/str_util.h
@@ -51,7 +51,6 @@ using tsl::str_util::StringReplace;
 using tsl::str_util::StripPrefix;
 using tsl::str_util::StripSuffix;
 using tsl::str_util::StripTrailingWhitespace;
-using tsl::str_util::Strnlen;
 using tsl::str_util::TitlecaseString;
 using tsl::str_util::Uppercase;
 // NOLINTEND(misc-unused-using-decls)
diff --git a/tensorflow/core/util/memmapped_file_system.cc b/tensorflow/core/util/memmapped_file_system.cc
@@ -269,8 +269,7 @@ bool MemmappedFileSystem::IsMemmappedPackageFilename(
 
 namespace {
 bool IsValidRegionChar(char c) {
-  return (c >= 'A' && c <= 'Z') || (c >= 'a' && c <= 'z') ||
-         (c >= '0' && c <= '9') || c == '_' || c == '.';
+  return absl::ascii_isalnum(c) || c == '_' || c == '.';
 }
 }  // namespace
 
diff --git a/tensorflow/core/util/semver_test.cc b/tensorflow/core/util/semver_test.cc
@@ -15,6 +15,7 @@ limitations under the License.
 
 #include <cstddef>
 
+#include "absl/strings/ascii.h"
 #include "absl/strings/strip.h"
 #include "tensorflow/core/lib/core/stringpiece.h"
 #include "tensorflow/core/lib/strings/str_util.h"
@@ -27,10 +28,7 @@ namespace {
 bool IsDotOrIdentifierChar(char c) {
   if (c == '.') return true;
   if (c == '-') return true;
-  if (c >= 'A' && c <= 'Z') return true;
-  if (c >= 'a' && c <= 'z') return true;
-  if (c >= '0' && c <= '9') return true;
-  return false;
+  return absl::ascii_isalnum(c);
 }
 
 bool ConsumeDotSeparatedIdentifiers(absl::string_view* s,
diff --git a/tensorflow/security/fuzzing/cc/BUILD b/tensorflow/security/fuzzing/cc/BUILD
@@ -33,6 +33,7 @@ tf_cc_fuzz_test(
     deps = [
         "//tensorflow/core/platform:str_util",
         "//tensorflow/core/platform:stringpiece",
+        "@com_google_absl//absl/strings",
     ],
 )
 
diff --git a/tensorflow/security/fuzzing/cc/arg_def_case_fuzz.cc b/tensorflow/security/fuzzing/cc/arg_def_case_fuzz.cc
@@ -17,6 +17,7 @@ limitations under the License.
 #include <string_view>
 
 #include "fuzztest/fuzztest.h"
+#include "absl/strings/ascii.h"
 #include "tensorflow/core/platform/str_util.h"
 #include "tensorflow/core/platform/stringpiece.h"
 
@@ -27,8 +28,8 @@ namespace {
 void FuzzTest(std::string_view data) {
   std::string ns = tensorflow::str_util::ArgDefCase(data);
   for (const auto &c : ns) {
-    const bool is_letter = 'a' <= c && c <= 'z';
-    const bool is_digit = '0' <= c && c <= '9';
+    const bool is_letter = absl::ascii_isalpha(c);
+    const bool is_digit = absl::ascii_isdigit(c);
     if (!is_letter && !is_digit) {
       assert(c == '_');
     }
diff --git a/third_party/xla/third_party/tsl/tsl/platform/str_util.cc b/third_party/xla/third_party/tsl/tsl/platform/str_util.cc
@@ -15,10 +15,11 @@ limitations under the License.
 
 #include "tsl/platform/str_util.h"
 
-#include <cctype>
+#include <charconv>
 #include <cstddef>
 #include <cstdint>
 #include <string>
+#include <system_error>  // NOLINT
 
 #include "absl/strings/ascii.h"
 #include "absl/strings/match.h"
@@ -50,28 +51,16 @@ size_t RemoveWhitespaceContext(absl::string_view* text) {
 }
 
 bool ConsumeLeadingDigits(absl::string_view* s, uint64_t* val) {
-  const char* p = s->data();
-  const char* limit = p + s->size();
-  uint64_t v = 0;
-  while (p < limit) {
-    const char c = *p;
-    if (c < '0' || c > '9') break;
-    uint64_t new_v = (v * 10) + (c - '0');
-    if (new_v / 8 < v) {
-      // Overflow occurred
-      return false;
-    }
-    v = new_v;
-    p++;
-  }
-  if (p > s->data()) {
-    // Consume some digits
-    s->remove_prefix(p - s->data());
-    *val = v;
-    return true;
-  } else {
+  uint64_t v;
+  auto [p, ec] =
+      std::from_chars(s->data(), s->data() + s->size(), v, /*base=*/10);
+  if (ec != std::errc{}) {
     return false;
   }
+  // Consume some digits
+  s->remove_prefix(p - s->data());
+  *val = v;
+  return true;
 }
 
 bool ConsumeNonWhitespace(absl::string_view* s, absl::string_view* val) {
@@ -122,14 +111,6 @@ string StringReplace(absl::string_view s, absl::string_view oldsub,
   return res;
 }
 
-size_t Strnlen(const char* str, const size_t string_max_len) {
-  size_t len = 0;
-  while (len < string_max_len && str[len] != '\0') {
-    ++len;
-  }
-  return len;
-}
-
 string ArgDefCase(absl::string_view s) {
   const size_t n = s.size();
 
diff --git a/third_party/xla/third_party/tsl/tsl/platform/str_util.h b/third_party/xla/third_party/tsl/tsl/platform/str_util.h
@@ -210,11 +210,6 @@ inline bool StrContains(absl::string_view haystack, absl::string_view needle) {
   return absl::StrContains(haystack, needle);
 }
 
