server/opensubtitles: init search_filesize

4 files changed, 335 insertions, 6 deletions

diff --git a/lib/attr.lua b/lib/attr.lua
new file mode 100644
index 0000000..8c1613a
--- /dev/null
+++ b/lib/attr.lua
@@ -0,0 +1,232 @@
+#!/usr/bin/env lua
+
+local util = require 'lib/util'
+
+local extract = function (name, patterns)
+	local r
+
+	for _, p in pairs(patterns) do
+		r = r or name:match(p)
+		name = name:gsub(p, '')
+	end
+
+	return name, r
+end
+
+local build_dlim = function (name, attrs)
+	local dlim ,r
+
+	attrs = attrs or {}
+	dlim = '[%-%.%s]?'
+	local vcodecs = {
+		'[M]m[Pp][Ee][Gg]' .. dlim .. '[1234]',
+	}
+	local acodecs = {
+		'[Dd][Tt][Ss]' .. dlim .. '[Hh][Dd]',
+		'[Dd][Dd]' .. dlim .. '[57]%.1',
+	}
+	local sources= {
+		"[Ww][Ee][Bb]" .. dlim .. "[Dd][Ll]",
+		"[Hh][Dd]" .. dlim .. "[Tt][Vv]",
+		"[Hh][Dd]" .. dlim .. "[Tt][Ss]",
+	}
+	local series = {
+		'[Ss]%d%d?' .. dlim .. '[Ee]%d%d?',
+	}
+	local sizes = {
+		'%d%d%d' .. dlim .. '[Mm][Bb]',
+		'%d%d?%.%d%d?' ..  dlim .. '[Gg][Bb]',
+	}
+	local depths = {
+		'1[02]' .. dlim .. '[Bb][Ii][Tt]'
+	}
+
+	name, attrs.vcodec = extract(name, vcodecs)
+	name, attrs.source = extract(name, sources)
+	name, attrs.acodecs  = extract(name, acodecs)
+	name, attrs.size  = extract(name, sizes)
+	name, attrs.depth  = extract(name, depths)
+
+	name, r = extract(name, series)
+	if r then
+		attrs.season = tonumber(r:match('%d+'))
+		attrs.episode = tonumber(r:match('%d+$'))
+	end
+
+	return name, attrs
+end
+
+local build_atom = function (name, attrs)
+	local r, year
+
+	attrs = attrs or {}
+	local vcodecs = {
+		"[Aa][Vv]1",
+		"[xXHh]26[345]",
+		"[Aa][Vv][Cc]",
+		"[Hh][Ee][Vv][Cc]",
+		"[Xx][Vv][Ii][Dd]",
+	}
+	local acodecs = {
+		"[Oo][Pp][Uu][Ss]",
+		"[Aa][Aa][Cc]",
+		"[Ee]?[Aa][Cc]3",
+		"[Dd][Tt][Ss]",
+	}
+	local sources= {
+		"[Bb][Ll][Uu][Rr][Aa][Yy]",
+		"[Bb][Rr][Rr][Ii][Pp]",
+		"[Dd][Vv][Dd][Rr][Ii][Pp]",
+		"[Ww][Ee][Bb][Rr][Ii][Pp]",
+		"[Hh][Dd][Rr][Ii][Pp]",
+		"[Rr][Ee][Rr][Ii][Pp]",
+	}
+	local reses = {
+		"2160[Pp]",
+		"1440[Pp]",
+		"1080[Pp]",
+		"720[Pp]",
+		"480[Pp]",
+		"[Uu][Hh][Dd]",
+		"4[Kk]"
+	}
+	local series = {
+		'%d%d[Xx]%d%d',
+	}
+	local channels = {
+		'6[Cc][Hh]',
+		'[57]%.1',
+	}
+
+	name, attrs.vcodec = extract(name, vcodecs)
+	name, attrs.source = extract(name, sources)
+	name, attrs.res  = extract(name, reses)
+	name, attrs.acodecs  = extract(name, acodecs)
