Seemingly working furigana/multi-hl parsing and some untested inline-fx parsing. Still no furigana layouting, and the parsing isn't integrated yet.

Originally committed to SVN as r1180.

automation/include/karaskel-auto4.lua

@@ -116,21 +116,118 @@ function karaskel.preproc_line_text(meta, styles, line)
 	line.kara = { n = 0 }
 	line.furi = { n = 0 }
 	
+	if styles[line.style] then
+		line.styleref = styles[line.style]
+	else
+		aegisub.debug.out(2, "WARNING: Style not found: " .. line.style .. "\n")
+		line.styleref = styles[1]
+	end
+	
 	line.text_stripped = ""
 	line.duration = line.end_time - line.start_time
 	
-	local worksyl = { }
+	local worksyl = { highlights = {n=0} }
 	local cur_inline_fx = ""
 	for i = 0, #kara do
 		local syl = kara[i]
 		
 		-- Detect any inline-fx tags
-		local inline_fx = syl.text:match("%{.*\\%-(.-)[}\\]")
+		local inline_fx = syl.text:match("%{.*\\%-([^}\\]-)")
 		if inline_fx then
 			cur_inline_fx = inline_fx
 		end
 		
+		-- Strip spaces (only basic ones, no fullwidth etc.)
+		local prespace, syltext, postspace = syl.text_stripped:match("^([ \t]*)(.-)([ \t]*)$")
+		
+		-- See if we've broken a (possible) multi-hl stretch
+		-- If we did it's time for a new worksyl (though never for the zero'th syllable)
+		local prefix = syltext:sub(1,unicode.charwidth(syltext,1))
+		if prefix ~= "#" and prefix ~= "＃" and i > 0 then
+			line.kara[line.kara.n] = worksyl
+			line.kara.n = line.kara.n + 1
+			worksyl = { highlights = {n=0} }
+		end
+
+		-- Detect furigana (both regular and fullwidth pipes work)
+		-- Furigana is stored independantly from syllables
+		if syltext:find("|") or syltext:find("｜") then
+			-- Replace fullwidth pipes, they aren't regex friendly
+			syltext = syltext:gsub("｜", "|")
+			-- Get before/after pipe text
+			local maintext, furitext = syltext:match("^(.-)|(.-)$")
+			syltext = maintext
+			
+			local furi = { }
+			furi.syl = worksyl
+			
+			-- Magic happens here
+			-- isbreak = Don't join this furi visually with previous furi, even if their main texts are adjacent
+			-- spillback = Allow this furi text to spill over the left edge of the main text
+			-- (Furi is always allowed to spill over the right edge of main text.)
+			local prefix = furitext:sub(1,unicode.charwidth(furitext,1))
+			if prefix == "!" or prefix == "！" then
+				furi.isbreak = true -- Don't join with furi in previous syllable
+				furi.spillback = false -- Allow to "spill" furi over the left edge of main text
+			elseif prefix == "<" or prefix == "＜" then
+				furi.isbreak = true
+				furi.spillback = true
+			else
+				furi.isbreak = false
+				furi.spillback = false
+			end
+			-- Remove the prefix character from furitext, if there was one
+			if furi.isbreak then
+				furitext = furitext:sub(unicode.charwidth(furitext,1)+1)
+			end
+			
+			furi.start_time = syl.start_time
+			furi.end_time = syl.end_time
+			furi.duration = syl.duration
+			furi.text = furitext
+			
+			line.furi.n = line.furi.n + 1
+			line.furi[line.furi.n] = furi
+		end
+		
+		-- Always add highlight data
+		local hl = {
+			start_time = syl.start_time,
+			end_time = syl.end_time,
+			duration = syl.duration
+		}
+		worksyl.highlights.n = worksyl.highlights.n + 1
+		worksyl.highlights[worksyl.highlights.n] = hl
+		
+		-- Syllables that aren't part of a multi-highlight generate a new output-syllable
+		if prefix ~= "#" and prefix ~= "＃" then
+			-- Update stripped line-text
+			line.text_stripped = line.text_stripped .. prespace .. syltext .. postspace
+			
+			-- Copy data from syl to worksyl
+			worksyl.text = syl.text
+			worksyl.duration = syl.duration
+			worksyl.kdur = syl.duration / 10
+			worksyl.start_time = syl.start_time
+			worksyl.end_time = syl.end_time
+			worksyl.tag = syl.tag
+			worksyl.line = line
+			worksyl.style = line.styleref
+			
+			-- And add new data to worksyl
+			worksyl.i = line.kara.n
+			worksyl.text_stripped = syltext
+			worksyl.inline_fx = cur_inline_fx
+		else
+			-- This is just an extra highlight
+			worksyl.duration = worksyl.duration + syl.duration
+			worksyl.end_time = syl.end_time
+		end
 	end
+
+	-- Add the last syllable
+	line.kara[line.kara.n] = worksyl
+	-- But don't increment n here, n should be the highest syllable index! (The zero'th syllable doesn't count.)
 end
 
