optimize url::can_parse method

.clang-tidy

@@ -5,12 +5,13 @@ Checks: >
   -bugprone-narrowing-conversions,
   -bugprone-suspicious-include,
   -bugprone-unhandled-exception-at-new,
+  -bugprone-throwing-static-initialization,
   clang-analyzer-*,
   -clang-analyzer-security.insecureAPI.DeprecatedOrUnsafeBufferHandling,
 
 # Turn all the warnings from the checks above into errors.
 WarningsAsErrors: '*'
 # Check first-party (non-system, non-vendored) headers.
 HeaderFilterRegex: '.*'
-ExcludeHeaderFilterRegex: 'build/_deps/'
+ExcludeHeaderFilterRegex: '(build[^/]*/_deps/|\.cpm-cache/|/expected\.h$)'
 SystemHeaders: false

.github/workflows/abi-check.yml

@@ -37,8 +37,21 @@ jobs:
       - name: Find latest release tag
         id: baseline
         run: |
-          # Find the most recent vX.Y.Z tag reachable from the current commit's history
-          LATEST_TAG=$(git tag --list 'v*.*.*' --sort=-version:refname | head -1)
+          # Find the most recent vX.Y.Z tag that does NOT point to the current HEAD.
+          # Excluding HEAD ensures we compare against a previous release even when a
+          # new release tag was just pushed to main alongside this workflow run.
+          CURRENT_SHA=$(git rev-parse HEAD)
+          LATEST_TAG=$(git tag --merged HEAD --list 'v*.*.*' --sort=-version:refname | while IFS= read -r tag; do
+            TAG_SHA=$(git rev-parse "${tag}^{}" 2>/dev/null)
+            if [ "$TAG_SHA" != "$CURRENT_SHA" ]; then
+              echo "$tag"
+              break
+            fi
+          done)
+          if [ -z "$LATEST_TAG" ]; then
+            echo "No previous release tag found — cannot establish a baseline."
+            exit 1
+          fi
           echo "Latest release tag: $LATEST_TAG"
           echo "tag=$LATEST_TAG" >> "$GITHUB_OUTPUT"
 

.github/workflows/lint_and_format_check.yml

@@ -29,7 +29,7 @@ jobs:
       - name: Run clang-format
         uses: jidicula/clang-format-action@6cd220de46c89139a0365edae93eee8eb30ca8fe # v4.16.0
         with:
-          clang-format-version: '17'
+          clang-format-version: '22'
           fallback-style: 'Google'
 
       - uses: chartboost/ruff-action@e18ae971ccee1b2d7bbef113930f00c670b78da4 # v1.0.0
@@ -38,11 +38,15 @@ jobs:
           version: 0.6.0
 
       - name: Install clang-tidy and libc++
-        run: sudo apt-get update && sudo apt-get install -y clang-tidy-20 libc++-20-dev libc++abi-20-dev
+        run: |
+          wget -qO- https://apt.llvm.org/llvm-snapshot.gpg.key | sudo tee /etc/apt/trusted.gpg.d/apt.llvm.org.asc
+          echo "deb http://apt.llvm.org/noble/ llvm-toolchain-noble-22 main" | sudo tee /etc/apt/sources.list.d/llvm.list
+          sudo apt-get update
+          sudo apt-get install -y clang-tidy-22 libc++-22-dev libc++abi-22-dev
 
       - name: Run clang-tidy
         run: >
-          cmake -B build -DADA_TESTING=ON -DADA_USE_UNSAFE_STD_REGEX_PROVIDER=ON -DCMAKE_CXX_CLANG_TIDY=clang-tidy-20 -DCMAKE_EXPORT_COMPILE_COMMANDS=ON -DCMAKE_CXX_FLAGS="-stdlib=libc++" &&
+          cmake -B build -DADA_TESTING=ON -DADA_USE_UNSAFE_STD_REGEX_PROVIDER=ON -DCMAKE_CXX_CLANG_TIDY=clang-tidy-22 -DCMAKE_EXPORT_COMPILE_COMMANDS=ON -DCMAKE_CXX_FLAGS="-stdlib=libc++" &&
           cmake --build build -j=4
         env:
-          CXX: clang++-20
+          CXX: clang++-22

.gitignore

@@ -4,6 +4,9 @@ build-*/
 *-build-*
 _fuzz_check/
 
+# CPM package cache (used by tools/run-clangcldocker.sh docker fallback)
+.cpm-cache/
+
 # Python cache
 __pycache__
 venv

CLAUDE.md → AGENTS.md

@@ -2,6 +2,19 @@
 
 This guide provides instructions for building, testing, and benchmarking the Ada URL parser library using CMake.
 
