Merge pull request #10 from slimgroup/gen-and-perf

Improve implementation for better performance with remotes

Author: mloubout

.github/workflows/main.yml

@@ -19,7 +19,7 @@ jobs:
       fail-fast: false
 
       matrix:
-        version: ['1.6', '1.7']
+        version: ['1.6', '1.7', '1.8']
         os: [ubuntu-latest, macos-latest]
 
     steps:

Project.toml

@@ -1,7 +1,7 @@
 name = "SegyIO"
 uuid = "157a0f19-4d44-4de5-a0d0-07e2f0ac4dfa"
 authors = ["Henryk Modzelewski <[email protected]>"]
-version = "0.7.7"
+version = "0.8.0"
 
 [deps]
 Distributed = "8ba89e20-285c-5b6f-9357-94700520ee1b"

src/SegyIO.jl

@@ -1,8 +1,9 @@
 module SegyIO
 
     myRoot=dirname(dirname(pathof(SegyIO)))
-    CHUNKSIZE=2048
-    MB2B=1024^2
+    CHUNKSIZE = 2048
+    TRACE_CHUNKSIZE = 2048
+    MB2B = 1024^2
 
     # what's being used
     using Distributed

src/read/read_block.jl

@@ -1,29 +1,7 @@
 export read_block!
 
 function read_block!(b::BlockScan, ns::Int, dsf::Int, tmp_data, tmp_headers)
-
-    f = open(b.file)
-    seek(f, b.startbyte)
-    brange = b.endbyte - b.startbyte
-    s = IOBuffer(read(f, brange))
-    ntraces = Int((brange)/(240 + ns*4))
-    fh = FileHeader()
-    fh.bfh.ns = ns; fh.bfh.DataSampleFormat = dsf
-
-    # Check dsf
-    datatype = Float32
-    if fh.bfh.DataSampleFormat == 1
-        datatype = IBMFloat32
-    elseif fh.bfh.DataSampleFormat != 5
-        @error "Data type not supported ($(fh.bfh.DataSampleFormat))"
-    end
-
-    th_b2s = th_byte2sample()
-    # Read each trace
-    for trace in 1:ntraces
-        read_trace!(s, fh.bfh, datatype, tmp_headers, tmp_data, trace, th_b2s)
-    end
-
+    read_block!(b, th_keys(), ns, dsf, tmp_data, tmp_headers)
 end
 
 function read_block!(b::BlockScan, keys::Array{String, 1}, ns::Int, dsf::Int, tmp_data, tmp_headers)
@@ -32,7 +10,8 @@ function read_block!(b::BlockScan, keys::Array{String, 1}, ns::Int, dsf::Int, tm
     seek(f, b.startbyte)
     brange = b.endbyte - b.startbyte
     s = IOBuffer(read(f, brange))
-    ntraces = Int((brange)/(240 + ns*4))
+    trace_size = (240 + ns*4)
+    ntraces = Int((brange)/trace_size)
     fh = FileHeader()
     fh.bfh.ns = ns; fh.bfh.DataSampleFormat = dsf
 
@@ -46,8 +25,10 @@ function read_block!(b::BlockScan, keys::Array{String, 1}, ns::Int, dsf::Int, tm
 
     th_b2s = th_byte2sample()
     # Read each trace
-    for trace in 1:ntraces
-        read_trace!(s, fh.bfh, datatype, tmp_headers, tmp_data, trace, keys, th_b2s)
+    for trace in 1:TRACE_CHUNKSIZE:ntraces
+        tracee = min(trace + TRACE_CHUNKSIZE - 1, ntraces)
+        chunk = length(trace:tracee)*trace_size
+        sloc = IOBuffer(read(s, chunk))
+        read_traces!(sloc, fh.bfh, datatype, tmp_headers, view(tmp_data, :, trace:tracee), trace, keys, th_b2s)
     end
-
 end

