Dict/Wiki lookup: less text cleanup on manual input

Don't cleanup input text as much when entered manually
(or when it's sane) than when coming from book text
selection.
This may allow looking up words like "-suffix", or
do more precise Wikipedia queries.

frontend/apps/reader/modules/readerdictionary.lua

@@ -214,7 +214,8 @@ function ReaderDictionary:addToMainMenu(menu_items)
                     os.date("%Y-%m-%d %H:%M:%S", value.time),
                     value.word,
                     callback = function()
-                        self:onLookupWord(value.word)
+                        -- Word had been cleaned before being added to history
+                        self:onLookupWord(value.word, true)
                     end
                 })
             end
@@ -385,10 +386,10 @@ function ReaderDictionary:addToMainMenu(menu_items)
     end
 end
 
-function ReaderDictionary:onLookupWord(word, box, highlight, link)
+function ReaderDictionary:onLookupWord(word, is_sane, box, highlight, link)
     logger.dbg("dict lookup word:", word, box)
     -- escape quotes and other funny characters in word
-    word = self:cleanSelection(word)
+    word = self:cleanSelection(word, is_sane)
     logger.dbg("dict stripped word:", word)
 
     self.highlight = highlight
@@ -609,7 +610,7 @@ local function tidyMarkup(results)
     return results
 end
 
-function ReaderDictionary:cleanSelection(text)
+function ReaderDictionary:cleanSelection(text, is_sane)
     -- Will be used by ReaderWikipedia too
     if not text then
         return ""
@@ -618,31 +619,33 @@ function ReaderDictionary:cleanSelection(text)
     -- some cleanup is still needed for selection we get from other engines
     -- (example: pdf selection "qu’autrefois," will be cleaned to "autrefois")
     --
-    -- Trim any space at start or end
-    text = text:gsub("^%s+", "")
-    text = text:gsub("%s+$", "")
-    -- Replace extended quote (included in the general puncturation range)
-    -- with plain ascii quote (for french words like "aujourd’hui")
-    text = text:gsub("\xE2\x80\x99", "'") -- U+2019 (right single quotation mark)
-    -- Strip punctuation characters around selection
-    text = util.stripPunctuation(text)
-    -- Strip some common english grammatical construct
-    text = text:gsub("'s$", '') -- english possessive
-    -- Strip some common french grammatical constructs
-    text = text:gsub("^[LSDMNTlsdmnt]'", '') -- french l' s' t'...
-    text = text:gsub("^[Qq][Uu]'", '') -- french qu'
     -- Replace no-break space with regular space
     text = text:gsub("\xC2\xA0", ' ') -- U+00A0 no-break space
-    -- There may be a need to remove some (all?) diacritical marks
-    -- https://en.wikipedia.org/wiki/Combining_character#Unicode_ranges
-    -- see discussion at https://github.com/koreader/koreader/issues/1649
-    -- Commented for now, will have to be checked by people who read
-    -- languages and texts that use them.
-    -- text = text:gsub("\204[\128-\191]", '') -- U+0300 to U+033F
-    -- text = text:gsub("\205[\128-\175]", '') -- U+0340 to U+036F
-    -- Trim any space now at start or end after above changes
+    -- Trim any space at start or end
     text = text:gsub("^%s+", "")
     text = text:gsub("%s+$", "")
+    if not is_sane then
+        -- Replace extended quote (included in the general puncturation range)
+        -- with plain ascii quote (for french words like "aujourd’hui")
+        text = text:gsub("\xE2\x80\x99", "'") -- U+2019 (right single quotation mark)
+        -- Strip punctuation characters around selection
+        text = util.stripPunctuation(text)
+        -- Strip some common english grammatical construct
+        text = text:gsub("'s$", '') -- english possessive
+        -- Strip some common french grammatical constructs
+        text = text:gsub("^[LSDMNTlsdmnt]'", '') -- french l' s' t'...
+        text = text:gsub("^[Qq][Uu]'", '') -- french qu'
+        -- There may be a need to remove some (all?) diacritical marks
+        -- https://en.wikipedia.org/wiki/Combining_character#Unicode_ranges
+        -- see discussion at https://github.com/koreader/koreader/issues/1649
+        -- Commented for now, will have to be checked by people who read
+        -- languages and texts that use them.
+        -- text = text:gsub("\204[\128-\191]", '') -- U+0300 to U+033F
+        -- text = text:gsub("\205[\128-\175]", '') -- U+0340 to U+036F
+        -- Trim any space now at start or end after above changes
+        text = text:gsub("^%s+", "")
+        text = text:gsub("%s+$", "")
+    end
     return text
 end
 
@@ -680,7 +683,8 @@ function ReaderDictionary:onShowDictionaryLookup()
                     is_enter_default = true,
                     callback = function()
                         UIManager:close(self.dictionary_lookup_dialog)
-                        self:onLookupWord(self.dictionary_lookup_dialog:getInputText())
+                        -- Trust that input text does not need any cleaning (allows querying for "-suffix")
+                        self:onLookupWord(self.dictionary_lookup_dialog:getInputText(), true)
                     end,
                 },
             }

