enhanced pre-layout -> layout algorithm

4 files changed, 128 insertions, 91 deletions

diff --git a/srchr/src/config.rs b/srchr/src/config.rs
index 9ed98d3..8a7faae 100644
--- a/srchr/src/config.rs
+++ b/srchr/src/config.rs
@@ -31,7 +31,7 @@ pub fn index_usage_fitness(index_perc: f32) -> f32 {
     }
 }
 
-pub const MAX_PINKY_USAGE_PERCENT: f32 = 0.092;
+pub const MAX_PINKY_USAGE_PERCENT: f32 = 0.095;
 pub fn pinky_col_penalty(lesser_perc: f32, greater_perc: f32) -> f32 {
     if greater_perc > MAX_PINKY_USAGE_PERCENT {
         if lesser_perc > MAX_PINKY_USAGE_PERCENT {
@@ -45,7 +45,12 @@ pub fn pinky_col_penalty(lesser_perc: f32, greater_perc: f32) -> f32 {
 }
 
 // Pre-layout to layout
-pub const MINIMISE_LSB_INSTEAD_OF_OPTIMISING_BALANCE: bool = true;
+#[allow(dead_code)]
+pub enum Algorithm {
+    MinimiseLSBThenBalance,
+    GreedyBalance,
+}
+pub const PRELAYOUT_TO_LAYOUT_ALGORITHM: Algorithm = Algorithm::MinimiseLSBThenBalance;
 
 // -----------------------------------------------------------------------------
 //     Do not change the following, the code makes assumptions about these
diff --git a/srchr/src/corpus.rs b/srchr/src/corpus.rs
index 846f611..bc6813e 100644
--- a/srchr/src/corpus.rs
+++ b/srchr/src/corpus.rs
@@ -1,4 +1,3 @@
-use std::fmt;
 use std::fs;
 
 use crate::config::*;
@@ -152,37 +151,3 @@ impl Corpus {
         });
     }
 }
-
-fn dump_bigrams(corpus: &Corpus) -> Vec<(String, f32)> {
-    let mut result: Vec<(String, f32)> = Vec::new();
-
-    for &x in KEY_CHARS.iter() {
-        for &y in KEY_CHARS.iter() {
-            let mut pair = String::from(x);
-            pair.push(y);
-            let perc = corpus.bigram_percs[pair_to_index(x as u8, y as u8)];
-            if perc > 0.0 {
-                result.push((pair, perc));
-            }
-        }
-    }
-
-    result.sort_by(|(_, c1), (_, c2)| {
-        if c2 < c1 {
-            std::cmp::Ordering::Less
-        } else {
-            std::cmp::Ordering::Greater
-        }
-    });
-
-    return result;
-}
-
-impl fmt::Display for Corpus {
-    fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result {
-        formatter.write_str(&format!(
-            "Corpus has top 5 bigrams {:?}",
-            &dump_bigrams(self)[0..5]
-        ))
-    }
-}
diff --git a/srchr/src/layout.rs b/srchr/src/layout.rs
index bc519c1..cc963c7 100644
--- a/srchr/src/layout.rs
+++ b/srchr/src/layout.rs
@@ -146,68 +146,136 @@ impl Layout {
             result
         }
 
-        let mut balance: f32 = 0.0;
-        let mut left_col: usize = 0;
-        let mut right_col: usize = 9;
-        let mut keys = ['x'; NUM_KEYS];
-
+        // Derive column orderings
+        let w_index_columns = weight_function(&pl.index_columns, corpus);
         let w_standard_columns = weight_function(&pl.standard_columns, corpus);
-        for (col, weight) in w_standard_columns.iter() {
-            let left: bool = ((balance >= 0.0) && (left_col <= 2)) || (right_col <= 6);
-            let ind = if left { left_col } else { right_col };
-            for i in 0..3 {
-                keys[ind + i * ROW_LENGTH] = col[i] as char;
-            }
-            if left {
-                left_col += 1;
-            } else {
-                right_col -= 1;
-            }
-            balance += if left { -weight } else { *weight };
-        }
 
