Merge pull request #19 from kerim371/add-segy-endian-check

Add segy endian check

Author: mloubout

src/SegyIO.jl

@@ -19,6 +19,7 @@ module SegyIO
     include("types/SeisCon.jl")
 
     #Reader
+    include("read/endianness.jl")
     include("read/read_fileheader.jl")
     include("read/read_traceheader.jl")
     include("read/read_trace.jl")

src/read/endianness.jl

@@ -0,0 +1,68 @@
+is_machine_littleendian() = Base.ENDIAN_BOM == 0x04030201
+is_machine_bigendian() = Base.ENDIAN_BOM == 0x01020304
+
+function is_segy_littleendian(s::IO)
+    bfh = BinaryFileHeader()
+    fh_b2s = fh_byte2sample()
+    
+    old_pos = position(s)
+    seek(s, fh_b2s["DataSampleFormat"])
+    dsf = read(s, typeof(bfh.DataSampleFormat))
+    dsf_bs = bswap(dsf)
+    seek(s, old_pos)
+
+    if (dsf >= 1 && dsf <= 16 && is_machine_littleendian()) ||
+       (dsf_bs >= 1 && dsf_bs <= 16 && is_machine_bigendian())
+        return true
+    else
+        return false
+    end
+end
+
+function is_segy_bigendian(s::IO)
+    bfh = BinaryFileHeader()
+    fh_b2s = fh_byte2sample()
+    
+    old_pos = position(s)
+    seek(s, fh_b2s["DataSampleFormat"])
+    dsf = read(s, typeof(bfh.DataSampleFormat))
+    dsf = bswap(read(s, typeof(bfh.DataSampleFormat)))
+    seek(s, old_pos)
+
+    if (dsf >= 1 && dsf <= 16 && is_machine_bigendian()) ||
+       (dsf_bs >= 1 && dsf_bs <= 16 && is_machine_littleendian())
+        return true
+    else
+        return false
+    end
+end
+
+
+"""
+bswap_needed(s::IO)
+
+Checks whether SEGY and host machine's byte order are same or not.
+
+NOTE: comparison is done based on binary file header value (Data Sample Format)
+thus stream (s::IO) must be opened from the begining of the file.
+
+# Examples
+
+s = open("/my/segy.segy", "r")
+swap_bytes = bswap_needed(s)
+"""
+function bswap_needed(s::IO)
+    bfh = BinaryFileHeader()
+    fh_b2s = fh_byte2sample()
+    
+    old_pos = position(s)
+    seek(s, fh_b2s["DataSampleFormat"])
+    dsf = read(s, typeof(bfh.DataSampleFormat))
+    seek(s, old_pos)
+
+    if dsf >= 1 && dsf <= 16
+        return false
+    else
+        return true
+    end
+end

src/read/read_block.jl

@@ -7,6 +7,7 @@ end
 function read_block!(b::BlockScan, keys::Array{String, 1}, ns::Int, dsf::Int, tmp_data, tmp_headers)
 
     f = open(b.file)
+    swap_bytes = bswap_needed(f)
     seek(f, b.startbyte)
     brange = b.endbyte - b.startbyte
     s = IOBuffer(read(f, brange))
@@ -29,6 +30,6 @@ function read_block!(b::BlockScan, keys::Array{String, 1}, ns::Int, dsf::Int, tm
         tracee = min(trace + TRACE_CHUNKSIZE - 1, ntraces)
         chunk = length(trace:tracee)*trace_size
         sloc = IOBuffer(read(s, chunk))
-        read_traces!(sloc, view(tmp_headers, trace:tracee), view(tmp_data, :, trace:tracee), keys, th_b2s)
+        read_traces!(sloc, view(tmp_headers, trace:tracee), view(tmp_data, :, trace:tracee), keys, th_b2s; swap_bytes=swap_bytes)
     end
 end

src/read/read_block_headers.jl

@@ -3,6 +3,7 @@ export read_block_headers!
 function read_block_headers!(b::BlockScan, keys::Array{String, 1}, ns::Int, dsf::Int, headers)
 
     f = open(b.file)
+    swap_bytes = bswap_needed(f)
     seek(f, b.startbyte)
     brange = b.endbyte - b.startbyte
     s = IOBuffer(read(f, brange))
@@ -23,7 +24,7 @@ function read_block_headers!(b::BlockScan, keys::Array{String, 1}, ns::Int, dsf:
     tmph = zeros(UInt8, 240)
     # Read each traceheader
     for trace in 1:ntraces
-        read_traceheader!(s, keys, th_b2s, headers[trace]; th=tmph)
+        read_traceheader!(s, keys, th_b2s, headers[trace]; swap_bytes=swap_bytes, th=tmph)
         skip(s, ns*4)
     end
 
