Merge pull request #2114 from chrox/fix_1705

split accient greek words with spacing character

frontend/apps/reader/modules/readerdictionary.lua

@@ -12,6 +12,7 @@ local T = require("ffi/util").template
 
 local ReaderDictionary = InputContainer:new{
     data_dir = nil,
+    dict_window_list = {},
 }
 
 function ReaderDictionary:init()
@@ -104,6 +105,7 @@ function ReaderDictionary:showDict(word, results, box)
     if results and results[1] then
         DEBUG("showing quick lookup window", word, results)
         self.dict_window = DictQuickLookup:new{
+            window_list = self.dict_window_list,
             ui = self.ui,
             highlight = self.highlight,
             dialog = self.dialog,
@@ -116,6 +118,7 @@ function ReaderDictionary:showDict(word, results, box)
             -- differentiate between dict and wiki
             wiki = self.wiki,
         }
+        table.insert(self.dict_window_list, self.dict_window)
         UIManager:show(self.dict_window)
     end
 end

frontend/ui/widget/closebutton.lua

@@ -52,10 +52,27 @@ function CloseButton:init()
         },
         doc = "Tap on close button",
     }
+
+    self.ges_events.HoldClose = {
+        GestureRange:new{
+            ges = "hold_release",
+            range = function() return self.dimen end,
+        },
+        doc = "Hold on close button",
+    }
 end
 
 function CloseButton:onClose()
-    self.window:onClose()
+    if self.window.onClose then
+        self.window:onClose()
+    end
+    return true
+end
+
+function CloseButton:onHoldClose()
+    if self.window.onHoldClose then
+        self.window:onHoldClose()
+    end
     return true
 end
 

frontend/ui/widget/dictquicklookup.lua

@@ -88,7 +88,7 @@ function DictQuickLookup:init()
                 -- callback function when HoldWord is handled as args
                 args = function(word)
                     self.ui:handleEvent(
-                        Event:new("LookupWord", word, self.word_box))
+                        Event:new("LookupWord", word, self.word_box, self.highlight))
                 end
             },
         }
@@ -384,8 +384,24 @@ end
 
 function DictQuickLookup:onClose()
     UIManager:close(self)
+    for i = #self.window_list, 1, -1 do
+        local window = self.window_list[i]
+        if window == self then
+            table.remove(self.window_list, i)
+        end
+    end
     if self.highlight then
-        self.highlight:handleEvent(Event:new("Tap"))
+        self.highlight:clear()
+    end
+    return true
+end
+
+function DictQuickLookup:onHoldClose()
+    self:onClose()
+    for i = #self.window_list, 1, -1 do
+        local window = self.window_list[i]
+        UIManager:close(window)
+        table.remove(self.window_list, i)
     end
     return true
 end

frontend/ui/widget/textboxwidget.lua

@@ -19,6 +19,7 @@ local RenderText = require("ui/rendertext")
 local Screen = require("device").screen
 local Geom = require("ui/geometry")
 local util = require("util")
+local DEBUG= require("dbg")
 
 local TextBoxWidget = Widget:new{
     text = nil,
@@ -282,7 +283,7 @@ function TextBoxWidget:onHoldWord(callback, ges)
     local x, y = ges.pos.x - self.dimen.x, ges.pos.y - self.dimen.y
     local line_num = math.ceil(y / self.line_height_px)
     local line = self.vertical_string_list[line_num]
-
+    DEBUG("holding on line", line)
     if line then
         local char_start = line.offset
         local char_end  -- char_end is non-inclusive
@@ -304,10 +305,10 @@ function TextBoxWidget:onHoldWord(callback, ges)
                 -- now find which word the character is in
                 local words = util.splitToWords(line.text)
                 local probe_idx = char_start
-                for _,w in ipairs(words) do
+                for _, w in ipairs(words) do
                     -- +1 for word separtor
-                    probe_idx = probe_idx + string.len(w)
-                    if idx <= probe_idx then
+                    probe_idx = probe_idx + #util.splitToChars(w)
+                    if idx <= probe_idx - 1 then
                         callback(w)
                         return
                     end

frontend/util.lua

@@ -121,10 +121,15 @@ end
 ---- @string text text to split
 ---- @treturn table list of words, spaces and punctuations
 function util.splitToWords(text)
-    -- TODO: write test
     local wlist = {}
-    for words in text:gmatch("[\32-\127\192-\255]+[\128-\191]*") do
-        for word in util.gsplit(words, "[%s%p]+", true) do
+    for word in util.gsplit(text, "[%s%p]+", true) do
+        -- if space splitted word contains CJK characters
+        if word:match("[\228-\234][\128-\191]+") then
+            -- split with CJK characters
+            for char in util.gsplit(word, "[\228-\234\192-\255][\128-\191]+", true) do
+                table.insert(wlist, char)
+            end
+        else
             table.insert(wlist, word)
         end
     end

spec/unit/util_spec.lua

@@ -52,4 +52,28 @@ describe("util module", function()
             "five",
         })
     end)
+
+    it("should split ancient greek words", function()
+        local words = util.splitToWords("Λαρισαῖος Λευκοθέα Λιγυαστάδης.")
+        assert.are_same(words, {
+            "Λαρισαῖος",
+            " ",
+            "Λευκοθέα",
+            " ",
+            "Λιγυαστάδης",
+            "."
+        })
+    end)
+
+    it("should split Chinese words", function()
+        local words = util.splitToWords("彩虹是通过太阳光的折射引起的。")
+        assert.are_same(words, {
+            "彩","虹","是","通","过","太","阳","光","的","折","射","引","起","的","。",
+        })
+    end)
+
+    it("should split words of multilingual text", function()
+        local words = util.splitToWords("BBC纪录片")
+        assert.are_same(words, {"BBC", "纪", "录", "片"})
+    end)
 end)