tod

rust-whisper-baked/src/main.rs

@@ -84,7 +84,7 @@ fn channel(flags: rust_whisper_lib::Flags) {
 }
 
 struct Destutterer {
-   prevs: Vec<Word>,
+   prev: Words,
 }
 
 fn new_destutterer() -> Destutterer {
@@ -97,59 +97,89 @@ impl Destutterer {
          return next;
       }
 
-      let nexts = Word::from_string(next.clone());
-
-      let mut n = self.prevs.len().clamp(0, nexts.len());
+      let next_words = Words::from_string(next.clone());
+      let mut n = self.prevs.len().clamp(0, next_words.len());
       while n > 0 {
-         let prev_s = Word::to_comparable_string(self.prevs, self.prevs.len(), n);
-         eprintln!("prevs: {:?} => '{}'", self.prevs[self.prevs.len() - n..].to_vec(), &prev_s);
-         let next_s = Word::to_comparable_string(nexts, 0, n); // TODO indexes skip stop words
-         eprintln!("nexts: {:?} => '{}'", nexts[..n].to_vec(), &next_s);
+         let prev_s = (self.prevs.len()-n..self.prevs.len()).map(|i| self.prevs.comparable_string(i)).collect().join(" ");
+         let next_s = (0-n).map(|i| next_words.comparable_string(i)).collect().join(" ");
+         eprintln!("prevs => '{}'", &prev_s);
+         eprintln!("nexts => '{}'", &next_s);
          if prev_s == next_s {
             break;
          }
          n -= 1;
       }
-      self.prevs = nexts.clone();
+      self.prevs = next_words;
       Word::to_string(nexts[n..].to_vec())
    }
 }
 
 #[derive(Clone, Debug)]
-struct Word {
-   raw: String,
+struct Words {
+   raw: Vec<String>,
 }
 
-impl Word {
-   fn from_string(s: String) -> Vec<Word> {
-      let mut result = vec![];
+impl Words {
+   fn from_string(s: String) -> Words {
+      let mut result = Words{raw: vec![]};
       for word in s.split(" ") {
          let word = word.trim();
          if word.len() > 0 {
-            result.push(Word{raw: word.to_string()});
+            result.raw.push(word.to_string());
          }
       }
       result
    }
 
-   fn to_comparable_string(v: Vec<Word>) -> String {
-      let skips = stop_words::get("en");
-      v.iter()
-         .map(|w| w.raw.to_lowercase())
-         .filter(|word| !skips.contains(word))
-         .map(|word| word.chars().filter(|c| c.is_ascii_alphanumeric()).collect::<String>())
-         .collect::<Vec<String>>()
-         .join(" ")
+   fn last_n_comparable_to_string(&self, n: usize) -> (String, usize) {
+      TODO
    }
 
-   fn to_string(v: Vec<Word>) -> String {
-      v.iter()
-         .map(|x| x.raw.clone())
+   fn first_n_comparable_to_string(&self, n: usize) -> (String, usize){
+      TODO
+   }
+
+   fn comparable_len(&self) -> usize {
+      self.to_comparable_words().len()
+   }
+
+   fn to_comparable_words(&self) -> Vec<Word> {
+      self.to_words().iter().filter(|x| x.s.is_some()).collect()
+   }
+
+   fn to_words(&self) -> Vec<Word> {
+      let skips = stop_words::get("en");
+      let strs = self.raw.iter()
+         .map(|w| w.to_lowercase())
+         .map(|word| word.chars().filter(|c| c.is_ascii_alphanumeric()).collect::<String>())
+         .collect::<Vec<String>>();
+      let mut result = vec![];
+      for i in 0..strs.len() {
+         result.push(Word{
+            s: match skips.contains(&strs[i]) {
+               true => None,
+               false => Some(strs[i]),
+            },
+            idx: i as usize,
+         });
+      }
+      result
+   }
+
+   fn to_string(&self) -> String {
+      self.raw.iter()
+         .map(|x| x.clone())
          .collect::<Vec<String>>()
          .join(" ")
    }
 }
 
+#[derive(Debug)]
+struct Word {
+   s: Option<String>,
+   idx: usize,
+}
+
 #[cfg(test)]
 mod tests {
    use super::*;