+	name, attrs.channel  = extract(name, channels)
+
+	name, r = extract(name, series)
+	if r then
+		attrs.season = tonumber(r:match('%d+'))
+		attrs.episode = tonumber(r:match('%d+$'))
+	end
+
+	for y in name:gmatch('%d%d%d%d') do
+		year = tonumber(y)
+		if year > 1900 and year <= tonumber(os.date('%Y')) then
+			attrs.year = y
+		end
+	end
+	if  attrs.year then
+		name = name:gsub(tostring(attrs.year), '')
+	end
+
+	return name, attrs
+end
+
+local build_low = function (name, attrs)
+	local low_attr, lows
+
+	lows = { 'SDH' }
+
+	low_attr = {}
+	for _, low in pairs(lows) do
+		low_attr[#low_attr + 1] = name:match(low)
+		name = name:gsub(low, '')
+	end
+
+	attrs = attrs or {}
+	if #low_attr > 0 then
+		attrs.low = low_attr
+	end
+
+	return name, attrs
+end
+
+local build_title = function (name, attrs)
+	attrs.title = {}
+	for w in name:gmatch('%w+') do
+		attrs.title[#attrs.title + 1] = w
+	end
+
+	if #attrs.title > 1 then
+		attrs.scene = attrs.title[#attrs.title]
+		attrs.title[#attrs.title] = nil
+	end
+
+	return attrs
+end
+
+local build = function (name)
+	local attrs = {}
+
+	name = build_dlim(name, attrs)
+	name = build_atom(name, attrs)
+	name = build_low(name, attrs)
+	build_title(name, attrs)
+
+	return attrs
+end
+
+local weigh = function (a1, a2)
+	local key_score, score
+
+	key_score = {
+		name = 10,
+		season = 10,
+		episode = 10,
+		source = 7,
+		scene = 5,
+		vcodec = 3,
+		acodec = 3,
+		rese = 2,
+		default = 1,
+	}
+
+	score = 0
+	for k, v in pairs(a1) do
+		if not a2[k] then
+			goto continue
+		end
+
+		if k == 'name' then
+			for _, name in pairs(v) do
+				if util.array_search(a2.name, name) then
+					score = score + key_score.name
+				end
+			end
+		else
+			if v == a2[k] then
+				score = score + (key_score[k] or key_score.default)
+			end
+		end
+
+	    ::continue::
+	end
+
+	return score
+end
+
+local fuzzy = function (name, tab)
+	local name_attr, high, score
+
+	high = {
+		score = 0,
+		name = next(tab)
+	}
+
+	name_attr = build(name)
+	for k in pairs(tab) do
+		score = weigh(name_attr, build(k))
+		if score > high.score then
+			high.score = score
+			high.name = k
+		end
+	end
+
+	return tab[high.name]
+end
+
+return {
+	build = build,
+	fuzzy = fuzzy,
+}
diff --git a/lib/util.lua b/lib/util.lua
index d52922d..6263507 100644
--- a/lib/util.lua
+++ b/lib/util.lua
@@ -70,6 +70,16 @@ local array_merge = function (a1, a2)
 	return a
 end
 
