Add option for custom hyphenation rules (#7787)

This is the successor of #7746.

frontend/apps/reader/modules/readerhighlight.lua

@@ -174,6 +174,20 @@ function ReaderHighlight:init()
         }
     end)
 
+    -- User hyphenation dict
+    self:addToHighlightDialog("11_user_dict", function(_self)
+        return {
+            text= _("Hyphenate"),
+            show_in_highlight_dialog_func = function()
+                return _self.ui.userhyph and _self.ui.userhyph:isAvailable() and not _self.selected_text.text:find("[ ,;-%.\n]")
+            end,
+            callback = function()
+                _self.ui.userhyph:modifyUserEntry(_self.selected_text.text)
+                _self:onClose()
+            end,
+        }
+    end)
+
     self.ui:registerPostInitCallback(function()
         self.ui.menu:registerToMainMenu(self)
     end)

frontend/apps/reader/modules/readertypography.lua

@@ -237,6 +237,7 @@ When the book's language tag is not among our presets, no specific features will
                 })
                 self.text_lang_tag = lang_tag
                 self.ui.document:setTextMainLang(lang_tag)
+                self.ui:handleEvent(Event:new("TypographyLanguageChanged"))
                 self.ui:handleEvent(Event:new("UpdatePos"))
             end,
             hold_callback = function(touchmenu_instance)
@@ -427,8 +428,8 @@ These settings will apply to all books with any hyphenation dictionary.
         enabled_func = function()
             return self.hyphenation and not self.hyph_soft_hyphens_only
         end,
-        separator = true,
     })
+    table.insert(hyphenation_submenu, self.ui.userhyph:getMenuEntry())
     table.insert(hyphenation_submenu, {
         text_func = function()
             -- Show the current language default hyph dict (ie: English_US for zh)
@@ -488,7 +489,7 @@ These settings will apply to all books with any hyphenation dictionary.
         end,
     })
     table.insert(hyphenation_submenu, {
-        text = _("Soft-hyphens only"),
+        text = _("Soft hyphens only"),
         callback = function()
             self.hyph_soft_hyphens_only = not self.hyph_soft_hyphens_only
             self.hyph_force_algorithmic = false
@@ -760,6 +761,7 @@ function ReaderTypography:onReadSettings(config)
         logger.dbg("Typography lang: no lang set, using", self.text_lang_tag)
     end
     self.ui.document:setTextMainLang(self.text_lang_tag)
+    self.ui:handleEvent(Event:new("TypographyLanguageChanged"))
 end
 
 function ReaderTypography:onPreRenderDocument(config)
@@ -780,6 +782,7 @@ function ReaderTypography:onPreRenderDocument(config)
             self.text_lang_tag = self.book_lang_tag
             self.ui.doc_settings:saveSetting("text_lang", self.text_lang_tag)
             self.ui.document:setTextMainLang(self.text_lang_tag)
+            self.ui:handleEvent(Event:new("TypographyLanguageChanged"))
             self.ui:handleEvent(Event:new("UpdatePos"))
         end,
         enabled_func = function()
@@ -809,6 +812,7 @@ function ReaderTypography:onPreRenderDocument(config)
         end
         self.text_lang_tag = self.book_lang_tag
         self.ui.document:setTextMainLang(self.text_lang_tag)
+        self.ui:handleEvent(Event:new("TypographyLanguageChanged"))
     end
 end
 

