2
0
mirror of https://github.com/koreader/koreader synced 2024-11-04 12:00:25 +00:00
koreader/frontend/document/djvudocument.lua
weijiuqiao edf7cc9a61
Vocabulary builder: support extracting context from pdfs (#9622)
Move getSelectedWordContext(), now document specific,
from ReaderHighlight into each document module.
2022-10-25 12:23:18 +02:00

167 lines
5.6 KiB
Lua

local Blitbuffer = require("ffi/blitbuffer")
local Document = require("document/document")
local DrawContext = require("ffi/drawcontext")
local DjvuDocument = Document:extend{
_document = false,
-- libdjvulibre manages its own additional cache, default value is hard written in c module.
is_djvu = true,
djvulibre_cache_size = nil,
dc_null = DrawContext.new(),
koptinterface = nil,
color_bb_type = Blitbuffer.TYPE_BBRGB24,
provider = "djvulibre",
provider_name = "DjVu Libre",
}
-- check DjVu magic string to validate
local function validDjvuFile(filename)
local f = io.open(filename, "r")
if not f then return false end
local magic = f:read(8)
f:close()
if not magic or magic ~= "AT&TFORM" then return false end
return true
end
function DjvuDocument:init()
self:updateColorRendering()
local djvu = require("libs/libkoreader-djvu")
self.koptinterface = require("document/koptinterface")
self.koptinterface:setDefaultConfigurable(self.configurable)
if not validDjvuFile(self.file) then
error("Not a valid DjVu file")
end
local ok
ok, self._document = pcall(djvu.openDocument, self.file, self.render_color, self.djvulibre_cache_size)
if not ok then
error(self._document) -- will contain error message
end
self.is_open = true
self.info.has_pages = true
self.info.configurable = true
self:_readMetadata()
end
function DjvuDocument:updateColorRendering()
Document.updateColorRendering(self) -- will set self.render_color
if self._document then
self._document:setColorRendering(self.render_color)
end
end
function DjvuDocument:getProps()
local props = self._document:getMetadata()
local _, _, docname = self.file:find(".*/(.*)")
docname = docname or self.file
-- According to djvused(1), the convention is that
-- BibTex keys are always lowercase and DocInfo capitalized
props.title = props.title or props.Title or docname:match("(.*)%.")
props.authors = props.author or props.Author
props.series = props.series or props.Series
props.language = props.language or props.Language
props.keywords = props.keywords or props.Keywords
props.description = props.description or props.Description
return props
end
function DjvuDocument:comparePositions(pos1, pos2)
return self.koptinterface:comparePositions(self, pos1, pos2)
end
function DjvuDocument:getPageTextBoxes(pageno)
return self._document:getPageText(pageno)
end
function DjvuDocument:getPanelFromPage(pageno, pos)
return self.koptinterface:getPanelFromPage(self, pageno, pos)
end
function DjvuDocument:getWordFromPosition(spos)
return self.koptinterface:getWordFromPosition(self, spos)
end
function DjvuDocument:getTextFromPositions(spos0, spos1)
return self.koptinterface:getTextFromPositions(self, spos0, spos1)
end
function DjvuDocument:getPageBoxesFromPositions(pageno, ppos0, ppos1)
return self.koptinterface:getPageBoxesFromPositions(self, pageno, ppos0, ppos1)
end
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
function DjvuDocument:getOCRText(pageno, tboxes)
return self.koptinterface:getOCRText(self, pageno, tboxes)
end
function DjvuDocument:getPageBlock(pageno, x, y)
return self.koptinterface:getPageBlock(self, pageno, x, y)
end
function DjvuDocument:getUsedBBox(pageno)
-- djvu does not support usedbbox, so fake it.
local used = {}
local native_dim = self:getNativePageDimensions(pageno)
used.x0, used.y0, used.x1, used.y1 = 0, 0, native_dim.w, native_dim.h
return used
end
function DjvuDocument:clipPagePNGFile(pos0, pos1, pboxes, drawer, filename)
return self.koptinterface:clipPagePNGFile(self, pos0, pos1, pboxes, drawer, filename)
end
function DjvuDocument:clipPagePNGString(pos0, pos1, pboxes, drawer)
return self.koptinterface:clipPagePNGString(self, pos0, pos1, pboxes, drawer)
end
function DjvuDocument:getPageBBox(pageno)
return self.koptinterface:getPageBBox(self, pageno)
end
function DjvuDocument:getPageDimensions(pageno, zoom, rotation)
return self.koptinterface:getPageDimensions(self, pageno, zoom, rotation)
end
function DjvuDocument:getCoverPageImage()
return self.koptinterface:getCoverPageImage(self)
end
function DjvuDocument:findText(pattern, origin, reverse, caseInsensitive, page)
return self.koptinterface:findText(self, pattern, origin, reverse, caseInsensitive, page)
end
function DjvuDocument:renderPage(pageno, rect, zoom, rotation, gamma, render_mode, hinting)
return self.koptinterface:renderPage(self, pageno, rect, zoom, rotation, gamma, render_mode, hinting)
end
function DjvuDocument:hintPage(pageno, zoom, rotation, gamma, render_mode)
return self.koptinterface:hintPage(self, pageno, zoom, rotation, gamma, render_mode)
end
function DjvuDocument:drawPage(target, x, y, rect, pageno, zoom, rotation, gamma, render_mode)
return self.koptinterface:drawPage(self, target, x, y, rect, pageno, zoom, rotation, gamma, render_mode)
end
function DjvuDocument:register(registry)
registry:addProvider("djvu", "image/vnd.djvu", self, 100)
registry:addProvider("djvu", "application/djvu", self, 100) -- Alternative mimetype for OPDS.
registry:addProvider("djvu", "image/x-djvu", self, 100) -- Alternative mimetype for OPDS.
registry:addProvider("djv", "image/vnd.djvu", self, 100)
end
return DjvuDocument