Split fields without regular expressions

Author: djc

Cargo.lock

@@ -13,7 +13,6 @@ edition = "2018"
 anyhow = "1"
 err-derive = "0.2.4"
 memmap = "0.7"
-regex = "1"
 structopt = "0.3"
 ahash = "0.3"
 rayon = "1.3"

Cargo.toml

@@ -1,7 +1,6 @@
 use std::str::FromStr;
 
 use anyhow::{anyhow, Error};
-use regex::Regex;
 use structopt::StructOpt;
 
 use topfew::{top_few_from_stream, KeyFinder};
@@ -14,8 +13,7 @@ fn main() -> Result<(), Error> {
             options.num
         ));
     }
-    let sep = Regex::new(&options.regex)?;
-    let kf = KeyFinder::new(Some(options.fields.indices), sep)?;
+    let kf = KeyFinder::new(Some(options.fields.indices));
     let top_list = top_few_from_stream(options.file.into(), &kf, options.num)?;
     for kc in top_list {
         println!("{} {}", kc.count, kc.key);
@@ -32,9 +30,6 @@ struct Options {
     /// Top number of matches to show
     #[structopt(long, short = "n", default_value = "10")]
     num: usize,
-    /// Regular expression used to split lines into fields
-    #[structopt(long, short = "e", default_value = "[ \\t]")]
-    regex: String,
     /// File to search
     file: String,
 }

src/bin/tf.rs

@@ -1,58 +1,62 @@
 use anyhow::{anyhow, Error};
-use regex::Regex;
 
 pub struct KeyFinder {
-    keys: Option<(usize, usize, Vec<bool>)>,
-    sep: Regex,
+    keys: Option<Vec<bool>>,
 }
 
 impl KeyFinder {
-    pub fn new(keys: Option<Vec<usize>>, sep: Regex) -> Result<Self, Error> {
+    pub fn new(keys: Option<Vec<usize>>) -> Self {
         let keys = keys.map(|mut keys| {
             keys.sort();
 
             let last = *keys.last().unwrap();
-            (
-                keys.len(),
-                last,
-                (0..=last)
-                    .map(|i| keys.contains(&(i + 1)))
-                    .collect::<Vec<_>>(),
-            )
+            (0..=last)
+                .map(|i| keys.contains(&(i + 1)))
+                .collect::<Vec<_>>()
         });
-        Ok(KeyFinder { keys, sep })
+        KeyFinder { keys }
     }
 
     pub fn key<'a>(&self, record: &'a str, s: &'a mut String) -> Result<&'a str, Error> {
-        let (num, last, keep) = match &self.keys {
+        let keep = match &self.keys {
             None => return Ok(record),
-            Some((num, _, _)) if *num == 0 => return Ok(record),
-            Some((num, last, keep)) => (num, last, keep),
+            Some(keep) if keep.len() == 0 => return Ok(record),
+            Some(keep) => keep,
         };
 
-        let mut fields = keep
-            .iter()
-            .zip(self.sep.splitn(record, last + 2))
-            .filter_map(|(keep, field)| if *keep { Some(field) } else { None });
-
-        if *num == 1 {
-            return match fields.next() {
-                Some(f) => Ok(f),
-                None => Err(anyhow!("not enough fields to make key")),
-            };
-        }
-
-        let mut found = 0;
-        for f in fields {
-            s.push(' ');
-            s.push_str(f);
-            found += 1;
+        let mut current = 0;
+        for c in record.chars() {
+            if c == ' ' || c == '\t' {
+                current += 1;
+            } else if current < keep.len() && keep[current] {
+                s.push(c);
+            } else if current >= keep.len() {
+                break;
+            }
         }
 
-        if found == *num {
+        if current >= keep.len() {
             Ok(s)
         } else {
             Err(anyhow!("not enough fields to make key"))
         }
     }
 }
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn key() {
+        let kf = KeyFinder::new(Some(vec![1]));
+        let mut s = String::new();
+        assert_eq!(kf.key(TEST, &mut s).unwrap(), "92.109.155.34");
+
+        s.clear();
+        let kf = KeyFinder::new(Some(vec![7]));
+        assert_eq!(kf.key(TEST, &mut s).unwrap(), "/");
+    }
+
+    const TEST: &str = "92.109.155.34 - - [09/Aug/2018:11:53:26 +0200] \"GET / HTTP/2.0\" 200 3219 \"https://www.facebook.com/\" \"Mozilla/5.0 (Macintosh; Intel Mac OS X 10_13_4) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/67.0.3396.99 Safari/537.36\"";
+}