From 0225cfa5933cbcb8c236b7cf9dbd05a6260ea911 Mon Sep 17 00:00:00 2001
From: Carl Hetherington <cth@carlh.net>
Date: Mon, 13 Oct 2025 22:59:22 +0200
Subject: fixup! WIP: SCC support.

---
 src/scc_reader.cc | 212 +++++++++++++++++++++++++++++++++++-------------------
 src/sub_time.cc   |  12 ++--
 src/sub_time.h    |   4 ++
 3 files changed, 151 insertions(+), 77 deletions(-)

(limited to 'src')

diff --git a/src/scc_reader.cc b/src/scc_reader.cc
index 10fde09..deb853b 100644
--- a/src/scc_reader.cc
+++ b/src/scc_reader.cc
@@ -27,16 +27,23 @@
 #include <boost/bind/bind.hpp>
 #include <boost/algorithm/string.hpp>
 #include <iostream>
+#include <set>
+#include <unordered_map>
 
 
 using std::pair;
+using std::set;
 using std::string;
+using std::unordered_map;
 using std::vector;
 using boost::function;
 using boost::optional;
 using namespace sub;
 
 
+static Rational const frame_rate{30000, 1001};
+
+
 /** @param s Subtitle string encoded in UTF-8 */
 SCCReader::SCCReader(string s)
 {
@@ -51,7 +58,7 @@ SCCReader::SCCReader(FILE* f)
 }
 
 
-boost::optional<Time>
+int
 convert_time(std::string time)
 {
 	vector<string> parts;
@@ -60,12 +67,8 @@ convert_time(std::string time)
 		throw SCCReader(String::compose("Unrecognised time %1", time));
 	}
 
-	return Time::from_hmsf(
-		raw_convert<int>(parts[0]),
-		raw_convert<int>(parts[1]),
-		raw_convert<int>(parts[2]),
-		raw_convert<int>(parts[3])
-		);
+	int seconds = raw_convert<int>(parts[0]) * 3600 + raw_convert<int>(parts[1]) * 60 + raw_convert<int>(parts[2]);
+	return std::round(seconds * frame_rate.fraction() + raw_convert<int>(parts[3]));
 }
 
 
@@ -142,22 +145,42 @@ preamble_address_code(string part)
 	PAC pac;
 
 	switch (bytes[1]) {
-	case 0x43:
-	case 0x45:
-	case 0x46:
-	case 0x49:
-	case 0x4a:
-	case 0x4c:
-	case 0x51:
+	/* Column 0 - from http://www.theneitherworld.com/mcpoodle/SCC_TOOLS/DOCS/SCC_FORMAT.HTML */
+	case 0xd0:
 	case 0x70:
+	case 0x51:
+	case 0xf1:
 	case 0xc2:
+	case 0x62:
+	case 0x43:
+	case 0xe3:
 	case 0xc4:
+	case 0x64:
+	case 0x45:
+	case 0xe5:
+	case 0x46:
+	case 0xe6:
 	case 0xc7:
+	case 0x67:
 	case 0xc8:
+	case 0x68:
+	case 0x49:
+	case 0xe9:
+	case 0x4a:
+	case 0xea:
 	case 0xcb:
+	case 0x6b:
+	case 0x4c:
+	case 0xec:
 	case 0xcd:
-	case 0xd0:
-	case 0xf1:
+	case 0x6d:
+
+	/* RESUMEDIRECTCAPTIONING in ccextractor? */
+	// case 0x29:
+	/* No documentation found for these... */
+	// case 0x2f:
+	// case 0x40:
+	// case 0xe0:
 		pac.row = ((bytes[1] & 0x20) >> 5) + 1;
 		pac.underline = bytes[1] & 0x1;
 		switch (bytes[1] & 0xf) {
@@ -197,34 +220,36 @@ preamble_address_code(string part)
 	}
 
