Support for Symbol and ZapfDingbats

Also fixes bugs related to octal character parsing in CosLiteralString

Author: sambitdash

431.txt

@@ -114,18 +114,14 @@ function parse_array(ps::BufferedInputStream)
 end
 
 function read_octal_escape!(c, ps)
-    local n::UInt16 = (c << 3)
+    local n::UInt8 = getnumval(c)
     for _ in 1:2
         b = peek(ps)
-        n = n << 3
-        if (ispdfodigit(b))
-            n += b
-            skip(ps,1)
-        else
-            break
-        end
+        !ispdfodigit(b) && return n
+        n = (n << 3) + getnumval(b)
+        skip(ps,1)
     end
-    n
+    return n
 end
 
 
@@ -139,7 +135,7 @@ function parse_string(ps::BufferedInputStream)
     if c == BACKSLASH
       c = advance!(ps)
       if ispdfodigit(c) #Read octal digits
-        append!(b, Vector{UInt8}(string(read_octal_escape!(c,ps))))
+        append!(b, read_octal_escape!(c,ps))
       elseif is_crorlf(c) #ignore the solidus, EOLs and move on
         chomp_space!(ps)
       else

src/CosReader.jl

@@ -57,6 +57,53 @@ endcmap
 CMapName currentdict /CMap defineresource pop
 end
 end
+
+Single byte ToUnicode CMap
+
+/CIDInit /ProcSet findresource begin 12 dict begin begincmap /CIDSystemInfo <<
+/Registry (F15+0) /Ordering (T1UV) /Supplement 0 >> def
+/CMapName /F15+0 def
+/CMapType 2 def
+1 begincodespacerange <01> <c9> endcodespacerange
+18 beginbfchar
+<05> <260E>
+<0a> <261B>
+<0b> <261E>
+<20> <0020>
+<29> <2605>
+<4d> <25CF>
+<4e> <274D>
+<4f> <25A0>
+<54> <25B2>
+<55> <25BC>
+<56> <25C6>
+<57> <2756>
+<58> <25D7>
+<75> <2663>
+<76> <2666>
+<77> <2665>
+<78> <2660>
+<a2> <2192>
+endbfchar
+15 beginbfrange
+<01> <04> <2701>
+<06> <09> <2706>
+<0c> <1f> <270C>
+<21> <28> <2720>
+<2a> <4c> <2729>
+<50> <53> <274F>
+<59> <5f> <2758>
+<60> <6d> <F8D7>
+<6e> <74> <2761>
+<79> <82> <2460>
+<83> <a1> <2776>
+<a3> <a4> <2194>
+<a5> <a7> <2798>
+<a8> <bb> <279C>
+<bc> <c9> <27B1>
+endbfrange
+endcmap CMapName currentdict /CMap defineresource pop end end
+
 =#
 const beginbfchar = b"beginbfchar"
 const endbfchar   = b"endbfchar"
@@ -139,18 +186,6 @@ function read_cmap(cmap::CosObject)
     return tcmap
 end
 
-get_encoded_string(s, font::CosNullType, page) = CDTextString(s)
-
-get_encoded_string(s, font, page::CosNullType) = CDTextString(s)
-
-# Simply applying ISO_8859-1. Not correct actually encoding tables to be consulted.
-# like: WinAnsiEncoding, MacRomanEncoding, MacExpertEncoding or PDFDocEncoding
-get_encoded_string(s::CosString, encoding::CosName) = CDTextString(s)
-
-# Differences should be specifically mapped.
-
-get_encoded_string(s::CosString, encoding::CosDict) = CDTextString(s)
-
 #=
 function get_encoded_string(s::CosXString, cmap::CosObject)
     cmap_vec = read_cmap(cmap)

