diff options
Diffstat (limited to 'tex/context/base/mkiv/lpdf-epd.lua')
| -rw-r--r-- | tex/context/base/mkiv/lpdf-epd.lua | 518 |
1 files changed, 452 insertions, 66 deletions
diff --git a/tex/context/base/mkiv/lpdf-epd.lua b/tex/context/base/mkiv/lpdf-epd.lua index cf02b5a22..86aa6f294 100644 --- a/tex/context/base/mkiv/lpdf-epd.lua +++ b/tex/context/base/mkiv/lpdf-epd.lua @@ -32,7 +32,7 @@ if not modules then modules = { } end modules ['lpdf-epd'] = { -- already was unicode). local setmetatable, rawset, rawget, type, next = setmetatable, rawset, rawget, type, next -local tostring, tonumber = tostring, tonumber +local tostring, tonumber, unpack = tostring, tonumber, unpack local lower, match, char, byte, find = string.lower, string.match, string.char, string.byte, string.find local abs = math.abs local concat = table.concat @@ -67,13 +67,27 @@ local xref = registry["epdf.XRef"] local catalog = registry["epdf.Catalog"] local pdfdoc = registry["epdf.PDFDoc"] +if not (object and dictionary and array and xref and catalog and pdfdoc) then + logs.report("fatal error","invalid pdf inclusion library (%s)",1) + os.exit() +end + local openPDF = epdf.open +local getMajorVersion = pdfdoc.getPDFMajorVersion +local getMinorVersion = pdfdoc.getPDFMinorVersion +local getXRef = pdfdoc.getXRef +local getRawCatalog = pdfdoc.getCatalog + +if not (openPDF and getMajorVersion and getMinorVersion and getXRef and getRawCatalog) then + logs.report("fatal error","invalid pdf inclusion library (%s)",2) + os.exit() +end + local getDict = object.getDict local getArray = object.getArray local getReal = object.getReal local getInt = object.getInt -local getNum = object.getNum local getString = object.getString local getBool = object.getBool local getName = object.getName @@ -81,59 +95,81 @@ local getRef = object.getRef local getRefNum = object.getRefNum local getType = object.getType -local getTypeName = object.getTypeName + +if not (getDict and getArray and getReal and getInt and getString and getBool and getName and getRef and getRefNum and getType) then + logs.report("fatal error","invalid pdf inclusion library (%s)",3) + os.exit() +end local streamReset = object.streamReset local streamGetDict = object.streamGetDict local streamGetChar = object.streamGetChar +local streamGetAll = object.streamGetAll + +if not (streamReset and streamGetDict and streamGetChar) then + logs.report("fatal error","invalid pdf inclusion library (%s)",3) + os.exit() +end local dictGetLength = dictionary.getLength local dictGetVal = dictionary.getVal local dictGetValNF = dictionary.getValNF local dictGetKey = dictionary.getKey +if not (dictGetLength and dictGetVal and dictGetValNF and dictGetKey) then + logs.report("fatal error","invalid pdf inclusion library (%s)",4) + os.exit() +end + local arrayGetLength = array.getLength local arrayGetNF = array.getNF local arrayGet = array.get +if not (arrayGetLength and arrayGetNF and arrayGet) then + logs.report("fatal error","invalid pdf inclusion library (%s)",5) + os.exit() +end + -- these are kind of weird as they can't be accessed by (root) object local getNumPages = catalog.getNumPages local getPageRef = catalog.getPageRef -local getXRef = pdfdoc.getXRef -local getRawCatalog = pdfdoc.getCatalog - local fetch = xref.fetch local getCatalog = xref.getCatalog local getDocInfo = xref.getDocInfo +if not (getNumPages and getPageRef and fetch and getCatalog and getDocInfo) then + logs.report("fatal error","invalid pdf inclusion library (%s)",6) + os.exit() +end + -- we're done with library shortcuts -local report_epdf = logs.reporter("epdf") - -local typenames = { [0] = - "boolean", - "integer", - "real", - "string", - "name", - "null", - "array", - "dictionary", - "stream", - "ref", - "cmd", - "error", - "eof", - "none", - "integer64", +local typenames = { [0] = + "boolean", + "integer", + "real", + "string", + "name", + "null", + "array", + "dictionary", + "stream", + "ref", + "cmd", + "error", + "eof", + "none", + "integer64", } -local typenumbers = table.swapped(typenames) +local typenumbers = table.swapped(typenames) + +local null_object_code = typenumbers.null +local ref_object_code = typenumbers.ref -local null_code = typenumbers.null -local ref_code = typenumbers.ref +local report_epdf = logs.reporter("epdf") local function fatal_error(...) report_epdf(...) @@ -206,32 +242,30 @@ local function prepare(document,d,t,n,k,mt,flags) if v then local r = dictGetValNF(d,i) local kind = getType(v) - if kind == null_code then + if kind == null_object_code then -- ignore - else + elseif kind then local key = dictGetKey(d,i) - if kind then - if r and getType(r) == ref_code then - local objnum = getRefNum(r) - local cached = document.__cache__[objnum] - if not cached then - cached = checked_access[kind](v,document,objnum,mt) - if cached then - document.__cache__[objnum] = cached - document.__xrefs__[cached] = objnum - end - end - t[key] = cached - else - local v, flag = checked_access[kind](v,document) - t[key] = v - if flag and flags then - flags[key] = flag -- flags + if r and getType(r) == ref_object_code then + local objnum = getRefNum(r) + local cached = document.__cache__[objnum] + if not cached then + cached = checked_access[kind](v,document,objnum,mt) + if cached then + document.__cache__[objnum] = cached + document.__xrefs__[cached] = objnum end end + t[key] = cached else - report_epdf("warning: nil value for key %a in dictionary",key) + local v, flag = checked_access[kind](v,document) + t[key] = v + if flag and flags then + flags[key] = flag -- flags + end end + else + report_epdf("warning: nil value for key %a in dictionary",key) end else fatal_error("error: invalid value at index %a in dictionary of %a",i,document.filename) @@ -245,6 +279,42 @@ local function prepare(document,d,t,n,k,mt,flags) return t[k] end +-- local function prepare(document,d,t,n,k,mt,flags) +-- for i=1,n do +-- local v = dictGetValNF(d,i) +-- if v then +-- local key = dictGetKey(d,i) +-- local kind = getType(v) +-- if kind == ref_object_code then +-- local objnum = getRefNum(v) +-- local cached = document.__cache__[objnum] +-- if not cached then +-- local v = dictGetVal(d,i) +-- local kind = getType(v) +-- cached = checked_access[kind](v,document,objnum,mt) +-- if cached then +-- document.__cache__[objnum] = cached +-- document.__xrefs__[cached] = objnum +-- end +-- end +-- t[key] = cached +-- else +-- local v, flag = checked_access[kind](v,document) +-- t[key] = v +-- if flag and flags then +-- flags[key] = flag -- flags +-- end +-- end +-- end +-- end +-- if mt then +-- setmetatable(t,mt) +-- else +-- getmetatable(t).__index = nil +-- end +-- return t[k] +-- end + local function some_dictionary(d,document) local n = d and dictGetLength(d) or 0 if n > 0 then @@ -293,11 +363,11 @@ local function prepare(document,a,t,n,k) local v = arrayGet(a,i) if v then local kind = getType(v) - if kind == null_code then + if kind == null_object_code then -- ignore elseif kind then local r = arrayGetNF(a,i) - if r and getType(r) == ref_code then + if r and getType(r) == ref_object_code then local objnum = getRefNum(r) local cached = document.__cache__[objnum] if not cached then @@ -326,6 +396,37 @@ local function prepare(document,a,t,n,k) end end +-- local function prepare(document,a,t,n,k) +-- for i=1,n do +-- local v = arrayGetNF(a,i) +-- if v then +-- local kind = getType(v) +-- if kind == ref_object_code then +-- local objnum = getRefNum(v) +-- local cached = document.