-
Notifications
You must be signed in to change notification settings - Fork 71
/
MAT_HDF5.jl
628 lines (569 loc) · 19.5 KB
/
MAT_HDF5.jl
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
# MAT_HDF5.jl
# Tools for reading MATLAB HDF5 (v7.3) files in Julia
#
# Copyright (C) 2012 Timothy E. Holy
#
# Permission is hereby granted, free of charge, to any person obtaining
# a copy of this software and associated documentation files (the
# "Software"), to deal in the Software without restriction, including
# without limitation the rights to use, copy, modify, merge, publish,
# distribute, sublicense, and/or sell copies of the Software, and to
# permit persons to whom the Software is furnished to do so, subject to
# the following conditions:
#
# The above copyright notice and this permission notice shall be
# included in all copies or substantial portions of the Software.
#
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
# EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
# MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
# NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
# LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
# OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
# WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
###########################################
## Reading and writing MATLAB .mat files ##
###########################################
module MAT_HDF5
using HDF5, SparseArrays
# deprecated for HDF5 v0.14+, but use deprecated binding to have common function with
# e.g. JLD.jl
import HDF5: exists
import Base: names, read, write, close
import HDF5: Reference
const HDF5Parent = Union{HDF5.File, HDF5.Group}
const HDF5BitsOrBool = Union{HDF5.BitsType,Bool}
mutable struct MatlabHDF5File <: HDF5.H5DataStore
plain::HDF5.File
toclose::Bool
writeheader::Bool
refcounter::Int
compress::Bool
function MatlabHDF5File(plain, toclose::Bool=true, writeheader::Bool=false, refcounter::Int=0, compress::Bool=false)
f = new(plain, toclose, writeheader, refcounter, compress)
if toclose
finalizer(close, f)
end
f
end
end
"""
close(matfile_handle)
Close a Matlab file.
"""
function close(f::MatlabHDF5File)
if f.toclose
close(f.plain)
if f.writeheader
magic = zeros(UInt8, 512)
identifier = "MATLAB 7.3 MAT-file" # minimal but sufficient
GC.@preserve magic identifier begin
magicptr = pointer(magic)
idptr = pointer(identifier)
unsafe_copyto!(magicptr, idptr, length(identifier))
end
magic[126] = 0x02
magic[127] = 0x49
magic[128] = 0x4d
rawfid = open(f.plain.filename, "r+")
write(rawfid, magic)
close(rawfid)
end
f.toclose = false
end
nothing
end
function matopen(filename::AbstractString, rd::Bool, wr::Bool, cr::Bool, tr::Bool, ff::Bool, compress::Bool)
local f
if ff && !wr
error("Cannot append to a write-only file")
end
if !cr && !isfile(filename)
error("File ", filename, " cannot be found")
end
pa = create_property(HDF5.H5P_FILE_ACCESS; fclose_degree = HDF5.H5F_CLOSE_STRONG)
if cr && (tr || !isfile(filename))
# We're truncating, so we don't have to check the format of an existing file
# Set the user block to 512 bytes, to save room for the header
p = create_property(HDF5.H5P_FILE_CREATE; userblock = 512)
f = HDF5.h5f_create(filename, HDF5.H5F_ACC_TRUNC, p.id, pa.id)
writeheader = true
else
f = HDF5.h5f_open(filename, wr ? HDF5.H5F_ACC_RDWR : HDF5.H5F_ACC_RDONLY, pa.id)
writeheader = false
end
close(pa)
fid = MatlabHDF5File(HDF5.File(f, filename), true, writeheader, 0, compress)
pathrefs = "/#refs#"
if haskey(fid.plain, pathrefs)
g = fid.plain[pathrefs]
fid.refcounter = length(g)-1
close(g)
end
fid
end
### Matlab file format specification ###
const name_type_attr_matlab = "MATLAB_class"
const empty_attr_matlab = "MATLAB_empty"
const sparse_attr_matlab = "MATLAB_sparse"
const int_decode_attr_matlab = "MATLAB_int_decode"
### Reading
function read_complex(dtype::HDF5.Datatype, dset::HDF5.Dataset, ::Type{T}) where T
if !check_datatype_complex(dtype)
close(dtype)
error("Unrecognized compound data type when reading ", HDF5.name(dset))
end
return read(dset, Complex{T})
end
function m_read(dset::HDF5.Dataset)
if haskey(dset, empty_attr_matlab)
# Empty arrays encode the dimensions as the dataset
dims = convert(Vector{Int}, read(dset))
mattype = read_attribute(dset, name_type_attr_matlab)
if mattype == "char"
return ""
elseif mattype == "struct"
# Not sure if this check is necessary but it is checked in
# `m_read(g::HDF5.Group)`
if haskey(dset, "MATLAB_fields")
return Dict{String,Any}(join(n)=>[] for n in read_attribute(dset, "MATLAB_fields"))
else
return Dict{String,Any}()
end
else
T = mattype == "canonical empty" ? Union{} : str2type_matlab[mattype]
return Array{T}(undef, dims...)