+local array_search = function (a, key)
+	for _, v in pairs(a) do
+		if v == key then
+			return true
+		end
+	end
+
+	return false
+end
+
 local table_print = function (t)
 	for k, v in pairs(t) do
 		print( '|'.. k .. '=' .. v .. '|')
@@ -186,6 +196,7 @@ return {
 	table_print = table_print,
 	table_match_or_any = table_match_or_any,
 	array_merge = array_merge,
+	array_search = array_search,
 	zip_ext_first = zip_ext_first,
 	string_vid_path_to_name = string_vid_path_to_name,
 	opensubtitles_hash = opensubtitles_hash,
diff --git a/main.lua b/main.lua
index f54c673..0f62464 100644
--- a/main.lua
+++ b/main.lua
@@ -42,7 +42,7 @@ local sub_needed = function ()
 end
 
 local sub_setup = function ()
-    local out, name, path, rc
+    local out, name, path, rc, filesize
 
     mp.osd_message('fetching subtitle')
 
@@ -61,8 +61,12 @@ local sub_setup = function ()
     if not util.file_exists(path) then
         name = mp.get_property_native('media-title')
     end
+    filesize = mp.get_property_native('file-size')
 
-    rc = opensubtitles.search(path, out, name)
+    rc = opensubtitles.search(path, out, {
+        name = name,
+        filesize = filesize
+    })
     if not rc then
         rc = subscene.search(path, out, name)
     end
diff --git a/server/opensubtitles.lua b/server/opensubtitles.lua
index b797f44..f0c6195 100644
--- a/server/opensubtitles.lua
+++ b/server/opensubtitles.lua
@@ -2,6 +2,7 @@
 
 local curl = require 'lib/curl'
 local util = require 'lib/util'
+local attr = require 'lib/attr'
 
 -- [[ languages supported by opensubtitles ]] --
 local languages = {
@@ -134,14 +135,94 @@ local search_ohash = function (ohash)
 	end
 end
 
-local search = function (path, out, name)
-	local ohash, link
+local ids_fetch = function (page)
+	local iter, no_name, line, id, name, tab
+
+	tab = {}
+	no_name = 0
+	iter = page:gmatch('[^\n\r]+')
+	while true do
+		line = iter()
+		if not line then
+			break
+		end
+
+		id = line:match('/en/subtitles/%d*')
+		if id then
+			id = id:match('%d+$')
+
+			line = iter() -- movie
+			if line:find('%.%.%.$') then
+				-- name cuts off...
+				name = line:gsub('"[^"]*$', '')
+				name = name:match('[^"]+$')
+			else
+				name = line:gsub('<br/><a rel.*$', '')
+				name = name:match('[^>]+$')
+			end
+
+			if not name then
+				line = iter()
+
+				if line:find('^%[S%d%dE%d%d%]$') then
+					-- it's a series
+					line = iter()
+					if line:find('%.%.%.$') then
+						name = line:gsub('^.*title="', '')
+						name = name:match('[^"]+')
+					else
+						name = line:match('[^<]+')
+					end
+				else
+					-- no name
+					name = tostring(no_name)
+					no_name = no_name + 1
+				end
+			end
+
+			tab[name] = id
+		end
+	end
+
+	return tab
+end
+
+local search_filesize = function (filesize, name)
+	local fetch, hcode, url, id, a
+
+	a = attr.build(name)
+
+	url = domain .. '/en' .. '/search/sublanguageid-' .. languages[language]
+	if a.season and a.episode then
+		url = url .. '/season-' .. a.season .. '/episode-' .. a.episode
+	end
+	url = url .. '/moviebytesize-' .. filesize
+
+	fetch, hcode = curl.get(url, nil, nil, tries)
+	if not hcode then
+		return nil
+	end
+
+	print(url)
+	util.table_print(ids_fetch(fetch))
+	id = attr.fuzzy(name, ids_fetch(fetch))
+	if id then
+		print(domain .. '/en/subtitleserve/sub/' .. id)
+		return domain .. '/en/subtitleserve/sub/' .. id
+	end
+end
+
+local search = function (path, out, info)
+	local ohash, link, name
 
 	if util.file_exists(path) then
 		ohash = util.opensubtitles_hash(path)
 		link = search_ohash(ohash)
-	else
-		name = name or util.string_vid_path_to_name(path)
+	end
+
+	if not link then
+		name = info.name or util.string_vid_path_to_name(path)
+		link = search_filesize(info.filesize, name)
 	end
 
 	if link then
@@ -151,5 +232,6 @@ end
 
 return {
 	search_ohash = search_ohash,
+	search_filesize = search_filesize,
 	search = search
 }