__cache__[objnum] +-- if not cached then +-- local v = arrayGet(a,i) +-- local kind = getType(v) +-- cached = checked_access[kind](v,document,objnum) +-- document.__cache__[objnum] = cached +-- document.__xrefs__[cached] = objnum +-- end +-- t[i] = cached +-- else +-- t[i] = checked_access[kind](v,document) +-- end +-- end +-- end +-- local m = getmetatable(t) +-- if m then +-- m.__index = nil +-- m.__len = nil +-- end +-- if k then +-- return t[k] +-- end +-- end + local function some_array(a,document) local n = a and arrayGetLength(a) or 0 if n > 0 then @@ -376,23 +477,53 @@ end -- todo: collect chunks -local function streamaccess(s,_,what) - if not what or what == "all" or what == "*all" then - local t, n = { }, 0 - streamReset(s) +-- local function streamaccess(s,_,what) +-- if not what or what == "all" or what == "*all" then +-- local t, n = { }, 0 +-- streamReset(s) +-- while true do +-- local c = streamGetChar(s) +-- if c < 0 then +-- break +-- else +-- n = n + 1 +-- t[n] = char(c) +-- end +-- end +-- return concat(t,"",1,n) +-- end +-- end + +local function getstream(s) + streamReset(s) + if streamGetAll then + return streamGetAll(s) + else + local t, b, n = { }, { }, 0 while true do local c = streamGetChar(s) if c < 0 then break else n = n + 1 - t[n] = char(c) + b[n] = c + end + if n == 2000 then + t[#t+1] = char(unpack(b,1,n)) + n = 1 end end + t[#t+1] = char(unpack(b,1,n)) return concat(t) end end +local function streamaccess(s,_,what) + if not what or what == "all" or what == "*all" then + return getstream(s) + end +end + local function get_stream(d,document) if d then streamReset(d) @@ -562,16 +693,19 @@ end -- with but it won't win a beauty contest. local function getpages(document,Catalog) - local __data__ = document.__data__ - local __xrefs__ = document.__xrefs__ - local __cache__ = document.__cache__ - local __xref__ = document.__xref__ + local __data__ = document.__data__ + local __xrefs__ = document.__xrefs__ + local __cache__ = document.__cache__ + local __xref__ = document.__xref__ + -- + local rawcatalog = getRawCatalog(__data__) + local nofpages = getNumPages(rawcatalog) -- - local rawcatalog = getRawCatalog(__data__) - local nofpages = getNumPages(rawcatalog) + local majorversion = getMajorVersion(__data__) + local minorversion = getMinorVersion(__data__) -- - local pages = { } - local metatable = { __index = Catalog.Pages } -- somewhat empty + local pages = { } + local metatable = { __index = Catalog.Pages } -- somewhat empty -- for pagenumber=1,nofpages do local pagereference = getPageRef(rawcatalog,pagenumber).num @@ -580,6 +714,7 @@ local function getpages(document,Catalog) if pagedata then -- rawset(pagedata,"number",pagenumber) pagedata.number = pagenumber + pagedata.object = pageobject pages[pagenumber] = pagedata __xrefs__[pagedata] = pagereference __cache__[pagereference] = pagedata @@ -590,7 +725,10 @@ local function getpages(document,Catalog) -- pages.n = nofpages -- - document.pages = pages + document.pages = pages + document.majorversion = majorversion + document.minorversion = minorversion + -- return pages end @@ -637,6 +775,8 @@ function lpdf_epdf.load(filename) document.Catalog = some_dictionary(getDict(getCatalog(__xref__)),document) document.Info = some_dictionary(getDict(getDocInfo(__xref__)),document) setmetatableindex(document,resolve) + -- + document.nofpages = getNumPages(getRawCatalog(__data__)) else document = false end @@ -735,14 +875,14 @@ local fromunicode = ( P(1) )^1 * Carg(1) -local function analyzefonts(document,resources) -- unfinished +local function analyzefonts(document,resources) -- unfinished, see mtx-pdf for better code local fonts = document.