140 lines
3.6 KiB
Ruby
140 lines
3.6 KiB
Ruby
require 'csv'
|
|
require 'redis/connection/hiredis'
|
|
require 'redis'
|
|
|
|
require 'logger'
|
|
require 'open-uri'
|
|
|
|
|
|
class Earthquakes
|
|
attr_accessor :id_set, :region_set, :magnitude_set
|
|
attr_accessor :magnitude_scores_set, :data_url
|
|
|
|
def initialize
|
|
@redis = Redis.new
|
|
@id_set = 'earthquake-ids'
|
|
@region_set = 'earthquake-regions'
|
|
@magnitude_set = 'earthquake-magnitudes'
|
|
@magnitude_scores_set = 'earthquakes-magnitudes-scores'
|
|
@data_url =
|
|
'http://earthquake.usgs.gov/earthquakes/catalogs/eqs7day-M1.txt'
|
|
|
|
@logger = Logger.new(STDOUT)
|
|
@logger.formatter = proc do |severity, datetime, progname, msg|
|
|
"#{datetime}: #{msg}\n"
|
|
end
|
|
end
|
|
|
|
def get_info(earthquake_id)
|
|
@redis.hgetall(with_prefix(earthquake_id, 'eqid'))
|
|
end
|
|
|
|
def in_region(region)
|
|
@redis.smembers(with_prefix(region, 'eqregion'))
|
|
end
|
|
|
|
def with_magnitude_in_range(min_magnitude, max_magnitude)
|
|
min_mag, max_mag =
|
|
magnitude_as_int(min_magnitude), magnitude_as_int(max_magnitude)
|
|
@logger.debug(
|
|
"Fetching magnitudes with scores in range #{min_mag}-#{max_mag}")
|
|
@redis.zrange(@magnitude_scores_set, min_mag, max_mag)
|
|
end
|
|
|
|
def all_ids
|
|
@redis.smembers(@id_set).map { |i| sans_prefix(i, 'eqid') }
|
|
end
|
|
|
|
def all_regions
|
|
@redis.smembers(@region_set).map { |r| sans_prefix(r, 'eqregion') }
|
|
end
|
|
|
|
def all_magnitudes
|
|
@redis.smembers(@magnitude_set).map { |m| sans_prefix(m, 'eqmagnitude') }
|
|
end
|
|
|
|
def drop_all
|
|
@logger.info('Removing data for each earthquake')
|
|
all_ids.each do |i|
|
|
@logger.info("Removing data for earthquake #{sans_prefix(i, 'eqid')}")
|
|
@redis.del(i)
|
|
end
|
|
|
|
@logger.info('Removing earthquake id set')
|
|
@redis.del(@id_set)
|
|
|
|
@logger.info('Removing earthquake region set')
|
|
@redis.del(@region_set)
|
|
|
|
@logger.info('Removing earthquake magnitude set')
|
|
@redis.del(@magnitude_set)
|
|
|
|
@logger.info('Removing earthquake magnitude scores set')
|
|
@redis.del(@magnitude_scores_set)
|
|
end
|
|
|
|
def load_data
|
|
data = CSV.parse(
|
|
open(@data_url).read, :headers => true, :header_converters => [:symbol]
|
|
)
|
|
|
|
data.each do |row|
|
|
@logger.info("Adding earthquake #{row[:eqid]}")
|
|
load_one_row(row)
|
|
end
|
|
end
|
|
|
|
private
|
|
def load_one_row(row)
|
|
@redis.multi do |redis_multi|
|
|
id_key = with_prefix(row[:eqid].downcase, 'eqid')
|
|
load_id_and_attrs_for_row(row, id_key, redis_multi)
|
|
load_region_for_row(row, id_key, redis_multi)
|
|
load_magnitude_for_row(row, id_key, redis_multi)
|
|
end
|
|
end
|
|
|
|
def load_id_and_attrs_for_row(row, id_key, redis_multi)
|
|
redis_multi.sadd(@id_set, id_key)
|
|
row.headers.each do |header|
|
|
redis_multi.hset(id_key, header.downcase, row[header].downcase)
|
|
end
|
|
end
|
|
|
|
def load_region_for_row(row, id_key, redis_multi)
|
|
region_key = with_prefix(row[:region].downcase, 'eqregion')
|
|
redis_multi.sadd(region_key, id_key)
|
|
redis_multi.sadd(@region_set, region_key)
|
|
end
|
|
|
|
def load_magnitude_for_row(row, id_key, redis_multi)
|
|
magnitude_key = with_prefix(row[:magnitude], 'eqmagnitude')
|
|
redis_multi.sadd(magnitude_key, id_key)
|
|
redis_multi.sadd(@magnitude_set, magnitude_key)
|
|
score = magnitude_as_int(row[:magnitude])
|
|
@logger.info("Setting score of #{score} for row #{row.inspect}")
|
|
redis_multi.zadd(@magnitude_scores_set, score, id_key)
|
|
end
|
|
|
|
def with_prefix(key, prefix)
|
|
if not /^#{prefix}:/.match(key)
|
|
return "#{prefix}:#{key}"
|
|
end
|
|
|
|
key
|
|
end
|
|
|
|
def sans_prefix(key, prefix)
|
|
key.gsub(/^#{prefix}:/, '')
|
|
end
|
|
|
|
def magnitude_as_int(magnitude)
|
|
magnitude.to_s.gsub(/\./, '').to_i
|
|
end
|
|
end
|
|
|
|
|
|
if $0 == __FILE__
|
|
Earthquakes.new.send(ARGV.first)
|
|
end
|