Read CPU Arpabet dictionary

dictionary/cmudict-0.7b

@@ -35471,7 +35471,7 @@ DZIUBA  JH IY0 UW1 B AH0
 DZIUK  JH IY0 UW1 K
 DZOKHAR  JH OW2 K AA1 R
 D_C  D IY1 S IY1
-DÉJÀ  D EY2 JH AA1
+DEJA  D EY2 JH AA1
 E  IY1
 E'S  IY1 Z
 E-MAIL  IY1 M EY2 L

src/arpabet.rs

@@ -0,0 +1,68 @@
+// Copyright (c) 2016 Brandon Thomas <bt@brand.io, echelon@gmail.com>
+
+use error::Error;
+use std::collections::HashMap;
+use regex::Regex;
+use std::fs::File;
+use std::io::BufReader;
+use std::io::BufRead;
+use std::io::Read;
+
+pub type Word = String;
+pub type Phoneme = String;
+pub type Polyphone = Vec<Phoneme>;
+
+pub struct ArpabetDictionary {
+  /// A map of lowercase words to polyphone breakdown (phones are uppercase).
+  /// eg. 'jungle' -> [JH, AH1, NG, G, AH0, L]
+  dictionary: HashMap<Word, Polyphone>
+}
+
+impl ArpabetDictionary {
+  /// Read CMU's Arpabet Dictionary.
+  /// See http://www.speech.cs.cmu.edu/cgi-bin/cmudict
+  pub fn load_from_file(filename: &str) -> Result<ArpabetDictionary, Error> {
+    let mut f = try!(File::open(filename));
+    let mut reader = BufReader::new(f);
+
+    let mut map = HashMap::new();
+    let mut buffer = String::new();
+
+    // Format resembles the following,
+    // ABBREVIATE  AH0 B R IY1 V IY0 EY2 T
+    let re = Regex::new(r"^(\w+)\s+(.*)\n$").unwrap();
+
+    while try!(reader.read_line(&mut buffer)) > 0 {
+      match re.captures(&buffer) {
+        None => {},
+        Some(caps) => {
+          let word_match = caps.at(1);
+          let phonemes_match = caps.at(2);
+
+          if word_match.is_some() && phonemes_match.is_some() {
+            let word = word_match.unwrap().to_lowercase();
+            let split = phonemes_match.unwrap().split(" ");
+            let v1 = split.collect::<Vec<&str>>();
+            let v2 = v1.iter().map(|s| s.to_string()).collect::<Vec<String>>();
+            map.insert(word, v2);
+          }
+        },
+      }
+      buffer.clear();
+    }
+
+    if map.is_empty() {
+      Err(Error::EmptyFile)
+    } else {
+      Ok(ArpabetDictionary { dictionary: map })
+    }
+  }
+
+  /// Get a polyphone from the dictionary.
+  pub fn get(&self, word: &str) -> Option<Polyphone> {
+    self.dictionary.get(word).and_then(|p| {
+      Some(p.iter().map(|s| s.to_string()).collect::<Vec<String>>())
+    })
+  }
+}
+

src/error.rs

@@ -0,0 +1,22 @@
+// Copyright (c) 2016 Brandon Thomas <bt@brand.io, echelon@gmail.com>
+
+use std::error;
+use std::io;
+use std::fmt;
+use std::convert::From;
+
+#[derive(Debug)]
+pub enum Error {
+  /// The file was empty or did not have parsable contents.
+  EmptyFile,
+
+  /// Wraps an IoError.
+  IoError { cause: io::Error },
+}
+
+impl From<io::Error> for Error {
+  fn from(error: io::Error) -> Error {
+    Error::IoError { cause: error }
+  }
+}
+

src/main.rs

@@ -6,6 +6,7 @@ extern crate hound;
 extern crate iron;
 #[macro_use]
 extern crate log;
+extern crate regex;
 extern crate resolve;
 extern crate router;
 extern crate rustc_serialize;
@@ -13,26 +14,28 @@ extern crate time;
 extern crate toml;
 extern crate urlencoded;
 
+pub mod arpabet;
+pub mod atom;
+pub mod error;
 pub mod config;
 pub mod dictionary;
 pub mod handlers;
 pub mod logger;
 pub mod words;
 
-use std::path::{Path, PathBuf};
-
+use arpabet::ArpabetDictionary;
 use clap::{App, Arg, ArgMatches};
-use iron::prelude::*;
-use router::Router;
-use resolve::hostname;
 use config::Config;
-
+use dictionary::VocabularyLibrary;
 use handlers::audio_synth_handler::AudioSynthHandler;
 use handlers::error_filter::ErrorFilter;
 use handlers::file_server_handler::FileServerHandler;
 use handlers::vocab_list_handler::VocabListHandler;
+use iron::prelude::*;
 use logger::SimpleLogger;
-use dictionary::VocabularyLibrary;
+use resolve::hostname;
+use router::Router;
+use std::path::{Path, PathBuf};
 
 fn main() {
   SimpleLogger::init().unwrap();
@@ -56,6 +59,8 @@ fn main() {
 
   get_hostname();
 
+  read_arpabet();
+
   start_server(&config, port);
 }
 
@@ -105,3 +110,13 @@ fn get_hostname() {
   let err = unsafe {gethostname (vec::raw::to_mut_ptr(buf) as *mut i8, len as u64)};
   if err != 0 { println("oops, gethostname failed"); return; }*/
 }
+
+fn read_arpabet() {
+  // TODO: Configurable path.
+  let dict = ArpabetDictionary::load_from_file("./dictionary/cmudict-0.7b").unwrap();
+
+  let result = dict.get("nintendo");
+
+  println!("Result: {:?}", result);
+}
+