Adding additional common datastructures for PDF translation.

Author: sambitdash

REQUIRE

@@ -2,4 +2,5 @@ julia 0.6
 Compat
 BufferedStreams 0.3
 Libz 0.2
+StringEncodings
 Documenter

src/CDObject.jl

@@ -0,0 +1,4 @@
+export CDTextString, CDDate
+
+const CDTextString = String
+const CDDate = DateTime

src/Common.jl

@@ -9,5 +9,6 @@ include("bytes.jl")
 include("errors.jl")
 include("Utils.jl")
 include("BufferParser.jl")
+include("CDObject.jl")
 
 end

src/Cos.jl

@@ -10,6 +10,7 @@ using ..Common
 using Compat
 
 include("CosObject.jl")
+include("CosObjectHelpers.jl")
 include("CosStream.jl")
 include("CosReader.jl")
 include("CosObjStream.jl")

src/CosObject.jl

@@ -4,6 +4,8 @@ export CosDict, CosString, CosNumeric, CosBoolean, CosTrue, CosFalse,
        CosObject, CosNull, CosNullType,CosFloat, CosInt, CosArray, CosName,
        CosDict, CosIndirectObjectRef, CosStream, get, set!
 
+using StringEncodings
+
 @compat abstract type CosObject end
 
 @inline get{T<:CosObject}(o::T)=o.val
@@ -61,25 +63,13 @@ end
   CosXString(arr::Vector{UInt8})=new(arr)
 end
 
-Base.convert(::Type{Vector{UInt8}}, xstr::CosXString)=
-  (xstr.val |> String |> hex2bytes)
-
-Base.convert(::Type{String}, xstr::CosXString)=
-  String(convert(Vector{UInt8},xstr))
-
-
 @compat struct CosLiteralString <: CosString
     val::Vector{UInt8}
     CosLiteralString(arr::Vector{UInt8})=new(arr)
 end
 
 CosLiteralString(str::AbstractString)=CosLiteralString(transcode(UInt8,str))
 
-Base.convert(::Type{Vector{UInt8}}, str::CosLiteralString)=copy(str.val)
-
-Base.convert(::Type{String}, str::CosLiteralString)=
-  String(convert(Vector{UInt8},str))
-
 @compat mutable struct CosArray <: CosObject
     val::Array{CosObject,1}
     function CosArray(arr::Array{T,1} where {T<:CosObject})

src/CosObjectHelpers.jl

@@ -0,0 +1,26 @@
+using ..Common
+
+function Base.convert(::Type{CDTextString}, xstr::CosXString)
+    const feff = [LATIN_F, LATIN_E, LATIN_F, LATIN_F]
+    const FEFF = [LATIN_UPPER_F, LATIN_UPPER_E, LATIN_UPPER_F, LATIN_UPPER_F]
+    prefix = xstr.val[1:4]
+    data = xstr.val
+    buffer = data |> String |> hex2bytes
+    if prefix == feff || prefix == FEFF
+        if (0x04030201 == ENDIAN_BOM)
+            len2 = div(length(buffer),2)
+            for i=1:len2
+                (buffer[2i-1], buffer[2i]) = (buffer[2i], buffer[2i-1])
+            end
+        end
+        utf_16_data = reinterpret(UInt16, buffer)
+        str = transcode(String, utf_16_data)
+    else
+        # Assume PDFDocEncoding (ISO-8859-1)
+        str = StringEncodings.decode(buffer, "ISO_8859-1")
+    end
+    return CDTextString(str)
+end
+
+Base.convert(::Type{CDTextString}, lstr::CosLiteralString) =
+    CDTextString(StringEncodings.decode(lstr.val, "ISO_8859-1"))

src/PDDoc.jl

@@ -2,10 +2,13 @@ export PDDoc,
        pdDocOpen,
        pdDocClose,
        pdDocGetCatalog,
+       pdDocGetInfo, pdDocGetProducers,
        pdDocGetCosDoc,
        pdDocGetPageCount,
        pdDocGetPage
 
+using ..Common
+
 @compat abstract type PDDoc end
 
 function pdDocOpen(fp::String)
@@ -37,7 +40,18 @@ end
 function pdDocGetPage(doc::PDDoc, name::String)
 end
 
+function pdDocGetInfo(doc::PDDoc)
+    ref = get(doc.cosDoc.trailer[1], CosName("Info"))
+    obj = cosDocGetObject(doc.cosDoc, ref)
+    return obj
+end
 
+function pdDocGetProducers(doc::PDDoc)
+    info = pdDocGetInfo(doc)
+    creator = CDTextString(get(info, CosName("Creator")))
+    producer = CDTextString(get(info, CosName("Producer")))
+    return Dict("creator" => creator, "producer" => producer)
+end
 
 @compat mutable struct PDDocImpl <: PDDoc
   cosDoc::CosDoc

test/runtests.jl

@@ -13,8 +13,11 @@ include("debugIO.jl")
       filename="files/1.pdf"
       println(filename)
       doc = pdDocOpen(filename)
+      println(pdDocGetCatalog(doc))
       cosDoc = pdDocGetCosDoc(doc)
-      println(cosDocGetRoot(cosDoc))
+      map(println, cosDoc.trailer)
+      println(pdDocGetInfo(doc))
+      println(pdDocGetProducers(doc))
       @assert pdDocGetPageCount(doc) == 2
       page = pdDocGetPage(doc, 1)
       @assert pdPageIsEmpty(page) == false