Combine domnet and sorta under evenmoreutils

src/bin/sorta.rs

@@ -0,0 +1,81 @@
+use std::io::{self, BufRead, Write};
+
+use lazy_static::lazy_static;
+use regex::{Regex, escape};
+
+lazy_static! {
+    static ref PARENS: Vec<(char, char)> = vec![
+        ('(', ')'),
+        ('[', ']'),
+        ('{', '}'),
+        ('【', '】'),
+    ];
+    static ref ARTICLES: Vec<&'static str> = vec![
+        "a",
+        "an",
+        "the",
+    ];
+    static ref CHAFF: Vec<&'static str> = vec![
+        "-",
+        "_",
+        ".",
+        "『",
+        "』",
+    ];
+    static ref EXTENSIONS: Vec<&'static str> = vec![
+        "mkv",
+        "mp4",
+        "webm",
+    ];
+}
+
+fn alternator_regex(choices: &Vec<&str>) -> String {
+    let mut result = String::new();
+    for (i, choice) in choices.iter().enumerate() {
+        if i > 0 {
+            result.push('|');
+        }
+        result.push_str(&escape(choice));
+    }
+    result
+}
+
+fn main() -> io::Result<()> {
+    let article_regex_raw = format!(r"\b({})\b", alternator_regex(&ARTICLES));
+    let extension_regex_raw = format!(r"\.({})", alternator_regex(&EXTENSIONS));
+
+    let mut space_paren_chaff_regex_raw = String::from(r"\s");
+    for (open, close) in PARENS.iter() {
+        let open = escape(&open.to_string());
+        let close = escape(&close.to_string());
+        space_paren_chaff_regex_raw.push_str(&format!("|{open}[^{close}]+{close}"));
+    }
+    for chaff in CHAFF.iter() {
+        space_paren_chaff_regex_raw.push('|');
+        space_paren_chaff_regex_raw.push_str(&escape(chaff));
+    }
+
+    let strip_start_regex = Regex::new(&format!("^(?i)({article_regex_raw}|{space_paren_chaff_regex_raw})*")).unwrap();
+    let strip_end_regex = Regex::new(&format!("(?i)({space_paren_chaff_regex_raw})*({extension_regex_raw})?$")).unwrap();
+
+    let mut results = Vec::new();
+
+    let stdin = io::stdin();
+    for line in stdin.lock().lines() {
+        let line = line?;
+        let strip_start_idx = strip_start_regex.find(&line).map(|m| m.end()).unwrap_or(0);
+        let strip_end_idx = strip_end_regex.find_at(&line, strip_start_idx).map(|m| m.start()).unwrap_or(line.len());
+        let item = ((&line[strip_start_idx..]).to_lowercase(), line, strip_start_idx, strip_end_idx);
+        let insertion_idx = match results.binary_search(&item) {
+            Ok(n) => n,
+            Err(n) => n,
+        };
+        results.insert(insertion_idx, item);
+    }
+
+    let mut stdout = io::stdout().lock();
+    for (_, line, strip_start_idx, strip_end_idx) in results.into_iter() {
+        writeln!(stdout, "{}\x1b[01;34m{}\x1b[0m{}", &line[..strip_start_idx], &line[strip_start_idx..strip_end_idx], &line[strip_end_idx..])?;
+    }
+    Ok(())
+}