n-way file merge in Ruby - ruby

Question

I have several files from Java application (Gigaspaces logs) from multiple hosts which I need to merge based on date/time value.
Since every log file is already sorted, I need to get a first record from every file into an array, decide which one have a key with minimum value, merge it to result file, get a new line from the same file & repeat.
Record's definition - first line have a key and all following lines have no key, example:
2015-04-05 02:33:42,135 GSC SEVERE [com.gigaspaces.lrmi] - LRMI Transport Protocol caught server exception caused by [/10.0.1.2:46949] client.; Caused by: java.lang.IllegalArgumentException
at java.nio.ByteBuffer.allocate(ByteBuffer.java:311)
at com.gigaspaces.lrmi.SmartByteBufferCache.get(SmartByteBufferCache.java:50)
at com.gigaspaces.lrmi.nio.Reader.readBytesFromChannelNoneBlocking(Reader.java:410)
at com.gigaspaces.lrmi.nio.Reader.readBytesNonBlocking(Reader.java:644)
at com.gigaspaces.lrmi.nio.Reader.bytesToStream(Reader.java:509)
at com.gigaspaces.lrmi.nio.Reader.readRequest(Reader.java:112)
at com.gigaspaces.lrmi.nio.ChannelEntry.readRequest(ChannelEntry.java:121)
at com.gigaspaces.lrmi.nio.Pivot.handleReadRequest(Pivot.java:445)
at com.gigaspaces.lrmi.nio.selector.handler.ReadSelectorThread.handleRead(ReadSelectorThread.java:81)
at com.gigaspaces.lrmi.nio.selector.handler.ReadSelectorThread.handleConnection(ReadSelectorThread.java:45)
at com.gigaspaces.lrmi.nio.selector.handler.AbstractSelectorThread.doSelect(AbstractSelectorThread.java:74)
at com.gigaspaces.lrmi.nio.selector.handler.AbstractSelectorThread.run(AbstractSelectorThread.java:50)
at java.lang.Thread.run(Thread.java:662)
Ideally, result file should contain key, directory/filename.log & rest of the record.
Questions:
How to get a record from file in Ruby?
How to open multiple files and iterate through them using algorithm described above?

Answer 1

Code
Read all lines from all files that begin with a date-time string into an array, then sort the array by the date-time strings:
require 'date'
def get_key_rows(*fnames)
fnames.flat_map do |fname|
IO.foreach(fname).with_object([]) do |s, arr|
dt = DateTime.strptime(s[0, 19], '%Y-%m-%d %H:%M:%S') rescue nil
arr << [s[0, 19], fname, s[19..-1].rstrip] if dt
end
end.sort_by(&:first)
end
This method returns an array of three-element arrays. Each three-element array corresponds to a key line in one of the files, comprised of the date/time string, the filename and the remainder of the part of the line that follows the date/time string. Note that it is not necessary for key lines to be ordered within each file. The method uses:
DateTime#strptime to identify key rows;
Enumerable#flat_map, rather than Enumerable#map followed by Array#flatten; and
Enumerable#sort_by to sort the key rows by date/time.
Regarding sort_by, note that the strings can be sorted by the date/time strings, rather than by corresponding DateTime objects, because the form of the date/time string is 'yyyy-mm-dd hh-mm-ss'.
Examples
Let's create some files:
IO.write("f0", "2015-04-05 02:33:42,135 more stuff in f0\n" +
"more in f0\n" +
"2015-04-05 04:33:42,135 more stuff in f0\n" +
"even more in f0")
#=> 108
IO.write("f1", "2015-04-04 02:33:42,135 more stuff in f1\n" +
"2015-04-06 02:33:42,135 more stuff in f1\n" +
"more in f1")
#=> 92
IO.write("f2", "something in f2\n" +
"2015-04-05 02:33:43,135 more stuff in f2\n" +
"even more in f2\n" +
"2015-04-04 02:23:42,135 more stuff in f2")
#=> 113
get_key_rows('f0', 'f1', 'f2')
#=> [["2015-04-04 02:23:42", "f2", ",135 more stuff in f2"],
# ["2015-04-04 02:33:42", "f1", ",135 more stuff in f1"],
# ["2015-04-05 02:33:42", "f0", ",135 more stuff in f0"],
# ["2015-04-05 02:33:43", "f2", ",135 more stuff in f2"],
# ["2015-04-05 04:33:42", "f0", ",135 more stuff in f0"],
# ["2015-04-06 02:33:42", "f1", ",135 more stuff in f1"]]