Merge pull request #89 from chrox/add_stardict

add a dictionary backend stardict

.gitignore

@@ -9,6 +9,7 @@ lua-*
 git-rev
 *.o
 tags
+test/*
 
 emu
 

Makefile

@@ -25,10 +25,10 @@ XGETTEXT_BIN=$(KOREADER_MISC_TOOL)/gettext/lua_xgettext.py
 MO_DIR=i18n
 
 
-all: $(KOR_BASE)/koreader-base $(KOR_BASE)/extr mo
+all: $(KOR_BASE)/koreader-base $(KOR_BASE)/extr $(KOR_BASE)/sdcv mo
 
-$(KOR_BASE)/koreader-base $(KOR_BASE)/extr:
-	make -C $(KOR_BASE) koreader-base extr
+$(KOR_BASE)/koreader-base $(KOR_BASE)/extr $(KOR_BASE)/sdcv:
+	make -C $(KOR_BASE) koreader-base extr sdcv
 
 fetchthirdparty:
 	git submodule init
@@ -50,6 +50,7 @@ bootstrapemu:
 	test -d $(EMU_DIR)/resources || (cd $(EMU_DIR) && ln -s ../resources ./)
 	test -e $(EMU_DIR)/koreader-base || (cd $(EMU_DIR) && ln -s ../$(KOR_BASE)/koreader-base ./)
 	test -e $(EMU_DIR)/extr || (cd $(EMU_DIR) && ln -s ../$(KOR_BASE)/extr ./)
+	test -e $(EMU_DIR)/sdcv || (cd $(EMU_DIR) && ln -s ../$(KOR_BASE)/sdcv ./)
 	test -e $(EMU_DIR)/reader.lua || (cd $(EMU_DIR) && ln -s ../reader.lua ./)
 	test -e $(EMU_DIR)/history || (mkdir $(EMU_DIR)/history)
 	test -e $(EMU_DIR)/$(MO_DIR) || (cd $(EMU_DIR) && ln -s ../$(MO_DIR) ./)
@@ -63,8 +64,8 @@ customupdate: all
 	rm -rf $(INSTALL_DIR)
 	# create new dir for package
 	mkdir -p $(INSTALL_DIR)/{history,screenshots,clipboard,libs}
-	cp -p README.md COPYING $(KOR_BASE)/{koreader-base,extr} koreader.sh $(LUA_FILES) $(INSTALL_DIR)
-	$(STRIP) --strip-unneeded $(INSTALL_DIR)/koreader-base $(INSTALL_DIR)/extr
+	cp -p README.md COPYING $(KOR_BASE)/{koreader-base,extr,sdcv} koreader.sh $(LUA_FILES) $(INSTALL_DIR)
+	$(STRIP) --strip-unneeded $(INSTALL_DIR)/koreader-base $(INSTALL_DIR)/extr $(INSTALL_DIR)/sdcv
 	mkdir $(INSTALL_DIR)/data
 	cp -L koreader-base/$(DJVULIB) $(KOR_BASE)/$(CRELIB) \
 		$(KOR_BASE)/$(LUALIB) $(KOR_BASE)/$(K2PDFOPTLIB) \

frontend/cache.lua

@@ -20,7 +20,7 @@ A global LRU cache
 ]]--
 Cache = {
 	-- cache configuration:
-	max_memsize = 1024*1024*15, -- 15MB cache size
+	max_memsize = 1024*1024*10, -- 10MB cache size
 	-- cache state:
 	current_memsize = 0,
 	-- associative cache

frontend/document/koptinterface.lua

@@ -7,7 +7,7 @@ require "ui/reader/readerconfig"
 KoptInterface = {
 	tessocr_data = "data",
 	ocr_lang = "eng",
-	ocr_type = 0, -- default, for more accuracy use 3
+	ocr_type = 3, -- default 0, for more accuracy use 3
 }
 
 ContextCacheItem = CacheItem:new{}
@@ -19,6 +19,15 @@ function ContextCacheItem:onFree()
 	end
 end
 
