Compare commits
2 Commits
96740f161a
...
e12c6af35e
Author | SHA1 | Date |
---|---|---|
Jeff Yates | e12c6af35e | |
Jeff Yates | 89bb9cf027 |
30
sorter.rb
30
sorter.rb
|
@ -5,7 +5,7 @@
|
|||
#Current Usage: #
|
||||
# Scans a file for groups of words and counts the totals #
|
||||
# Input file: ./tester.txt #
|
||||
# Wroud group file: ./bins.json #
|
||||
# Word group file: ./bins.json #
|
||||
# Output file: ./output.json #
|
||||
# #
|
||||
# The input file is human readable, easy to edit. #
|
||||
|
@ -13,7 +13,7 @@
|
|||
# #
|
||||
#To Do: #
|
||||
# * use STDIN to accept file to scan #
|
||||
# * Support scaning multiple files #
|
||||
# * Support scanning multiple files #
|
||||
# * Output to csv #
|
||||
# * Strip out header and footer text #
|
||||
############################################################################
|
||||
|
@ -25,7 +25,7 @@ require 'optparse'
|
|||
#bin_counter expects:
|
||||
# bin - a hash with strings to search for
|
||||
# test - the document text to search through
|
||||
#This method returns a hash containg the strings and their frequency
|
||||
#This method returns a hash containing the strings and their frequency
|
||||
def bin_counter (bin, text)
|
||||
ret = Hash.new
|
||||
bin.each do |word|
|
||||
|
@ -63,18 +63,38 @@ def strip_text (text, start, fin)
|
|||
text.lines(start,chomp: true)[1].lines(fin,chomp: true)[0]
|
||||
end
|
||||
|
||||
def split_text (text, start, fin)
|
||||
split1 = text.lines(start, chomp: true)[1..-1]
|
||||
split2 = []
|
||||
ret = ""
|
||||
split1.each do |section|
|
||||
split2 += section.lines(fin, chomp: true)
|
||||
end
|
||||
split2.each do |section|
|
||||
ret += section
|
||||
end
|
||||
return ret
|
||||
end
|
||||
|
||||
options = Hash.new
|
||||
OptionParser.new do |opts|
|
||||
opts.banner = 'sorter.rb --options'
|
||||
opts.on("-f", "--file file", "Name of the file to process") do |file|
|
||||
options[:file] = file
|
||||
end
|
||||
opts.on("-t", "--type type", "Type of file. Must be \"iat\" or \"pn\"") do |type|
|
||||
options[:type] = type
|
||||
end
|
||||
opts.on("-b", "--bin-file binfile", "Name of the bin file") do |binfile|
|
||||
options[:binfile] = binfile
|
||||
end
|
||||
end.parse!
|
||||
|
||||
|
||||
csv = CSV.read('./bins.csv')
|
||||
csv = CSV.read(options[:binfile])
|
||||
text = File.read(options[:file])
|
||||
text = strip_text(text,'PLOVEINTAKE','PLOVECLOSING')
|
||||
text = strip_text(text, 'PLOVEINTAKE', 'PLOVECLOSING') if options[:type] == 'iat'
|
||||
text = split_text(text, 'Narrative:', 'Signatures:') if options[:type] == 'pn'
|
||||
output = Hash.new #Creating the output storage object
|
||||
bins = Hash.new #This hash stores the bins
|
||||
outfile = options[:file] + '-out.json'
|
||||
|
|
Loading…
Reference in New Issue