__fonts__ if resources then local fontlist = resources.Font if fontlist then for id, data in expanded(fontlist) do if not fonts[id] then - -- a quck hack ... I will look into it more detail if I find a real + -- a quick hack ... I will look into it more detail if I find a real -- -application for it local tounicode = data.ToUnicode() if tounicode then @@ -836,7 +976,7 @@ function lpdf_epdf.getpagecontent(document,pagenumber) end --- This is also an experiment. When I really neet it I can improve it, fo rinstance +-- This is also an experiment. When I really need it I can improve it, for instance -- with proper position calculating. It might be usefull for some search or so. local softhyphen = utfchar(0xAD) .. "$" @@ -925,3 +1065,249 @@ end -- local destination = document.__data__:findDest(name) -- return destination and destination.number -- end + +-- This is experimental code that we need for testing the transition from +-- poppler to a new lightweight library. Don't rely on this code to remain +-- as it is now. Interesting is that performance of this variant is the same +-- as the natural page includer. + +if img then do + + local copydictionary = nil + local copyarray = nil + + local ref_object_code = typenumbers.ref + local boolean_object_code = typenumbers.boolean + local integer_object_code = typenumbers.integer + local real_object_code = typenumbers.real + local string_object_code = typenumbers.string + local name_object_code = typenumbers.name + local null_object_code = typenumbers.null + local array_object_code = typenumbers.array + local dictionary_object_code = typenumbers.dictionary + local stream_object_code = typenumbers.stream + local cmd_object_code = typenumbers.cmd + + local pdfreserveobject = lpdf.reserveobject + local pdfflushobject = lpdf.flushobject + local pdfflushstreamobject = lpdf.flushstreamobject + local pdfreference = lpdf.reference + local pdfconstant = lpdf.constant + local pdfarray = lpdf.array + local pdfdictionary = lpdf.dictionary + local pdfunicode = lpdf.unicode + local pdfstring = lpdf.string + local pdfnull = lpdf.null + + local report = logs.reporter("backend","xobjects") + + local factor = 65536 / (7200/7227) -- 1/number.dimenfactors.bp + + local createimage = images.create + + local function scaledbbox(b) + return { b[1]*factor, b[2]*factor, b[3]*factor, b[4]*factor } + end + + local function copyobject(xref,copied,kind,r,v) + if kind == null_object_code then + return pdfnull() + elseif r and getType(r) == ref_object_code then + local objnum = getRefNum(r) + local r = copied[objnum] + if r then + -- report("%s object %i is reused",kind,objnum) + else + local o + r = pdfreserveobject() + copied[objnum] = r + if kind == array_object_code then + local a = copyarray(xref,copied,fetch(xref,objnum,0)) + pdfflushobject(r,tostring(a)) + elseif kind == dictionary_object_code then + local d = copydictionary(xref,copied,fetch(xref,objnum,0)) + pdfflushobject(r,tostring(d)) + elseif kind == stream_object_code then + local f = fetch(xref,objnum,0) + local d = copydictionary(xref,copied,false,streamGetDict(f)) + local s = getstream(f) + -- + d.Filter = nil + d.Length = nil + d.DecodeParms = nil + d.DL = nil + -- + pdfflushstreamobject(s,d,true,r) + else + report("reference not done: %s", kind) + end + end + return pdfreference(r) + elseif kind == array_object_code then + return copyarray(xref,copied,v) + elseif kind == dictionary_object_code then + return copydictionary(xref,copied,v) + elseif kind == integer_object_code then + return getInt(v) + elseif kind == real_object_code then + return getReal(v) + elseif kind == name_object_code then + return