Add and use new string-cache-codegen crate

Author: aidanhs

.travis.yml

@@ -13,6 +13,7 @@ script:
   - cargo test --features log-events
   - "if [ $TRAVIS_RUST_VERSION = nightly ]; then cargo test --features unstable; fi"
   - cargo test --features heapsize
+  - "cd string-cache-codegen/ && cargo build && cd .."
   - "cd examples/event-log/ && cargo build && cd ../.."
   - "cd examples/summarize-events/ && cargo build && cd ../.."
 notifications:

Cargo.toml

@@ -9,6 +9,12 @@ repository = "https://github.com/servo/string-cache"
 documentation = "https://docs.rs/string_cache/"
 build = "build.rs"
 
+# Do not `exclude` ./string-cache-codegen because we want to include
+# ./string-cache-codegen/shared.rs, and `include` is a pain to use
+# (It has to be exhaustive.)
+# This means that packages for this crate include some unused files,
+# but they’re not too big so that shouldn’t be a problem.
+
 [lib]
 name = "string_cache"
 
@@ -36,4 +42,4 @@ heapsize = { version = "0.3", optional = true }
 rand = "0.3"
 
 [build-dependencies]
-phf_generator = "0.7.4"
+string_cache_codegen = { version = "0.3", path = "./string-cache-codegen" }

build.rs

@@ -1,62 +1,14 @@
-extern crate phf_generator;
+extern crate string_cache_codegen;
 
-#[path = "src/shared.rs"] #[allow(dead_code)] mod shared;
-#[path = "src/static_atom_list.rs"] mod static_atom_list;
+#[path = "src/static_atom_list.rs"]
+mod static_atom_list;
 
 use std::env;
-use std::fs::File;
-use std::io::{BufWriter, Write};
 use std::path::Path;
 
 fn main() {
-    let hash_state = generate();
-    write_static_atom_set(&hash_state);
-    write_atom_macro(&hash_state);
-}
-
-fn generate() -> phf_generator::HashState {
-    let mut set = std::collections::HashSet::new();
-    for atom in static_atom_list::ATOMS {
-        if !set.insert(atom) {
-            panic!("duplicate static atom `{:?}`", atom);
-        }
-    }
-    phf_generator::generate_hash(static_atom_list::ATOMS)
-}
-
-fn write_static_atom_set(hash_state: &phf_generator::HashState) {
-    let path = Path::new(&std::env::var("OUT_DIR").unwrap()).join("static_atom_set.rs");
-    let mut file = BufWriter::new(File::create(&path).unwrap());
-    macro_rules! w {
-        ($($arg: expr),+) => { (writeln!(&mut file, $($arg),+).unwrap()) }
-    }
-    w!("pub static STATIC_ATOM_SET: PhfStrSet = PhfStrSet {{");
-    w!("    key: {},", hash_state.key);
-    w!("    disps: &[");
-    for &(d1, d2) in &hash_state.disps {
-        w!("        ({}, {}),", d1, d2);
-    }
-    w!("    ],");
-    w!("    atoms: &[");
-    for &idx in &hash_state.map {
-        w!("        {:?},", static_atom_list::ATOMS[idx]);
-    }
-    w!("    ],");
-    w!("}};");
-}
-
-fn write_atom_macro(hash_state: &phf_generator::HashState) {
-    let path = Path::new(&env::var("OUT_DIR").unwrap()).join("atom_macro.rs");
-    let mut file = BufWriter::new(File::create(&path).unwrap());
-    writeln!(file, r"#[macro_export]").unwrap();
-    writeln!(file, r"macro_rules! atom {{").unwrap();
-    for (i, &idx) in hash_state.map.iter().enumerate() {
-        writeln!(
-            file,
-            r"({:?}) => {{ $crate::Atom {{ unsafe_data: 0x{:x}, phantom: ::std::marker::PhantomData }} }};",
-            static_atom_list::ATOMS[idx],
-            shared::pack_static(i as u32),
-        ).unwrap();
-    }
-    writeln!(file, r"}}").unwrap();
+    string_cache_codegen::AtomType::new("atom::tests::TestAtom", "test_atom!")
+        .atoms(static_atom_list::ATOMS)
+        .write_to_file(&Path::new(&env::var("OUT_DIR").unwrap()).join("test_atom.rs"))
+        .unwrap()
 }

examples/summarize-events/src/main.rs

@@ -12,7 +12,7 @@ extern crate string_cache;
 extern crate rustc_serialize;
 extern crate phf_shared;
 
