adding libbase64 competitor

lemire · Apr 1, 2024 · Mar 20, 2024 · Mar 20, 2024 · Mar 22, 2024 · Mar 22, 2024
commit 7b304d3cd21efb484a814ba9dcf42ff6808f6e73
diff --git a/benchmarks/base64/benchmark_base64.cpp b/benchmarks/base64/benchmark_base64.cpp
@@ -10,7 +10,9 @@
 #include <vector>
 
 #include "libbase64.h"
+#include "libbase64_spaces.h"
 #include "node_base64.h"
+
 #include "simdutf.h"
 
 #include "event_counter.h"
@@ -261,7 +263,7 @@ void bench(std::vector<std::vector<char>> &data, uint8_t mode) {
     bool spaces = contains_spaces(data);
     if (spaces) {
       printf("# the base64 data contains spaces, so we cannot use straigth "
-             "libbase64::base64_decode\n");
+             "libbase64::base64_decode directly\n");
     } else {
       pretty_print(data.size(), volume, "libbase64",
                    bench([&data, &buffer1, &buffer2]() {
@@ -279,6 +281,21 @@ void bench(std::vector<std::vector<char>> &data, uint8_t mode) {
                      }
                    }));
     }
+    pretty_print(
+        data.size(), volume, "libbase64_space_decode",
+        bench([&data, &buffer1, &buffer2]() {
+          for (const std::vector<char> &source : data) {
+
+            size_t outlen;
+            bool ok = libbase64_space_decode(source.data(), source.size(),
+                                             buffer1.data(), &outlen);
+            if (!ok) {
+              std::cerr << "Error: "
+                        << " failed to decode base64 " << std::endl;
+              throw std::runtime_error("Error: failed to decode base64 ");
+            }
+          }
+        }));
     pretty_print(
         data.size(), volume, "node", bench([&data, &buffer1, &buffer2]() {
           for (const std::vector<char> &source : data) {
@@ -373,11 +390,12 @@ int bench_bun() {
   std::vector<std::pair<std::string, std::string>> tests = {
       {"big hello world", bigBuffer}, {"random 16 bytes", crypto}};
   // Could be nicer with C++20
-  for (auto & i : tests) {
+  for (auto &i : tests) {
     printf("# %s\n", i.first.c_str());
     std::string source = i.second;
     volatile size_t base64_size;
-    std::vector<char> buffer1(simdutf::base64_length_from_binary(source.size()));
+    std::vector<char> buffer1(
+        simdutf::base64_length_from_binary(source.size()));
     pretty_print(1, source.size(), "libbase64",
                  bench([&source, &buffer1, &base64_size]() {
                    size_t outlen;

diff --git a/benchmarks/base64/libbase64_spaces.h b/benchmarks/base64/libbase64_spaces.h
@@ -0,0 +1,50 @@
+
+// https://github.com/aklomp/base64/blob/b20a31a997e0b48274fa09e58b65ee9202531e4f/bin/base64.c#L392
+static inline size_t libbase64_find_space(const char *p, const size_t avail) {
+  for (size_t len = 0; len < avail; len++) {
+    if (p[len] == '\n' || p[len] == '\r' || p[len] == ' ' || p[len] == '\t') {
+      return len;
+    }
+  }
+
+  return avail;
+}
+
+// Inspired by
+// https://github.com/aklomp/base64/blob/b20a31a997e0b48274fa09e58b65ee9202531e4f/bin/base64.c#L405
+
+static bool libbase64_space_decode(const char *start, size_t avail, char *outbuf,
+                   size_t *outlen) {
+  struct base64_state state;
+  *outlen = 0;
+
+  // Initialize the decoder's state structure.
+  base64_stream_decode_init(&state, 0);
+
+  while (avail > 0) {
+    size_t len = libbase64_find_space(start, avail);
+    if (len == 0) {
+      start++;
+      avail--;
+      continue;
+    }
+
+    // Decode the chunk into the raw buffer.
+    size_t outlen = 0;
+    if (base64_stream_decode(&state, start, len, outbuf, &outlen) == 0) {
+      // decoding error
+      return false;
+    }
+
+    // Update the output buffer pointer and total size.
+    outbuf += outlen;
+    outlen += outlen;
+    if(avail == len) {
+      break;
+    }
+
+    start += len + 1;
+    avail -= len + 1;
+  }
+  return true;
+}