de_serialize works. Variables names improved. Code beautified. TODO: clustering tests
[epclust.git] / epclust / R / de_serialize.R
diff --git a/epclust/R/de_serialize.R b/epclust/R/de_serialize.R
new file mode 100644 (file)
index 0000000..682db53
--- /dev/null
@@ -0,0 +1,73 @@
+#data: matrix of double or connection
+serialize = function(data_ascii, data_bin_file, nb_per_chunk,
+       sep=",", nbytes=4, endian=.Platform$endian)
+{
+       if (is.character(data_ascii))
+               data_ascii = file(data_ascii, open="r")
+       else if (is(data_ascii,"connection") && !isOpen(data_ascii))
+               open(data_ascii)
+
+       first_write = (!file.exists(data_bin_file) || file.info(data_bin_file)$size == 0)
+       data_bin = file(data_bin_file, open=ifelse(first_write,"wb","ab"))
+
+       #write data length on first call
+       if (first_write)
+       {
+               #number of items always on 8 bytes
+               writeBin(0L, data_bin, size=8) #,endian="little")
+               if (is.matrix(data_ascii))
+                       data_length = ncol(data_ascii)
+               else #if (is(data, "connection"))
+               {
+                       data_line = scan(data_ascii, double(), sep=sep, nlines=1, quiet=TRUE)
+                       writeBin(data_line, data_bin, size=nbytes)
+                       data_length = length(data_line)
+               }
+       }
+
+       if (is.matrix(data_ascii))
+               index = 1
+       repeat
+       {
+               if (is.matrix(data_ascii))
+               {
+                       range = index:min(nrow(data_ascii),index+nb_per_chunk)
+                       data_chunk =
+                               if (range[1] <= nrow(data_ascii))
+                                       as.double(t(data_ascii[range,]))
+                               else
+                                       integer(0)
+                       index = index + nb_per_chunk
+               }
+               else
+                       data_chunk = scan(data_ascii, double(), sep=sep, nlines=nb_per_chunk, quiet=TRUE)
+               if (length(data_chunk)==0)
+                       break
+               writeBin(data_chunk, data_bin, size=nbytes)
+       }
+
+       if (first_write)
+       {
+               #ecrire file_size-1 / (nbytes*nbWritten) en 0 dans bin_data ! ignored == file_size
+               ignored = seek(data_bin, 0)
+               writeBin(data_length, data_bin, size=8)
+       }
+       close(data_bin)
+
+       if (is(data_ascii,"connection"))
+               close(data_ascii)
+}
+
+#read in binary file, always same structure
+getDataInFile = function(indices, data_bin_file, nbytes=4, endian=.Platform$endian)
+{
+       data_bin = file(data_bin_file, "rb")
+       data_length = readBin(data_bin, "integer", 1, 8, endian)
+       #Ou t(sapply(...)) (+ rapide ?)
+       data_ascii = do.call( rbind, lapply( indices, function(i) {
+               ignored = seek(data_bin, 8+((i-1)*data_length*nbytes))
+               readBin(data_bin, "double", n=data_length, size=nbytes)
+       } ) )
+       close(data_bin)
+       data_ascii
+}