@@ -32,6 +33,7 @@ end
 function read_block_headers!(b::BlockScan, ns::Int, dsf::Int, headers)
 
     f = open(b.file)
+    swap_bytes = bswap_needed(f)
     seek(f, b.startbyte)
     brange = b.endbyte - b.startbyte
     s = IOBuffer(read(f, brange))
@@ -50,7 +52,7 @@ function read_block_headers!(b::BlockScan, ns::Int, dsf::Int, headers)
     th_b2s = th_byte2sample()
     # Read each traceheader
     for trace in 1:ntraces
-        read_traceheader!(s, th_b2s, headers[trace])
+        read_traceheader!(s, th_b2s, headers[trace]; swap_bytes=swap_bytes)
         skip(s, ns*4)
     end
 

src/read/read_file.jl

@@ -18,8 +18,11 @@ Read entire SEGY file from stream 's', only reading the header values in 'keys'.
 """
 function read_file(s::IO, keys::Array{String, 1}, warn_user::Bool; 
                    start_byte::Int = 3600, end_byte::Int = position(seekend(s)))
+
+    swap_bytes = bswap_needed(s)
+
     # Read File Header
-    fh = read_fileheader(s)
+    fh = read_fileheader(s; swap_bytes=swap_bytes)
 
     # Move to start of block
     seek(s, start_byte)
@@ -53,7 +56,7 @@ function read_file(s::IO, keys::Array{String, 1}, warn_user::Bool;
         tracee = min(trace + TRACE_CHUNKSIZE - 1, ntraces)
         chunk = length(trace:tracee)*trace_size
         sloc = IOBuffer(read(s, chunk))
-        read_traces!(sloc, view(headers, trace:tracee), view(data, :, trace:tracee), keys, th_b2s)
+        read_traces!(sloc, view(headers, trace:tracee), view(data, :, trace:tracee), keys, th_b2s; swap_bytes=swap_bytes)
     end
 
     return SeisBlock(fh, headers, data)

src/read/read_fileheader.jl

@@ -4,17 +4,17 @@ export read_fileheader
 """
 # Info
 
-Use: fileheader = read_fileheader(s::IO; bigendian::Bool = true)
+Use: fileheader = read_fileheader(s::IO; swap_bytes::Bool = bswap_needed(s))
 
 Returns a binary file header formed using bytes 3200-3600 from the stream 's'.
 """
-function read_fileheader(s::IO; bigendian::Bool = true)
-    return read_fileheader(s, fh_keys(); bigendian=bigendian)
+function read_fileheader(s::IO; swap_bytes::Bool = bswap_needed(s))
+    return read_fileheader(s, fh_keys(); swap_bytes=swap_bytes)
 end
 
 
 """
-Use: fileheader = read_fileheader(s::IO, keys::Array{String,1}; bigendian::Bool = true)
+Use: fileheader = read_fileheader(s::IO, keys::Array{String,1}; swap_bytes::Bool = bswap_needed(s))
 
 Return a fileheader from stream 's' with the fields defined in 'keys'.
 
@@ -41,7 +41,7 @@ Read only the sample interval and number of traces from the file header.
     0, 0, 0, 0, 0, 0, 0, 0, 0, 0, Dict("expf"=>3226,"sfe"=>3234,"rgc"=>3250,"jobid"=>3200,
     "dt"=>3216,"nsfr"=>3222,"slen"=>3236,"vpol"=>3258,"renum"=>3208,"dsf"=>3224…))
 """
-function read_fileheader(s::IO, keys::Array{String,1}; bigendian::Bool = true)
+function read_fileheader(s::IO, keys::Array{String,1}; swap_bytes::Bool = bswap_needed(s))
 
     # Return to start of stream
     seekstart(s)
@@ -52,7 +52,7 @@ function read_fileheader(s::IO, keys::Array{String,1}; bigendian::Bool = true)
     # Initialize binary file header
     bfh = BinaryFileHeader()
     fh_b2s = fh_byte2sample()
-    swp(x) = bigendian ? bswap(x) : x
+    swp(x) = swap_bytes ? bswap(x) : x
 
     for k in keys
 

src/read/read_trace.jl

@@ -1,54 +1,53 @@
-export read_trace
+export read_traces
 
 """
-Use:    read_trace!(s::IO,
+Use:    read_traces!(s::IO,
                     fh::BinaryFileHeader,
                     datatype::Type,
                     headers::AbstractArray{BinaryTraceHeader,1},
                     data::AbstractArray{<:Union{IBMFloat32, Float32}, 2},
                     trace::Int, 
-                    th_byte2sample::Dict{String,Int32})
+                    th_byte2sample::Dict{String,Int32};
+                    swap_bytes::Bool)
 
