From: Mikko Rasa <tdb@tdb.fi>
Date: Mon, 12 Apr 2021 21:53:03 +0000 (+0300)
Subject: Add support for base64-encoded strings in text format
X-Git-Url: http://git.tdb.fi/?a=commitdiff_plain;h=242191b162cd16cc1e67e38069eff3147827a8b6;p=libs%2Fdatafile.git

Add support for base64-encoded strings in text format

They're much more efficient both space and decoding wise than using C
escapes to encode binary data.
---

diff --git a/source/textparser.cpp b/source/textparser.cpp
index 7dc4bde..7ea0600 100644
--- a/source/textparser.cpp
+++ b/source/textparser.cpp
@@ -113,6 +113,7 @@ Token TextParser::parse_token()
 		FLOATEXPSIGN,
 		STRING,
 		STRING_ESCAPE,
+		STRING_BASE64,
 		ACCEPT,
 		ZERO,
 		DECIMAL,
@@ -121,6 +122,7 @@ Token TextParser::parse_token()
 		FLOAT,
 		FLOATEXP,
 		STRING_END,
+		STRING_BASE64_PAD,
 		IDENTIFIER
 	};
 
@@ -133,6 +135,7 @@ Token TextParser::parse_token()
 		Token::SPECIAL,
 		Token::SPECIAL,
 		Token::SPECIAL,
+		Token::SPECIAL,
 		Token::INTEGER,
 		Token::INTEGER,
 		Token::INTEGER,
@@ -140,6 +143,7 @@ Token TextParser::parse_token()
 		Token::FLOAT,
 		Token::FLOAT,
 		Token::STRING,
+		Token::STRING,
 		Token::IDENTIFIER
 	};
 
@@ -165,6 +169,8 @@ Token TextParser::parse_token()
 				state = FLOAT;
 			else if(c=='"')
 				state = STRING;
+			else if(c=='=')
+				state = STRING_BASE64;
 			else if(c=='{' || c=='}' || c==';')
 				return Token(Token::SPECIAL, string(1, c));
 			else if(isdigit(c))
@@ -255,6 +261,18 @@ Token TextParser::parse_token()
 			state = STRING;
 			break;
 
+		case STRING_BASE64:
+			if(c=='=')
+				state = STRING_BASE64_PAD;
+			else if(!isalnum(c) && c!='+' && c!='/')
+				throw parse_error(buf);
+			break;
+
+		case STRING_BASE64_PAD:
+			if(c!='=')
+				throw parse_error(buf);
+			break;
+
 		case IDENTIFIER:
 			if(!isalpha(c) && !isdigit(c) && c!='_' && c!='-' && c!='/')
 				throw parse_error(buf);
@@ -273,6 +291,8 @@ Token TextParser::parse_token()
 				return Token(Token::IDENTIFIER, buf.substr(1));
 			else if(state==STRING_END)
 				return Token(Token::STRING, c_unescape(buf.substr(1, buf.size()-2)));
+			else if(state==STRING_BASE64 || state==STRING_BASE64_PAD)
+				return Token(Token::STRING, base64_decode(buf));
 			else
 				return Token(token_type[state], buf);
 		}
@@ -291,5 +311,41 @@ bool TextParser::isodigit(int c)
 	return (c>='0' && c<='7');
 }
 
+string TextParser::base64_decode(const string &data)
+{
+	string bin;
+	unsigned accum = 0;
+	unsigned a_bits = 0;
+	for(string::const_iterator i=data.begin(); i!=data.end(); ++i)
+	{
+		unsigned d;
+		if(*i>='A' && *i<='Z')
+			d = *i-'A';
+		else if(*i>='a' && *i<='z')
+			d = 26+(*i-'a');
+		else if(*i>='0' && *i<='9')
+			d = 52+(*i-'0');
+		else if(*i=='+')
+			d = 62;
+		else if(*i=='/')
+			d = 63;
+		else if(*i=='=')
+			continue;
+		else
+			throw invalid_argument("TextParser::base64_decode");
+
+		accum = (accum<<6)|d;
+		a_bits += 6;
+
+		if(a_bits>=8)
+		{
+			bin += (accum>>(a_bits-8))&0xFF;
+			a_bits -= 8;
+		}
+	}
+
+	return bin;
+}
+
 } // namespace DataFile
 } // namespace Msp
diff --git a/source/textparser.h b/source/textparser.h
index e8c8922..1eec9c6 100644
--- a/source/textparser.h
+++ b/source/textparser.h
@@ -19,6 +19,7 @@ protected:
 	Token parse_token();
 	bool is_delimiter(int);
 	bool isodigit(int);
+	static std::string base64_decode(const std::string &);
 };
 
 } // namespace DataFile