frontend/apps/reader/modules/readerhighlight.lua

@@ -937,14 +937,14 @@ function ReaderHighlight:lookup(selected_word, selected_link)
     -- if we extracted text directly
     if selected_word.word then
         local word_box = self.view:pageToScreenTransform(self.hold_pos.page, selected_word.sbox)
-        self.ui:handleEvent(Event:new("LookupWord", selected_word.word, word_box, self, selected_link))
+        self.ui:handleEvent(Event:new("LookupWord", selected_word.word, false, word_box, self, selected_link))
     -- or we will do OCR
     elseif selected_word.sbox and self.hold_pos then
         local word = self.ui.document:getOCRWord(self.hold_pos.page, selected_word)
         logger.dbg("OCRed word:", word)
         if word and word ~= "" then
             local word_box = self.view:pageToScreenTransform(self.hold_pos.page, selected_word.sbox)
-            self.ui:handleEvent(Event:new("LookupWord", word, word_box, self, selected_link))
+            self.ui:handleEvent(Event:new("LookupWord", word, false, word_box, self, selected_link))
         else
             UIManager:show(InfoMessage:new{
                 text = info_message_ocr_text,

frontend/apps/reader/modules/readerlink.lua

@@ -711,7 +711,7 @@ function ReaderLink:onGoToExternalLink(link_url)
             callback = function()
                 UIManager:nextTick(function()
                     UIManager:close(dialog)
-                    self.ui:handleEvent(Event:new("LookupWikipedia", wiki_page, false, true, wiki_lang))
+                    self.ui:handleEvent(Event:new("LookupWikipedia", wiki_page, true, false, true, wiki_lang))
                 end)
             end,
         })

frontend/apps/reader/modules/readerwikipedia.lua

@@ -53,7 +53,8 @@ function ReaderWikipedia:lookupInput()
                     is_enter_default = true,
                     callback = function()
                         UIManager:close(self.input_dialog)
-                        self:onLookupWikipedia(self.input_dialog:getInputText())
+                        -- Trust that input text does not need any cleaning (allows querying for "-suffix")
+                        self:onLookupWikipedia(self.input_dialog:getInputText(), true)
                     end,
                 },
             }
@@ -98,7 +99,8 @@ function ReaderWikipedia:addToMainMenu(menu_items)
                     os.date("%Y-%m-%d %H:%M:%S", value.time),
                     text,
                     callback = function()
-                        self:onLookupWikipedia(value.word, nil, value.page, value.lang)
+                        -- Word had been cleaned before being added to history
+                        self:onLookupWikipedia(value.word, true, nil, value.page, value.lang)
                     end
                 })
             end
@@ -375,16 +377,16 @@ function ReaderWikipedia:initLanguages(word)
     end
 end
 
-function ReaderWikipedia:onLookupWikipedia(word, box, get_fullpage, forced_lang)
+function ReaderWikipedia:onLookupWikipedia(word, is_sane, box, get_fullpage, forced_lang)
     -- Wrapped through Trapper, as we may be using Trapper:dismissableRunInSubprocess() in it
     Trapper:wrap(function()
-        self:lookupWikipedia(word, box, get_fullpage, forced_lang)
+        self:lookupWikipedia(word, is_sane, box, get_fullpage, forced_lang)
     end)
     return true
 end
 
-function ReaderWikipedia:lookupWikipedia(word, box, get_fullpage, forced_lang)
-    if NetworkMgr:willRerunWhenOnline(function() self:lookupWikipedia(word, box, get_fullpage, forced_lang) end) then
+function ReaderWikipedia:lookupWikipedia(word, is_sane, box, get_fullpage, forced_lang)
+    if NetworkMgr:willRerunWhenOnline(function() self:lookupWikipedia(word, is_sane, box, get_fullpage, forced_lang) end) then
         -- Not online yet, nothing more to do here, NetworkMgr will forward the callback and run it once connected!
         return
     end
@@ -404,7 +406,7 @@ function ReaderWikipedia:lookupWikipedia(word, box, get_fullpage, forced_lang)
     -- no need to clean word if get_fullpage, as it is the exact wikipetia page title
     if word and not get_fullpage then
         -- escape quotes and other funny characters in word
-        word = self:cleanSelection(word)
+        word = self:cleanSelection(word, is_sane)
         -- no need to lower() word with wikipedia search
     end
     logger.dbg("stripped word:", word)

frontend/ui/widget/dictquicklookup.lua

@@ -1102,7 +1102,8 @@ function DictQuickLookup:inputLookup()
         else
             event = "LookupWord"
         end
-        self.ui:handleEvent(Event:new(event, word))
+        -- Trust that input text does not need any cleaning (allows querying for "-suffix")
+        self.ui:handleEvent(Event:new(event, word, true))
     end
 end
 
@@ -1131,18 +1132,21 @@ end
 
 function DictQuickLookup:lookupWikipedia(get_fullpage)
     local word
+    local is_sane
     if get_fullpage then
         -- we use the word of the displayed result's definition, which
         -- is the exact title of the full wikipedia page
         word = self.lookupword
+        is_sane = true
     else
         -- we use the original word that was querried
         word = self.word
+        is_sane = false
     end
     self:resyncWikiLanguages()
-    -- strange : we need to pass false instead of nil if word_box is nil,
-    -- otherwise get_fullpage is not passed
-    self.ui:handleEvent(Event:new("LookupWikipedia", word, self.word_box and self.word_box or false, get_fullpage))
+    -- (With Event, we need to pass false instead of nil if word_box is nil,
+    -- otherwise next arguments are discarded)
+    self.ui:handleEvent(Event:new("LookupWikipedia", word, is_sane, self.word_box and self.word_box or false, get_fullpage))
 end
 
 return DictQuickLookup