-        let w_index_columns = weight_function(&pl.index_columns, corpus);
+        let mut standard_column_ordering = [0; 6];
+        let mut left_ind;
+        let mut right_ind;
+
+        match PRELAYOUT_TO_LAYOUT_ALGORITHM {
+            Algorithm::MinimiseLSBThenBalance => {
+                // first we still want left pinky to have lowest load
+                let mut balance: f32 = -w_index_columns[0].1 + w_index_columns[1].1;
+                standard_column_ordering[0] = 0;
+                standard_column_ordering[5] = 1;
 
-        let (left_ind, right_ind);
-        if MINIMISE_LSB_INSTEAD_OF_OPTIMISING_BALANCE {
-            // Choose index order to minimise LSB (middle-index only) even though we
-            // don't directly optimise for this parameter
-            let mut lsbs = [0.0; 2];
-            for option in [0, 1] {
-                for left in [true, false] {
-                    let mind = if left { 2 } else { 7 };
-                    let side = if left { option } else { 1 - option };
-                    let m1 = keys[mind + 10 * 0] as u8;
-                    let m2 = keys[mind + 10 * 1] as u8;
-                    let m3 = keys[mind + 10 * 2] as u8;
-                    let i1 = w_index_columns[side].0[3];
-                    let i2 = w_index_columns[side].0[4];
-                    let i3 = w_index_columns[side].0[5];
-                    lsbs[option] += corpus.get_bigram_perc(m1, i1)
-                        + corpus.get_bigram_perc(m1, i2)
-                        + corpus.get_bigram_perc(m1, i3)
-                        + corpus.get_bigram_perc(m2, i1)
-                        + corpus.get_bigram_perc(m2, i2)
-                        + corpus.get_bigram_perc(m2, i3)
-                        + corpus.get_bigram_perc(m3, i1)
-                        + corpus.get_bigram_perc(m3, i2)
-                        + corpus.get_bigram_perc(m3, i3);
+                // next find minimum lsb pairing
+                let mut lsbs = [0.0; 2];
+                for swap in [false, true] {
+                    for first_pairing in [true, false] {
+                        let mind = match (swap, first_pairing) {
+                            (true, true) => 5,
+                            (true, false) => 4,
+                            (false, true) => 4,
+                            (false, false) => 5,
+                        };
+                        let m1 = w_standard_columns[mind].0[0];
+                        let m2 = w_standard_columns[mind].0[1];
+                        let m3 = w_standard_columns[mind].0[2];
+                        let iind = if first_pairing { 0 } else { 1 };
+                        let i1 = w_index_columns[iind].0[3];
+                        let i2 = w_index_columns[iind].0[4];
+                        let i3 = w_index_columns[iind].0[5];
+                        lsbs[if swap { 1 } else { 0 }] += corpus.get_bigram_perc(m1, i1)
+                            + corpus.get_bigram_perc(m1, i2)
+                            + corpus.get_bigram_perc(m1, i3)
+                            + corpus.get_bigram_perc(m2, i1)
+                            + corpus.get_bigram_perc(m2, i2)
+                            + corpus.get_bigram_perc(m2, i3)
+                            + corpus.get_bigram_perc(m3, i1)
+                            + corpus.get_bigram_perc(m3, i2)
+                            + corpus.get_bigram_perc(m3, i3);
+                    }
                 }
+
+                // if lsbs[0] < lsbs[1] then (4, 0) and (5, 1) are paired, else (5, 0) and (4, 1)
+                // record weights of pairings and whether we swapped 4 <-> 5 for min lsb
+                let (swap, w1, w2) = if lsbs[0] < lsbs[1] {
+                    (
+                        false,
+                        w_standard_columns[4].1 + w_index_columns[0].1,
+                        w_standard_columns[5].1 + w_index_columns[1].1,
+                    )
+                } else {
+                    (
+                        true,
+                        w_standard_columns[5].1 + w_index_columns[0].1,
+                        w_standard_columns[4].1 + w_index_columns[1].1,
+                    )
+                };
