Rewrite printable.py codegen to emit C++

support/printable.py

@@ -1,8 +1,11 @@
 #!/usr/bin/env python3
 
+# This script is based on
+# https://github.com/rust-lang/rust/blob/master/library/core/src/unicode/printable.py
+# distributed under https://github.com/rust-lang/rust/blob/master/LICENSE-MIT.
+
 # This script uses the following Unicode tables:
 # - UnicodeData.txt
-# Script license: https://github.com/rust-lang/rust/blob/master/LICENSE-MIT
 
 
 from collections import namedtuple
@@ -112,23 +115,20 @@ def compress_normal(normal):
     return compressed
 
 def print_singletons(uppers, lowers, uppersname, lowersname):
-    print("#[rustfmt::skip]")
-    print("const {}: &[(u8, u8)] = &[".format(uppersname))
+    print("  static constexpr singleton {}[] = {{".format(uppersname))
     for u, c in uppers:
-        print("    ({:#04x}, {}),".format(u, c))
-    print("];")
-    print("#[rustfmt::skip]")
-    print("const {}: &[u8] = &[".format(lowersname))
+        print("    {{{:#04x}, {}}},".format(u, c))
+    print("  };")
+    print("  static constexpr unsigned char {}[] = {{".format(lowersname))
     for i in range(0, len(lowers), 8):
         print("    {}".format(" ".join("{:#04x},".format(l) for l in lowers[i:i+8])))
-    print("];")
+    print("  };")
 
 def print_normal(normal, normalname):
-    print("#[rustfmt::skip]")
-    print("const {}: &[u8] = &[".format(normalname))
+    print("  static constexpr unsigned char {}[] = {{".format(normalname))
     for v in normal:
         print("    {}".format(" ".join("{:#04x},".format(i) for i in v)))
-    print("];")
+    print("  };")
 
 def main():
     file = get_file("https://www.unicode.org/Public/UNIDATA/UnicodeData.txt")
@@ -171,67 +171,66 @@ def main():
     normal1 = compress_normal(normal1)
 
     print("""\
-// NOTE: The following code was generated by "src/libcore/unicode/printable.py",
-//       do not edit directly!
+struct singleton {
+  unsigned char upper;
+  unsigned char lowercount;
+};
 
-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;
+inline auto check(uint16_t x, const singleton* singletonuppers,
+                  size_t singletonuppers_size,
+                  const unsigned char* singletonlowers,
+                  const unsigned char* normal, size_t normal_size) -> bool {
+  auto xupper = x >> 8;
+  auto lowerstart = 0;
+  for (size_t i = 0; i < singletonuppers_size; ++i) {
+    auto su = singletonuppers[i];
+    auto lowerend = lowerstart + su.lowercount;
+    if (xupper < su.upper) break;
+    if (xupper == su.upper) {
+      for (auto j = lowerstart; j < lowerend; ++j) {
+        if (singletonlowers[j] == x) return false;
+      }
     }
+    lowerstart = lowerend;
+  }
 
-    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;
-    }
-    current
+  auto xsigned = static_cast<int>(x);
+  auto current = true;
+  for (size_t i = 0; i < normal_size; ++i) {
+    auto v = static_cast<int>(normal[i]);
+    auto len = v & 0x80 != 0 ? (v & 0x7f) << 8 | normal[i++] : v;
+    xsigned -= len;
+    if (xsigned < 0) break;
+    current = !current;
+  }
+  return current;
 }
 
-pub(crate) fn is_printable(x: char) -> bool {
-    let x = x as u32;
-    let lower = x as u16;
-    if x < 0x10000 {
-        check(lower, SINGLETONS0U, SINGLETONS0L, NORMAL0)
-    } else if x < 0x20000 {
-        check(lower, SINGLETONS1U, SINGLETONS1L, NORMAL1)
-    } else {\
+inline auto is_printable(uint32_t cp) -> bool {\
+""")
+    print_singletons(singletons0u, singletons0l, 'singletons0u', 'singletons0l')
+    print_singletons(singletons1u, singletons1l, 'singletons1u', 'singletons1l')
+    print_normal(normal0, 'normal0')
+    print_normal(normal1, 'normal1')
+    print("""\
+  auto lower = static_cast<uint16_t>(cp);
+  if (cp < 0x10000) {
+    return check(lower, singletons0u,
+                 sizeof(singletons0u) / sizeof(*singletons0u), singletons0l,
+                 normal0, sizeof(normal0));
+  }
+  if (cp < 0x20000) {
+    return check(lower, singletons1u,
+                 sizeof(singletons1u) / sizeof(*singletons1u), singletons1l,
+                 normal1, sizeof(normal1));
+  }\
 """)
     for a, b in extra:
-        print("        if 0x{:x} <= x && x < 0x{:x} {{".format(a, a + b))
-        print("            return false;")
-        print("        }")
+        print("  if (0x{:x} <= cp && cp < 0x{:x}) return false;".format(a, a + b))
     print("""\
-        true
-    }
+  return true;
 }\
 """)
-    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()