src/read/read_file.jl

@@ -6,43 +6,9 @@ read_file(s)
 Read entire SEGY file from stream 's'.
 """
 function read_file(s::IO, warn_user::Bool; start_byte::Int = 3600,
-                                           end_byte::Int = position(seekend(s)))
-    
-    # Read File Header
-    fh = read_fileheader(s)
-    
-    # Move to start of block
-    seek(s, start_byte)
-
-    # Check datatype of file
-    datatype = Float32
-    if fh.bfh.DataSampleFormat == 1
-        datatype = IBMFloat32
-    elseif fh.bfh.DataSampleFormat != 5
-        @error "Data type not supported ($(fh.bfh.DataSampleFormat))"
-    end
-
-    # Check fixed length trace flag
-    (fh.bfh.FixedLengthTraceFlag!=1 & warn_user) && @warn "Fixed length trace flag set in stream: $s"
-    
-    ## Check for extended text header
-
-    # Read traces
-    ntraces = Int((end_byte - start_byte)/(240 + fh.bfh.ns*4))
-
-    # Preallocate memory
-    headers = Array{BinaryTraceHeader, 1}(undef, ntraces)
-    data = Array{datatype, 2}(undef, fh.bfh.ns, ntraces)
-    th_b2s = th_byte2sample()
-
-    # Read each trace
-    for trace in 1:ntraces
-
-        read_trace!(s, fh.bfh, datatype, headers, data, trace, th_b2s)
-
-    end
-
-    return SeisBlock(fh, headers, data)
+                   end_byte::Int = position(seekend(s)))
+    # Read with all keys
+    return read_file(s, th_keys(), warn_user; start_byte=start_byte, end_byte=end_byte)
 end
 
 """
@@ -51,8 +17,7 @@ read_file(s, keys)
 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)))
+                   start_byte::Int = 3600, end_byte::Int = position(seekend(s)))
 
     # Read File Header
     fh = read_fileheader(s)
@@ -64,7 +29,7 @@ function read_file(s::IO, keys::Array{String, 1}, warn_user::Bool;
     datatype = Float32
     if fh.bfh.DataSampleFormat == 1
         datatype = IBMFloat32
-    else
+    elseif fh.bfh.DataSampleFormat != 5
         @error "Data type not supported ($(fh.bfh.DataSampleFormat))"
     end
 
@@ -74,18 +39,20 @@ function read_file(s::IO, keys::Array{String, 1}, warn_user::Bool;
     ## Check for extended text header
 
     # Read traces
-    ntraces = Int((end_byte - start_byte)/(240 + fh.bfh.ns*4))
+    trace_size = (240 + fh.bfh.ns*4)
+    ntraces = Int((end_byte - start_byte)/trace_size)
 
     # Preallocate memory
     headers = Array{BinaryTraceHeader, 1}(undef, ntraces)
     data = Array{datatype, 2}(undef, fh.bfh.ns, ntraces)
     th_b2s = th_byte2sample()
 
     # Read each trace
-    for trace in 1:ntraces
-
-        read_trace!(s, fh.bfh, datatype, headers, data, trace, keys, th_b2s)
-
+    for trace in 1:TRACE_CHUNKSIZE:ntraces
+        tracee = min(trace + TRACE_CHUNKSIZE - 1, ntraces)
+        chunk = length(trace:tracee)*trace_size
+        sloc = IOBuffer(read(s, chunk))
+        read_traces!(sloc, fh.bfh, datatype, headers, view(data, :, trace:tracee), trace, keys, th_b2s)
     end
 
     return SeisBlock(fh, headers, data)

src/read/read_fileheader.jl

@@ -9,45 +9,7 @@ Use: fileheader = read_fileheader(s::IO; bigendian::Bool = true)
 Returns a binary file header formed using bytes 3200-3600 from the stream 's'.
 """
 function read_fileheader(s::IO; bigendian::Bool = true)
-    
-    # Return to start of stream
-    seekstart(s)
-
-    # Read text header
-    th = String(read(s, 3200))
-
-    # Initialize binary file header
-    bfh = BinaryFileHeader()
-    fh_b2s = fh_byte2sample()
-
-
-    if bigendian
-        # Read all header values and put in fileheader
-        for k in keys(fh_b2s)
-
-           # Seek to this header value location
-           seek(s, fh_b2s[k])
-
-           # Read into file header
-           sym = Symbol(k)
-           setfield!(bfh, sym, bswap(read(s, typeof(getfield(bfh, sym)))))
-        end
-    else
-        # Read all header values and put in fileheader
-        for k in keys(fh_b2s)
-
-           # Seek to this header value location
-           seek(s, fh_b2s[k])
-
-           # Read into file header
-           sym = Symbol(k)
-           setfield!(bfh, sym, read(s, typeof(getfield(bfh, sym))))
-        end
-    end
-    
-    # Move to end of file header
-    seek(s, 3600)
-    return FileHeader(th, bfh)
+    return read_fileheader(s, fh_keys(); bigendian=bigendian)
 end
 