 Reads 'trace' from the current position of stream 's' into 'headers' and
 'data'. 
 """
 function read_traces!(s::IO, headers::AbstractVector{BinaryTraceHeader},
                       data::AbstractMatrix{<:Union{IBMFloat32, Float32}},
-                      th_byte2sample::Dict{String,Int32})
+                      th_byte2sample::Dict{String,Int32};
+                      swap_bytes::Bool)
 
-    return read_traces!(s, headers, data, collect(keys(th_byte2sample)), th_byte2sample)
+    return read_traces!(s, headers, data, collect(keys(th_byte2sample)), th_byte2sample; swap_bytes=swap_bytes)
 end
 
 """
-Use:    read_trace!(s::IO,
+Use:    read_traces!(s::IO,
                     fh::BinaryFileHeader,
                     datatype::Type,
                     headers::AbstractArray{BinaryTraceHeader,1},
                     data::AbstractArray{<:Union{IBMFloat32, Float32}, 2},
                     trace::Int,
                     keys::Array{String,1},
-                    th_byte2sample::Dict{String,Int32})
+                    th_byte2sample::Dict{String,Int32};
+                    swap_bytes::Bool)
 
 Reads 'trace' from the current position of stream 's' into 'headers' and
 'data'. Only the header values in 'keys' and read.
 """
 function read_traces!(s::IO, headers::AbstractVector{BinaryTraceHeader},
                       data::AbstractMatrix{DT}, keys::Array{String,1},
-                      th_byte2sample::Dict{String,Int32}) where {DT<:Union{IBMFloat32, Float32}}
+                      th_byte2sample::Dict{String,Int32}; swap_bytes::Bool) where {DT<:Union{IBMFloat32, Float32}}
 
     ntrace = size(data, 2)
     ntrace == 0 && return
-    swp = swp_func(DT)
     tmph = zeros(UInt8, 240)
     for trace_loc=0:ntrace-1
         # Read trace header
-        read_traceheader!(s, keys, th_byte2sample, headers[trace_loc+1]; th=tmph)
+        read_traceheader!(s, keys, th_byte2sample, headers[trace_loc+1]; swap_bytes=swap_bytes, th=tmph)
         # Read trace
         read!(s, view(data, :, trace_loc+1))
     end
-    map!(swp, data, data)
+    (swap_bytes) && (data = bswap.(data))
     nothing
 end
-
-swp_func(::Type{Float32}) = bswap
-swp_func(::Any) = x -> x

src/read/read_traceheader.jl

@@ -3,34 +3,34 @@ export read_traceheader
 """
 # Info
 
-Use: fileheader = read_fileheader(s::IO; bigendian::Bool = true)
+Use: fileheader = read_fileheader(s::IO; swap_bytes::Bool)
 
 
 Returns a binary trace header formed from the current position in the stream 's'.
 """
-function read_traceheader(s::IO, th_byte2sample::Dict{String,Int32}; bigendian::Bool=true, th=zeros(UInt8, 240))
-    return read_traceheader(s, collect(keys(th_byte2sample)), th_byte2sample; bigendian=bigendian, th=th)
+function read_traceheader(s::IO, th_byte2sample::Dict{String,Int32}; swap_bytes::Bool, th=zeros(UInt8, 240))
+    return read_traceheader(s, collect(keys(th_byte2sample)), th_byte2sample; swap_bytes=swap_bytes, th=th)
 end
 
 """
-Use: fileheader = read_traceheader(s::IO, keys = Array{String,1}; bigendian::Bool = true)
+Use: fileheader = read_traceheader(s::IO, keys = Array{String,1}; swap_bytes::Bool)
 
 Returns a binary trace header formed from the current position in the stream 's', only reading
 header values denoted in 'keys'.
 """
 function read_traceheader(s::IO, keys::Array{String,1}, th_byte2sample::Dict{String, Int32};
-                          bigendian::Bool = true, th=zeros(UInt8, 240))
+                          swap_bytes::Bool, th=zeros(UInt8, 240))
 
