242e23a0b496b8ccf33134facf3c617c7c56af19
[epclust.git] / epclust / R / de_serialize.R
1 #data: matrix of double or connection
2 serialize = function(data_ascii, data_bin_file, nb_per_chunk,
3 sep=",", nbytes=4, endian=.Platform$endian)
4 {
5 if (is.character(data_ascii))
6 data_ascii = file(data_ascii, open="r")
7 else if (is(data_ascii,"connection") && !isOpen(data_ascii))
8 open(data_ascii)
9
10 first_write = (!file.exists(data_bin_file) || file.info(data_bin_file)$size == 0)
11 data_bin = file(data_bin_file, open=ifelse(first_write,"wb","ab"))
12
13 #write data length on first call
14 if (first_write)
15 {
16 #number of items always on 8 bytes
17 writeBin(0L, data_bin, size=8) #,endian="little")
18 if (is.matrix(data_ascii))
19 data_length = ncol(data_ascii)
20 else #if (is(data, "connection"))
21 {
22 data_line = scan(data_ascii, double(), sep=sep, nlines=1, quiet=TRUE)
23 writeBin(data_line, data_bin, size=nbytes)
24 data_length = length(data_line)
25 }
26 }
27
28 if (is.matrix(data_ascii))
29 index = 1
30 repeat
31 {
32 if (is.matrix(data_ascii))
33 {
34 range = index:min(nrow(data_ascii),index+nb_per_chunk)
35 data_chunk =
36 if (range[1] <= nrow(data_ascii))
37 as.double(t(data_ascii[range,]))
38 else
39 integer(0)
40 index = index + nb_per_chunk
41 }
42 else
43 data_chunk = scan(data_ascii, double(), sep=sep, nlines=nb_per_chunk, quiet=TRUE)
44 if (length(data_chunk)==0)
45 break
46 writeBin(data_chunk, data_bin, size=nbytes)
47 }
48
49 if (first_write)
50 {
51 #ecrire file_size-1 / (nbytes*nbWritten) en 0 dans bin_data ! ignored == file_size
52 ignored = seek(data_bin, 0)
53 writeBin(data_length, data_bin, size=8)
54 }
55 close(data_bin)
56
57 if (is(data_ascii,"connection"))
58 close(data_ascii)
59 }
60
61 #read in binary file, always same structure
62 getDataInFile = function(indices, data_bin_file, nbytes=4, endian=.Platform$endian)
63 {
64 data_bin = file(data_bin_file, "rb")
65 data_size = file.info(data_bin)$size
66 data_length = readBin(data_bin, "integer", 1, 8, endian)
67 #Ou t(sapply(...)) (+ rapide ?)
68 data_ascii = do.call( rbind, lapply( indices, function(i) {
69 offset = 8+(i-1)*data_length*nbytes
70 if (offset > data_size)
71 return (vector("double",0))
72 ignored = seek(data_bin, offset)
73 readBin(data_bin, "double", n=data_length, size=nbytes)
74 } ) )
75 close(data_bin)
76 if (ncol(data_ascii)>0) data_ascii else NULL
77 }