Reduce a table used for Debug impl of str.

This commit shrinks the size of the aforementioned table from
2,102 bytes to 1,197 bytes. This is achieved by an observation that
most u16 entries are common in its upper byte. Specifically:

- SINGLETONS now uses two tables, one for (upper byte, lower count)
  and another for a series of lower bytes. For each upper byte given
  number of lower bytes are read and compared.

- NORMAL now uses a variable length format for the count of "true"
  codepoints and "false" codepoints (one byte with MSB unset, or
  two big-endian bytes with the first MSB set).

The code size and relative performance roughly remains same as this
commit tries to optimize for both. The new table and algorithm has
been verified for the equivalence to older ones.

Author: lifthrasiir

src/etc/char_private.py

@@ -76,6 +76,66 @@ def get_codepoints(f):
     for c in range(prev_codepoint + 1, NUM_CODEPOINTS):
         yield Codepoint(c, None)
 
+def compress_singletons(singletons):
+    uppers = [] # (upper, # items in lowers)
+    lowers = []
+
+    for i in singletons:
+        upper = i >> 8
+        lower = i & 0xff
+        if len(uppers) == 0 or uppers[-1][0] != upper:
+            uppers.append((upper, 1))
+        else:
+            upper, count = uppers[-1]
+            uppers[-1] = upper, count + 1
+        lowers.append(lower)
+
+    return uppers, lowers
+
+def compress_normal(normal):
+    # lengths 0x00..0x7f are encoded as 00, 01, ..., 7e, 7f
+    # lengths 0x80..0x7fff are encoded as 80 80, 80 81, ..., ff fe, ff ff
+    compressed = [] # [truelen, (truelenaux), falselen, (falselenaux)]
+
+    prev_start = 0
+    for start, count in normal:
+        truelen = start - prev_start
+        falselen = count
+        prev_start = start + count
+
+        assert truelen < 0x8000 and falselen < 0x8000
+        entry = []
+        if truelen > 0x7f:
+            entry.append(0x80 | (truelen >> 8))
+            entry.append(truelen & 0xff)
+        else:
+            entry.append(truelen & 0x7f)
+        if falselen > 0x7f:
+            entry.append(0x80 | (falselen >> 8))
+            entry.append(falselen & 0xff)
+        else:
+            entry.append(falselen & 0x7f)
+
+        compressed.append(entry)
+
+    return compressed
+
+def print_singletons(uppers, lowers, uppersname, lowersname):
+    print("const {}: &'static [(u8, u8)] = &[".format(uppersname))
+    for u, c in uppers:
+        print("    ({:#04x}, {}),".format(u, c))
+    print("];")
+    print("const {}: &'static [u8] = &[".format(lowersname))
+    for i in range(0, len(lowers), 8):
+        print("    {}".format(" ".join("{:#04x},".format(l) for l in lowers[i:i+8])))
+    print("];")
+
+def print_normal(normal, normalname):
+    print("const {}: &'static [u8] = &[".format(normalname))
+    for v in normal:
+        print("    {}".format(" ".join("{:#04x},".format(i) for i in v)))
+    print("];")
+
 def main():
     file = get_file("http://www.unicode.org/Public/UNIDATA/UnicodeData.txt")
 
@@ -111,6 +171,11 @@ def main():
             else:
                 normal0.append((a, b - a))
 
+    singletons0u, singletons0l = compress_singletons(singletons0)
+    singletons1u, singletons1l = compress_singletons(singletons1)
+    normal0 = compress_normal(normal0)
+    normal1 = compress_normal(normal1)
+
     print("""\
 // Copyright 2012-2016 The Rust Project Developers. See the COPYRIGHT
 // file at the top-level directory of this distribution and at
@@ -125,38 +190,49 @@ def main():
 // NOTE: The following code was generated by "src/etc/char_private.py",
 //       do not edit directly!
 
-use slice::SliceExt;
-
-fn check(x: u16, singletons: &[u16], normal: &[u16]) -> bool {
-    for &s in singletons {
-        if x == s {
-            return false;
-        } else if x < s {
+fn check(x: u16, singletonuppers: &[(u8, u8)], singletonlowers: &[u8],
+         normal: &[u8]) -> bool {
+    let xupper = (x >> 8) as u8;
+    let mut lowerstart = 0;
+    for &(upper, lowercount) in singletonuppers {
+        let lowerend = lowerstart + lowercount as usize;
+        if xupper == upper {
+            for &lower in &singletonlowers[lowerstart..lowerend] {
+                if lower == x as u8 {
+                    return false;
+                }
+            }
+        } else if xupper < upper {
             break;
         }
+        lowerstart = lowerend;
     }
-    for w in normal.chunks(2) {
-        let start = w[0];
-        let len = w[1];
-        let difference = (x as i32) - (start as i32);
-        if 0 <= difference {
-            if difference < len as i32 {
-                return false;
-            }
+
+    let mut x = x as i32;
+    let mut normal = normal.iter().cloned();
+    let mut current = true;
+    while let Some(v) = normal.next() {
+        let len = if v & 0x80 != 0 {
+            ((v & 0x7f) as i32) << 8 | normal.next().unwrap() as i32
         } else {
+            v as i32
+        };
+        x -= len;
+        if x < 0 {
             break;
         }
+        current = !current;
     }
-    true
+    current
 }
 
 pub fn is_printable(x: char) -> bool {
     let x = x as u32;
     let lower = x as u16;
     if x < 0x10000 {
-        check(lower, SINGLETONS0, NORMAL0)
+        check(lower, SINGLETONS0U, SINGLETONS0L, NORMAL0)
     } else if x < 0x20000 {
-        check(lower, SINGLETONS1, NORMAL1)
+        check(lower, SINGLETONS1U, SINGLETONS1L, NORMAL1)
     } else {\
 """)
     for a, b in extra:
@@ -169,22 +245,10 @@ def main():
 }\
 """)
     print()
-    print("const SINGLETONS0: &'static [u16] = &[")
-    for s in singletons0:
-        print("    0x{:x},".format(s))
-    print("];")
-    print("const SINGLETONS1: &'static [u16] = &[")
-    for s in singletons1:
-        print("    0x{:x},".format(s))
-    print("];")
-    print("const NORMAL0: &'static [u16] = &[")
-    for a, b in normal0:
-        print("    0x{:x}, 0x{:x},".format(a, b))
-    print("];")
-    print("const NORMAL1: &'static [u16] = &[")
-    for a, b in normal1:
-        print("    0x{:x}, 0x{:x},".format(a, b))
-    print("];")
+    print_singletons(singletons0u, singletons0l, 'SINGLETONS0U', 'SINGLETONS0L')
+    print_singletons(singletons1u, singletons1l, 'SINGLETONS1U', 'SINGLETONS1L')
+    print_normal(normal0, 'NORMAL0')
+    print_normal(normal1, 'NORMAL1')
 
 if __name__ == '__main__':
     main()