src/PDFonts.jl

@@ -65,7 +65,7 @@ end
     pdPageExtractText(io::IO, page::PDPage) -> IO
 ```
 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. 
+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))
@@ -157,9 +157,19 @@ end
 # for type 0 use cmap.
 # for symbol and zapfdingbats - use font encoding
 # for others use STD Encoding
+# Reading encoding from the font files in case of Symbolic fonts are not supported.
+# Font subset is addressed with font name identification.
 function merge_encoding!(pdfont::PDFont, encoding::CosNullType,
                         page::PDPage, font::CosObject)
-    merge!(pdfont.encoding, STDEncoding_to_Unicode)
+    subtype  = cosDocGetObject(page.doc.cosDoc, font, cn"Subtype")
+    (subtype != cn"Type1") && (subtype != cn"MMType1") && return pdfont
+    basefont = cosDocGetObject(page.doc.cosDoc, font, cn"BaseFont")
+    basefont_with_subset = CDTextString(basefont)
+    basefont_str = rsplit(basefont_with_subset, '+';limit=2)[end]
+    enc = (basefont_str == "Symbol") ? SYMEncoding_to_Unicode :
+          (basefont_str == "ZapfDigbats") ? ZAPEncoding_to_Unicode :
+          STDEncoding_to_Unicode
+    merge!(pdfont.encoding, enc)
     return pdfont
 end
 

src/PDPage.jl

@@ -62,7 +62,7 @@ function NativeEncodingToUnicode(barr::Vector{UInt8}, mapping::Dict)
     l = length(barr)
     carr = Vector{Char}(l)
     for i = 1:l
-        carr[i] = get(mapping, barr[i], Char(0x0))
+        carr[i] = get(mapping, barr[i], zero(Char))
     end
     return carr
 end

src/Utils.jl

@@ -9,7 +9,7 @@ export BACKSPACE, NULL,TAB, LINE_FEED,NEWLINE, FORM_FEED, RETURN,
        PERCENT,PERIOD, NUMBER_SIGN, BANG, TILDE, LATIN_UPPER_D,STREAM,ENDSTREAM,
        LATIN_UPPER_E,LATIN_UPPER_F, LATIN_UPPER_O, LATIN_UPPER_P,
        LATIN_UPPER_R,XREF, TRAILER, STARTXREF, EOF, OBJ, ENDOBJ, ispdfspace,
-       ispdfdelimiter,ispdfdigit, ispdfodigit, ispdfxdigit, gethexval, is_crorlf
+       ispdfdelimiter, ispdfdigit, ispdfodigit, ispdfxdigit, gethexval, getnumval, is_crorlf
 
 
 
@@ -146,4 +146,7 @@ gethexval(b::UInt8) =   (DIGIT_ZERO <= b <= DIGIT_NINE) ? b - DIGIT_ZERO :
                         (LATIN_A <= b <= LATIN_F) ? b - LATIN_A + 0xa :
                         throw(ErrorException(E_BAD_NUMBER))
 
+getnumval(b::UInt8) =   (DIGIT_ZERO <= b <= DIGIT_NINE) ? b - DIGIT_ZERO :
+                        throw(ErrorException(E_BAD_NUMBER))
+
 is_crorlf(b::UInt8) = ((b == RETURN) ||(b == LINE_FEED))

src/bytes.jl

@@ -171,6 +171,26 @@ include("debugIO.jl")
         end
     end
 
+    @testset "Symbol Fonts test" begin
+        @test begin
+            filename="431.pdf"
+            DEBUG && println(filename)
+            isfile(filename) ||
+                download("http://www.stillhq.com/pdfdb/000431/data.pdf",filename)
+            doc = pdDocOpen(filename)
+            (npage = pdDocGetPageCount(doc)) == 54
+            for i=1:npage
+                page = pdDocGetPage(doc, i)
+                if pdPageIsEmpty(page) == false
+                    pdPageGetContentObjects(page)
+                    pdPageExtractText(IOBuffer(), page)
+                end
+            end
+            pdDocClose(doc)
+            length(utilPrintOpenFiles()) == 0
+        end
+    end
+
     files=readdir(get_tempdir())
     @assert length(files) == 0
 end