Fix for #99. Also removed some code redundant now.

Author: sambitdash

.gitignore

@@ -31,6 +31,7 @@ test/*.pdf
 test/*.res
 test/pvt
 test/PDFTest*
+test/pdftest*
 file.txt
 data/fonts/Arial.afm
 test/sample01.pem

Project.toml

@@ -31,6 +31,7 @@ julia = "1.6"
 [extras]
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
 ZipFile = "a5390f91-8eb1-5f08-bee0-b1d1ffed6cea"
+Downloads = "f43a241f-c20a-4ad4-852c-f6b1247861c6"
 
 [targets]
-test = ["Test", "ZipFile"]
+test = ["Test", "ZipFile", "Downloads"]

src/Inflate.jl

@@ -45,13 +45,7 @@ const Z_MEM_ERROR     = -4
 const Z_BUF_ERROR     = -5
 const Z_VERSION_ERROR = -6
 
-@static if Base.VERSION > v"1.3-" 
-    using Zlib_jll: libz
-else
-    isfile(joinpath(dirname(@__FILE__),"..","deps","deps.jl")) ||
-        error("PDFIO not properly installed. Please run Pkg.build(\"PDFIO\")")
-    include("../deps/deps.jl")
-end
+using Zlib_jll: libz
 
 _zlibVersion() = ccall((:zlibVersion, libz), Ptr{Cstring}, ())
 

src/LibCrypto.jl

@@ -1,10 +1,4 @@
-@static if Base.VERSION > v"1.3-"
-    using OpenSSL_jll: libcrypto
-else
-    isfile(joinpath(dirname(@__FILE__), "..", "deps", "deps.jl")) ||
-        error("PDFIO not properly installed. Please run Pkg.build(\"PDFIO\")")
-    include("../deps/deps.jl")
-end
+using OpenSSL_jll: libcrypto
 
 using Base: SecretBuffer, SecretBuffer!
 import Base: copy

src/PDFonts.jl

@@ -335,7 +335,7 @@ cmap_command(b::Vector{UInt8}) =
     nothing : Symbol(String(b))
 
 function on_cmap_command!(stm::IO, command::Symbol,
-                         params::Vector{CosInt}, cmap::CMap)
+                         params::Vector{CosInt}, cmap::CMap) 
     n = get(pop!(params))
     o1, o2, o3 = CosNull, CosNull, CosNull
     for i = 1:n
@@ -358,12 +358,30 @@ function on_cmap_command!(stm::IO, command::Symbol,
             end
         else
             l = length(d1)
+            @assert (d1[1] <= d2[1]) E_INVALID_CODESPACERANGE
             if l == 1
                 cmap.code_space[Interval(d1[1], d2[1])] = CosNull
             else
-                imap = IntervalTree{UInt8, CosNullType}()
-                imap[Interval(d1[2], d2[2])] = CosNull
-                cmap.code_space[Interval(d1[1], d2[1])] = imap
+                if d1[2] < d2[2]
+                    imap = IntervalTree{UInt8, CosNullType}()
+                    imap[Interval(d1[2], d2[2])] = CosNull
+                    cmap.code_space[Interval(d1[1], d2[1])] = imap
+                else
+                    @warn "Corrupt CMap file. Repairing... Some encodings may not map properly."
+                    imap = IntervalTree{UInt8, CosNullType}()
+                    imap[Interval(d1[2], 0xff)] = CosNull
+                    cmap.code_space[Interval(d1[1], d1[1])] = imap
+
+                    imap = IntervalTree{UInt8, CosNullType}()
+                    imap[Interval(0x00, d2[2])] = CosNull
+                    cmap.code_space[Interval(d2[1], d2[1])] = imap
+
+                    if d2[1] - d1[1] > 1
+                        imap = IntervalTree{UInt8, CosNullType}()
+                        imap[Interval(0x00, 0xff)] = CosNull
+                        cmap.code_space[Interval(d1[1]+1, d2[1]-1)] = imap    
+                    end
+                end
             end
         end
     end