+
+                // now we try to balance hands according to combined weight, place
+                // on left and check if we need to swap
+                let w: f32;
+                if w1 < w2 {
+                    (
+                        (standard_column_ordering[2], left_ind),
+                        (standard_column_ordering[3], right_ind),
+                    ) = if swap {
+                        ((5, 0), (4, 1))
+                    } else {
+                        ((4, 0), (5, 1))
+                    };
+                    w = -w1 + w2;
+                } else {
+                    (
+                        (standard_column_ordering[2], left_ind),
+                        (standard_column_ordering[3], right_ind),
+                    ) = if swap {
+                        ((4, 1), (5, 0))
+                    } else {
+                        ((5, 1), (4, 0))
+                    };
+                    w = -w2 + w1;
+                }
+                if (balance + w).abs() > (balance - w).abs() {
+                    // swapping would be better
+                    (left_ind, right_ind) = (right_ind, left_ind);
+                    (standard_column_ordering[2], standard_column_ordering[3]) =
+                        (standard_column_ordering[3], standard_column_ordering[2]);
+                    balance -= w;
+                } else {
+                    balance += w;
+                }
+
+                // finally we do ring finger
+                (standard_column_ordering[1], standard_column_ordering[4]) =
+                    if balance >= 0.0 { (3, 2) } else { (2, 3) };
             }
-            (left_ind, right_ind) = if lsbs[0] < lsbs[1] { (0, 1) } else { (1, 0) };
-            // if this disagrees with the balancing order, let's consider ring
-            // column swaps
-            if balance >= 0.0 && left_ind == 0 && w_standard_columns[2].0[0] == keys[1] as u8 {
-                for i in 0..3 {
-                    let t = keys[8 + 10 * i];
-                    keys[8 + 10 * i] = keys[1 + 10 * i];
-                    keys[1 + 10 * i] = t;
+            Algorithm::GreedyBalance => {
+                // greedy on weight, not provably correct in *all* cases but good
+                // enough in practice
+                let mut balance: f32 = 0.0;
+                let mut left_col: usize = 0;
+                let mut right_col: usize = 9;
+
+                for (src, (_, weight)) in w_standard_columns.iter().enumerate() {
+                    let left: bool = ((balance >= 0.0) && (left_col <= 2)) || (right_col <= 6);
+                    let ind = if left { left_col } else { right_col };
+                    standard_column_ordering[ind] = src;
+                    if left {
+                        left_col += 1;
+                    } else {
+                        right_col -= 1;
+                    }
+                    balance += if left { -weight } else { *weight };
                 }
+                (left_ind, right_ind) = if balance >= 0.0 { (0, 1) } else { (1, 0) }
             }
-        } else {
-            (left_ind, right_ind) = if balance >= 0.0 { (0, 1) } else { (1, 0) }
         }
 
+        // write keys
+        let mut keys = ['x'; NUM_KEYS];
+        // standard
+        for (ind, src) in standard_column_ordering.into_iter().enumerate() {
+            let target = if ind < 3 { ind } else { 4 + ind };
+            for i in 0..3 {
+                keys[target + i * ROW_LENGTH] = w_standard_columns[src].0[i] as char;
+            }
+        }
+        // index
         for j in 0..3 {
             keys[3 + j * ROW_LENGTH] = w_index_columns[left_ind].0[j] as char;
             keys[6 + j * ROW_LENGTH] = w_index_columns[right_ind].0[j] as char;
diff --git a/srchr/src/main.rs b/srchr/src/main.rs
index 628dbdb..af1d0d2 100644
--- a/srchr/src/main.rs
+++ b/srchr/src/main.rs
@@ -88,7 +88,6 @@ fn main() {
     } else {
         corpus = Corpus::load_from_text_file(CORPUS_FILE_NAME).unwrap();
     }
-    println!("{}", corpus);
 
     let seed_layout = Layout::from_verbose(STARTING_LAYOUT_STRING).unwrap();
     let seed_evl = Evaluation::evaluate_layout(&corpus, &seed_layout);