From 3735e825d81463e22ee9f76d4d0f5b7bc9ec323e Mon Sep 17 00:00:00 2001
From: hlorenzi <hlorenzi@users.noreply.github.com>
Date: Sat, 7 Oct 2023 19:05:53 -0300
Subject: [PATCH] fix #181: fix optimized matcher algorithm to query for all
 prefixes

---
 src/asm/defs/ruledef_map.rs | 79 ++++++++++++++++++++-----------------
 src/asm/matcher/mod.rs      |  5 +--
 tests/issue181/ok.asm       | 17 ++++++++
 3 files changed, 61 insertions(+), 40 deletions(-)
 create mode 100644 tests/issue181/ok.asm

diff --git a/src/asm/defs/ruledef_map.rs b/src/asm/defs/ruledef_map.rs
index 40f0bf48..7c10d062 100644
--- a/src/asm/defs/ruledef_map.rs
+++ b/src/asm/defs/ruledef_map.rs
@@ -1,21 +1,20 @@
 use crate::*;
 
 
-pub const RULEDEF_MAP_PREFIX_SIZE: usize = 4;
+const MAX_PREFIX_SIZE: usize = 4;
 
 
-pub type RuledefMapPrefix = [char; RULEDEF_MAP_PREFIX_SIZE];
+pub type RuledefMapPrefix = [char; MAX_PREFIX_SIZE];
 
 
 #[derive(Debug)]
 pub struct RuledefMap
 {
-    /// Instructions that start with a mnemonic
-    prefixed: std::collections::HashMap<RuledefMapPrefix, Vec<RuledefMapEntry>>,
-    
-    /// All other instructions, e.g. the ones that start
-    /// with a parameter
-    unprefixed: Vec<RuledefMapEntry>,
+    /// Each rule is considered to have a prefix
+    /// consisting of the first few "exact" pattern-part characters,
+    /// stopping before parameter and subrule slots, and also
+    /// cut off by the constant maximum prefix size declared above.
+    prefixes_to_rules: std::collections::HashMap<RuledefMapPrefix, Vec<RuledefMapEntry>>,
 }
 
 
@@ -32,8 +31,7 @@ impl RuledefMap
     pub fn new() -> RuledefMap
     {
         RuledefMap {
-            prefixed: std::collections::HashMap::new(),
-            unprefixed: Vec::new(),
+            prefixes_to_rules: std::collections::HashMap::new(),
         }
     }
 
@@ -71,12 +69,12 @@ impl RuledefMap
         rule_ref: util::ItemRef<asm::Rule>,
         rule: &asm::Rule)
     {
-        let mut prefix: RuledefMapPrefix = ['\0'; RULEDEF_MAP_PREFIX_SIZE];
+        let mut prefix: RuledefMapPrefix = ['\0'; MAX_PREFIX_SIZE];
         let mut prefix_index = 0;
 
         for part in &rule.pattern
         {
-            if prefix_index >= RULEDEF_MAP_PREFIX_SIZE
+            if prefix_index >= MAX_PREFIX_SIZE
             {
                 break;
             }
@@ -96,17 +94,10 @@ impl RuledefMap
             rule_ref,
         };
 
-        if prefix_index > 0
-        {
-            self.prefixed
-                .entry(prefix)
-                .or_insert_with(|| Vec::new())
-                .push(entry);
-        }
-        else
-        {
-            self.unprefixed.push(entry);
-        }
+        self.prefixes_to_rules
+            .entry(prefix)
+            .or_insert_with(|| Vec::new())
+            .push(entry);
     }
 
 
@@ -114,12 +105,12 @@ impl RuledefMap
         walker: &syntax::TokenWalker)
         -> RuledefMapPrefix
     {
-        let mut prefix: RuledefMapPrefix = ['\0'; RULEDEF_MAP_PREFIX_SIZE];
+        let mut prefix: RuledefMapPrefix = ['\0'; MAX_PREFIX_SIZE];
         let mut prefix_index = 0;
 
         let mut walker_index = 0;
 
-        while prefix_index < RULEDEF_MAP_PREFIX_SIZE
+        while prefix_index < MAX_PREFIX_SIZE
         {
             let token = walker.next_nth(walker_index);
             walker_index += 1;
@@ -128,7 +119,7 @@ impl RuledefMap
             {
                 for c in token.text().chars()
                 {
-                    if prefix_index >= RULEDEF_MAP_PREFIX_SIZE
+                    if prefix_index >= MAX_PREFIX_SIZE
                     {
                         break;
                     }
@@ -150,20 +141,34 @@ impl RuledefMap
     pub fn query_prefixed(
         &self,
         prefix: RuledefMapPrefix)
-        -> &[RuledefMapEntry]
+        -> [&[RuledefMapEntry]; MAX_PREFIX_SIZE + 1]
     {
-        match self.prefixed.get(&prefix)
+        // Try querying for every possible prefix,
+        // including the empty prefix,
+        // i.e. "abcde" will query for "", "a", "ab", "abc", and "abcd".
+        let mut results: [&[RuledefMapEntry]; MAX_PREFIX_SIZE + 1] =
+            [&[]; MAX_PREFIX_SIZE + 1];
+
+        for i in 0..(MAX_PREFIX_SIZE + 1)
         {
-            Some(entries) => entries,
-            None => &[],
-        }
-    }
+            let mut subprefix = prefix;
+            for j in i..MAX_PREFIX_SIZE
+            {
+                subprefix[j] = '\0';
+            }
 
+            if let Some(entries) = self.prefixes_to_rules.get(&subprefix)
+            {
+                results[i] = entries;
+            }
 
-    pub fn query_unprefixed(
-        &self)
-        -> &[RuledefMapEntry]
-    {
-        &self.unprefixed
+            if i < MAX_PREFIX_SIZE &&
+                prefix[i] == '\0'
+            {
+                break;
+            }
+        }
+
+        results
     }
 }
\ No newline at end of file
diff --git a/src/asm/matcher/mod.rs b/src/asm/matcher/mod.rs
index 169bf65a..f0b8b4aa 100644
--- a/src/asm/matcher/mod.rs
+++ b/src/asm/matcher/mod.rs
@@ -422,10 +422,9 @@ fn match_with_ruledef_map<'tokens>(
     let mut matches = WorkingMatches::new();
 
     let prefix = asm::RuledefMap::parse_prefix(&walker);
-    let prefixed_entries = defs.ruledef_map.query_prefixed(prefix);
-    let unprefixed_entries = defs.ruledef_map.query_unprefixed();
+    let entries = defs.ruledef_map.query_prefixed(prefix);
 
-    for entry in prefixed_entries.iter().chain(unprefixed_entries)
+    for entry in entries.iter().flat_map(|e| e.iter())
     {
         let ruledef = defs.ruledefs.get(entry.ruledef_ref);
         let rule = ruledef.get_rule(entry.rule_ref);
diff --git a/tests/issue181/ok.asm b/tests/issue181/ok.asm
new file mode 100644
index 00000000..d1ba422f
--- /dev/null
+++ b/tests/issue181/ok.asm
@@ -0,0 +1,17 @@
+#subruledef SELECT
+{
+	C => 0xc
+}
+
+#ruledef
+{
+	A{sel:SELECT}    => 0x1 @ sel
+	AB{sel:SELECT}   => 0x2 @ sel
+	ABB{sel:SELECT}	 => 0x3 @ sel
+	ABBB{sel:SELECT} => 0x4 @ sel
+}
+
+AC    ; = 0x1c
+ABC   ; = 0x2c
+ABBC  ; = 0x3c
+ABBBC ; = 0x4c
\ No newline at end of file