candidate sets for the fill

cruciverb-core/src/solver/candidate.rs

@@ -0,0 +1,107 @@
+use crate::dict::Dictionary;
+
+pub fn score_candidate(
+    word_idx: usize,
+    dict: &Dictionary,
+    used_words: &[bool],
+    placed_texts: &[String],
+) -> f64 {
+    if used_words[word_idx] {
+        return -1.0;
+    }
+
+    let word = dict.get(word_idx);
+    let text = &word.text;
+    let mut score = word.frequency;
+
+    // reject words that are too similar to already-placed words
+    for placed in placed_texts {
+        if is_related(text, placed) {
+            return -1.0;
+        }
+    }
+
+    // bonus for letter variety
+    let bytes = text.as_bytes();
+    let mut seen = [false; 26];
+    let mut unique = 0;
+    for &b in bytes {
+        let idx = (b - b'A') as usize;
+        if !seen[idx] {
+            seen[idx] = true;
+            unique += 1;
+        }
+    }
+    let variety = unique as f64 / bytes.len() as f64;
+    score += variety * 0.1;
+
+    // bonus for common letters
+    let common = b"ETAOINSHRDLU";
+    let common_count = bytes.iter().filter(|b| common.contains(b)).count();
+    let common_ratio = common_count as f64 / bytes.len() as f64;
+    score += common_ratio * 0.05;
+
+    score
+}
+
+pub fn is_related(a: &str, b: &str) -> bool {
+    // one is a prefix of the other (SEAL / SEALS, RUN / RUNNING)
+    if a.len() >= 3 && b.len() >= 3 {
+        let (short, long) = if a.len() <= b.len() { (a, b) } else { (b, a) };
+        if long.starts_with(short) {
+            let suffix = &long[short.len()..];
+            // common inflection suffixes
+            if matches!(suffix, "S" | "ES" | "ED" | "ER" | "ING" | "LY" | "TION" | "NESS" | "MENT" | "ABLE" | "ISH" | "IST" | "IZE") {
+                return true;
+            }
+        }
+        // handle e-dropping: MAKE/MAKING, BAKE/BAKING
+        if short.len() >= 4 && long.ends_with("ING") {
+            let stem = &long[..long.len() - 3];
+            if short.len() == stem.len() + 1 && short.starts_with(stem) && short.ends_with('E') {
+                return true;
+            }
+        }
+        // consonant doubling: RUN/RUNNING, SIT/SITTING
+        if long.ends_with("ING") && long.len() >= 6 {
+            let before_ing = &long[..long.len() - 3];
+            let bb = before_ing.as_bytes();
+            if bb.len() >= 4 && bb[bb.len() - 1] == bb[bb.len() - 2] {
+                let stem = &before_ing[..before_ing.len() - 1];
+                if short == stem {
+                    return true;
+                }
+            }
+        }
+        // also ED doubling: STOP/STOPPED
+        if long.ends_with("ED") && long.len() >= 5 {
+            let before_ed = &long[..long.len() - 2];
+            let bb = before_ed.as_bytes();
+            if bb.len() >= 4 && bb[bb.len() - 1] == bb[bb.len() - 2] {
+                let stem = &before_ed[..before_ed.len() - 1];
+                if short == stem {
+                    return true;
+                }
+            }
+        }
+    }
+    false
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn related_words() {
+        assert!(is_related("SEAL", "SEALS"));
+        assert!(is_related("RUN", "RUNNING"));
+        assert!(is_related("MAKE", "MAKING"));
+        assert!(is_related("PLAY", "PLAYED"));
+        assert!(is_related("PLAY", "PLAYER"));
+        assert!(is_related("QUICK", "QUICKLY"));
+        assert!(!is_related("CAT", "DOG"));
+        assert!(!is_related("SEA", "SEAL"));
+        assert!(!is_related("THE", "THEM"));
+    }
+}