-#[path = "../../../src/shared.rs"]
+#[path = "../../../string-cache-codegen/shared.rs"]
 #[allow(dead_code)]
 mod shared;
 

src/atom.rs

@@ -37,8 +37,6 @@ use self::UnpackedAtom::{Dynamic, Inline, Static};
 #[cfg(feature = "log-events")]
 use event::Event;
 
-include!(concat!(env!("OUT_DIR"), "/static_atom_set.rs"));
-
 #[cfg(not(feature = "log-events"))]
 macro_rules! log (($e:expr) => (()));
 
@@ -199,7 +197,7 @@ impl StaticAtomSet for EmptyStaticAtomSet {
 pub type DefaultAtom = Atom<EmptyStaticAtomSet>;
 
 pub struct Atom<Static: StaticAtomSet> {
-    /// This field is public so that the `atom!()` macro can use it.
+    /// This field is public so that the `atom!()` macros can use it.
     /// You should not otherwise access this field.
     #[doc(hidden)]
     pub unsafe_data: u64,
@@ -234,7 +232,7 @@ impl<Static: StaticAtomSet> Atom<Static> {
 
 impl<Static: StaticAtomSet> Default for Atom<Static> {
     fn default() -> Self {
-        atom!("")
+        Self::from("")
     }
 }
 
@@ -581,26 +579,17 @@ fn copy_memory(src: &[u8], dst: &mut [u8]) {
     }
 }
 
-#[cfg(all(test, feature = "unstable"))]
-#[path = "bench.rs"]
-mod bench;
-
 #[cfg(test)]
+#[macro_use]
 mod tests {
     use std::mem;
     use std::thread;
-    use super::Atom as GenericAtom;
-    use super::{StaticAtomSet, StringCacheEntry, STATIC_ATOM_SET, PhfStrSet};
+    use super::{StaticAtomSet, StringCacheEntry};
     use super::UnpackedAtom::{Dynamic, Inline, Static};
     use shared::ENTRY_ALIGNMENT;
 
-    pub type Atom = GenericAtom<DefaultStatic>;
-    pub struct DefaultStatic;
-    impl StaticAtomSet for DefaultStatic {
-        fn get() -> &'static PhfStrSet {
-            &STATIC_ATOM_SET
-        }
-    }
+    include!(concat!(env!("OUT_DIR"), "/test_atom.rs"));
+    pub type Atom = TestAtom;
 
     #[test]
     fn test_as_slice() {
@@ -741,17 +730,17 @@ mod tests {
             assert_eq_fmt!("0x{:016X}", x.unsafe_data, Atom::from(s).unsafe_data);
             assert_eq!(0x2, x.unsafe_data & 0xFFFF_FFFF);
             // The index is unspecified by phf.
-            assert!((x.unsafe_data >> 32) <= DefaultStatic::get().atoms.len() as u64);
+            assert!((x.unsafe_data >> 32) <= TestAtomStaticSet::get().atoms.len() as u64);
         }
 
         // This test is here to make sure we don't change atom representation
         // by accident.  It may need adjusting if there are changes to the
         // static atom table, the tag values, etc.
 
         // Static atoms
-        check_static("a",       atom!("a"));
-        check_static("address", atom!("address"));
-        check_static("area",    atom!("area"));
+        check_static("a",       test_atom!("a"));
+        check_static("address", test_atom!("address"));
+        check_static("area",    test_atom!("area"));
 
         // Inline atoms
         check("e",       0x0000_0000_0000_6511);
@@ -789,27 +778,27 @@ mod tests {
 
     #[test]
     fn atom_macro() {
-        assert_eq!(atom!("body"), Atom::from("body"));
-        assert_eq!(atom!("font-weight"), Atom::from("font-weight"));
+        assert_eq!(test_atom!("body"), Atom::from("body"));
+        assert_eq!(test_atom!("font-weight"), Atom::from("font-weight"));
     }
 
