perf: Split the uts46 table for better cache locality (-4%)

Author: Markus Westerlind

idna/src/make_uts46_mapping_table.py

@@ -23,8 +23,6 @@
 // except according to those terms.
 
 // Generated by make_idna_table.py
-
-static TABLE: &'static [Range] = &[
 ''')
 
 txt = open("IdnaMappingTable.txt")
@@ -123,12 +121,23 @@ def mergeable_key(r):
     unicode_str = group[0][3]
     optimized_ranges.append((first, last, mapping, unicode_str))
 
+
+print("static TABLE: &'static [Range] = &[")
+
+for (first, last, mapping, unicode_str) in optimized_ranges:
+    if unicode_str is not None:
+        mapping += rust_slice(strtab_slice(unicode_str))
+    print("    Range { from: '%s', to: '%s', }," % (escape_char(char(first)),
+                                                                escape_char(char(last))))
+
+print("];\n")
+
+print("static MAPPING_TABLE: &'static [Mapping] = &[")
+
 for (first, last, mapping, unicode_str) in optimized_ranges:
     if unicode_str is not None:
         mapping += rust_slice(strtab_slice(unicode_str))
-    print("    Range { from: '%s', to: '%s', mapping: %s }," % (escape_char(char(first)),
-                                                                escape_char(char(last)),
-                                                                mapping))
+    print("    %s," % mapping)
 
 print("];\n")
 

idna/src/uts46.rs

@@ -55,7 +55,6 @@ enum Mapping {
 struct Range {
     from: char,
     to: char,
-    mapping: Mapping,
 }
 
 fn find_char(codepoint: char) -> &'static Mapping {
@@ -68,7 +67,7 @@ fn find_char(codepoint: char) -> &'static Mapping {
             Equal
         }
     });
-    r.ok().map(|i| &TABLE[i].mapping).unwrap()
+    r.ok().map(|i| &MAPPING_TABLE[i]).unwrap()
 }
 
 fn map_char(codepoint: char, flags: Flags, output: &mut String, errors: &mut Vec<Error>) {