Adding common data structures for PDF

Author: sambitdash

REQUIRE

@@ -3,4 +3,5 @@ Compat
 BufferedStreams 0.3
 Libz 0.2
 StringEncodings
+TimeZones
 Documenter

src/CDObject.jl

@@ -1,4 +1,38 @@
-export CDTextString, CDDate
+export CDTextString, CDDate, CDRect
 
 const CDTextString = String
-const CDDate = DateTime
+
+using TimeZones
+
+"""
+PDF files support the string format: (D:YYYYMMDDHHmmSSOHH'mm)
+
+"""
+@compat struct CDDate
+    d::ZonedDateTime
+    CDDate(d::ZonedDateTime) = new(d)
+end
+
+function CDDate(s::CDTextString)
+    s = ascii(s)
+    if startswith(s, "D:")
+        s = s[3:end]
+    end
+    s = *(split(s,'\'')...)
+    format = "yyyymmddHHMMSS"
+    if endswith(s, 'Z')
+        s = s[1:end-1]
+    else
+        format *= "zzzz"
+    end
+    CDDate(ZonedDateTime(s, format))
+end
+
+Base.show(io::IO, dt::CDDate) = show(io, dt.d)
+
+@compat struct CDRect{T <: Number}
+    llx::T
+    lly::T
+    urx::T
+    ury::T
+end

src/CosObject.jl

@@ -4,8 +4,6 @@ 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
@@ -191,7 +189,7 @@ showref(io::IO, o::CosObject) = show(io, o)
 
 show(io::IO, o::CosNullType) = print(io, "null")
 
-show(io::IO, o::CosName) = @printf io "/%s" split(String(o.val),'_')[2]
+show(io::IO, o::CosName) = @printf io "/%s" String(o)
 
 show(io::IO, o::CosXString) =  @printf "%s" "<"*String(copy(o.val))*">"
 

src/CosObjectHelpers.jl

@@ -1,6 +1,10 @@
 using ..Common
 
-function Base.convert(::Type{CDTextString}, xstr::CosXString)
+import Base: convert
+
+using StringEncodings
+
+function 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]
@@ -13,7 +17,7 @@ function Base.convert(::Type{CDTextString}, xstr::CosXString)
                 (buffer[2i-1], buffer[2i]) = (buffer[2i], buffer[2i-1])
             end
         end
-        utf_16_data = reinterpret(UInt16, buffer)
+        utf_16_data = reinterpret(UInt16, buffer[3:end])
         str = transcode(String, utf_16_data)
     else
         # Assume PDFDocEncoding (ISO-8859-1)
@@ -22,5 +26,15 @@ function Base.convert(::Type{CDTextString}, xstr::CosXString)
     return CDTextString(str)
 end
 
-Base.convert(::Type{CDTextString}, lstr::CosLiteralString) =
+convert(::Type{CDTextString}, lstr::CosLiteralString) =
     CDTextString(StringEncodings.decode(lstr.val, "ISO_8859-1"))
+
+convert{T <: Number}(::Type{T}, i::CosInt) = T(get(i))
+
+convert{T <: Number}(::Type{T}, f::CosFloat) = T(get(f))
+
+convert(::Type{CDRect}, a::CosArray) = CDRect(a...)
+
+convert{T <: CosString}(::Type{CDDate}, ls::T) = CDDate(CDTextString(ls))
+
+convert(::Type{CDTextString}, name::CosName) = CDTextString(split(String(name.val),'_')[2])

src/PDDoc.jl

@@ -2,7 +2,7 @@ export PDDoc,
        pdDocOpen,
        pdDocClose,
        pdDocGetCatalog,
-       pdDocGetInfo, pdDocGetProducers,
+       pdDocGetInfo,
        pdDocGetCosDoc,
        pdDocGetPageCount,
        pdDocGetPage
@@ -43,14 +43,13 @@ 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)
+    dInfo = Dict{CDTextString, Union{CDTextString, CDDate}}()
+    for (index, data) in enumerate(get(obj))
+        skey = CDTextString(data[1])
+        dInfo[skey] = (skey == "CreationDate") || (skey == "ModDate") ?
+                      CDDate(data[2]) : CDTextString(data[2])
+    end
+    return dInfo
 end
 
 @compat mutable struct PDDocImpl <: PDDoc

test/debugIO.jl

@@ -33,8 +33,8 @@ IODebug=[0,Vector{Tuple{AbstractString,IOStream}}()]
 function get_tempfilepath()
   global IODebug
   IODebug[1]+=1
-  path=get_tempdir()*"/"*string(IODebug[1])
-  return (path,util_open(path,"w"))
+  path = joinpath(get_tempdir(), string(IODebug[1]))
+  return (path, util_open(path,"w"))
 end
 
 function util_open(filename, mode)

test/runtests.jl

@@ -16,8 +16,8 @@ include("debugIO.jl")
       println(pdDocGetCatalog(doc))
       cosDoc = pdDocGetCosDoc(doc)
       map(println, cosDoc.trailer)
-      println(pdDocGetInfo(doc))
-      println(pdDocGetProducers(doc))
+      info = pdDocGetInfo(doc)
+      @assert info["Producer"] == "LibreOffice 5.3" && info["Creator"] == "Writer"
       @assert pdDocGetPageCount(doc) == 2
       page = pdDocGetPage(doc, 1)
       @assert pdPageIsEmpty(page) == false