end
end
mattype = haskey(dset, name_type_attr_matlab) ? read_attribute(dset, name_type_attr_matlab) : "cell"
if mattype == "cell"
# Cell arrays, represented as an array of refs
refs = read(dset, Reference)
out = Array{Any}(undef, size(refs))
f = HDF5.file(dset)
for i = 1:length(refs)
dset = f[refs[i]]
try
out[i] = m_read(dset)
finally
close(dset)
end
end
return out
elseif !haskey(str2type_matlab,mattype)
@warn "MATLAB $mattype values are currently not supported"
return missing
end
# Regular arrays of values
# Convert to Julia type
T = str2type_matlab[mattype]
# Check for a COMPOUND data set, and if so handle complex numbers specially
dtype = datatype(dset)
try
class_id = HDF5.h5t_get_class(dtype.id)
d = class_id == HDF5.H5T_COMPOUND ? read_complex(dtype, dset, T) : read(dset, T)
length(d) == 1 ? d[1] : d
finally
close(dtype)
end
end
function add!(A, x)
for i = 1:length(A)
@inbounds A[i] += x
end
A
end
# reading a struct, struct array, or sparse matrix
function m_read(g::HDF5.Group)
mattype = read_attribute(g, name_type_attr_matlab)
if mattype != "struct"
# Check if this is a sparse matrix.
fn = keys(g)
if haskey(attributes(g), sparse_attr_matlab)
# This is a sparse matrix.
# ir is the row indices, jc is the column boundaries.
# We add one to account for the zero-based (MATLAB) to one-based (Julia) transition
jc = add!(convert(Vector{Int}, read(g, "jc")), 1)
if "data" in fn && "ir" in fn && "jc" in fn
# This matrix is not empty.
ir = add!(convert(Vector{Int}, read(g, "ir")), 1)
dset = g["data"]
T = str2type_matlab[mattype]
try
dtype = datatype(dset)
class_id = HDF5.h5t_get_class(dtype.id)
try
data = class_id == HDF5.H5T_COMPOUND ? read_complex(dtype, dset, T) : read(dset, T)
finally
close(dtype)
end
finally
close(dset)
end
else
# This matrix is empty.
ir = Int[]
data = str2type_matlab[mattype][]
end
return SparseMatrixCSC(convert(Int, read_attribute(g, sparse_attr_matlab)), length(jc)-1, jc, ir, data)
elseif mattype == "function_handle"
@warn "MATLAB $mattype values are currently not supported"
return missing
else
error("Cannot read from a non-struct group, type was $mattype")
end
end
if haskey(g, "MATLAB_fields")
fn = [join(f) for f in read_attribute(g, "MATLAB_fields")]
else
fn = keys(g)
end
s = Dict{String, Any}()
for i = 1:length(fn)
dset = g[fn[i]]
try
s[fn[i]] = m_read(dset)
finally
close(dset)
end
end
s
end
"""
read(matfile_handle, varname) -> value
Read a variable from an opened Matlab file and return its value.