-// Returns the length of the given null-terminated byte string 'str'.
-// Returns 'string_max_len' if the null character was not found in the first
-// 'string_max_len' bytes of 'str'.
-size_t Strnlen(const char* str, const size_t string_max_len);
-
 //   ----- NON STANDARD, TF SPECIFIC METHOD -----
 // Converts "^2ILoveYou!" to "i_love_you_". More specifically:
 // - converts all non-alphanumeric characters to underscores
diff --git a/third_party/xla/third_party/tsl/tsl/platform/str_util_test.cc b/third_party/xla/third_party/tsl/tsl/platform/str_util_test.cc
@@ -367,12 +367,4 @@ TEST(StringReplace, EmptyStringReplaceAll) {
   EXPECT_EQ("", str_util::StringReplace("", "a", "X", /*replace_all=*/true));
 }
 
-TEST(Strnlen, Basic) {
-  EXPECT_EQ(0, str_util::Strnlen("ab", 0));
-  EXPECT_EQ(1, str_util::Strnlen("a", 1));
-  EXPECT_EQ(2, str_util::Strnlen("abcd", 2));
-  EXPECT_EQ(3, str_util::Strnlen("abc", 10));
-  EXPECT_EQ(4, str_util::Strnlen("a \t\n", 10));
-}
-
 }  // namespace tsl
diff --git a/third_party/xla/xla/service/llvm_ir/llvm_util.cc b/third_party/xla/xla/service/llvm_ir/llvm_util.cc
@@ -31,6 +31,7 @@ limitations under the License.
 #include "absl/log/check.h"
 #include "absl/log/log.h"
 #include "absl/status/statusor.h"
+#include "absl/strings/ascii.h"
 #include "absl/strings/str_cat.h"
 #include "absl/strings/str_format.h"
 #include "absl/strings/string_view.h"
@@ -626,23 +627,20 @@ std::string SanitizeFunctionName(std::string function_name) {
   // are illegal.
 
   // Sanitize chars in function_name.
-  std::transform(function_name.begin(), function_name.end(),
-                 function_name.begin(), [](char c) {
-                   if (('a' <= c && c <= 'z') || ('A' <= c && c <= 'Z') ||
-                       ('0' <= c && c <= '9') || c == '_' || c == '$') {
-                     return c;
-                   }
-                   return '_';
-                 });
+  absl::c_transform(function_name, function_name.begin(), [](char c) {
+    if (absl::ascii_isalnum(c) || c == '_' || c == '$') {
+      return c;
+    }
+    return '_';
+  });
 
   // Ensure the name isn't empty.
   if (function_name.empty()) {
     function_name = "__unnamed";
   }
 
   // Ensure the name doesn't start with a number.
-  if (!function_name.empty() && function_name[0] >= '0' &&
-      function_name[0] <= '9') {
+  if (!function_name.empty() && absl::ascii_isdigit(function_name[0])) {
     function_name.insert(function_name.begin(), '_');
   }
 
diff --git a/third_party/xla/xla/sort_json.cc b/third_party/xla/xla/sort_json.cc
@@ -104,7 +104,7 @@ absl::StatusOr<std::unique_ptr<T>> ParseSequence(absl::string_view outer_json,
 
 absl::Status EnsureValidLiteralStart(char c) {
   if (c != '"' && c != '+' && c != '-' && c != 'f' && c != 't' && c != 'n' &&
-      (c < '0' || c > '9')) {
+      !absl::ascii_isdigit(c)) {
     return absl::InvalidArgumentError(absl::StrCat(
         "Invalid first character of literal: '", std::string(1, c), "'."));
   }
diff --git a/third_party/xla/xla/tsl/util/BUILD b/third_party/xla/xla/tsl/util/BUILD
@@ -278,6 +278,8 @@ cc_library(
     deps = [
         "//xla/tsl/platform:errors",
         "//xla/tsl/platform:status",
+        "@com_google_absl//absl/strings",
+        "@com_google_absl//absl/strings:string_view",
         "@local_tsl//tsl/platform:str_util",
         "@local_tsl//tsl/platform:stringpiece",
     ],
diff --git a/third_party/xla/xla/tsl/util/device_name_utils.cc b/third_party/xla/xla/tsl/util/device_name_utils.cc

Original file line number	Diff line number	Diff line change
`@@ -269,8 +269,7 @@ bool MemmappedFileSystem::IsMemmappedPackageFilename(`
`269`	`269`
`270`	`270`	`namespace {`
`271`	`271`	`bool IsValidRegionChar(char c) {`
`272`		`- return (c >= 'A' && c <= 'Z') \|\| (c >= 'a' && c <= 'z') \|\|`
`273`		`- (c >= '0' && c <= '9') \|\| c == '_' \|\| c == '.';`
	`272`	`+ return absl::ascii_isalnum(c) \|\| c == '_' \|\| c == '.';`
`274`	`273`	`}`
`275`	`274`	`} // namespace`
`276`	`275`
Original file line number	Diff line number	Diff line change
`@@ -33,6 +33,7 @@ tf_cc_fuzz_test(`
`33`	`33`	`deps = [`
`34`	`34`	`"//tensorflow/core/platform:str_util",`
`35`	`35`	`"//tensorflow/core/platform:stringpiece",`
	`36`	`+ "@com_google_absl//absl/strings",`
`36`	`37`	`],`
`37`	`38`	`)`
`38`	`39`