More lookups moved to perfect hashing

raphlinus · raphlinus · commit db57ffc5e767 · 2019-04-09T16:46:28.000-07:00
diff --git a/scripts/unicode.py b/scripts/unicode.py
@@ -287,7 +287,6 @@ def _compute_stream_safe_tables(self):
 hexify = lambda c: '{:04X}'.format(c)
 
 def gen_combining_class(combining_classes, out):
-    out.write("#[inline]\n")
     (salt, keys) = minimal_perfect_hash(combining_classes)
     out.write("pub fn canonical_combining_class(c: char) -> u8 {\n")
     out.write("    mph_lookup(c.into(), &[\n")
@@ -299,7 +298,7 @@ def gen_combining_class(combining_classes, out):
         kv = int(combining_classes[k]) | (k << 8)
         out.write("        0x{:x},\n".format(kv))
     out.write("    ],\n")
-    out.write("    u8_lookup_fk, u8_lookup_fv, 0) as u8\n")
+    out.write("    u8_lookup_fk, u8_lookup_fv, 0)\n")
     out.write("}\n")
 
 def gen_composition_table(canon_comp, out):
@@ -376,16 +375,20 @@ def gen_nfkd_qc(prop_tables, out):
 
 def gen_combining_mark(general_category_mark, out):
     out.write("#[inline]\n")
+    (salt, keys) = minimal_perfect_hash(general_category_mark)
     out.write("pub fn is_combining_mark(c: char) -> bool {\n")
-    out.write("    match c {\n")
-
-    for char in sorted(general_category_mark):
-        out.write("        '\\u{%s}' => true,\n" % hexify(char))
-
-    out.write("        _ => false,\n")
-    out.write("    }\n")
+    out.write("    mph_lookup(c.into(), &[\n")
+    for s in salt:
+        out.write("        0x{:x},\n".format(s))
+    out.write("    ],\n")
+    out.write("    &[\n")
+    for k in keys:
+        out.write("        0x{:x},\n".format(k))
+    out.write("    ],\n")
+    out.write("    bool_lookup_fk, bool_lookup_fv, false)\n")
     out.write("}\n")
 
+
 def gen_stream_safe(leading, trailing, out):
     out.write("#[inline]\n")
     out.write("pub fn stream_safe_leading_nonstarters(c: char) -> usize {\n")
@@ -399,15 +402,18 @@ def gen_stream_safe(leading, trailing, out):
     out.write("}\n")
     out.write("\n")
 
-    out.write("#[inline]\n")
+    (salt, keys) = minimal_perfect_hash(trailing)
     out.write("pub fn stream_safe_trailing_nonstarters(c: char) -> usize {\n")
-    out.write("    match c {\n")
-
-    for char, num_trailing in sorted(trailing.items()):
-        out.write("        '\\u{%s}' => %d,\n" % (hexify(char), num_trailing))
-
-    out.write("        _ => 0,\n")
-    out.write("    }\n")
+    out.write("    mph_lookup(c.into(), &[\n")
+    for s in salt:
+        out.write("        0x{:x},\n".format(s))
+    out.write("    ],\n")
+    out.write("    &[\n")
+    for k in keys:
+        kv = int(trailing[k]) | (k << 8)
+        out.write("        0x{:x},\n".format(kv))
+    out.write("    ],\n")
+    out.write("    u8_lookup_fk, u8_lookup_fv, 0) as usize\n")
     out.write("}\n")
 
 def gen_tests(tests, out):
diff --git a/src/perfect_hash.rs b/src/perfect_hash.rs
@@ -12,6 +12,7 @@
 
 // This function is based on multiplication being fast and is "good enough". Also
 // it can share some work between the unsalted and salted versions.
+#[inline]
 fn my_hash(key: u32, salt: u32, n: usize) -> usize {
     let y = key.wrapping_add(salt).wrapping_mul(2654435769);
     let y = y ^ key.wrapping_mul(0x31415926);
@@ -33,11 +34,25 @@ pub(crate) fn mph_lookup<KV, V, FK, FV>(x: u32, salt: &[u16], kv: &[KV], fk: FK,
 }
 
 /// Extract the key in a 24 bit key and 8 bit value packed in a u32.
+#[inline]
 pub(crate) fn u8_lookup_fk(kv: u32) -> u32 {
     kv >> 8
 }
 
 /// Extract the value in a 24 bit key and 8 bit value packed in a u32.
-pub(crate) fn u8_lookup_fv(kv: u32) -> u32 {
-    kv & 0xff
+#[inline]
+pub(crate) fn u8_lookup_fv(kv: u32) -> u8 {
+    (kv & 0xff) as u8
+}
+
+/// Extract the key for a boolean lookup.
+#[inline]
+pub(crate) fn bool_lookup_fk(kv: u32) -> u32 {
+    kv
+}
+
+/// Extract the value for a boolean lookup.
+#[inline]
+pub(crate) fn bool_lookup_fv(_kv: u32) -> bool {
+    true
 }
diff --git a/src/tables.rs b/src/tables.rs