 	switch (bytes[1]) {
+	/* Columns 4 - 28 */
 	case 0x52:
-	case 0x54:
-	case 0x57:
-	case 0x58:
-	case 0x5b:
-	case 0x5d:
-	case 0x5e:
-	case 0x73:
-	case 0x75:
-	case 0x76:
-	case 0x79:
-	case 0x7a:
-	case 0x7c:
-	case 0x7f:
+	case 0xf2:
 	case 0xd3:
+	case 0x73:
+	case 0x54:
+	case 0xf4:
 	case 0xd5:
+	case 0x75:
 	case 0xd6:
-	case 0xd9:
-	case 0xda:
-	case 0xdc:
-	case 0xdf:
-	case 0xf2:
-	case 0xf4:
+	case 0x76:
+	case 0x57:
 	case 0xf7:
+	case 0x58:
 	case 0xf8:
+	case 0xd9:
+	case 0x79:
+	case 0xda:
+	case 0x7a:
+	case 0x5b:
 	case 0xfb:
+	case 0xdc:
+	case 0x7c:
+	case 0x5d:
 	case 0xfd:
+	case 0x5e:
 	case 0xfe:
+	case 0xdf:
+	case 0x7f:
+		pac.row = ((bytes[1] & 0x20) >> 5) + 1;
 		pac.underline = bytes[1] & 0x1;
 		pac.column = (bytes[1] & 0xe) * 2;
 		if (adjust(pac, bytes[0])) {
@@ -239,18 +264,19 @@ preamble_address_code(string part)
 void
 SCCReader::read(function<optional<string> ()> get_line)
 {
-	bool got_header = false;
-
-	RawSubtitle displayed;
-	RawSubtitle non_displayed;
+	set<string> ignore;
+	ignore.insert("102f");
+	ignore.insert("9420"); // RCL: resume caption loading; the next caption is "pop-on"
+	ignore.insert("9440"); // should be a line break?
 
-	displayed.vertical_position.line = 0;
-	displayed.vertical_position.reference = TOP_OF_SUBTITLE;
+	set<string> line_break;
+	line_break.insert("94e0");
 
-	non_displayed.vertical_position.line = 0;
-	non_displayed.vertical_position.reference = TOP_OF_SUBTITLE;
+	unordered_map<string, char> replace;
+	replace["9229"] = '\'';
 
-	optional<sub::Time> time;
+	bool got_header = false;
+	int to_fix = 0;
 
 	while (true) {
 		auto line = get_line ();
@@ -258,8 +284,6 @@ SCCReader::read(function<optional<string> ()> get_line)
 			break;
 		}
 
-		std::cout << *line << "\n";
-
 		trim_right_if(*line, boost::is_any_of ("\n\r"));
 
 		if (*line == "Scenarist_SCC V1.0" && !got_header) {
@@ -280,53 +304,79 @@ SCCReader::read(function<optional<string> ()> get_line)
 			throw SCCError(String::compose("No tab character found in line %1", *line));
 		}
 
-		time = convert_time(line->substr(0, tab));
+		/* XXX: don't fuck about with this stuff, we just add a frame per two-bytes to the initial time,
+		 * and probably do this NDF adjustment.
+		 */
+		int time_in_frames = convert_time(line->substr(0, tab));
 
 		vector<string> parts;
 		boost::algorithm::split(parts, line->substr(tab + 1), boost::is_any_of(" "));
 
