Move state splitting algorithm into its own file

cli/src/generate/build_tables/build_lex_table.rs

@@ -1,7 +1,7 @@
 use super::coincident_tokens::CoincidentTokenIndex;
 use super::item::TokenSet;
-use super::split_state_id_groups;
 use super::token_conflicts::TokenConflictMap;
+use crate::generate::dedup::split_state_id_groups;
 use crate::generate::grammars::{LexicalGrammar, SyntaxGrammar};
 use crate::generate::nfa::{CharacterSet, NfaCursor};
 use crate::generate::rules::Symbol;

cli/src/generate/build_tables/minimize_parse_table.rs

@@ -1,6 +1,6 @@
 use super::item::TokenSet;
-use super::split_state_id_groups;
 use super::token_conflicts::TokenConflictMap;
+use crate::generate::dedup::split_state_id_groups;
 use crate::generate::grammars::{LexicalGrammar, SyntaxGrammar, VariableType};
 use crate::generate::rules::{AliasMap, Symbol};
 use crate::generate::tables::{ParseAction, ParseState, ParseStateId, ParseTable, ParseTableEntry};

cli/src/generate/build_tables/mod.rs

@@ -349,67 +349,3 @@ fn all_chars_are_alphabetical(cursor: &NfaCursor) -> bool {
         }
     })
 }
-
-fn split_state_id_groups<S>(
-    states: &Vec<S>,
-    state_ids_by_group_id: &mut Vec<Vec<usize>>,
-    group_ids_by_state_id: &mut Vec<usize>,
-    start_group_id: usize,
-    mut f: impl FnMut(&S, &S, &Vec<usize>) -> bool,
-) -> bool {
-    let mut result = false;
-
-    let mut group_id = start_group_id;
-    while group_id < state_ids_by_group_id.len() {
-        let state_ids = &state_ids_by_group_id[group_id];
-        let mut split_state_ids = Vec::new();
-
-        let mut i = 0;
-        while i < state_ids.len() {
-            let left_state_id = state_ids[i];
-            if split_state_ids.contains(&left_state_id) {
-                i += 1;
-                continue;
-            }
-
-            let left_state = &states[left_state_id];
-
-            // Identify all of the other states in the group that are incompatible with
-            // this state.
-            let mut j = i + 1;
-            while j < state_ids.len() {
-                let right_state_id = state_ids[j];
-                if split_state_ids.contains(&right_state_id) {
-                    j += 1;
-                    continue;
-                }
-                let right_state = &states[right_state_id];
-
-                if f(left_state, right_state, &group_ids_by_state_id) {
-                    split_state_ids.push(right_state_id);
-                }
-
-                j += 1;
-            }
-
-            i += 1;
-        }
-
-        // If any states were removed from the group, add them all as a new group.
-        if split_state_ids.len() > 0 {
-            result = true;
-            state_ids_by_group_id[group_id].retain(|i| !split_state_ids.contains(&i));
-
-            let new_group_id = state_ids_by_group_id.len();
-            for id in &split_state_ids {
-                group_ids_by_state_id[*id] = new_group_id;
-            }
-
-            state_ids_by_group_id.push(split_state_ids);
-        }
-
-        group_id += 1;
-    }
-
-    result
-}

cli/src/generate/dedup.rs

@@ -0,0 +1,63 @@
+pub(crate) fn split_state_id_groups<S>(
+    states: &Vec<S>,
+    state_ids_by_group_id: &mut Vec<Vec<usize>>,
+    group_ids_by_state_id: &mut Vec<usize>,
+    start_group_id: usize,
+    mut f: impl FnMut(&S, &S, &Vec<usize>) -> bool,
+) -> bool {
+    let mut result = false;
+
+    let mut group_id = start_group_id;
+    while group_id < state_ids_by_group_id.len() {
+        let state_ids = &state_ids_by_group_id[group_id];
+        let mut split_state_ids = Vec::new();
+
+        let mut i = 0;
+        while i < state_ids.len() {
+            let left_state_id = state_ids[i];
+            if split_state_ids.contains(&left_state_id) {
+                i += 1;
+                continue;
+            }
+
+            let left_state = &states[left_state_id];
+
+            // Identify all of the other states in the group that are incompatible with
+            // this state.
+            let mut j = i + 1;
+            while j < state_ids.len() {
+                let right_state_id = state_ids[j];
+                if split_state_ids.contains(&right_state_id) {
+                    j += 1;
+                    continue;
+                }
+                let right_state = &states[right_state_id];
+
+                if f(left_state, right_state, &group_ids_by_state_id) {
+                    split_state_ids.push(right_state_id);
+                }
+
+                j += 1;
+            }
+
+            i += 1;
+        }
+
+        // If any states were removed from the group, add them all as a new group.
+        if split_state_ids.len() > 0 {
+            result = true;
+            state_ids_by_group_id[group_id].retain(|i| !split_state_ids.contains(&i));
+
+            let new_group_id = state_ids_by_group_id.len();
+            for id in &split_state_ids {
+                group_ids_by_state_id[*id] = new_group_id;
+            }
+
+            state_ids_by_group_id.push(split_state_ids);
+        }
+
+        group_id += 1;
+    }
+
+    result
+}

cli/src/generate/mod.rs

@@ -11,6 +11,7 @@ use std::path::{Path, PathBuf};
 use std::process::{Command, Stdio};
 
 mod build_tables;
+mod dedup;
 mod grammars;
 mod nfa;
 mod node_types;