     #[test]
     fn match_atom() {
         assert_eq!(2, match Atom::from("head") {
-            atom!("br") => 1,
-            atom!("html") | atom!("head") => 2,
+            test_atom!("br") => 1,
+            test_atom!("html") | test_atom!("head") => 2,
             _ => 3,
         });
 
         assert_eq!(3, match Atom::from("body") {
-            atom!("br") => 1,
-            atom!("html") | atom!("head") => 2,
+            test_atom!("br") => 1,
+            test_atom!("html") | test_atom!("head") => 2,
             _ => 3,
         });
 
         assert_eq!(3, match Atom::from("zzzzzz") {
-            atom!("br") => 1,
-            atom!("html") | atom!("head") => 2,
+            test_atom!("br") => 1,
+            test_atom!("html") | test_atom!("head") => 2,
             _ => 3,
         });
     }
@@ -868,3 +857,7 @@ mod tests {
         assert!(Atom::from("camembert".to_owned()) == Atom::from("camembert"));
     }
 }
+
+#[cfg(all(test, feature = "unstable"))]
+#[path = "bench.rs"]
+mod bench;

src/bench.rs

@@ -27,20 +27,12 @@ and cheap to move around, which isn't reflected in these tests.
 
 */
 
-use atom::tests::Atom;
+use atom::tests::TestAtom;
 use test::{Bencher, black_box};
 
-macro_rules! test_atom {
-    ($tt: tt) => {{
-        // Add type annotation to help inference
-        let atom: Atom = atom!($tt);
-        atom
-    }}
-}
-
 // Just shorthand
-fn mk(x: &str) -> Atom {
-    Atom::from(x)
+fn mk(x: &str) -> TestAtom {
+    TestAtom::from(x)
 }
 
 macro_rules! check_type (($name:ident, $x:expr, $p:pat) => (
@@ -89,7 +81,7 @@ macro_rules! bench_one (
         fn intern(b: &mut Bencher) {
             let x = $x.to_string();
             b.iter(|| {
-                black_box(Atom::from(&*x));
+                black_box(TestAtom::from(&*x));
             });
         }
     );
@@ -142,7 +134,7 @@ macro_rules! bench_all (
             use std::string::ToString;
             use std::iter::repeat;
 
-            use atom::tests::Atom;
+            use atom::tests::TestAtom;
             use atom::UnpackedAtom::{Static, Inline, Dynamic};
 
             use super::mk;
@@ -213,7 +205,7 @@ macro_rules! bench_rand ( ($name:ident, $len:expr) => (
                 *n = (*n % 0x40) + 0x20;
             }
             let s = str::from_utf8(&buf[..]).unwrap();
-            black_box(Atom::from(s));
+            black_box(TestAtom::from(s));
         });
     }
 ));

src/lib.rs

@@ -24,11 +24,17 @@ extern crate phf_shared;
 
 pub use atom::{Atom, StaticAtomSet, PhfStrSet, EmptyStaticAtomSet, DefaultAtom};
 
-include!(concat!(env!("OUT_DIR"), "/atom_macro.rs"));
-
 #[cfg(feature = "log-events")]
 #[macro_use]
 pub mod event;
 
 pub mod atom;
-pub mod shared;
+
+#[path = "../string-cache-codegen/shared.rs"]
+mod shared;
+
+// Make test_atom! macro work in this crate.
+// `$crate` would not be appropriate for other crates creating such macros
+mod string_cache {
+    pub use {Atom, StaticAtomSet, PhfStrSet};
+}

string-cache-codegen/Cargo.toml

@@ -0,0 +1,16 @@
+[package]
+
+name = "string_cache_codegen"
+version = "0.3.0"
+authors = [ "The Servo Project Developers" ]
+description = "A codegen library for string-cache, developed as part of the Servo project."
+license = "MIT / Apache-2.0"
+repository = "https://github.com/servo/string-cache"
+documentation = "https://docs.rs/string_cache_codegen/"
+
+[lib]
+name = "string_cache_codegen"
+path = "lib.rs"
+
+[dependencies]
+phf_generator = "0.7.15"