+		vector<RawSubtitle> current_line;
+		optional<Time> from;
+		RawSubtitle current;
+		current.vertical_position.line = 0;
+		current.vertical_position.reference = TOP_OF_SUBTITLE;
+
+		auto maybe_add = [&]() {
+			if (!current.text.empty()) {
+				std::cerr << "Add " << current.from << " " << current.to << " " << current.text << "\n";
+				current_line.push_back(current);
+			}
+			current.text = "";
+		};
+
 		for (auto part: parts) {
-			if (part == "9420") {
-				/* RCL: resume caption loading
-				 * The next caption is "pop-on"
-				 */
-				std::cout << "-->resume caption loading<--\n";
+			if (ignore.find(part) != ignore.end()) {
+				++time_in_frames;
+				continue;
+			}
+			std::cerr << "___" << part << "___\n";
+			auto replace_iter = replace.find(part);
+			if (replace_iter != replace.end()) {
+				current.text = current.text.substr(0, current.text.length() - 1) + replace_iter->second;
+				++time_in_frames;
 				continue;
+			}
+
+			if (line_break.find(part) != line_break.end()) {
+				maybe_add();
+				current.vertical_position.line = current.vertical_position.line.get_value_or(0) + 1;
 			} else if (part == "97a1") {
 				/* TO: tab over
 				 * Move one column over
 				 */
-				non_displayed.text += " ";
+				current.text += " ";
 			} else if (part == "97a2") {
 				/* TO: tab over
 				 * Move two columns over
 				 */
-				non_displayed.text += "  ";
+				current.text += "  ";
 			} else if (part == "9723") {
 				/* TO: tab over
 				 * Move three columns over
 				 */
-				non_displayed.text += "   ";
+				current.text += "   ";
 			} else if (part == "94ae") {
 				/* ENM: erase non-displayed memory */
-				non_displayed.text = "";
-				std::cout << "-->erase non-displayed<--\n";
+				current.text = "";
 			} else if (part == "942f") {
 				/* EOC: end of caption */
-				displayed = non_displayed;
-				SUB_ASSERT(time);
-				displayed.from = *time;
-				non_displayed.text = "";
-				std::cout << "-->end of caption<--\n";
+				from = Time::from_frames(time_in_frames, frame_rate);
 			} else if (part == "942c") {
 				/* EDM: erase displayed memory */
-				SUB_ASSERT(time);
-				displayed.to = *time;
-				/* XXX: last subtitle finished */
-				std::cout << "-->erase displayed<--\n";
+				from = Time::from_frames(time_in_frames, frame_rate);
+			} else if (part == "91ae" || part == "946e" || part == "94ce") {
+				maybe_add();
+				current.italic = true;
+			} else if (part == "9120") {
+				maybe_add();
+				current.italic = false;
 			} else if (auto pac = preamble_address_code(part)) {
-				std::cout << "-->PAC<-- " << part << "\n";
-				/* XXX */
+				// std::cerr << "-" << part << "-PAC row=" << pac->row << "\n";
+				maybe_add();
+				current.vertical_position.line = pac->row;
+				current.vertical_position.lines = 16;
 			} else {
 				int text[2];
 				if (sscanf(part.c_str(), "%2x%2x", text, text + 1) < 2) {
@@ -334,13 +384,31 @@ SCCReader::read(function<optional<string> ()> get_line)
 				}
 				for (int i = 0; i < 2; ++i) {
 					if (text[i] != 0x80) {
-						non_displayed.text += static_cast<unsigned char>(text[i] & 0x7f);
+						current.text += static_cast<unsigned char>(text[i] & 0x7f);
 					}
 				}
-				std::cout << "part-as-chars " << part << " " << non_displayed.text << "\n";
+				std::cerr << ">" << current.text << "\n";
 			}
-			// XXX
-			// time->add(sub::Time::from_hmsf(0, 0, 0, 1));
+			time_in_frames++;
 		}
+
+		maybe_add();
+
+		for (auto i = _subs.rbegin(); i != _subs.rend(); ++i) {
+			if (to_fix == 0) {
+				break;
+			}
+			SUB_ASSERT(from);
+			i->to = *from;
+			--to_fix;
+		}
+
+		for (auto sub: current_line) {
+			SUB_ASSERT(from);
+			sub.from = *from;
+			_subs.push_back(sub);
+		}
+
+		to_fix = current_line.size();
 	}
 }
diff --git a/src/sub_time.cc b/src/sub_time.cc
index 763b4e4..6773e67 100644
--- a/src/sub_time.cc
+++ b/src/sub_time.cc
@@ -155,15 +155,17 @@ Time::from_hms (int h, int m, int s, int ms)
 	return Time (h * 3600 + m * 60 + s, ms, Rational (1000, 1));
 }
 
-/** Create a Time from a number of frames.
- *  rate must be integer.
- */
+/** Create a Time from a number of frames */
 Time
 Time::from_frames (int f, Rational rate)
 {
 	SUB_ASSERT (rate.denominator != 0);
-	SUB_ASSERT (rate.integer ());
-	return Time (f / rate.integer_fraction(), f % rate.integer_fraction(), rate);
+	if (rate.integer()) {
+		return Time(f / rate.integer_fraction(), f % rate.integer_fraction(), rate);
+	} else {
+		auto const seconds = std::floor(f / rate.fraction());
+		return Time(seconds, std::round(f - seconds * rate.fraction()), rate);
+	}
 }
 
 double
diff --git a/src/sub_time.h b/src/sub_time.h
index 93088cb..0ee8559 100644
--- a/src/sub_time.h
+++ b/src/sub_time.h
@@ -45,6 +45,10 @@ public:
 	void add (Time t);
 	void scale (float f);
 
+	boost::optional<Rational> frame_rate() const {
+		return _rate;
+	}
+
 	static Time from_hmsf (int h, int m, int s, int f, boost::optional<Rational> rate = boost::optional<Rational> ());
 	static Time from_hms (int h, int m, int s, int ms);
 	static Time from_frames (int frames, Rational rate);
-- 
cgit v1.2.3