See `matopen` and `matread`.
"""
function read(f::MatlabHDF5File, name::String)
local val
obj = f.plain[name]
try
val = m_read(obj)
finally
close(obj)
end
val
end
"""
names(matfile_handle) -> Vector{String}
Return a list of variables in an opened Matlab file.
See `matopen`.
"""
names(f::MatlabHDF5File) = keys(f)
"""
exists(matfile_handle, varname) -> Bool
Return true if a variable is present in an opened Matlab file.
See `matopen`.
"""
exists(p::MatlabHDF5File, path::String) = haskey(p, path)
# HDF5v0.14+ H5DataStore uses keys/haskey
Base.keys(f::MatlabHDF5File) = filter!(x -> x!="#refs#" && x!="#subsystem#", keys(f.plain))
Base.haskey(p::MatlabHDF5File, path::String) = haskey(p.plain, path)
### Writing
# Check whether a varname is valid for MATLAB
check_valid_varname(s::AbstractString) = if match(r"^[a-zA-Z][a-zA-Z0-9_]*$", s) == nothing
error("Invalid variable name or key \"$s\": variable names must start with a letter and contain only alphanumeric characters and underscore")
elseif length(s) > 63
error("Invalid variable name or key \"$s\": variable names must be less than 64 characters")
end
toarray(x::Array) = x
toarray(x::Array{Bool}) = reinterpret(UInt8, x)
toarray(x::Bool) = UInt8[x]
toarray(x) = [x]
# Write the MATLAB type string for dset
m_writetypeattr(dset, ::Type{Complex{T}}) where T = m_writetypeattr(dset, T)
function m_writetypeattr(dset, T)
if !haskey(type2str_matlab, T)
error("Type ", T, " is not (yet) supported")
end
typename = type2str_matlab[T]
# Write the attribute
write_attribute(dset, name_type_attr_matlab, typename)
if T == Bool
write_attribute(dset, int_decode_attr_matlab, Int32(1))
end
end
# Writes an empty scalar or array
function m_writeempty(parent::HDF5Parent, name::String, data::AbstractArray)
adata = [size(data)...]
dset, dtype = create_dataset(parent, name, adata)
try
write_attribute(dset, empty_attr_matlab, 0x01)
m_writetypeattr(dset, eltype(data))
write_dataset(dset, dtype, adata)
finally
close(dset)
close(dtype)
end
end
# Write an array to a dataset in a MATLAB file, returning the dataset
function m_writearray(parent::HDF5Parent, name::String, adata::AbstractArray{T}, compress::Bool) where {T<:HDF5BitsOrBool}
if compress
dset, dtype = create_dataset(parent, name, adata;
compress = 3, chunk = HDF5.heuristic_chunk(adata))
else
dset, dtype = create_dataset(parent, name, adata)
end
try
write_dataset(dset, dtype, adata)
dset
catch e
close(dset)
rethrow(e)
finally
close(dtype)
end
end
function m_writearray(parent::HDF5Parent, name::String, adata::AbstractArray{Complex{T}}, compress::Bool) where {T<:HDF5BitsOrBool}
dtype = build_datatype_complex(T)
try
stype = dataspace(adata)
if compress
obj_id = create_dataset(parent, name, dtype, stype;
compress = 3, chunk = HDF5.heuristic_chunk(adata))
else
obj_id = create_dataset(parent, name, dtype, stype)
end
dset = HDF5.Dataset(obj_id, HDF5.file(parent))
try
arr = reshape(reinterpret(T, adata), tuple(2, size(adata)...))
