Cleaner handling of Object Streams and XRef tables

Author: sambitdash

src/CosDoc.jl

@@ -168,6 +168,18 @@ end
 function cosDocGetObject(doc::CosDocImpl, stmref::CosIndirectObjectRef,
                          ref::CosIndirectObjectRef, locObj::CosObjectLoc)
     objstm = cosDocGetObject(doc, stmref)
+    if (objstm === CosNull)
+        #= This is not really needed but PDF specification is kind of equivocal if
+           object streams should be referenced in the XRef stream.
+
+           An object stream itself, like any stream, shall be an indirect object, and
+           therefore, there shall be an entry for it in a cross-reference table or
+           cross-reference stream (see 7.5.8, "Cross-Reference Streams"), although there
+           might not be any references to it (of the form 243 0 R).
+        =#
+        objstm = scan_object_stream(doc, stmref)
+        attach_object(doc, objstm)
+    end
     (objstm === CosNull) && return CosNull
     if (locObj.obj == CosNull)
         locObj.obj = cosObjectStreamGetObject(objstm, ref, locObj.loc)
@@ -176,6 +188,25 @@ function cosDocGetObject(doc::CosDocImpl, stmref::CosIndirectObjectRef,
     return locObj.obj
 end
 
+function scan_object_stream(doc::CosDocImpl, stmref::CosIndirectObjectRef)
+    look_ahead = 2048
+    loc = doc.startxref - look_ahead
+    if loc < 0
+        loc = 0
+    end
+    seek(doc.ps, loc)
+    look_ahead = doc.startxref - loc
+    ref = get(stmref)
+    keyword = "$(ref[1]) $(ref[2]) obj"
+    loc1 = locate_keyword!(doc.ps, transcode(UInt8, keyword), look_ahead)
+    loc1 < 0 && return CosNull
+    seek(doc.ps, loc + loc1)
+    obj = parse_indirect_obj(doc.ps, doc.xref)
+    obj === CosNull && return CosNull
+    doc.xref[stmref] = CosObjectLoc(loc + loc1, CosNull, obj)
+    return obj
+end
+
 function read_header(ps)
   skipv(ps,PERCENT)
   b = UInt8[]

src/CosObjStream.jl

@@ -1,24 +1,22 @@
 using BufferedStreams
 
-# The xref stream may be accessed later. There is no point encrypting this data
-#Ideal will be to remove the filter.
 function make_number(data, start, nbytes)
-  sum = 0
-  for ii = 1:nbytes
-    sum *= 256
-    sum += data[start+ii]
-  end
-  return sum
+    sum = 0
+    for ii = 1:nbytes
+        sum *= 256
+        sum += data[start+ii]
+    end
+    return sum
 end
 
 function get_xref_record(data, start, w)
-  v=Vector{Int}()
-  for tw in w
-    n = make_number(data, start, tw)
-    start += tw
-    push!(v, n)
-  end
-  return v
+    v=Vector{Int}()
+    for tw in w
+        n = make_number(data, start, tw)
+        start += tw
+        push!(v, n)
+    end
+    return v
 end
 
 function createObjectStreams(stm::CosStream)
@@ -31,62 +29,59 @@ function createObjectStreams(stm::CosStream)
 end
 
 function read_xref_stream(xrefstm::CosObject,
-  xref::Dict{CosIndirectObjectRef, CosObjectLoc})
+    xref::Dict{CosIndirectObjectRef, CosObjectLoc})
 
-  @assert get(xrefstm, CosName("Type"))==CosName("XRef")
-  size = get(xrefstm, CosName("Size"))
-  @assert size !=CosNull
+    @assert get(xrefstm, cn"Type") == cn"XRef"
+    size = get(xrefstm, cn"Size")
+    @assert size !=CosNull
 
-  w = get(xrefstm, CosName("W"))
-  @assert w != CosNull
-  @assert length(w) == 3
+    w = get(xrefstm, cn"W")
+    @assert w != CosNull
+    @assert length(w) == 3
 
-  index = get(xrefstm, CosName("Index"))
+    index = get(xrefstm, cn"Index")
 
-  if (index == CosNull)
-    index = CosArray([CosInt(0),size])
-  end
+    if (index == CosNull)
+        index = CosArray([CosInt(0),size])
+    end
 
