Base64 split implementation in separate file (#23313)

CHANGELOG.md

@@ -12,6 +12,8 @@ All notable changes to this project will be documented in this file.
 ### Changed
 
 ### Fixed
+- Berry `bytes().asstring()` now truncates a string if buffer contains NULL
+- Berry string literals containing NULL are truncated
 
 ### Removed
 

lib/default/base64-1.1.1/src/base64.cpp

@@ -0,0 +1,128 @@
+/**
+ * Base64 encoding and decoding of strings. Uses '+' for 62, '/' for 63, '=' for padding
+ */
+
+ #include "base64.hpp"
+
+unsigned char binary_to_base64(unsigned char v) {
+  // Capital letters - 'A' is ascii 65 and base64 0
+  if(v < 26) return v + 'A';
+  
+  // Lowercase letters - 'a' is ascii 97 and base64 26
+  if(v < 52) return v + 71;
+  
+  // Digits - '0' is ascii 48 and base64 52
+  if(v < 62) return v - 4;
+  
+  // '+' is ascii 43 and base64 62
+  if(v == 62) return '+';
+  
+  // '/' is ascii 47 and base64 63
+  if(v == 63) return '/';
+  
+  return 64;
+}
+
+unsigned char base64_to_binary(unsigned char c) {
+  // Capital letters - 'A' is ascii 65 and base64 0
+  if('A' <= c && c <= 'Z') return c - 'A';
+  
+  // Lowercase letters - 'a' is ascii 97 and base64 26
+  if('a' <= c && c <= 'z') return c - 71;
+  
+  // Digits - '0' is ascii 48 and base64 52
+  if('0' <= c && c <= '9') return c + 4;
+  
+  // '+' is ascii 43 and base64 62
+  if(c == '+') return 62;
+  
+  // '/' is ascii 47 and base64 63
+  if(c == '/') return 63;
+  
+  return 255;
+}
+
+unsigned int encode_base64_length(unsigned int input_length) {
+  return (input_length + 2)/3*4;
+}
+
+unsigned int decode_base64_length(unsigned char input[]) {
+  unsigned char *start = input;
+  
+  while(base64_to_binary(input[0]) < 64) {
+    ++input;
+  }
+  
+  unsigned int input_length = input - start;
+  
+  unsigned int output_length = input_length/4*3;
+  
+  switch(input_length % 4) {
+    default: return output_length;
+    case 2: return output_length + 1;
+    case 3: return output_length + 2;
+  }
+}
+
+unsigned int encode_base64(unsigned char input[], unsigned int input_length, unsigned char output[]) {
+  unsigned int full_sets = input_length/3;
+  
+  // While there are still full sets of 24 bits...
+  for(unsigned int i = 0; i < full_sets; ++i) {
+    output[0] = binary_to_base64(                         input[0] >> 2);
+    output[1] = binary_to_base64((input[0] & 0x03) << 4 | input[1] >> 4);
+    output[2] = binary_to_base64((input[1] & 0x0F) << 2 | input[2] >> 6);
+    output[3] = binary_to_base64( input[2] & 0x3F);
+    
+    input += 3;
+    output += 4;
+  }
+  
+  switch(input_length % 3) {
+    case 0:
+      output[0] = '\0';
+      break;
+    case 1:
+      output[0] = binary_to_base64(                         input[0] >> 2);
+      output[1] = binary_to_base64((input[0] & 0x03) << 4);
+      output[2] = '=';
+      output[3] = '=';
+      output[4] = '\0';
+      break;
+    case 2:
+      output[0] = binary_to_base64(                         input[0] >> 2);
+      output[1] = binary_to_base64((input[0] & 0x03) << 4 | input[1] >> 4);
+      output[2] = binary_to_base64((input[1] & 0x0F) << 2);
+      output[3] = '=';
+      output[4] = '\0';
+      break;
+  }
+  
+  return encode_base64_length(input_length);
+}
+
+unsigned int decode_base64(unsigned char input[], unsigned char output[]) {
+  unsigned int output_length = decode_base64_length(input);
+  
+  // While there are still full sets of 24 bits...
+  for(unsigned int i = 2; i < output_length; i += 3) {
+    output[0] = base64_to_binary(input[0]) << 2 | base64_to_binary(input[1]) >> 4;
+    output[1] = base64_to_binary(input[1]) << 4 | base64_to_binary(input[2]) >> 2;
+    output[2] = base64_to_binary(input[2]) << 6 | base64_to_binary(input[3]);
+    
+    input += 4;
+    output += 3;
+  }
+  
+  switch(output_length % 3) {
+    case 1:
+      output[0] = base64_to_binary(input[0]) << 2 | base64_to_binary(input[1]) >> 4;
+      break;
+    case 2:
+      output[0] = base64_to_binary(input[0]) << 2 | base64_to_binary(input[1]) >> 4;
+      output[1] = base64_to_binary(input[1]) << 4 | base64_to_binary(input[2]) >> 2;
+      break;
+  }
+  
+  return output_length;
+}

lib/default/base64-1.1.1/src/base64.hpp

@@ -69,127 +69,4 @@ unsigned int encode_base64(unsigned char input[], unsigned int input_length, uns
  */
 unsigned int decode_base64(unsigned char input[], unsigned char output[]);
 
