Merge pull request #340 from chrox/master

set auto detect word gap in reflowing by default

defaults.lua

@@ -48,13 +48,13 @@ DKOPTREADER_CONFIG_PAGE_MARGIN = 0.10	-- range from 0.0 to 1.0
 DKOPTREADER_CONFIG_LINE_SPACING = 1.2	-- range from 0.5 to 2.0
 DKOPTREADER_CONFIG_RENDER_QUALITY = 1.0	-- range from 0.5 to 2.0
 DKOPTREADER_CONFIG_AUTO_STRAIGHTEN = 0	-- range from 0 to 10
-DKOPTREADER_CONFIG_JUSTIFICATION = -1	-- -1 = auto, 0 = left, 1 = center, 2 = right, 3 = full
+DKOPTREADER_CONFIG_JUSTIFICATION = 3	-- -1 = auto, 0 = left, 1 = center, 2 = right, 3 = full
 DKOPTREADER_CONFIG_MAX_COLUMNS = 2		-- range from 1 to 4
 DKOPTREADER_CONFIG_CONTRAST = 1.0		-- range from 0.2 to 2.0
 
 -- word spacing for reflow
-DKOPTREADER_CONFIG_WORD_SAPCINGS = {0.05, 0.15, 0.375}  -- range from 0.05 to 0.5
-DKOPTREADER_CONFIG_DEFAULT_WORD_SAPCING = 0.15			-- range from 0.05 to 0.5
+DKOPTREADER_CONFIG_WORD_SAPCINGS = {0.05, -1, 0.375}    -- range from 0.05 to 0.5
+DKOPTREADER_CONFIG_DEFAULT_WORD_SAPCING = -1			-- range from 0.05 to 0.5
 -- document languages for OCR
 DKOPTREADER_CONFIG_DOC_LANGS_TEXT = {"English", "Chinese"}
 DKOPTREADER_CONFIG_DOC_LANGS_CODE = {"eng", "chi_sim"}	-- language code, make sure you have corresponding training data

frontend/document/koptinterface.lua

@@ -58,7 +58,6 @@ function KoptInterface:createContext(doc, pageno, bbox)
 		lang == "jpn" or lang == "kor" then
 		kc:setCJKChar()
 	end
-	DEBUG("configurable", doc.configurable)
 	kc:setLanguage(lang)
 	kc:setTrim(doc.configurable.trim_page)
 	kc:setWrap(doc.configurable.text_wrap)
@@ -339,9 +338,8 @@ function KoptInterface:getTextBoxes(doc, pageno)
 	else
 		if doc.configurable.text_wrap == 1 then
 			return self:getNativeTextBoxes(doc, pageno)
-			--return self:getTextBoxesFromScratch(doc, pageno)
 		else
-			return self:getTextBoxesFromScratch(doc, pageno)
+			return self:getNativeTextBoxesFromScratch(doc, pageno)
 		end
 	end
 end
@@ -394,12 +392,39 @@ function KoptInterface:getNativeTextBoxes(doc, pageno)
 	end
 end
 
+--[[
+get text boxes in reflowed page via optical method, 
+i.e. OCR pre-processing in Tesseract and Leptonica.
+--]]
+function KoptInterface:getReflowedTextBoxesFromScratch(doc, pageno)
+	local bbox = doc:getPageBBox(pageno)
+	local context_hash = self:getContextHash(doc, pageno, bbox)
+	local hash = "scratchrfpgboxes|"..context_hash
+	local cached = Cache:check(hash)
+	if not cached then
+		local kctx_hash = "kctx|"..context_hash
+		local cached = Cache:check(kctx_hash)
+		if cached then
+			local reflowed_kc = self:waitForContext(cached.kctx)
+			local fullwidth, fullheight = reflowed_kc:getPageDim()
+			local kc = self:createContext(doc, pageno)
+			kc:copyDestBMP(reflowed_kc)
+			local boxes = kc:getNativeWordBoxes(0, 0, fullwidth, fullheight)
+			Cache:insert(hash, CacheItem:new{ scratchrfpgboxes = boxes })
+			kc:free()
+			return boxes
+		end
+	else
+		return cached.scratchrfpgboxes
+	end
+end
+
 --[[
 get text boxes in native page via optical method, 
 i.e. OCR pre-processing in Tesseract and Leptonica.
 --]]
-function KoptInterface:getTextBoxesFromScratch(doc, pageno)
-	local hash = "pgboxes|"..doc.file.."|"..pageno
+function KoptInterface:getNativeTextBoxesFromScratch(doc, pageno)
+	local hash = "scratchnativepgboxes|"..doc.file.."|"..pageno
 	local cached = Cache:check(hash)
 	if not cached then
 		local page_size = Document.getNativePageDimensions(doc, pageno)
@@ -413,12 +438,12 @@ function KoptInterface:getTextBoxesFromScratch(doc, pageno)
 		local page = doc._document:openPage(pageno)
 		page:getPagePix(kc)
 		local boxes = kc:getNativeWordBoxes(0, 0, page_size.w, page_size.h)
-		Cache:insert(hash, CacheItem:new{ pgboxes = boxes })
+		Cache:insert(hash, CacheItem:new{ scratchnativepgboxes = boxes })
 		page:close()
 		kc:free()
 		return boxes
 	else
-		return cached.pgboxes
+		return cached.scratchnativepgboxes
 	end
 end
 
@@ -644,7 +669,7 @@ get word and word box from position in reflowed page
 ]]--
 function KoptInterface:getWordFromReflowPosition(doc, boxes, pos)
 	local pageno = pos.page
-	local reflowed_page_boxes = self:getReflowedTextBoxes(doc, pageno)
+	local reflowed_page_boxes = self:getReflowedTextBoxesFromScratch(doc, pageno)
 	local reflowed_word_box = self:getWordFromBoxes(reflowed_page_boxes, pos)
 	local reflowed_pos = reflowed_word_box.box:center()
 	local native_pos = self:reflowToNativePosTransform(doc, pageno, reflowed_pos)

frontend/ui/data/koptoptions.lua

@@ -182,7 +182,7 @@ local KoptOptions = {
 			{
 				name = "word_spacing",
 				name_text = S.WORD_GAP,
-				toggle = {S.SMALL, S.MEDIUM, S.LARGE},
+				toggle = {S.SMALL, S.AUTO, S.LARGE},
 				values = DKOPTREADER_CONFIG_WORD_SAPCINGS,
 				default_value = DKOPTREADER_CONFIG_DEFAULT_WORD_SAPCING,
 			},