pdfconstant(getName(v)) + elseif kind == string_object_code then + local s = getString(v) + if not s or s == "" then + return "" + end + local u = lpegmatch(u_pattern,s) + if u then + return pdfunicode(s) + end + return pdfstring(s) + elseif kind == boolean_object_code then + return getBool(v) + elseif kind == stream_object_code then + -- hm ... + return getStream(v) + else + report("object not done: %s", kind) + end + end + + copyarray = function (xref,copied,object) + local a = getArray(object) + local n = a and arrayGetLength(a) or 0 + if n > 0 then + local target = pdfarray() + for i=1,n do + local v = arrayGet(a,i) + if v then + local kind = getType(v) + local r = arrayGetNF(a,i) + target[i] = copyobject(xref,copied,kind,r,v) + end + end + return target + end + end + + copydictionary = function (xref,copied,object,d) + local d = d or getDict(object) + local n = d and dictGetLength(d) or 0 + if n > 0 then + local target = pdfdictionary() + for i=1,n do + local v = dictGetVal(d,i) + if v then + local kind = getType(v) + local key = dictGetKey(d,i) + local r = dictGetValNF(d,i) + target[key] = copyobject(xref,copied,kind,r,v) + end + end + return target + end + end + + local function copy_resources(pdfdoc,xref,copied,pagedata) + local object = pagedata.object + if object then + local d = getDict(object) + local n = d and dictGetLength(d) or 0 + for i=1,n do + local k = dictGetKey(d,i) + if v and k == "Resources" then + local v = dictGetVal(d,i) + local kind = getType(v) + local r = dictGetValNF(d,i) + return copyobject(xref,copied,kind,r,v) + end + end + end + end + + local function openpdf(filename) + local pdfdoc = lpdf_epdf.load(filename) + if pdfdoc then + pdfdoc.__copied__ = pdfdoc.__copied__ or { } + pdfdoc.filename = filename + return pdfdoc + end + end + + local function closepdf(pdfdoc) + if pdfdoc then + lpdf_epdf.unload(pdfdoc.filename) + end + end + + local function querypdf(pdfdoc,pagenumber) + if pdfdoc then + if not pagenumber then + pagenumber = 1 + end + local root = pdfdoc.Catalog + local page = pdfdoc.pages[pagenumber] + if page then + local mediabox = page.MediaBox or { 0, 0, 0, 0 } + local cropbox = page.CropBox or mediabox + return { + filename = pdfdoc.filename, + pagenumber = pagenumber, + nofpages = pdfdoc.nofpages, + boundingbox = scaledbbox(cropbox), + cropbox = cropbox, + mediabox = mediabox, + bleedbox = page.BleedBox or cropbox, + trimbox = page.TrimBox or cropbox, + artbox = page.ArtBox or cropbox, + } + end + end + end + + local function copypage(pdfdoc,pagenumber,attributes) + if pdfdoc then + local root = pdfdoc.Catalog + local page = pdfdoc.pages[pagenumber or 1] + local pageinfo = querypdf(pdfdoc,pagenumber) + local contents = page.Contents + local xref = pdfdoc.__xref__ + local copied = pdfdoc.__copied__ + -- + local xobject = pdfdictionary { + Type = pdfconstant("XObject"), + Subtype = pdfconstant("Form"), + -- image attributes + FormType = 1, + BBox = pageinfo.cropbox, + -- Metadata = copy(xref,copied,root,"Metadata"), + -- Group = copy(xref,copied,page,"Group"), + -- LastModified = copy(xref,copied,page,"LastModified"), + -- Metadata = copy(xref,copied,page,"Metadata"), + -- PieceInfo = copy(xref,copied,page,"PieceInfo"), + Resources = copy_resources(pdfdoc,xref,copied,page), + -- SeparationInfo = copy(xref,copied,page,"SeparationInfo"), + } + if attributes then + for k, v in next, expand(attributes) do + page[k] = v -- maybe nested + end + end + return createimage { + bbox = pageinfo.boundingbox, + stream = contents(), + attr = xobject(), + } + end + end + + -- todo: codeinjections + + lpdf_epdf.image = { + open = openpdf, + close = closepdf, + query = querypdf, + copy = copypage, + } + +end end |