ugly-rust-text-parser.rs

extern crate glob;
extern crate regex;
extern crate csv;

use std::vec::Vec;
use std::env;
use glob::glob;
use std::path::{Path, PathBuf};

use std::io::prelude::*;
use std::fs::File;
use regex::Regex;

use csv::Writer;

fn main() {

  let sreg = Regex::new(r"(?i)section").unwrap();
  let seid = Regex::new(r"\d{9}").unwrap();
  let baid = Regex::new(r"\d{13}").unwrap();

  let path = env::current_exe().unwrap();
  let base = path.parent().unwrap();
  env::set_current_dir(base).ok();

  let txts: Vec<PathBuf> = glob("*.txt").unwrap().map(|r| r.unwrap()).collect();

  let csv_out = Path::new("output.csv");
  let mut writer = Writer::from_file(&csv_out).unwrap();

  writer.encode( ("Items", "Section") );

  for txt in txts {
    let mut f = File::open(txt.as_path()).unwrap();
    let mut s = String::new();
    f.read_to_string(&mut s);

    let mut csid = "";
    let mut lines = s.lines_any().collect::<Vec<&str>>().into_iter().peekable();

    loop {
      if lines.peek().is_none() { break; }
      let line = lines.next().unwrap().trim();
      if sreg.is_match(line) {
        loop {
          if lines.peek().is_none() { break; }
          if baid.is_match( lines.peek().unwrap() ) { break; }
          let line = lines.next().unwrap().trim();
          if seid.is_match(line) { 
            csid = line;
            break;
          } else {
            continue;
          }
        }
      } else if baid.is_match(line) {
        let item = format!("=\"{}\"", line);
        let sect = format!("=\"{}\"", csid);
        writer.encode( (item, sect) );
      }    
    }

  }

}