ansaurus

Question

Answer 1

+1 A:

Use this psuedocode

File.open("filename") do |l|
    one,two,three,four,five,six = l.split(" ")
    three = three[0,1]
    four = four[0,1]
    five = five[0,1]
    six = six[0,1]
    // code to create output format
end

so for example this line:

64 => [25, 25, 25, 25]

is formed by
one => [three,four,five,six]

ennuikiller 2009-09-06 20:33:31

I have added some sample advisories to the main question. If it helps with your answer, take a look at those to see what other contents are in the files. Its (unfortunately) not only what I post, as what I posted to parse was just an excerpt. :/

Josh 2009-09-06 20:40:33

Answer 2

A:

If you want to keep it generic, you can use STDIN, e.g

forecastparser.rb < forecast.txt

and read each line via

input = $<.read.split
input.each do |line|
...
end

Omar Qureshi 2009-09-06 20:37:58

I think its less about how to import the file, and more about how to parse those blocks of text by themselves, going through the file and "catching" each one somehow (in order as they appear)

Josh 2009-09-06 20:39:38

Answer 3

A:

It should be reasonably easy to build a right-recursive grammar for this, e.g.:

forecast : FORECAST VALID <int>/<int>Z <int><direction> <int><direction> \n <maxwind> <forecast_list>

direction : N
          | S
          | E
          | W
          | NE
          | NW (etc etc)

maxwind : MAX WIND <int> KT...GUSTS <int> KT

forecast_list : forecast_line \n forecast_list
              | 

forecast_line : <int> KT... <int><direction> <int><direction> <int><direction> <int><direction>

With a grammar like that, you can write (by hand) a recursive descent parser, which should be pretty simple. The benefit of this is that your production rules are context-free, so you should be able to deal with minor format shifts or new types of data files fairly easily.

2009-09-06 22:05:32

Answer 4

A:

Taking a quick glance on those files you linked it seems the "blocks" of information are the same -- same type of information -- between the files, even if the format is widely different?

So if I were to do this I would get a list of possible values for each block and then test/parse each block out of that. If it's a hurricane warning I know there aren't any important numbers, but tropical depression probably has something I'm interested in. (On a side note a tropical depression sounds really funny to me as a swede who hasn't heard of any weather being called depressing officially ^^)


@block_no = 0
[..]
File.open('forecast') do |f|
  block = []
  line = file.readline.strip
  block << line unless line.strip == ''

  Forecast.parse(block) # which has the current block_no and knows what kind of possible values there are to read out
  @block_no += 1
end

This feels as a very generic answer, but if I were to try to do this I would need to know of all possible formats the information could show up in before I could come up with a good solution. Possible just using a whole bunch of String#scan calls will be the best. :)

Good luck

ba 2009-09-07 12:45:07

Answer 5

+1 A:

I am posting an answer to this because I feel that the answers really didn't satisfy the requirements posted in the original question. Basically, there are multiple blocks of text with the same starting line, as so:

FORECAST VALID 23/1800Z 25.4N  75.6W
...
FORECAST VALID 24/0600Z 25.5N  78.8W
...
FORECAST VALID 24/1800Z 25.8N  81.8W

What I ended up doing is creating a regular expression for this line:

/^(FORECAST|OUTLOOK)\sVALID\s(\d+)\/(\d+)Z\s([\d\.]+)N\s+([\d\.]+)W/

Now, I needed to loop through every block of text until there were no more left. Since these forecasts are typically at the end of the advisory, I did it like so:

forecast_data = []

# Grab the rest of the forecast data
until data.eof?
  forecast_data << data.readline.strip
end

forecast_times = [12,24,36,48,72,96,120]
forecasts ||= {}
current_forecast = {}

until forecast_data.empty?
  line = forecast_data.shift

  if line =~ regular_expression
    # Start a new "current_forecast" array, which
    # contains the current block of text's data,
    # and parse it...
    forecasts.merge!(hour => current_forecast)
  end

  # Additional parsing for this block here...
end

# Merge the final block in with the rest
forecasts.merge!(hour => current_forecast) unless current_forecast.empty?

This seems to work. If anyone else has any idea on how to refactor this, or do it better using another method, please feel free to add another answer or comment and I'll change the answer! Thanks to everyone who posted; its truly appreciated.

Josh 2009-09-10 07:51:28

ansaurus

tags:

views:

answers:

Match text in loops in Ruby

related questions