+## Pre-commit Checklist
+
+Always run the clang-format and clang-tidy script before committing:
+
+```bash
+bash tools/run-clangcldocker.sh
+```
+
+This runs clang-format on all tracked source files and clang-tidy on `src/ada.cpp`
+(the single translation unit that includes all first-party code). The script uses
+the locally installed LLVM 22 toolchain when available, otherwise falls back to the
+`xianpengshen/clang-tools:22` Docker image automatically.
+
 ## Quick Reference
 
 ```bash

benchmarks/bench_protocol.cpp

@@ -21,8 +21,8 @@
 #include "counters/bench.h"
 
 template <class Function1, class Function2>
-counters::event_aggregate shuffle_bench(Function1 &&function1,
-                                        Function2 &&function2,
+counters::event_aggregate shuffle_bench(Function1&& function1,
+                                        Function2&& function2,
                                         size_t min_repeat = 300,
                                         size_t min_time_ns = 400'000'000,
                                         size_t max_repeat = 1000000,
@@ -77,7 +77,7 @@ constexpr uint64_t scheme_keys[] = {
 
 // branchless load of up to 5 characters into a uint64_t, padding with zeros if
 // n < 5
-inline uint64_t branchless_load5(const char *p, size_t n) {
+inline uint64_t branchless_load5(const char* p, size_t n) {
   uint64_t input = (uint8_t)p[0];
   input |= ((uint64_t)(uint8_t)p[n > 1] << 8) & (0 - (uint64_t)(n > 1));
   input |= ((uint64_t)(uint8_t)p[(n > 2) * 2] << 16) & (0 - (uint64_t)(n > 2));
@@ -131,7 +131,7 @@ std::optional<SchemeType> get_scheme_type(std::string_view scheme) noexcept {
   return std::nullopt;
 }
 
-double pretty_print(const std::string &name, size_t num_values,
+double pretty_print(const std::string& name, size_t num_values,
                     counters::event_aggregate agg) {
   printf("%-50s : ", name.c_str());
   printf(" %5.3f ns ", agg.fastest_elapsed_ns() / double(num_values));
@@ -269,7 +269,7 @@ void collect_benchmark_results(size_t number_strings) {
   gen.seed(42);  // reset seed to ensure same shuffle for all benchmarks
 }
 
-int main(int argc, char **argv) {
+int main(int argc, char** argv) {
   if (!counters::has_performance_counters()) {
     printf(
         "Performance counters not available, you may need to run with sudo.\n");

benchmarks/competitors/servo-url/servo_url.h

@@ -16,13 +16,13 @@ struct Url;
 
 extern "C" {
 
-Url *parse_url(const char *raw_input, size_t raw_input_length);
+Url* parse_url(const char* raw_input, size_t raw_input_length);
 
-void free_url(Url *raw);
+void free_url(Url* raw);
 
-const char *parse_url_to_href(const char *raw_input, size_t raw_input_length);
+const char* parse_url_to_href(const char* raw_input, size_t raw_input_length);
 
-void free_string(const char *);
+void free_string(const char*);
 }  // extern "C"
 
 }  // namespace servo_url

fuzz/can_parse.cc

@@ -7,7 +7,7 @@
 #include "ada.cpp"
 #include "ada.h"
 
-extern "C" int LLVMFuzzerTestOneInput(const uint8_t *data, size_t size) {
+extern "C" int LLVMFuzzerTestOneInput(const uint8_t* data, size_t size) {
   FuzzedDataProvider fdp(data, size);
   std::string source = fdp.ConsumeRandomLengthString(256);
   std::string base_source = fdp.ConsumeRandomLengthString(256);

fuzz/idna.cc

@@ -6,7 +6,7 @@
 #include "ada.cpp"
 #include "ada.h"
 
-extern "C" int LLVMFuzzerTestOneInput(const uint8_t *data, size_t size) {
+extern "C" int LLVMFuzzerTestOneInput(const uint8_t* data, size_t size) {
   FuzzedDataProvider fdp(data, size);
   std::string source = fdp.ConsumeRandomLengthString(256);
   std::string source2 = fdp.ConsumeRandomLengthString(64);

fuzz/parse.cc

@@ -8,8 +8,8 @@
 #include "ada.cpp"
 #include "ada.h"
 
-bool is_valid_utf8_string(const char *buf, size_t len) {
-  const uint8_t *data = reinterpret_cast<const uint8_t *>(buf);
+bool is_valid_utf8_string(const char* buf, size_t len) {
+  const uint8_t* data = reinterpret_cast<const uint8_t*>(buf);
   uint64_t pos = 0;
   uint32_t code_point = 0;
   while (pos < len) {