src/errors.jl

@@ -2,7 +2,7 @@ export  E_EXPECTED_EOF, E_UNEXPECTED_EOF, E_UNEXPECTED_CHAR, E_BAD_KEY,
         E_BAD_ESCAPE, E_BAD_CONTROL, E_LEADING_ZERO, E_BAD_NUMBER, E_BAD_HEADER,
         E_BAD_TRAILER, E_BAD_TYPE, E_NOT_IMPLEMENTED,
         E_INVALID_OBJECT, E_INVALID_PAGE_NUMBER, E_INVALID_PAGE_LABEL,
-        E_NOT_TAGGED_PDF, E_INVALID_PASSWORD
+        E_NOT_TAGGED_PDF, E_INVALID_PASSWORD, E_INVALID_CODESPACERANGE
 
 # The following errors may be thrown by the reader
 const E_EXPECTED_EOF    = "Expected end of input"
@@ -28,3 +28,4 @@ const E_INVALID_PASSWORD    = "The password supplied to open the document is inv
 const E_INVALID_CRYPT       = "The crypt handler is not supported"
 const E_NOT_TAGGED_PDF      = "PDF file is not tagged"
 const E_NOT_IMPLEMENTED     = "Not Implemented"
+const E_INVALID_CODESPACERANGE = "Invalid code space range in CMap"

test/runtests.jl

@@ -5,15 +5,16 @@ using PDFIO.Cos
 using PDFIO.Common
 using ZipFile
 using AbstractTrees
+using Downloads
 
 # Internal methods for testing only
 using PDFIO.Cos: parse_indirect_ref, decode_ascii85, CosXString, parse_value
-using PDFIO.PD: openssl_error
+using PDFIO.PD: openssl_error, read_cmap, get_encoded_string
 using PDFIO.Common: read_pkcs12
 
 include("debugIO.jl")
 
-pdftest_ver  = "0.0.9"
+pdftest_ver  = "0.0.10"
 pdftest_link = "https://github.com/sambitdash/PDFTest/archive/v"*pdftest_ver
 
 zipfile = "pdftest-"*pdftest_ver
@@ -22,7 +23,7 @@ zipfile *= ".zip"
 pdftest_dir="PDFTest-"*pdftest_ver*"/"
 
 if !isdir(pdftest_dir)
-    isfile(zipfile) || download(pdftest_link, zipfile)
+    isfile(zipfile) || Downloads.download(pdftest_link, zipfile)
     r = ZipFile.Reader(zipfile)
     buf = Vector{UInt8}(undef, 64*1024)
     for f in r.files
@@ -57,6 +58,8 @@ function local_testfiles(filename, filesdir="files")
             joinpath(@__DIR__, pdftest_dir, filesdir, filename))
 end
 
+local_files(filename, filesdir="files") = joinpath(@__DIR__, pdftest_dir, filesdir, filename)
+
 @testset "PDFIO tests" begin
     @testset "Miscellaneous" begin
         @test_throws ErrorException skipv(IOBuffer([UInt8(65), UInt8(66)]),
@@ -416,6 +419,23 @@ end
             @test files_equal(resfile, template)
             length(utilPrintOpenFiles()) == 0
         end
+
+        @test begin
+            filename="bad.cmap"
+            DEBUG && println(filename)
+            path = local_files(filename)
+            io = util_open(path, "r")
+            cmap = nothing
+            try
+                cmap = read_cmap(io)
+            finally
+                util_close(io)
+            end
+            @test cmap !== nothing
+            @test all(get_encoded_string(UInt8[0x00, 0xff], cmap).== [Char(0x0111)])
+            @test all(get_encoded_string(UInt8[0x01, 0x08], cmap).== [Char(0x0110)])
+            all(get_encoded_string(UInt8[0x00, 0xfb, 0x00, 0xfe], cmap) .== [Char(0x0106), Char(0x010D)])
+        end
     end
 
     @testset "Corrupt File" begin