write_dataset(dset, dtype, arr)
catch e
close(dset)
rethrow(e)
finally
close(stype)
end
dset
finally
close(dtype)
end
end
# Write a scalar or array
function m_write(mfile::MatlabHDF5File, parent::HDF5Parent, name::String, data::Union{T, Complex{T}, Array{T}, Array{Complex{T}}}) where {T<:HDF5BitsOrBool}
if isempty(data)
m_writeempty(parent, name, data)
return
end
dset = m_writearray(parent, name, toarray(data), mfile.compress)
try
m_writetypeattr(dset, T)
finally
close(dset)
end
end
# Write sparse arrays
function m_write(mfile::MatlabHDF5File, parent::HDF5Parent, name::String, data::SparseMatrixCSC{T}) where T
g = create_group(parent, name)
try
m_writetypeattr(g, T)
write_attribute(g, sparse_attr_matlab, UInt64(size(data, 1)))
if !isempty(data.nzval)
close(m_writearray(g, "data", toarray(data.nzval), mfile.compress))
close(m_writearray(g, "ir", add!(isa(data.rowval, Vector{UInt64}) ? copy(data.rowval) : convert(Vector{UInt64}, data.rowval), typemax(UInt64)), mfile.compress))
end
close(m_writearray(g, "jc", add!(isa(data.colptr, Vector{UInt64}) ? copy(data.colptr) : convert(Vector{UInt64}, data.colptr), typemax(UInt64)), mfile.compress))
finally
close(g)
end
end
# Write BitArray as Array{Bool}. Would be better not to require the conversion, but this is easy
m_write(mfile::MatlabHDF5File, parent::HDF5Parent, name::String, s::BitArray) =
m_write(mfile, parent, name, convert(Array{Bool}, s))
# Write a string
function m_write(mfile::MatlabHDF5File, parent::HDF5Parent, name::String, str::AbstractString)
if isempty(str)
data = UInt64[0, 0]
# Create the dataset
dset, dtype = create_dataset(parent, name, data)
try
write_attribute(dset, name_type_attr_matlab, "char")
write_attribute(dset, empty_attr_matlab, 0x01)
write_dataset(dset, dtype, data)
finally
close(dset)
close(dtype)
end
else
# Here we assume no UTF-16
data = zeros(UInt16, 1, length(str))
i = 1
for c in str
data[i] = c
i += 1
end
# Create the dataset
dset, dtype = create_dataset(parent, name, data)
try
write_attribute(dset, name_type_attr_matlab, "char")
write_attribute(dset, int_decode_attr_matlab, Int32(2))
write_dataset(dset, dtype, data)
finally
close(dset)
close(dtype)
end
end
end
# Write cell arrays
function m_write(mfile::MatlabHDF5File, parent::HDF5Parent, name::String, data::Array{T}) where T
pathrefs = "/#refs#"
fid = HDF5.file(parent)
local g
local refs
if !haskey(fid, pathrefs)
g = create_group(fid, pathrefs)
else
g = fid[pathrefs]
end
try
# If needed, create the "empty" item
if !haskey(g, "a")
edata = zeros(UInt64, 2)
eset, etype = create_dataset(g, "a", edata)
try
write_dataset(eset, etype, edata)
write_attribute(eset, name_type_attr_matlab, "canonical empty")
write_attribute(eset, "MATLAB_empty", 0x00)
finally
close(etype)
close(eset)
end
else
a = g["a"]
if !haskey(attributes(a), "MATLAB_empty")
error("Must create the empty item, with name a, first")
end
close(a)
end
# Write the items to the reference group
refs = Array{Reference}(undef, size(data))
for i = 1:length(data)
mfile.refcounter += 1
itemname = string(mfile.refcounter)
m_write(mfile, g, itemname, data[i])
# Extract references
tmp = g[itemname]
refs[i] = Reference(tmp, pathrefs*"/"*itemname)
close(tmp)
end
finally
close(g)
end
# Write the references as the chosen variable
cset, ctype = create_dataset(parent, name, refs)
try
