Utils: is_base64url.

GitOrigin-RevId: b58daeb9fa82d37e9c653d72ef702bcb15a5392b
genesem · Jan 19, 2018 · 22ad865 · 22ad865
1 parent c816837
commit 22ad865
Show file tree

Hide file tree

Showing 3 changed files with 56 additions and 23 deletions.
diff --git a/tdutils/td/utils/base64.cpp b/tdutils/td/utils/base64.cpp
@@ -188,11 +188,8 @@ Result<string> base64url_decode(Slice base64) {
   return output;
 }
 
-bool is_base64(Slice input) {
-  if ((input.size() & 3) != 0) {
-    return false;
-  }
-
+template <bool is_url>
+static bool is_base64_impl(Slice input) {
   size_t padding_length = 0;
   while (!input.empty() && input.back() == '=') {
     input.remove_suffix(1);
@@ -201,22 +198,35 @@ bool is_base64(Slice input) {
   if (padding_length >= 3) {
     return false;
   }
+  if ((!is_url || padding_length > 0) && ((input.size() + padding_length) & 3) != 0) {
+    return false;
+  }
+  if (is_url && (input.size() & 3) == 1) {
+    return false;
+  }
 
-  init_base64_table();
-  for (size_t i = 0; i < input.size(); i++) {
-    if (char_to_value[input.ubegin()[i]] == 64) {
+  unsigned char *table;
+  if (is_url) {
+    init_base64url_table();
+    table = url_char_to_value;
+  } else {
+    init_base64_table();
+    table = char_to_value;
+  }
+  for (auto c:input) {
+    if (table[static_cast<unsigned char>(c)] == 64) {
       return false;
     }
   }
 
   if ((input.size() & 3) == 2) {
-    auto value = char_to_value[input.back()];
+    auto value = table[input.back()];
     if ((value & 15) != 0) {
       return false;
     }
   }
   if ((input.size() & 3) == 3) {
-    auto value = char_to_value[input.back()];
+    auto value = table[input.back()];
     if ((value & 3) != 0) {
       return false;
     }
@@ -225,13 +235,21 @@ bool is_base64(Slice input) {
   return true;
 }
 
+bool is_base64(Slice input) {
+  return is_base64_impl<false>(input);
+}
+
+bool is_base64url(Slice input) {
+  return is_base64_impl<true>(input);
+}
+
 string base64_filter(Slice input) {
   string res;
   res.reserve(input.size());
   init_base64_table();
-  for (size_t i = 0; i < input.size(); i++) {
-    if (char_to_value[input.ubegin()[i]] != 64 || input[i] == '=') {
-      res += input[i];
+  for (auto c : input) {
+    if (char_to_value[static_cast<unsigned char>(c)] != 64 || c == '=') {
+      res += c;
     }
   }
   return res;

diff --git a/tdutils/td/utils/base64.h b/tdutils/td/utils/base64.h
@@ -19,6 +19,7 @@ string base64url_encode(Slice input);
 Result<string> base64url_decode(Slice base64);
 
 bool is_base64(Slice input);
+bool is_base64url(Slice input);
 
 string base64_filter(Slice input);
 

diff --git a/tdutils/test/misc.cpp b/tdutils/test/misc.cpp
@@ -115,6 +115,30 @@ TEST(Misc, errno_tls_bug) {
 }
 
 TEST(Misc, base64) {
+  ASSERT_TRUE(is_base64("dGVzdA==") == true);
+  ASSERT_TRUE(is_base64("dGVzdB==") == false);
+  ASSERT_TRUE(is_base64("dGVzdA=") == false);
+  ASSERT_TRUE(is_base64("dGVzdA") == false);
+  ASSERT_TRUE(is_base64("dGVz") == true);
+  ASSERT_TRUE(is_base64("") == true);
+  ASSERT_TRUE(is_base64("ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/") == true);
+  ASSERT_TRUE(is_base64("ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/=") == false);
+  ASSERT_TRUE(is_base64("ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-/") == false);
+  ASSERT_TRUE(is_base64("ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_") == false);
+  ASSERT_TRUE(is_base64("====") == false);
+
+  ASSERT_TRUE(is_base64url("dGVzdA==") == true);
+  ASSERT_TRUE(is_base64url("dGVzdB==") == false);
+  ASSERT_TRUE(is_base64url("dGVzdA=") == false);
+  ASSERT_TRUE(is_base64url("dGVzdA") == true);
+  ASSERT_TRUE(is_base64url("dGVz") == true);
+  ASSERT_TRUE(is_base64url("") == true);
+  ASSERT_TRUE(is_base64url("ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_") == true);
+  ASSERT_TRUE(is_base64url("ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_=") == false);
+  ASSERT_TRUE(is_base64url("ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-/") == false);
+  ASSERT_TRUE(is_base64url("ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/") == false);
+  ASSERT_TRUE(is_base64url("====") == false);
+
   for (int l = 0; l < 300000; l += l / 20 + l / 1000 * 500 + 1) {
     for (int t = 0; t < 10; t++) {
       string s = rand_string(std::numeric_limits<char>::min(), std::numeric_limits<char>::max(), l);
@@ -139,16 +163,6 @@ TEST(Misc, base64) {
   ASSERT_TRUE(base64_encode("      /'.;.';≤.];,].',[.;/,.;/]/..;!@#!*(%?::;!%\";") ==
               "ICAgICAgLycuOy4nO+KJpC5dOyxdLicsWy47LywuOy9dLy4uOyFAIyEqKCU/"
               "Ojo7ISUiOw==");
-  ASSERT_TRUE(is_base64("dGVzdA==") == true);
-  ASSERT_TRUE(is_base64("dGVzdB==") == false);
-  ASSERT_TRUE(is_base64("dGVzdA=") == false);
-  ASSERT_TRUE(is_base64("dGVzdA") == false);
-  ASSERT_TRUE(is_base64("dGVz") == true);
-  ASSERT_TRUE(is_base64("") == true);
-  ASSERT_TRUE(is_base64("ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/") == true);
-  ASSERT_TRUE(is_base64("ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/=") == false);
-  ASSERT_TRUE(is_base64("ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-/") == false);
-  ASSERT_TRUE(is_base64("====") == false);
 }
 
 TEST(Misc, to_integer) {