frontend/apps/reader/modules/readeruserhyph.lua

@@ -0,0 +1,228 @@
+local DataStorage = require("datastorage")
+local Event = require("ui/event")
+local FFIUtil = require("ffi/util")
+local InfoMessage = require("ui/widget/infomessage")
+local InputDialog = require("ui/widget/inputdialog")
+local UIManager = require("ui/uimanager")
+local WidgetContainer = require("ui/widget/container/widgetcontainer")
+local lfs = require("libs/libkoreader-lfs")
+local logger = require("logger")
+local _ = require("gettext")
+local T = require("ffi/util").template
+
+local ReaderUserHyph = WidgetContainer:new{
+    -- return values from setUserHyphenationDict (crengine's UserHyphDict::init())
+    USER_DICT_RELOAD = 0,
+    USER_DICT_NOCHANGE = 1,
+    USER_DICT_MALFORMED = 2,
+    USER_DICT_ERROR_NOT_SORTED = 3,
+}
+
+-- returns path to the user dictionary
+function ReaderUserHyph:getDictionaryPath()
+    return FFIUtil.joinPath(DataStorage:getSettingsDir(),
+        "user-" .. tostring(self.ui.document:getTextMainLangDefaultHyphDictionary():gsub(".pattern$", "")) .. ".hyph")
+end
+
+-- Load the user dictionary suitable for the actual language
+-- if reload==true, force a reload
+-- Unload is done automatically when a new dictionary is loaded.
+function ReaderUserHyph:loadDictionary(name, reload)
+    if G_reader_settings:isTrue("hyph_user_dict") and lfs.attributes(name, "mode") == "file" then
+        local ret = self.ui.document:setUserHyphenationDict(name, reload)
+        -- this should only happen, if a user edits a dictionary by hand or the user messed
+        -- with the dictionary file by hand. -> Warning and disable.
+        if ret == self.USER_DICT_ERROR_NOT_SORTED then
+            UIManager:show(InfoMessage:new{
+                text = T(_("The user dictionary\n%1\nis not alphabetically sorted.\n\nIt has been disabled."), name),
+            })
+            logger.warn("UserHyph: Dictionary " .. name .. " is not sorted alphabetically.")
+            G_reader_settings:makeFalse("hyph_user_dict")
+        elseif ret == self.USER_DICT_MALFORMED then
+            UIManager:show(InfoMessage:new{
+                text = T(_("The user dictionary\n%1\nhas corrupted entries.\n\nOnly valid entries will be used."), name),
+            })
+            logger.warn("UserHyph: Dictionary " .. name .. " has corrupted entries.")
+        end
+    else
+        self.ui.document:setUserHyphenationDict() -- clear crengine user hyph dict
+    end
+end
+
+-- Reload on change of the hyphenation language
+function ReaderUserHyph:onTypographyLanguageChanged()
+    self:loadUserDictionary()
+end
+
+-- Reload on "ChangedUserDictionary" event,
+-- doesn't load dictionary if filesize and filename haven't changed
+-- if reload==true reload
+function ReaderUserHyph:loadUserDictionary(reload)
+    self:loadDictionary(self:isAvailable() and self:getDictionaryPath() or "", reload and true or false)
+    self.ui:handleEvent(Event:new("UpdatePos"))
+end
+
+-- Functions to use with the UI
+
+function ReaderUserHyph:isAvailable()
+    return G_reader_settings:isTrue("hyph_user_dict") and self:_enabled()
+end
+
+function ReaderUserHyph:_enabled()
+    return self.ui.typography.hyphenation
+end
+
+-- add Menu entry
+function ReaderUserHyph:getMenuEntry()
+    return {
+        text = _("Custom hyphenation rules"),
+        help_text = _("The hyphenation of a word can be changed from its default by long pressing for 3 seconds and selecting 'Hyphenate'."),
+        callback = function()
+            local hyph_user_dict =  not G_reader_settings:isTrue("hyph_user_dict")
+            G_reader_settings:saveSetting("hyph_user_dict", hyph_user_dict)
+            self:loadUserDictionary() -- not needed to force a reload here
+        end,
+        checked_func = function()
+            return self:isAvailable()
+        end,
+        enabled_func = function()
+            return self:_enabled()
+        end,
+        separator = true,
+    }
+end
+
+-- Helper functions for dictionary entries-------------------------------------------
+
+-- checks if suggestion is well formated
+function ReaderUserHyph:checkHyphenation(suggestion, word)
+    if suggestion:find("%-%-") then
+        return false -- two or more consecutive '-'
+    end
+
+    suggestion = suggestion:gsub("-","")
+    if self.ui.document:getLowercasedWord(suggestion) == self.ui.document:getLowercasedWord(word) then
+        return true -- characters match (case insensitive)
+    end
+    return false
+end
+
+function ReaderUserHyph:updateDictionary(word, hyphenation)
+    local dict_file = self:getDictionaryPath()
+    local new_dict_file = dict_file .. ".new"
+
+    local new_dict = io.open(new_dict_file, "w")
+    if not new_dict then
+        logger.err("UserHyph: could not open " .. new_dict_file)
+        return
+    end
+
+    local word_lower = self.ui.document:getLowercasedWord(word)
+    local line
+
+    local dict = io.open(dict_file, "r")
+    if dict then
+        line = dict:read()
+        --search entry
+        while line and self.ui.document:getLowercasedWord(line:sub(1, line:find(";") - 1)) < word_lower do
+            new_dict:write(line .. "\n")
+            line = dict:read()
+        end
+
+        -- last word = nil if EOF, else last_word=word if found in file, else last_word is word after the new entry
+        if line then
+            local last_word = self.ui.document:getLowercasedWord(line:sub(1, line:find(";") - 1))
+            if last_word == self.ui.document:getLowercasedWord(word) then
+                line = nil -- word found
+            end
+        else
+            line = nil -- EOF
+        end
+    end
+
+    -- write new entry
+    if hyphenation and hyphenation ~= "" then
+        new_dict:write(string.format("%s;%s\n", word, hyphenation))
+    end
+
+    -- write old entry if there was one
+    if line then
+        new_dict:write(line .. "\n")
+    end
+
+    if dict then
+        repeat
+            line = dict:read()
+            if line then
+                new_dict:write(line .. "\n")
+            end
+        until (not line)
+        dict:close()
+        os.remove(dict_file)
+    end
+
+    new_dict:close()
+    os.rename(new_dict_file, dict_file)
+
+    self:loadUserDictionary(true) -- dictionary has changed, force a reload here
+end
+
+function ReaderUserHyph:modifyUserEntry(word)
+    if word:find("[ ,;-%.]") then return end -- no button if more than one word
+
+    if not self.ui.document then return end
+
+    local suggested_hyphenation = self.ui.document:getHyphenationForWord(word)
+
+    local input_dialog
+    input_dialog = InputDialog:new{
+        title = T(_("Hyphenate: %1"), word),
+        description = _("Add hyphenation positions with hyphens ('-') or spaces (' ')."),
+        input = suggested_hyphenation,
+        old_hyph_lowercase = self.ui.document:getLowercasedWord(suggested_hyphenation),
+        input_type = "string",
+        buttons = {
+            {
+                {
+                    text = _("Cancel"),
+                    callback = function()
+                        UIManager:close(input_dialog)
+                    end,
+                },
+                {
+                    text = _("Remove"),
+                    callback = function()
+                        UIManager:close(input_dialog)
+                        self:updateDictionary(word)
+                    end,
+                },
+                {
+                    text = _("Save"),
+                    is_enter_default = true,
+                    callback = function()
+                        local new_suggestion = input_dialog:getInputText()
+                        new_suggestion = new_suggestion:gsub(" ","-") -- replace spaces with hyphens
+                        new_suggestion = new_suggestion:gsub("^-","") -- remove leading hypenations
+                        new_suggestion = new_suggestion:gsub("-$","") -- remove trailing hypenations
+
+                        if self:checkHyphenation(new_suggestion, word) then
+                            -- don't save if no changes
+                            if self.ui.document:getLowercasedWord(new_suggestion) ~= input_dialog.old_hyph_lowercase then
+                                self:updateDictionary(word, new_suggestion)
+                            end
+                            UIManager:close(input_dialog)
+                        else
+                            UIManager:show(InfoMessage:new{
+                                text = T(_("Invalid hyphenation!"), self.dict_file),
+                            })
+                        end
+                    end,
+                },
+            },
+        },
+    }
+    UIManager:show(input_dialog)
+    input_dialog:onShowKeyboard()
+end
+
+return ReaderUserHyph

