renun/koreader
1
0
Fork 0
mirror of https://github.com/koreader/koreader.git synced 2025-08-10 00:52:38 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

Vocabulary builder: support extracting context from pdfs (#9622)

Browse Source 
Move getSelectedWordContext(), now document specific,
from ReaderHighlight into each document module.
This commit is contained in:
weijiuqiao
2022-10-25 18:23:18 +08:00
committed by GitHub
parent 58613d66e7
commit edf7cc9a61
6 changed files with 82 additions and 20 deletions
Download Patch File Download Diff File Expand all files Collapse all files

24
frontend/apps/reader/modules/readerhighlight.lua
View File

@@ -1282,26 +1282,12 @@ dbg:guard(ReaderHighlight, "lookup",
end)
function ReaderHighlight:getSelectedWordContext(nb_words)
if not self.ui.rolling or not self.selected_text then return nil end
local pos_start = self.selected_text.pos0
local pos_end = self.selected_text.pos1
for i=0, nb_words do
local ok, start = pcall(self.ui.document.getPrevVisibleWordStart, self.ui.document, pos_start)
if ok then pos_start = start
else break end
if not self.selected_text then return end
local ok, prev_context, next_context = pcall(self.ui.document.getSelectedWordContext, self.ui.document,
self.selected_text.text, nb_words, self.selected_text.pos0, self.selected_text.pos1)
if ok then
return prev_context, next_context
end
for i=0, nb_words do
local ok, ending = pcall(self.ui.document.getNextVisibleWordEnd, self.ui.document, pos_end)
if ok then pos_end = ending
else break end
end
local ok_prev, prev = pcall(self.ui.document.getTextFromXPointers, self.ui.document, pos_start, self.selected_text.pos0)
local ok_next, next = pcall(self.ui.document.getTextFromXPointers, self.ui.document, self.selected_text.pos1, pos_end)
return ok_prev and prev, ok_next and next
end
function ReaderHighlight:viewSelectionHTML(debug_view, no_css_files_buttons)

22
frontend/document/credocument.lua
View File

@@ -716,6 +716,28 @@ function CreDocument:getNextVisibleChar(xp)
return self._document:getNextVisibleChar(xp)
end
function CreDocument:getSelectedWordContext(word, nb_words, pos0, pos1)
local pos_start = pos0
local pos_end = pos1
for i=0, nb_words do
local start = self:getPrevVisibleWordStart(pos_start)
if start then pos_start = start
else break end
end
for i=0, nb_words do
local ending = self:getNextVisibleWordEnd(pos_end)
if ending then pos_end = ending
else break end
end
local prev = self:getTextFromXPointers(pos_start, pos0)
local next = self:getTextFromXPointers(pos1, pos_end)
return prev, next
end
function CreDocument:drawCurrentView(target, x, y, rect, pos)
if self.buffer and (self.buffer.w ~= rect.w or self.buffer.h ~= rect.h) then
self.buffer:free()

4
frontend/document/djvudocument.lua
View File

@@ -96,6 +96,10 @@ function DjvuDocument:nativeToPageRectTransform(pageno, rect)
return self.koptinterface:nativeToPageRectTransform(self, pageno, rect)
end
function DjvuDocument:getSelectedWordContext(word, nb_words, pos)
return self.koptinterface:getSelectedWordContext(word, nb_words, pos)
end
function DjvuDocument:getOCRWord(pageno, wbox)
return self.koptinterface:getOCRWord(self, pageno, wbox)
end

46
frontend/document/koptinterface.lua
View File

@@ -1037,6 +1037,7 @@ Get word and word box from `doc` position.
function KoptInterface:getWordFromPosition(doc, pos)
local text_boxes = self:getTextBoxes(doc, pos.page)
if text_boxes then
self.last_text_boxes = text_boxes
if doc.configurable.text_wrap == 1 then
return self:getWordFromReflowPosition(doc, text_boxes, pos)
else
@@ -1094,6 +1095,51 @@ function KoptInterface:getWordFromNativePosition(doc, boxes, pos)
return word_box
end
function KoptInterface:getSelectedWordContext(word, nb_words, pos)
local boxes = self.last_text_boxes
if not pos or not boxes or #boxes == 0 then return end
local i, j = getWordBoxIndices(boxes, pos)
if boxes[i][j].word ~= word then return end
local li, wi = i, j
local prev_count, next_count = 0, 0
local prev_text, next_text = {}, {}
while prev_count < nb_words do
if li == 1 and wi == 1 then
break
elseif wi == 1 then
li = li - 1
wi = #boxes[li]
else
wi = wi - 1
end
local current_word = boxes[li][wi].word
if #current_word > 0 then
table.insert(prev_text, 1, current_word)
prev_count = prev_count + 1
end
end
li, wi = i, j
while next_count < nb_words do
if li == #boxes and wi == #boxes[li] then
break
elseif wi == #boxes[li] then
li = li + 1
wi = 1
else
wi = wi + 1
end
local current_word = boxes[li][wi].word
if #current_word > 0 then
table.insert(next_text, current_word)
next_count = next_count + 1
end
end
if #prev_text == 0 and #next_text == 0 then return end
return table.concat(prev_text, " "), table.concat(next_text, " ")
end
--[[--
Get link from position in screen page.
]]--

4
frontend/document/pdfdocument.lua
View File

@@ -129,6 +129,10 @@ function PdfDocument:nativeToPageRectTransform(pageno, rect)
return self.koptinterface:nativeToPageRectTransform(self, pageno, rect)
end
function PdfDocument:getSelectedWordContext(word, nb_words, pos)
return self.koptinterface:getSelectedWordContext(word, nb_words, pos)
end
function PdfDocument:getOCRWord(pageno, wbox)
return self.koptinterface:getOCRWord(self, pageno, wbox)
end

2
plugins/vocabbuilder.koplugin/main.lua
View File

@@ -517,7 +517,7 @@ function WordInfoDialog:init()
VerticalSpan:new{width= Size.padding.default},
has_context and
TextBoxWidget:new{
text = "..." .. self.prev_context:gsub("\n", " ") .. "【" ..self.title.."】" .. self.next_context:gsub("\n", " ") .. "...",
text = "..." .. (self.prev_context or ""):gsub("\n", " ") .. "【" ..self.title.."】" .. (self.next_context or ""):gsub("\n", " ") .. "...",
width = width,
face = Font:getFace("smallffont"),
alignment = self.title_align or "left",