From f278c35f3f27c107e4c49c36c2a195cdf1bbe252 Mon Sep 17 00:00:00 2001
From: Niels Martin Hansen <nielsm@indvikleren.dk>
Date: Fri, 25 Apr 2014 16:04:08 +0200
Subject: [PATCH] Dynamically use either inline_string escaping or uuencoding
 for extradata

Since luabins generates binary data which grows up to 3x by escaping,
it's more efficient to uuencode that instead.
A marker is placed as the first character of the value field, either
'e' for inline_string escaped text, or 'u' for uuencoded binary data.
The key is always inline_string escaped, as it will typically be
human readable.
---
 libaegisub/include/libaegisub/ass/uuencode.h |  4 ++--
 src/ass_parser.cpp                           | 18 ++++++++++++++++--
 src/subtitle_format_ass.cpp                  | 14 +++++++++++++-
 3 files changed, 31 insertions(+), 5 deletions(-)

diff --git a/libaegisub/include/libaegisub/ass/uuencode.h b/libaegisub/include/libaegisub/ass/uuencode.h
index fbc957c3d..b6b30e442 100644
--- a/libaegisub/include/libaegisub/ass/uuencode.h
+++ b/libaegisub/include/libaegisub/ass/uuencode.h
@@ -28,7 +28,7 @@ namespace agi { namespace ass {
 
 /// Encode a blob of data, using ASS's nonstandard variant
 template<typename RandomAccessRange>
-std::string UUEncode(RandomAccessRange const& data) {
+std::string UUEncode(RandomAccessRange const& data, bool insert_linebreaks=true) {
 	using std::begin;
 	using std::end;
 
@@ -51,7 +51,7 @@ std::string UUEncode(RandomAccessRange const& data) {
 		for (size_t i = 0; i < std::min<size_t>(size - pos + 1, 4u); ++i) {
 			ret += dst[i] + 33;
 
-			if (++written == 80 && pos + 3 < size) {
+			if (insert_linebreaks && ++written == 80 && pos + 3 < size) {
 				written = 0;
 				ret += "\r\n";
 			}
diff --git a/src/ass_parser.cpp b/src/ass_parser.cpp
index 478704eb5..4e2e9061b 100644
--- a/src/ass_parser.cpp
+++ b/src/ass_parser.cpp
@@ -22,6 +22,8 @@
 #include "string_codec.h"
 #include "subtitle_format.h"
 
+#include <libaegisub/ass/uuencode.h>
+
 #include <algorithm>
 #include <boost/algorithm/string/case_conv.hpp>
 #include <boost/algorithm/string/predicate.hpp>
@@ -115,13 +117,25 @@ void AssParser::ParseGraphicsLine(std::string const& data) {
 }
 
 void AssParser::ParseExtradataLine(std::string const &data) {
-	static const boost::regex matcher("Data:[[:space:]]*(\\d+),([^,]+),(.*)");
+	static const boost::regex matcher("Data:[[:space:]]*(\\d+),([^,]+),(.)(.*)");
 	boost::match_results<std::string::const_iterator> mr;
 
 	if (boost::regex_match(data, mr, matcher)) {
 		auto id = boost::lexical_cast<uint32_t>(mr.str(1));
 		auto key = inline_string_decode(mr.str(2));
-		auto value = inline_string_decode(mr.str(3));
+		auto valuetype = mr.str(3);
+		auto value = mr.str(4);
+		if (valuetype == "e") {
+			// escaped/inline_string encoded
+			value = inline_string_decode(value);
+		} else if (valuetype == "u") {
+			// ass uuencoded
+			auto valuedata = agi::ass::UUDecode(value);
+			value = std::string(valuedata.begin(), valuedata.end());
+		} else {
+			// unknown, error?
+			value = "";
+		}
 
 		// ensure next_extradata_id is always at least 1 more than the largest existing id
 		target->next_extradata_id = std::max(id+1, target->next_extradata_id);
diff --git a/src/subtitle_format_ass.cpp b/src/subtitle_format_ass.cpp
index 3d5972a36..c5d1800df 100644
--- a/src/subtitle_format_ass.cpp
+++ b/src/subtitle_format_ass.cpp
@@ -27,6 +27,7 @@
 #include "text_file_writer.h"
 #include "version.h"
 
+#include <libaegisub/ass/uuencode.h>
 #include <libaegisub/fs.h>
 
 DEFINE_SIMPLE_EXCEPTION(AssParseError, SubtitleFormatParseError, "subtitle_io/parse/ass")
@@ -129,7 +130,18 @@ struct Writer {
 			line += ",";
 			line += inline_string_encode(edi.second.first);
 			line += ",";
-			line += inline_string_encode(edi.second.second);
+			std::string encoded_data = inline_string_encode(edi.second.second);
+			if (4*edi.second.second.size() < 3*encoded_data.size()) {
+				// the inline_string encoding grew the data by more than uuencoding would
+				// so base64 encode it instead
+				line += "u"; // marker for uuencoding
+				encoded_data = agi::ass::UUEncode(edi.second.second, false);
+				printf("did uuencoding, original size=%lu, encoded size=%lu\n", edi.second.second.size(), encoded_data.size());
+				line += encoded_data;
+			} else {
+				line += "e"; // marker for inline_string encoding (escaping)
+				line += encoded_data;
+			}
 			file.WriteLineToFile(line);
 		}
 	}