@@ -93,7 +93,7 @@ bool is_valid_utf8_string(const char *buf, size_t len) {
 }
 
 // Exercise all getters and boolean predicates on ada::url
-static void exercise_url_predicates(const ada::url &u) {
+static void exercise_url_predicates(const ada::url& u) {
   volatile size_t length = 0;
   length += u.get_href().size();
   length += u.get_origin().size();
@@ -119,7 +119,7 @@ static void exercise_url_predicates(const ada::url &u) {
 }
 
 // Exercise all getters and boolean predicates on ada::url_aggregator
-static void exercise_aggregator_predicates(const ada::url_aggregator &u) {
+static void exercise_aggregator_predicates(const ada::url_aggregator& u) {
   volatile size_t length = 0;
   length += u.get_href().size();
   length += u.get_origin().size();
@@ -150,7 +150,7 @@ static void exercise_aggregator_predicates(const ada::url_aggregator &u) {
   (void)u.to_diagram();
 }
 
-extern "C" int LLVMFuzzerTestOneInput(const uint8_t *data, size_t size) {
+extern "C" int LLVMFuzzerTestOneInput(const uint8_t* data, size_t size) {
   FuzzedDataProvider fdp(data, size);
   std::string source = fdp.ConsumeRandomLengthString(256);
   std::string base = fdp.ConsumeRandomLengthString(256);

fuzz/url_search_params.cc

@@ -6,7 +6,7 @@
 #include "ada.cpp"
 #include "ada.h"
 
-extern "C" int LLVMFuzzerTestOneInput(const uint8_t *data, size_t size) {
+extern "C" int LLVMFuzzerTestOneInput(const uint8_t* data, size_t size) {
   FuzzedDataProvider fdp(data, size);
   std::string source = fdp.ConsumeRandomLengthString(256);
   std::string base_source = fdp.ConsumeRandomLengthString(256);
@@ -33,7 +33,7 @@ extern "C" int LLVMFuzzerTestOneInput(const uint8_t *data, size_t size) {
       length += val->size();
     }
     auto all_vals = initialized.get_all(source);
-    for (const auto &v : all_vals) {
+    for (const auto& v : all_vals) {
       length += v.size();
     }
   }
@@ -42,7 +42,7 @@ extern "C" int LLVMFuzzerTestOneInput(const uint8_t *data, size_t size) {
   initialized.sort();
 
   // Test C++ range-for iteration; also verify has(k) and has(k,v) consistency.
-  for (const auto &pair : initialized) {
+  for (const auto& pair : initialized) {
     length += pair.first.size();
     length += pair.second.size();
 
@@ -114,7 +114,7 @@ extern "C" int LLVMFuzzerTestOneInput(const uint8_t *data, size_t size) {
   {
     auto all_vals = search_params.get_all(source);
     length += all_vals.size();
-    for (const auto &v : all_vals) {
+    for (const auto& v : all_vals) {
       length += v.size();
     }
   }
@@ -170,7 +170,7 @@ extern "C" int LLVMFuzzerTestOneInput(const uint8_t *data, size_t size) {
   }
 
   // Test C++ range-for on the mutated params
-  for (const auto &pair : search_params) {
+  for (const auto& pair : search_params) {
     length += pair.first.size();
     length += pair.second.size();
   }
@@ -181,7 +181,7 @@ extern "C" int LLVMFuzzerTestOneInput(const uint8_t *data, size_t size) {
   length += search_params.size();
 
   // Test that reset() followed by iteration doesn't crash
-  for (const auto &pair : search_params) {
+  for (const auto& pair : search_params) {
     length += pair.first.size();
     length += pair.second.size();
   }
@@ -270,7 +270,7 @@ extern "C" int LLVMFuzzerTestOneInput(const uint8_t *data, size_t size) {
 
       // Size and iteration must not crash.
       length += sp_from_url.size();
-      for (const auto &pair : sp_from_url) {
+      for (const auto& pair : sp_from_url) {
         length += pair.first.size();
         length += pair.second.size();
       }

include/ada/common_defs.h

@@ -44,10 +44,10 @@
 #endif
 
 // Align to N-byte boundary
-#define ADA_ROUNDUP_N(a, n) (((a) + ((n)-1)) & ~((n)-1))
-#define ADA_ROUNDDOWN_N(a, n) ((a) & ~((n)-1))
+#define ADA_ROUNDUP_N(a, n) (((a) + ((n) - 1)) & ~((n) - 1))
+#define ADA_ROUNDDOWN_N(a, n) ((a) & ~((n) - 1))
 
-#define ADA_ISALIGNED_N(ptr, n) (((uintptr_t)(ptr) & ((n)-1)) == 0)
+#define ADA_ISALIGNED_N(ptr, n) (((uintptr_t)(ptr) & ((n) - 1)) == 0)
 
 #if defined(ADA_REGULAR_VISUAL_STUDIO)