-  cosStreamRemoveFilters(xrefstm)
-
-
-  input = get(xrefstm)
-  data = read(input)
-  close(input)
-
-  w_n = get(w,true) #This size is 3
-  recsize = sum(w_n)
-
-  lenidx = length(index)
-  @assert rem(lenidx,2) == 0
-  idx_int=get(index,true)
-
-  it = 0 #iterator for data
-  count_record = 0
-  for i = 1:div(lenidx,2)
-    for j = 0:idx_int[2*i]-2
-      oid = idx_int[2*i-1]+j
-      record = get_xref_record(data,it,w_n)
-      @assert length(record) == 3
-      @assert record[1] in 0:2
-
-      loc = (record[1] == 1) ? record[2] :
-            (record[1] == 2) ? record[3] : 0
-      stm = (record[1] == 2) ? CosIndirectObjectRef(record[2],0) : CosNull
-      ref = (record[1] == 1) ? CosIndirectObjectRef(oid, record[3]) :
-            (record[1] == 2) ? CosIndirectObjectRef(oid, 0) :
-                              CosIndirectObjectRef(0,0)
-
-      it += recsize
-      if (record[1] != 0)
-        count_record +=1
-        if !haskey(xref,ref)
-          xref[ref]=CosObjectLoc(loc,stm)
+    # The xref stream may be accessed later. There is no point encrypting this data
+    # Ideal will be to remove the filter.
+    cosStreamRemoveFilters(xrefstm)
+
+    input = get(xrefstm)
+    data = read(input)
+    close(input)
+    datasize = length(data)
+
+    w_n = get(w,true) #This size is 3
+    recsize = sum(w_n)
+
+    lenidx = length(index)
+    @assert rem(lenidx,2) == 0
+    idx_int=get(index,true)
+
+    it = 0 #iterator for data
+    for i = 1:div(lenidx,2)
+        for j = 0:idx_int[2*i] - 1
+            oid = idx_int[2*i-1]+j
+            record = get_xref_record(data, it, w_n)
+            @assert length(record) == 3
+            @assert record[1] in 0:2
+
+            loc = (record[1] == 1) ? record[2] :
+                  (record[1] == 2) ? record[3] : 0
+            stm = (record[1] == 2) ? CosIndirectObjectRef(record[2],0) : CosNull
+            ref = (record[1] == 1) ? CosIndirectObjectRef(oid, record[3]) :
+                  (record[1] == 2) ? CosIndirectObjectRef(oid, 0) :
+                                     CosIndirectObjectRef(0,0)
+            it += recsize
+            record[1] != 0 && !haskey(xref,ref) && (xref[ref] = CosObjectLoc(loc,stm))
+            it >= datasize && break
         end
-      end
+        it >= datasize && break
     end
-  end
-  return xref
+    return xref
 end
 
 function read_object_info_from_stm(stm::CosStream,
@@ -105,31 +100,25 @@ function read_object_info_from_stm(stm::CosStream,
   end
 end
 
-function cosObjectStreamGetObject(stm::CosIndirectObject{CosObjectStream},
-  ref::CosIndirectObjectRef, loc::Int)
-  return cosObjectStreamGetObject(stm.obj, ref, loc)
-end
+cosObjectStreamGetObject(stm::CosIndirectObject{CosObjectStream},
+    ref::CosIndirectObjectRef, loc::Int) = cosObjectStreamGetObject(stm.obj, ref, loc)
 
-function cosObjectStreamGetObject(stm::CosObjectStream,
-  ref::CosIndirectObjectRef, loc::Int)
-  objtuple = get(ref)
-  if (stm.oids[loc+1] != objtuple[1])
-    return CosNull
-  end
-  dirobj = cosObjectStreamGetObject(stm, CosNull, loc)
-  return CosIndirectObject(objtuple[1], objtuple[2], dirobj)
+function cosObjectStreamGetObject(stm::CosObjectStream, ref::CosIndirectObjectRef, loc::Int)
+    objtuple = get(ref)
+    stm.oids[loc+1] != objtuple[1] && return CosNull
+    dirobj = cosObjectStreamGetObject(stm, CosNull, loc)
+    return CosIndirectObject(objtuple[1], objtuple[2], dirobj)
 end
 
-function cosObjectStreamGetObject(stm::CosObjectStream,
-  ref::CosNullType, loc::Int)
-  filename = get(stm, CosName("F"))
-  io = util_open(String(filename),"r")
-  ps = BufferedInputStream(io)
-  try
-    seek(ps, stm.oloc[loc+1])
-    obj = parse_value(ps)
-    return obj
-  finally
-    close(ps)
-  end
+function cosObjectStreamGetObject(stm::CosObjectStream, ref::CosNullType, loc::Int)
+    filename = get(stm, CosName("F"))
+    io = util_open(String(filename),"r")
+    ps = BufferedInputStream(io)
+    try
+        seek(ps, stm.oloc[loc+1])
+        obj = parse_value(ps)
+        return obj
+    finally
+        close(ps)
+    end
 end