-unsigned char binary_to_base64(unsigned char v) {
-  // Capital letters - 'A' is ascii 65 and base64 0
-  if(v < 26) return v + 'A';
-  
-  // Lowercase letters - 'a' is ascii 97 and base64 26
-  if(v < 52) return v + 71;
-  
-  // Digits - '0' is ascii 48 and base64 52
-  if(v < 62) return v - 4;
-  
-  // '+' is ascii 43 and base64 62
-  if(v == 62) return '+';
-  
-  // '/' is ascii 47 and base64 63
-  if(v == 63) return '/';
-  
-  return 64;
-}
-
-unsigned char base64_to_binary(unsigned char c) {
-  // Capital letters - 'A' is ascii 65 and base64 0
-  if('A' <= c && c <= 'Z') return c - 'A';
-  
-  // Lowercase letters - 'a' is ascii 97 and base64 26
-  if('a' <= c && c <= 'z') return c - 71;
-  
-  // Digits - '0' is ascii 48 and base64 52
-  if('0' <= c && c <= '9') return c + 4;
-  
-  // '+' is ascii 43 and base64 62
-  if(c == '+') return 62;
-  
-  // '/' is ascii 47 and base64 63
-  if(c == '/') return 63;
-  
-  return 255;
-}
-
-unsigned int encode_base64_length(unsigned int input_length) {
-  return (input_length + 2)/3*4;
-}
-
-unsigned int decode_base64_length(unsigned char input[]) {
-  unsigned char *start = input;
-  
-  while(base64_to_binary(input[0]) < 64) {
-    ++input;
-  }
-  
-  unsigned int input_length = input - start;
-  
-  unsigned int output_length = input_length/4*3;
-  
-  switch(input_length % 4) {
-    default: return output_length;
-    case 2: return output_length + 1;
-    case 3: return output_length + 2;
-  }
-}
-
-unsigned int encode_base64(unsigned char input[], unsigned int input_length, unsigned char output[]) {
-  unsigned int full_sets = input_length/3;
-  
-  // While there are still full sets of 24 bits...
-  for(unsigned int i = 0; i < full_sets; ++i) {
-    output[0] = binary_to_base64(                         input[0] >> 2);
-    output[1] = binary_to_base64((input[0] & 0x03) << 4 | input[1] >> 4);
-    output[2] = binary_to_base64((input[1] & 0x0F) << 2 | input[2] >> 6);
-    output[3] = binary_to_base64( input[2] & 0x3F);
-    
-    input += 3;
-    output += 4;
-  }
-  
-  switch(input_length % 3) {
-    case 0:
-      output[0] = '\0';
-      break;
-    case 1:
-      output[0] = binary_to_base64(                         input[0] >> 2);
-      output[1] = binary_to_base64((input[0] & 0x03) << 4);
-      output[2] = '=';
-      output[3] = '=';
-      output[4] = '\0';
-      break;
-    case 2:
-      output[0] = binary_to_base64(                         input[0] >> 2);
-      output[1] = binary_to_base64((input[0] & 0x03) << 4 | input[1] >> 4);
-      output[2] = binary_to_base64((input[1] & 0x0F) << 2);
-      output[3] = '=';
-      output[4] = '\0';
-      break;
-  }
-  
-  return encode_base64_length(input_length);
-}
-
-unsigned int decode_base64(unsigned char input[], unsigned char output[]) {
-  unsigned int output_length = decode_base64_length(input);
-  
-  // While there are still full sets of 24 bits...
-  for(unsigned int i = 2; i < output_length; i += 3) {
-    output[0] = base64_to_binary(input[0]) << 2 | base64_to_binary(input[1]) >> 4;
-    output[1] = base64_to_binary(input[1]) << 4 | base64_to_binary(input[2]) >> 2;
-    output[2] = base64_to_binary(input[2]) << 6 | base64_to_binary(input[3]);
-    
-    input += 4;
-    output += 3;
-  }
-  
-  switch(output_length % 3) {
-    case 1:
-      output[0] = base64_to_binary(input[0]) << 2 | base64_to_binary(input[1]) >> 4;
-      break;
-    case 2:
-      output[0] = base64_to_binary(input[0]) << 2 | base64_to_binary(input[1]) >> 4;
-      output[1] = base64_to_binary(input[1]) << 4 | base64_to_binary(input[2]) >> 2;
-      break;
-  }
-  
-  return output_length;
-}
-
 #endif // ifndef

lib/libesp32/berry/src/be_byteslib.c

@@ -806,7 +806,8 @@ static int m_asstring(bvm *vm)
 {
     buf_impl attr = bytes_check_data(vm, 0);
     check_ptr(vm, &attr);
-    be_pushnstring(vm, (const char*) attr.bufptr, attr.len);
+    size_t safe_len = strnlen((const char*) attr.bufptr, attr.len);
+    be_pushnstring(vm, (const char*) attr.bufptr, safe_len);
     be_return(vm);
 }
 

lib/libesp32/berry/src/be_lexer.c

@@ -284,7 +284,8 @@ static void tr_string(blexer *lexer)
             break;
         }
     }
-    lexer->buf.len = dst - lexbuf(lexer);
+    size_t len = dst - lexbuf(lexer);
+    lexer->buf.len = strnlen(lexbuf(lexer), len);
 }
 
 static int skip_newline(blexer *lexer)

lib/libesp32/berry/tests/bytes.be

@@ -354,3 +354,9 @@ assert(b.appendb64(c, 2) == bytes("AABBCC49673D3D"))
 b = bytes("AABBCC")
 assert(bytes().fromstring(bytes("11").tob64()) == bytes('45513D3D'))
 assert(b.appendb64(c, 1, 1) == bytes("AABBCC45513D3D"))
+
+#- asstring truncates if NULL is present -#
+s=bytes("414243").asstring()
+assert(size(s) == 3)
+s=bytes("410000").asstring()
+assert(size(s) == 1)

lib/libesp32/berry/tests/lexer.be

@@ -84,3 +84,6 @@ var malformed_numbers = [
 for i : malformed_numbers
     test_source(i, 'malformed number')
 end
+
+#- ensure that string literal with NULL character is truncated -#
+assert(size('aa\000bb\000cc') == 2)