     # Initialize binary file header
     traceheader = BinaryTraceHeader()
 
-    read_traceheader!(s, keys, th_byte2sample, traceheader; bigendian=bigendian, th=th)
+    read_traceheader!(s, keys, th_byte2sample, traceheader; swap_bytes=swap_bytes, th=th)
     return traceheader
 end
 
 function read_traceheader!(s::IO, keys::Array{String,1}, th_byte2sample::Dict{String, Int32}, hdr::BinaryTraceHeader;
-                          bigendian::Bool = true, th=zeros(UInt8, 240))
-
+                          swap_bytes::Bool, th=zeros(UInt8, 240))
+    
     # read full trace header then split
     read!(s, th)
 
@@ -40,11 +40,11 @@ function read_traceheader!(s::IO, keys::Array{String,1}, th_byte2sample::Dict{St
         nb = sizeof(getfield(hdr, sym)) - 1
         bst = th_byte2sample[k]+1
         val = reinterpret(typeof(getfield(hdr, sym)), th[bst:bst+nb])[1]
-        bigendian && (val = bswap(val))
+        swap_bytes && (val = bswap(val))
         setfield!(hdr, sym, val)
     end
     nothing
 end
 
-read_traceheader!(s::IO, thb::Dict{String, Int32}, hdr::BinaryTraceHeader; be::Bool = true, th=zeros(UInt8, 240)) =
-    read_traceheader!(s, collect(keys(thb)), thb, hdr; be=be, th=th)
+read_traceheader!(s::IO, thb::Dict{String, Int32}, hdr::BinaryTraceHeader; swap_bytes::Bool, th=zeros(UInt8, 240)) =
+    read_traceheader!(s, collect(keys(thb)), thb, hdr; swap_bytes=swap_bytes, th=th)

src/scan/scan_block.jl

@@ -2,7 +2,11 @@ export scan_block
 
 function scan_block(buf::IO, mem_block::Int, mem_trace::Int, keys::Array{String,1},
                     chunk_start::Int, file::String, th_byte2sample::Dict{String, Int32})
-    
+
+    f = open(file)
+    swap_bytes = bswap_needed(f)
+    close(f)
+
     # Calc info about this block
     startbyte = position(buf) + chunk_start
     ntraces_block = Int(mem_block/mem_trace)
@@ -12,7 +16,7 @@ function scan_block(buf::IO, mem_block::Int, mem_trace::Int, keys::Array{String,
     # Read all headers and record end byte
     while !eof(buf) && count<ntraces_block
         count += 1
-        read_traceheader!(buf, keys, th_byte2sample,  headers[count] )
+        read_traceheader!(buf, keys, th_byte2sample,  headers[count]; swap_bytes=swap_bytes)
         skip(buf, mem_trace-240)
     end 
     endbyte = position(buf) + chunk_start

src/scan/scan_shots.jl

@@ -3,6 +3,10 @@ export scan_shots
 function scan_shots!(s::IO, mem_chunk::Int, mem_trace::Int,
                     keys::Array{String,1}, file::AbstractString, scan::Array{BlockScan,1}, fl_eof::Bool)
 
+    f = open(file)
+    swap_bytes = bswap_needed(f)
+    close(f)
+
     # Load chunk into memory
     chunk_start = position(s)
     buf = IOBuffer(read(s, mem_chunk))
@@ -14,7 +18,7 @@ function scan_shots!(s::IO, mem_chunk::Int, mem_trace::Int,
     # Get headers from chunk
     th = zeros(UInt8, 240)
     for i in 1:ntraces
-        read_traceheader!(buf, keys, SegyIO.th_b2s, headers[i]; th=th)
+        read_traceheader!(buf, keys, SegyIO.th_b2s, headers[i]; swap_bytes=swap_bytes, th=th)
         skip(buf, mem_trace-240)
     end
 

src/types/IBMFloat32.jl

@@ -4,7 +4,7 @@ primitive type IBMFloat32 32 end
 ieeeOfPieces(fr::UInt32, exp::Int32, sgn::UInt32) = 
     reinterpret(Float32, convert(UInt32,fr >>> 9) | convert(UInt32,exp << 23) | sgn) :: Float32
 
-import Base.convert, Base.Float32
+import Base.convert, Base.Float32, Base.bswap
 
 function convert(::Type{Float32}, ibm::IBMFloat32)
     local fr::UInt32 = ntoh(reinterpret(UInt32, ibm))
@@ -29,3 +29,7 @@ function convert(::Type{Float32}, ibm::IBMFloat32)
 end
 Float32(ibm::IBMFloat32) = convert(Float32,ibm)
 ## From JuliaSeis
+
+function bswap(ibm::IBMFloat32)
+    return reinterpret(IBMFloat32, bswap(reinterpret(UInt32, ibm)))
+end