add document language option in config dialog

Document language setting is used for better text extraction and OCR.

defaults.lua

@@ -41,6 +41,10 @@ DKOPTREADER_CONFIG_JUSTIFICATION = -1	-- -1 = auto, 0 = left, 1 = center, 2 = ri
 DKOPTREADER_CONFIG_MAX_COLUMNS = 2		-- range from 1 to 4
 DKOPTREADER_CONFIG_CONTRAST = 1.0		-- range from 0.2 to 2.0
 
+-- document languages for OCR
+DKOPTREADER_CONFIG_DOC_LANGS_TEXT = {"English", "Chinese_S", "Chinese_T"}
+DKOPTREADER_CONFIG_DOC_LANGS_CODE = {"eng", "chi_sim", "chi_tra"}
+DKOPTREADER_CONFIG_DOC_DEFAULT_LANG_CODE = "eng"
 
 -- ####################################################################
 -- following features are not supported right now

frontend/document/koptinterface.lua

@@ -62,6 +62,7 @@ function KoptInterface:createContext(doc, pageno, bbox)
 	kc:setDefectSize(doc.configurable.defect_size)
 	kc:setLineSpacing(doc.configurable.line_spacing)
 	kc:setWordSpacing(doc.configurable.word_spacing)
+	kc:setLanguage(doc.configurable.doc_language)
 	kc:setBBox(bbox.x0, bbox.y0, bbox.x1, bbox.y1)
 	if Dbg.is_on then kc:setDebug() end
 	return kc
@@ -129,6 +130,13 @@ function KoptInterface:getReflewTextBoxes(doc, pageno)
 		local cached = Cache:check(kctx_hash)
 		if cached then
 			local kc = self:waitForContext(cached.kctx)
+			--kc:setDebug()
+			local lang = doc.configurable.doc_language
+			if lang == "chi_sim" or lang == "chi_tra" or 
+				lang == "jpn" or lang == "kor" then
+				kc:setCJKChar()
+			end
+			kc:setLanguage(lang)
 			local fullwidth, fullheight = kc:getPageDim()
 			local boxes = kc:getWordBoxes(0, 0, fullwidth, fullheight)
 			Cache:insert(hash, CacheItem:new{ rfpgboxes = boxes })
@@ -146,6 +154,12 @@ function KoptInterface:getTextBoxes(doc, pageno)
 		local kc_hash = "kctx|"..doc.file.."|"..pageno
 		local kc = KOPTContext.new()
 		kc:setDebug()
+		local lang = doc.configurable.doc_language
+		if lang == "chi_sim" or lang == "chi_tra" or 
+			lang == "jpn" or lang == "kor" then
+			kc:setCJKChar()
+		end
+		kc:setLanguage(lang)
 		local page = doc._document:openPage(pageno)
 		page:getPagePix(kc)
 		local fullwidth, fullheight = kc:getPageDim()
@@ -167,6 +181,7 @@ function KoptInterface:getReflewOCRWord(doc, pageno, rect)
 		local dummy = KOPTContext.new()
 		Cache:insert(ocrengine, OCREngine:new{ ocrengine = dummy })
 	end
+	self.ocr_lang = doc.configurable.doc_language
 	local bbox = doc:getPageBBox(pageno)
 	local context_hash = self:getContextHash(doc, pageno, bbox)
 	local hash = "rfocrword|"..context_hash..rect.x..rect.y..rect.w..rect.h
@@ -197,6 +212,7 @@ function KoptInterface:getOCRWord(doc, pageno, rect)
 		local dummy = KOPTContext.new()
 		Cache:insert(ocrengine, OCREngine:new{ ocrengine = dummy })
 	end
+	self.ocr_lang = doc.configurable.doc_language
 	local hash = "ocrword|"..doc.file.."|"..pageno..rect.x..rect.y..rect.w..rect.h
 	local cached = Cache:check(hash)
 	if not cached then

