Support for multiple content streams per page

Author: sambitdash

src/CosStream.jl

@@ -4,8 +4,9 @@ import Libz:BufferedStreams.readbytes!
 
 import BufferedStreams:readbytes!
 
-export cosStreamRemoveFilters,
-       decode
+export  cosStreamRemoveFilters,
+        merge_streams,
+        decode
 
 function _not_implemented(input)
   error(E_NOT_IMPLEMENTED)
@@ -83,6 +84,27 @@ function cosStreamRemoveFilters(stm::CosObject)
   return stm
 end
 
+function merge_streams(stms::CosArray)
+    (path,io) = get_tempfilepath()
+    try
+        dict = CosDict()
+        println(path)
+        set!(dict, cn"F", CosLiteralString(path))
+        ret = CosStream(dict, false)
+        v = get(stms)
+        for stm in v
+            bufstm = decode(stm)
+            data = read(bufstm)
+            close(bufstm)
+            write(io, data)
+        end
+        return ret
+    finally
+        util_close(io)
+    end
+    return CosNull
+end
+
 """
 Reads the filter data and decodes the stream.
 """

src/PDPage.jl

@@ -101,31 +101,29 @@ converted to an actual stream object
 get_page_content_ref(page::PDPageImpl) = get(page.cospage, cn"Contents")
 
 function get_page_contents(page::PDPageImpl, contents::CosArray)
-  len = length(contents)
-  for i = 1:len
-    ref = splice!(get(contents), 1)
-    cosstm = get_page_contents(page,ref)
-    if (cosstm != CosNull)
-      push!(get(contents),cosstm)
+    len = length(contents)
+    arr = get(contents)
+    for i = 1:len
+        ref = splice!(arr, 1)
+        cosstm = get_page_contents(page, ref)
+        if (cosstm !== CosNull)
+            push!(arr, cosstm)
+        end
     end
-  end
-  return contents
+    return merge_streams(contents)
 end
 
-function get_page_contents(page::PDPageImpl, contents::CosIndirectObjectRef)
-  return cosDocGetObject(page.doc.cosDoc, contents)
-end
+get_page_contents(page::PDPageImpl, contents::CosIndirectObjectRef) =
+    cosDocGetObject(page.doc.cosDoc, contents)
 
-function get_page_contents(page::PDPage, contents::CosObject)
-  return CosNull
-end
+get_page_contents(page::PDPage, obj::CosObject) = obj
 
 function load_page_objects(page::PDPageImpl)
     contents = pdPageGetContents(page)
     if (isnull(page.content_objects))
-        page.content_objects=Nullable(PDPageObjectGroup())
+        page.content_objects = Nullable(PDPageObjectGroup())
     end
-    load_page_objects(page, contents)
+    return load_page_objects(page, contents)
 end
 
 load_page_objects(page::PDPageImpl, stm::CosNullType) = nothing
@@ -137,12 +135,13 @@ function load_page_objects(page::PDPageImpl, stm::CosObject)
     finally
         close(bufstm)
     end
+    return nothing
 end
 
-function load_page_objects(page::PDPageImpl, stm::CosArray)
-  for s in get(stm)
-    load_page_objects(page,s)
-  end
+function load_page_objects(page::PDPageImpl, stms::CosArray)
+    stm = merge_streams(stms)
+    page.contents = stm
+    return load_page_objects(page, stm)
 end
 
 

src/PDPageElement.jl

@@ -101,11 +101,9 @@ function populate_element(grp::PDPageObjectGroup, elem::PDPageElement)
             unshift!(elem.operands,operand)
         end
     else
-        len=endof(grp.objs)
-        while(isa(grp.objs[len],CosObject))
+        while(isa(grp.objs[end],CosObject))
             operand=pop!(grp.objs)
             unshift!(elem.operands,operand)
-            len = endof(grp.objs)
         end
     end
 end