 

automation/tests/test-auto3-parsing.auto3

@@ -0,0 +1,10 @@
+name = "Test auto3 parsing"
+description = "Make sure it works"
+version, kind = 3, "basic_ass"
+
+include "karaskel-base.auto3"
+
+function process_lines(meta, styles, lines, config)
+	karaskel.parse_syllable_data(meta, styles, lines)
+	return lines
+end

automation/tests/test-furi.lua

@@ -0,0 +1,42 @@
+script_name = "Test furigana parsing"
+script_description = "Tests the Auto4/Lua karaskel furigana and multi-highlight parsing code by running it and dumping the result"
+script_author = "jfs"
+
+include "karaskel.lua"
+
+function test_furi(subs)
+	aegisub.progress.task("Collecting header data")
+	local meta, styles = karaskel.collect_head(subs, true) -- make sure to create furigana styles
+	
+	aegisub.progress.task("Preprocessing lines")
+	for i = 1, #subs do
+		local l = subs[i]
+		if l.class == "dialogue" then
+			aegisub.progress.task(l.text)
+			karaskel.preproc_line_text(meta, styles, l)
+			
+			-- Dump the thing
+			aegisub.debug.out(4, "Line: %s\nStripped: %s\nDuration: %d\n", l.text, l.text_stripped, l.duration)
+			aegisub.debug.out(4, "Karaoke syllables: (%d)\n", l.kara.n)
+			for s = 0, l.kara.n do
+				local syl = l.kara[s]
+				aegisub.debug.out(4, "\tSyllable: text='%s' stripped='%s' duration=%d, start/end_time=%d/%d, inline_fx='%s', highlights=%d\n", syl.text, syl.text_stripped, syl.duration, syl.start_time, syl.end_time, syl.inline_fx, syl.highlights.n)
+				aegisub.debug.out(4, "\t\tHighlights:")
+				for h = 1, syl.highlights.n do
+					local hl = syl.highlights[h]
+					aegisub.debug.out(4, " %d-%d=%d", hl.start_time, hl.end_time, hl.duration)
+				end
+				aegisub.debug.out(4, "\n")
+			end
+			aegisub.debug.out(4, "Furigana parts: (%d)\n", l.furi.n)
+			for f = 1, l.furi.n do
+				local furi = l.furi[f]
+				aegisub.debug.out(4, "\tFurigana: text='%s', duration=%d, start/end_time=%d/%d, flags=%s%s, syl='%s'\n", furi.text, furi.duration, furi.start_time, furi.end_time, furi.isbreak and "b" or "", furi.spillback and "s" or "", furi.syl.text_stripped)
+			end
+			aegisub.debug.out(4, "  - - - - - -\n")
+		end
+	end
+	aegisub.debug.out(4, "Done dumping!")
+end
+
+aegisub.register_macro(script_name, script_description, test_furi)