frontend/apps/reader/readerui.lua

@@ -48,6 +48,7 @@ local ReaderStyleTweak = require("apps/reader/modules/readerstyletweak")
 local ReaderToc = require("apps/reader/modules/readertoc")
 local ReaderTypeset = require("apps/reader/modules/readertypeset")
 local ReaderTypography = require("apps/reader/modules/readertypography")
+local ReaderUserHyph = require("apps/reader/modules/readeruserhyph")
 local ReaderView = require("apps/reader/modules/readerview")
 local ReaderWikipedia = require("apps/reader/modules/readerwikipedia")
 local ReaderZooming = require("apps/reader/modules/readerzooming")
@@ -314,6 +315,12 @@ function ReaderUI:init()
             view = self.view,
             ui = self
         })
+        -- user hyphenation (must be registered before typography)
+        self:registerModule("userhyph", ReaderUserHyph:new{
+            dialog = self.dialog,
+            view = self.view,
+            ui = self
+        })
         -- typography menu (replaces previous hyphenation menu / ReaderHyphenation)
         self:registerModule("typography", ReaderTypography:new{
             dialog = self.dialog,

frontend/document/credocument.lua

@@ -980,6 +980,25 @@ function CreDocument:setTextHyphenationSoftHyphensOnly(toggle)
     self._document:setStringProperty("crengine.textlang.hyphenation.soft.hyphens.only", toggle and 1 or 0)
 end
 
+function CreDocument:setUserHyphenationDict(dict, reload)
+    logger.dbg("CreDocument: set textlang hyphenation dict", dict or "none")
+    return self._document:setUserHyphenationDict(dict or "", reload or false)
+end
+
+function CreDocument:getHyphenationForWord(word)
+    if word then
+        return self._document:getHyphenationForWord(word)
+    end
+    return word
+end
+
+function CreDocument:getLowercasedWord(word)
+    if word then
+        return self._document:getLowercasedWord(word)
+    end
+    return word
+end
+
 function CreDocument:setTextHyphenationForceAlgorithmic(toggle)
     logger.dbg("CreDocument: set textlang hyphenation force algorithmic", toggle)
     self._document:setStringProperty("crengine.textlang.hyphenation.force.algorithmic", toggle and 1 or 0)