frontend/ui/data/koptoptions.lua

@@ -160,6 +160,13 @@ KoptOptions = {
 					},
 				}
 			},
+			{
+				name="doc_language",
+				name_text = DOC_LANG_STR,
+				toggle = DKOPTREADER_CONFIG_DOC_LANGS_TEXT,
+				values = DKOPTREADER_CONFIG_DOC_LANGS_CODE,
+				default_value = DKOPTREADER_CONFIG_DOC_DEFAULT_LANG_CODE,
+			},
 			{
 				name="screen_rotation",
 				name_text = VERTICAL_TEXT_STR,
@@ -195,6 +202,7 @@ KoptOptions = {
 				toggle = {ZERO_DEG_STR, FIVE_DEG_STR, TEN_DEG_STR},
 				values = {0, 5, 10},
 				default_value = DKOPTREADER_CONFIG_AUTO_STRAIGHTEN,
+				show = false,
 			},
 			{
 				name = "detect_indent",

frontend/ui/data/strings.lua

@@ -10,6 +10,7 @@ TEXT_ALIGN_STR = _("Text Align")
 FONTSIZE_FINE_TUNING_STR = _("Fine Tuning")
 CONTRAST_STR = _("Contrast")
 REFLOW_STR = _("Reflow")
+DOC_LANG_STR = _("Document Language")
 VERTICAL_TEXT_STR = _("Vertical Text")
 WORD_GAP_STR = _("Word Gap")
 DEFECT_SIZE_STR = _("Defect Size")

frontend/ui/reader/readerconfig.lua

@@ -6,7 +6,8 @@ function Configurable:hash(sep)
 	local hash = ""
 	local excluded = {multi_threads = true,}
 	for key,value in pairs(self) do
-		if type(value) == "number" and not excluded[key] then
+		if type(value) == "number" or type(value) == "string" 
+			and not excluded[key] then
 			hash = hash..sep..value
 		end
 	end
@@ -28,8 +29,9 @@ end
 
 function Configurable:loadSettings(settings, prefix)
 	for key,value in pairs(self) do
-		if type(value) == "number" or type(value) == "table" then
-			saved_value = settings:readSetting(prefix..key)
+		if type(value) == "number" or type(value) == "string"
+			or type(value) == "table" then
+			local saved_value = settings:readSetting(prefix..key)
 			self[key] = (saved_value == nil) and self[key] or saved_value
 			--Debug("Configurable:loadSettings", "key", key, "saved value", saved_value,"Configurable.key", self[key])
 		end
@@ -39,7 +41,8 @@ end
 
 function Configurable:saveSettings(settings, prefix)
 	for key,value in pairs(self) do
-		if type(value) == "number" or type(value) == "table" then
+		if type(value) == "number" or type(value) == "string"
+			or type(value) == "table" then
 			settings:saveSetting(prefix..key, value)
 		end
 	end

frontend/ui/widget/config.lua

@@ -204,6 +204,16 @@ function ConfigOption:init()
 			end
 			-- make current index according to configurable table
 			local current_item = nil
+			local function value_diff(val1, val2, name)
+				if type(val1) ~= type(val2) then
+					error("different data types in option", name)
+				end
+				if type(val1) == "number" then
+					return math.abs(val1 - val2)
+				elseif type(val1) == "string" then
+					return val1 == val2 and 0 or 1
+				end
+			end
 			if self.options[c].name then
 				if self.options[c].values then
 					-- check if current value is stored in configurable or calculated in runtime
@@ -211,18 +221,18 @@ function ConfigOption:init()
 								or self.config.configurable[self.options[c].name]
 					local min_diff = nil
 					if type(val) == "table" then
-						min_diff = math.abs(val[1] - self.options[c].values[1][1])
+						min_diff = value_diff(val[1], self.options[c].values[1][1])
 					else
-						min_diff = math.abs(val - self.options[c].values[1])
+						min_diff = value_diff(val, self.options[c].values[1])
 					end
 						
 					local diff = nil
 					for index, val_ in pairs(self.options[c].values) do
 						local diff = nil
 						if type(val) == "table" then
-							diff = math.abs(val[1] - val_[1])
+							diff = value_diff(val[1], val_[1])
 						else
-							diff = math.abs(val - val_)
+							diff = value_diff(val, val_)
 						end
 						if val == val_ then
 							current_item = index