 
@@ -85,39 +47,23 @@ function read_fileheader(s::IO, keys::Array{String,1}; bigendian::Bool = true)
     seekstart(s)
 
     # Read text header
-    th = String(read(s, 3200))
+    th = read(s, 3600)
 
     # Initialize binary file header
     bfh = BinaryFileHeader()
     fh_b2s = fh_byte2sample()
+    swp(x) = bigendian ? bswap(x) : x
 
+    for k in keys
 
-    if bigendian
-        # Read all header values and put in fileheader
-        for k in keys
-
-           # Seek to this header value location
-           seek(s, fh_b2s[k])
-
-           # Read into file header
-           sym = Symbol(k)
-           setfield!(bfh, sym, bswap(read(s, typeof(getfield(bfh, sym)))))
-        end
-    else
-        # Read all header values and put in fileheader
-        for k in keys
-
-           # Seek to this header value location
-           seek(s, fh_b2s[k])
-
-           # Read into file header
-           sym = Symbol(k)
-           setfield!(bfh, sym, read(s, typeof(getfield(bfh, sym))))
-        end
+        # Read into file header
+        sym = Symbol(k)
+        nb = sizeof(getfield(bfh, sym))-1
+        bst = fh_b2s[k]+1
+        val = swp(reinterpret(typeof(getfield(bfh, sym)), th[bst:bst+nb])[1])
+        setfield!(bfh, sym, val)
     end
 
-    # Move to end of file header
     seek(s, 3600)
-
-    return FileHeader(th, bfh)
+    return FileHeader(String(th[1:3200]), bfh)
 end

src/read/read_trace.jl

@@ -12,20 +12,12 @@ Use:    read_trace!(s::IO,
 Reads 'trace' from the current position of stream 's' into 'headers' and
 'data'. 
 """
-function read_trace!(s::IO, fh::BinaryFileHeader,
-                    datatype::Type,
-                    headers::AbstractArray{BinaryTraceHeader,1},
-                    data::AbstractArray{<:Union{IBMFloat32, Float32}, 2},
-                    trace::Int, 
-                    th_byte2sample::Dict{String,Int32})
-    
-    # Read trace header
-    setindex!(headers, read_traceheader(s, th_byte2sample), trace)
-    
-    # Read trace
-    setindex!(data, read_tracedata(s, fh, datatype), :, trace)
-    
-    nothing
+function read_traces!(s::IO, fh::BinaryFileHeader, datatype::Type,
+                    headers::AbstractVector{BinaryTraceHeader},
+                    data::AbstractMatrix{<:Union{IBMFloat32, Float32}},
+                    trace::Int, th_byte2sample::Dict{String,Int32})
+
+    return read_traces!(s, fh, datatype, headers, data, trace, collect(keys(th_byte2sample)), th_byte2sample)
 end
 
 """
@@ -41,20 +33,23 @@ Use:    read_trace!(s::IO,
 Reads 'trace' from the current position of stream 's' into 'headers' and
 'data'. Only the header values in 'keys' and read.
 """
-function read_trace!(s::IO, fh::BinaryFileHeader,
+function read_traces!(s::IO, fh::BinaryFileHeader,
                     datatype::Type,
-                    headers::AbstractArray{BinaryTraceHeader,1},
-                    data::AbstractArray{<:Union{IBMFloat32, Float32}, 2},
+                    headers::AbstractVector{BinaryTraceHeader},
+                    data::AbstractMatrix{<:Union{IBMFloat32, Float32}},
                     trace::Int,
                     keys::Array{String,1},
                     th_byte2sample::Dict{String,Int32})
-    
-    # Read trace header
-    setindex!(headers, read_traceheader(s, keys, th_byte2sample), trace)
-    
-    # Read trace
-    setindex!(data, read_tracedata(s, fh, datatype), :, trace)
-    
+
+    ntrace = size(data, 2)
+    ntrace == 0 && return
+    for trace_loc=0:ntrace-1
+        # Read trace header
+        setindex!(headers, read_traceheader(s, keys, th_byte2sample), trace+trace_loc)
+
+        # Read trace
+        setindex!(data, read_tracedata(s, fh, datatype), :, trace_loc+1)
+    end
     nothing
 end