+OCREngine = CacheItem:new{}
+
+function OCREngine:onFree()
+	if self.ocrengine.freeOCR then
+		DEBUG("free OCREngine", self.ocrengine)
+		self.ocrengine:freeOCR()
+	end
+end
+
 function KoptInterface:waitForContext(kc)
 	-- if koptcontext is being processed in background thread
 	-- the isPreCache will return 1.
@@ -107,6 +116,11 @@ function KoptInterface:getPageText(doc, pageno)
 end
 
 function KoptInterface:getOCRWord(doc, pageno, rect)
+	local ocrengine = "ocrengine"
+	if not Cache:check(ocrengine) then
+		local dummy = KOPTContext.new()
+		Cache:insert(ocrengine, OCREngine:new{ ocrengine = dummy })
+	end
 	local bbox = doc:getPageBBox(pageno)
 	local context_hash = self:getContextHash(doc, pageno, bbox)
 	local hash = "ocrword|"..context_hash..rect.x..rect.y..rect.w..rect.h
@@ -117,14 +131,10 @@ function KoptInterface:getOCRWord(doc, pageno, rect)
 		if cached then
 			local kc = self:waitForContext(cached.kctx)
 			local fullwidth, fullheight = kc:getPageDim()
-			--os.execute("echo 3 > /proc/sys/vm/drop_caches")
 			local ok, word = pcall(
-				kc.getOCRWord, kc,
-				self.tessocr_data,
-				self.ocr_lang,
-				self.ocr_type, 
-				rect.x, rect.y,
-				rect.w, rect.h)
+				kc.getTOCRWord, kc,
+				rect.x, rect.y, rect.w, rect.h,
+				self.tessocr_data, self.ocr_lang, self.ocr_type, 0, 1)
 			Cache:insert(hash, CacheItem:new{ ocrword = word })
 			return word
 		end

frontend/ui/device.lua

@@ -7,14 +7,15 @@ Device = {
 }
 
 function Device:getModel()
+	if self.model then return self.model end
 	local std_out = io.popen("grep 'MX' /proc/cpuinfo | cut -d':' -f2 | awk {'print $2'}", "r")
 	local cpu_mod = std_out:read()
 	if not cpu_mod then
 		local ret = os.execute("grep 'Hardware : Mario Platform' /proc/cpuinfo", "r")
 		if ret ~= 0 then
-			return nil
+			self.model = nil
 		else
-			return "KindleDXG"
+			self.model = "KindleDXG"
 		end
 	end
 	if cpu_mod == "MX50" then
@@ -25,20 +26,21 @@ function Device:getModel()
 		-- another special file for KT is Neonode zForce touchscreen:
 		-- /sys/devices/platform/zforce.0/
 		if pw_test_fd then
-			return "KindlePaperWhite"
+			self.model = "KindlePaperWhite"
 		elseif kt_test_fd then
-			return "KindleTouch"
+			self.model = "KindleTouch"
 		else
-			return "Kindle4"
+			self.model = "Kindle4"
 		end
 	elseif cpu_mod == "MX35" then
 		-- check if we are running on Kindle 3 (additional volume input)
-		return "Kindle3"
+		self.model = "Kindle3"
 	elseif cpu_mod == "MX3" then
-		return "Kindle2"
+		self.model = "Kindle2"
 	else
-		return nil
+		self.model = nil
 	end
+	return self.model
 end
 
 function Device:isKindle4()

frontend/ui/reader/readerdictionary.lua

@@ -7,33 +7,52 @@ function ReaderDictionary:init()
 	local dev_mod = Device:getModel()
 	if dev_mod == "KindlePaperWhite" or dev_mod == "KindleTouch" then
 		require("liblipclua")
-		JSON = require("JSON")
 		DEBUG("init lipc handler com.github.koreader.dictionary")
 		self.lipc_handle = lipc.init("com.github.koreader.dictionary")
 	end
+	JSON = require("JSON")
 end
 
 function ReaderDictionary:onLookupWord(word)
+	self:stardictLookup(word)
+end
+
+function ReaderDictionary:nativeDictLookup(word)
 	DEBUG("lookup word:", word)
 	--self:quickLookup()
-	if self.lipc_handle and JSON and word then
+	if self.lipc_handle and word then
 		self.lipc_handle:set_string_property(
 			"com.github.koreader.kpvbooklet.dict", "lookup", word)
 		local results_str = self.lipc_handle:get_string_property(
 			"com.github.koreader.kpvbooklet.word", word)
 		if results_str then
-			--DEBUG("def str:", word, definitions)
+			--DEBUG("result str:", word, results_str)
+			local ok, results_tab = pcall(JSON.decode, JSON, results_str)
+			--DEBUG("lookup result table:", word, results_tab)
+			if results_tab and results_tab[1] then
+				self:showDict(results_tab[1])
+			end
+		end
+	end
+end
+
+function ReaderDictionary:stardictLookup(word)
+	DEBUG("lookup word:", word)
+	if word then
+		local std_out = io.popen("./sdcv -nj "..'\"'..word..'\"', "r")
+		local results_str = std_out:read("*all")
+		if results_str then
+			--DEBUG("result str:", word, results_str)
 			local ok, results_tab = pcall(JSON.decode, JSON, results_str)
-			DEBUG("lookup result table:", word, results_tab)
-			if results_tab[1] then
-				self:quickLookup(results_tab[1])
+			--DEBUG("lookup result table:", word, results_tab)
+			if results_tab and results_tab[1] then
+				self:showDict(results_tab[1])
 			end
 		end
 	end
-	return true
 end
 
