Few code cleanup and font encoding differences.

Author: sambitdash

REQUIRE

@@ -6,3 +6,4 @@ TimeZones
 Documenter
 LabelNumerals
 IntervalTrees
+agl_aglfn

docs/src/index.md

@@ -66,6 +66,7 @@ pdPageGetContents
 pdPageIsEmpty
 pdPageGetCosObject
 pdPageGetContentObjects
+pdPageExtractText
 ```
 ## PDF Page objects
 ```@docs

src/CosDoc.jl

@@ -126,6 +126,20 @@ to go through checking the type of the objects before accessing the contents.
 """
 cosDocGetObject(doc::CosDoc, obj::CosObject) = CosNull
 
+"""
+```
+    cosDocGetObject(doc::CosDoc, dict::CosObject, key::CosName) -> CosObject
+```
+Returns the object referenced inside the `dict` dictionary. `dict` can be a PDF dictionary
+object reference or an indirect object or a direct `CosDict` object.
+"""
+function cosDocGetObject(doc::CosDoc, dict::CosObject, key::CosName)
+    if dict isa CosIndirectObjectRef
+        dict = cosDocGetObject(doc, dict)
+    end
+    dict === CosNull && return CosNull
+    return cosDocGetObject(doc, get(dict, key))
+end
 
 function cosDocGetRoot(doc::CosDocImpl)
     root = doc.hasNativeXRefStm ? get(doc.xrefstm[1], CosName("Root")) :

src/CosObject.jl

@@ -3,7 +3,7 @@ import Base:get, length, show
 export CosDict, CosString, CosNumeric, CosBoolean, CosTrue, CosFalse,
        CosObject, CosNull, CosNullType,CosFloat, CosInt, CosArray, CosName,
        CosDict, CosIndirectObjectRef, CosStream, get, set!, @cn_str,
-       createTreeNode, CosTreeNode
+       createTreeNode, CosTreeNode, CosIndirectObject
 
 """
 ```

src/PDPage.jl

@@ -64,10 +64,11 @@ end
 ```
     pdPageExtractText(io::IO, page::PDPage) -> IO
 ```
-Extracts the text from the `page`. This extraction works only for tagged PDF files only.
+Extracts the text from the `page`. This extraction works best for tagged PDF files only.
+For PDFs not tagged, some line and word breaks will not be extracted properly. 
 """
 function pdPageExtractText(io::IO, page::PDPage)
-    page.doc.isTagged != :tagged && throw(ErrorException(E_NOT_TAGGED_PDF))
+    # page.doc.isTagged != :tagged && throw(ErrorException(E_NOT_TAGGED_PDF))
     state = Dict()
     state[:page] = page
     showtext(io, pdPageGetContentObjects(page), state)
@@ -144,16 +145,47 @@ function load_page_objects(page::PDPageImpl, stm::CosArray)
   end
 end
 
-function merge_encoding(pdfont::PDFont, encoding::CosName, page::PDPage, font::CosObject)
-    pdfont.encoding = encoding == cn"WinAnsiEncoding"   ? WINEncoding_to_Unicode :
-                      encoding == cn"MacRomanEncoding"  ? MACEncoding_to_Unicode :
-                      encoding == cn"MacExpertEncoding" ? MEXEncoding_to_Unicode :
-                      STDEncoding_to_Unicode
+function merge_encoding!(pdfont::PDFont, encoding::CosName, page::PDPage, font::CosObject)
+    encoding_mapping =  encoding == cn"WinAnsiEncoding"   ? WINEncoding_to_Unicode :
+                        encoding == cn"MacRomanEncoding"  ? MACEncoding_to_Unicode :
+                        encoding == cn"MacExpertEncoding" ? MEXEncoding_to_Unicode :
+                        STDEncoding_to_Unicode
+    merge!(pdfont.encoding, encoding_mapping)
+    return pdfont
 end
 
-function merge_encoding(pdfont::PDFont, encoding::CosNullType,
+# for type 0 use cmap.
+# for symbol and zapfdingbats - use font encoding
+# for others use STD Encoding
+function merge_encoding!(pdfont::PDFont, encoding::CosNullType,
                         page::PDPage, font::CosObject)
-    pdfont.encoding = STDEncoding_to_Unicode
+    merge!(pdfont.encoding, STDEncoding_to_Unicode)
+    return pdfont
+end
+
+function merge_encoding!(pdfont::PDFont,
+                        encoding::Union{CosDict, CosIndirectObject{CosDict}},
+                        page::PDPage, font::CosObject)
+    baseenc = cosDocGetObject(page.doc.cosDoc, get(encoding, cn"BaseEncoding"))
+    baseenc !==  CosNull && merge_encoding!(pdfont, baseenc, page, font)
+    # Add the Differences
+    diff = cosDocGetObject(page.doc.cosDoc, get(encoding, cn"Differences"))
+    diff === CosNull && return pdfont
+    values = get(diff)
+    d = Dict()
+    cid = 0
+    for v in values
+        if v isa CosInt
+            cid = get(v)
+        else
+            @assert cid != 0
+            d[cid] = v
+            cid += 1
+        end
+    end
+    dict_to_unicode = dict_remap(d, AGL_Glyph_to_Unicode)
+    merge!(pdfont.encoding, dict_to_unicode)
+    return pdfont
 end
 
 function populate_font_encoding(page, font, fontname)
@@ -163,7 +195,7 @@ function populate_font_encoding(page, font, fontname)
         #diff = cosDocGetObject(page.doc.cosDoc, get(font, cn"Differences"))
         toUnicode = cosDocGetObject(page.doc.cosDoc, get(font, cn"ToUnicode"))
         #pdfont.toUnicode = read_cmap(toUnicode)
-        merge_encoding(pdfont, encoding, page, font)
+        merge_encoding!(pdfont, encoding, page, font)
         page.fonts[fontname] = pdfont
     end
 end

src/Utils.jl

@@ -42,9 +42,6 @@ function dict_remap(ab, bc)
     d = Dict()
     for (a, b) in ab
         c = get(bc, b, zero(valtype(bc)))
-        if a == 0x20
-            println("a $a: b $b :c -$c-")
-        end
         d[a] = c
     end
     return d

test/runtests.jl

@@ -30,11 +30,6 @@ include("debugIO.jl")
             pdDocClose(doc)
             length(utilPrintOpenFiles()) == 0
         end
-        @test_throws ErrorException begin
-            doc = pdDocOpen("files/1.pdf")
-            page = pdDocGetPage(doc, 1)
-            pdPageExtractText(IOBuffer(), page)
-        end
   end
 
     @testset "PDF File with ObjectStreams" begin
@@ -81,7 +76,6 @@ include("debugIO.jl")
             try
                 npage= pdDocGetPageCount(doc)
                 for i=1:npage
-                    println(i)
                     page = pdDocGetPage(doc, i)
                     if pdPageIsEmpty(page) == false
                         pdPageGetContentObjects(page)