write_dataset(cset, ctype, refs)
write_attribute(cset, name_type_attr_matlab, "cell")
finally
close(ctype)
close(cset)
end
end
# Check that keys are valid for a struct, and convert them to an array of ASCIIStrings
function check_struct_keys(k::Vector)
asckeys = Vector{String}(undef, length(k))
for i = 1:length(k)
key = k[i]
if !isa(key, AbstractString)
error("Only Dicts with string keys may be saved as MATLAB structs")
end
check_valid_varname(key)
asckeys[i] = convert(String, key)
end
asckeys
end
# Write a struct from arrays of keys and values
function m_write(mfile::MatlabHDF5File, parent::HDF5Parent, name::String, k::Vector{String}, v::Vector)
g = create_group(parent, name)
write_attribute(g, name_type_attr_matlab, "struct")
for i = 1:length(k)
m_write(mfile, g, k[i], v[i])
end
write_attribute(g, "MATLAB_fields", HDF5.VLen(k))
end
# Write Associative as a struct
m_write(mfile::MatlabHDF5File, parent::HDF5Parent, name::String, s::AbstractDict) =
m_write(mfile, parent, name, check_struct_keys(collect(keys(s))), collect(values(s)))
# Write generic CompositeKind as a struct
function m_write(mfile::MatlabHDF5File, parent::HDF5Parent, name::String, s)
if isbits(s)
error("This is the write function for CompositeKind, but the input doesn't fit")
end
T = typeof(s)
m_write(mfile, parent, name, check_struct_keys([string(x) for x in fieldnames(T)]), [getfield(s, x) for x in fieldnames(T)])
end
# Check whether a variable name is valid, then write it
"""
write(matfile_handle, varname, value)
Write the value into an opened Matlab file as the specified variable.
See `matopen` and `matwrite`.
"""
function write(parent::MatlabHDF5File, name::String, thing)
check_valid_varname(name)
m_write(parent, parent.plain, name, thing)
end
## Type conversion operations ##
struct MatlabString end
const str2type_matlab = Dict(
"canonical empty" => nothing,
"int8" => Int8,
"uint8" => UInt8,
"int16" => Int16,
"uint16" => UInt16,
"int32" => Int32,
"uint32" => UInt32,
"int64" => Int64,
"uint64" => UInt64,
"single" => Float32,
"double" => Float64,
"cell" => Any,
"char" => MatlabString,
"logical" => Bool
)
const type2str_matlab = Dict(
Int8 => "int8",
UInt8 => "uint8",
Int16 => "int16",
UInt16 => "uint16",
Int32 => "int32",
UInt32 => "uint32",
Int64 => "int64",
UInt64 => "uint64",
Float32 => "single",
Float64 => "double",
Bool => "logical"
)
function read(obj::Union{HDF5.Dataset,HDF5.Attribute}, ::Type{MatlabString})
T = HDF5.get_jl_type(obj)
data = read(obj, T)
if size(data, 1) == 1
sz = size(data)
data = reshape(data, sz[2:end])
end
if ndims(data) == 1
return String(convert(Vector{Char}, data))
elseif ndims(data) == 2
return datap = String[rstrip(String(convert(Vector{Char}, vec(data[i, :])))) for i = 1:size(data, 1)]
else
return data
end
end
## Utilities for handling complex numbers
function build_datatype_complex(T::Type)
memtype_id = create_datatype(HDF5.H5T_COMPOUND, 2*sizeof(T))
HDF5.h5t_insert(memtype_id, "real", 0, HDF5.hdf5_type_id(T))
HDF5.h5t_insert(memtype_id, "imag", sizeof(T), HDF5.hdf5_type_id(T))
HDF5.Datatype(memtype_id)
end
function check_datatype_complex(dtype::HDF5.Datatype)
n = HDF5.h5t_get_nmembers(dtype.id)
if n != 2
return false
end
if HDF5.h5t_get_member_name(dtype.id, 0) != "real" ||
HDF5.h5t_get_member_name(dtype.id, 1) != "imag"
return false
end
true
end
end