-function ReaderDictionary:quickLookup(result)
+function ReaderDictionary:showDict(result)
 --	UIManager:show(DictQuickLookup:new{
 --		dict = "Oxford Dictionary of English",
 --		definition = "coordination n. [mass noun] 1 the organization of the different elements of a \

frontend/ui/reader/readerhighlight.lua

@@ -68,6 +68,10 @@ function ReaderHighlight:onHold(arg, ges)
 		local screen_rect = self.view:pageToScreenTransform(self.pos.page, self.word_info.box)
 		DEBUG("highlight word rect", screen_rect)
 		if screen_rect then
+			screen_rect.x = screen_rect.x - screen_rect.h * 0.2
+			screen_rect.y = screen_rect.y - screen_rect.h * 0.2
+			screen_rect.w = screen_rect.w + screen_rect.h * 0.4
+			screen_rect.h = screen_rect.h + screen_rect.h * 0.4
 			self.view.highlight.rect = screen_rect
 			UIManager:setDirty(self.dialog, "partial")
 			-- if we extracted text directly
@@ -77,11 +81,11 @@ function ReaderHighlight:onHold(arg, ges)
 			else
 				UIManager:scheduleIn(0.1, function()
 					local word_box = self.word_info.box
-					word_box.x = word_box.x - math.floor(word_box.h * 0.1)
-					word_box.y = word_box.y - math.floor(word_box.h * 0.1)
-					word_box.w = word_box.w + math.floor(word_box.h * 0.2)
-					word_box.h = word_box.h + math.floor(word_box.h * 0.2)
-					--local word = self.ui.document:getOCRWord(self.pos.page, word_box)
+					word_box.x = word_box.x - math.floor(word_box.h * 0.2)
+					word_box.y = word_box.y - math.floor(word_box.h * 0.4)
+					word_box.w = word_box.w + math.floor(word_box.h * 0.4)
+					word_box.h = word_box.h + math.floor(word_box.h * 0.6)
+					local word = self.ui.document:getOCRWord(self.pos.page, word_box)
 					DEBUG("OCRed word:", word)
 					self.ui:handleEvent(Event:new("LookupWord", word))
 				end)

frontend/ui/reader/readerview.lua

@@ -289,12 +289,7 @@ function ReaderView:drawScrollView(bb, x, y)
 end
 
 function ReaderView:drawHightlight(bb, x, y, rect)
-	-- slightly enlarge the highlight box
-	-- for better viewing experience
-	local x = rect.x - rect.h * 0.01
-	local y = rect.y - rect.h * 0.01
-	local w = rect.w + rect.h * 0.02
-	local h = rect.h + rect.h * 0.02
+	local x, y, w, h = rect.x, rect.y, rect.w, rect.h
 	
 	self.highlight.drawer = self.highlight.drawer or "underscore"
 	if self.highlight.drawer == "underscore" then

frontend/ui/widget/dict.lua

@@ -11,7 +11,6 @@ DictQuickLookup = InputContainer:new{
 	
 	title_face = Font:getFace("tfont", 20),
 	content_face = Font:getFace("cfont", 18),
-	width = Screen:getWidth() - 100
 }
 
 function DictQuickLookup:init()
@@ -44,13 +43,13 @@ function DictQuickLookup:init()
 				TextBoxWidget:new{
 					text = self.dict,
 					face = self.title_face,
-					width = self.width,
+					width = Screen:getWidth() - 100,
 				},
 				VerticalSpan:new{ width = 20 },
 				TextBoxWidget:new{
 					text = self.definition,
 					face = self.content_face,
-					width = self.width,
+					width = Screen:getWidth() - 100,
 				}
 			}
 		}

koreader.sh

@@ -12,6 +12,9 @@ cd /mnt/us/koreader/
 # export trained OCR data directory
 export TESSDATA_PREFIX="data"
 
+# export dict directory
+export STARDICT_DATA_DIR="data/dict"
+
 # bind-mount system fonts
 if ! grep /mnt/us/koreader/fonts/host /proc/mounts; then
 	mount -o bind /usr/java/lib/fonts /mnt/us/koreader/fonts/host