Merge pull request #280 from epage/bsearch

chore: Reduce code-gen memory usage

Cargo.lock

@@ -222,8 +222,6 @@ version = "0.4.0"
 dependencies = [
  "codegenrs",
  "itertools 0.10.0",
- "phf",
- "phf_codegen",
  "structopt",
  "unicase",
 ]
@@ -233,7 +231,6 @@ name = "codespell-dict"
 version = "0.4.0"
 dependencies = [
  "log",
- "phf",
  "unicase",
 ]
 
@@ -808,8 +805,6 @@ version = "0.4.0"
 dependencies = [
  "codegenrs",
  "itertools 0.10.0",
- "phf",
- "phf_codegen",
  "regex",
  "structopt",
  "unicase",
@@ -820,7 +815,6 @@ name = "misspell-dict"
 version = "0.4.0"
 dependencies = [
  "log",
- "phf",
  "unicase",
 ]
 
@@ -1778,8 +1772,6 @@ version = "0.4.0"
 dependencies = [
  "codegenrs",
  "itertools 0.10.0",
- "phf",
- "phf_codegen",
  "structopt",
  "unicase",
 ]
@@ -1789,7 +1781,6 @@ name = "wikipedia-dict"
 version = "0.4.0"
 dependencies = [
  "log",
- "phf",
  "unicase",
 ]
 

crates/codespell-dict/Cargo.toml

@@ -15,7 +15,9 @@ publish = false
 azure-devops = { project = "crate-ci", pipeline = "typos" }
 codecov = { repository = "crate-ci/typos" }
 
+[package.metadata.release]
+disable-release = true
+
 [dependencies]
-phf = { version = "0.8", features = ["unicase"] }
 unicase = "2.5"
 log = "0.4"

crates/codespell-dict/codegen/Cargo.toml

@@ -14,9 +14,10 @@ publish = false
 azure-devops = { project = "crate-ci", pipeline = "typos" }
 codecov = { repository = "crate-ci/typos" }
 
+[package.metadata.release]
+disable-release = true
+
 [dependencies]
-phf = { version = "0.8", features = ["unicase"] }
-phf_codegen = "0.8"
 unicase = "2.5"
 itertools = "0.10"
 codegenrs = "1.0"

crates/codespell-dict/codegen/src/main.rs

@@ -32,24 +32,18 @@ fn generate<W: std::io::Write>(file: &mut W) {
     .unwrap();
     writeln!(file, "#![allow(clippy::unreadable_literal)]",).unwrap();
     writeln!(file).unwrap();
-    writeln!(file, "use unicase::UniCase;").unwrap();
 
     let dict = parse_dict(DICT);
 
-    writeln!(
+    writeln!(file, "pub static WORD_DICTIONARY: &[(&str, &[&str])] = &[").unwrap();
-        file,
-        "pub static WORD_DICTIONARY: phf::Map<unicase::UniCase<&'static str>, &[&'static str]> = ",
-    )
-    .unwrap();
-    let mut builder = phf_codegen::Map::new();
     for (typo, corrections) in dict {
         let value = itertools::join(corrections.iter().map(|s| format!("{:?}", s)), ", ");
         let value = format!("&[{}]", value);
-        builder.entry(unicase::UniCase::new(typo), &value);
+
+        let key = format!("{:?}", typo);
+        writeln!(file, "  ({}, {}),", key, &value).unwrap();
     }
-    let codegenned = builder.build();
+    writeln!(file, "];").unwrap();
-    writeln!(file, "{}", codegenned).unwrap();
-    writeln!(file, ";").unwrap();
 }
 
 #[derive(Debug, StructOpt)]

crates/misspell-dict/Cargo.toml

@@ -15,7 +15,9 @@ publish = false
 azure-devops = { project = "crate-ci", pipeline = "typos" }
 codecov = { repository = "crate-ci/typos" }
 
+[package.metadata.release]
+disable-release = true
+
 [dependencies]
-phf = { version = "0.8", features = ["unicase"] }
 unicase = "2.5"
 log = "0.4"

crates/misspell-dict/codegen/Cargo.toml

@@ -14,9 +14,10 @@ publish = false
 azure-devops = { project = "crate-ci", pipeline = "typos" }
 codecov = { repository = "crate-ci/typos" }
 
+[package.metadata.release]
+disable-release = true
+
 [dependencies]
-phf = { version = "0.8", features = ["unicase"] }
-phf_codegen = "0.8"
 unicase = "2.5"
 itertools = "0.10"
 codegenrs = "1.0"

crates/misspell-dict/codegen/src/main.rs

@@ -64,60 +64,70 @@ fn generate<W: std::io::Write>(file: &mut W) {
     .unwrap();
     writeln!(file, "#![allow(clippy::unreadable_literal)]",).unwrap();
     writeln!(file).unwrap();
-    writeln!(file, "use unicase::UniCase;").unwrap();
 
     let Words {
         main,
         american,
         british,
     } = parse_dict(DICT);
+    let mut main: Vec<_> = main.into_iter().collect();
+    main.sort_unstable_by(|a, b| {
+        unicase::UniCase::new(a.0)
+            .partial_cmp(&unicase::UniCase::new(b.0))
+            .unwrap()
+    });
+    let mut american: Vec<_> = american.into_iter().collect();
+    american.sort_unstable_by(|a, b| {
+        unicase::UniCase::new(a.0)
+            .partial_cmp(&unicase::UniCase::new(b.0))
+            .unwrap()
+    });
+    let mut british: Vec<_> = british.into_iter().collect();
+    british.sort_unstable_by(|a, b| {
+        unicase::UniCase::new(a.0)
+            .partial_cmp(&unicase::UniCase::new(b.0))
+            .unwrap()
+    });
 
-    writeln!(
+    writeln!(file, "pub static MAIN_DICTIONARY: &[(&str, &[&str])] = &[").unwrap();
-        file,
+    for (typo, corrections) in main.into_iter() {
-        "pub static MAIN_DICTIONARY: phf::Map<unicase::UniCase<&'static str>, &[&'static str]> = ",
-    )
-    .unwrap();
-    let mut builder = phf_codegen::Map::new();
-    for (typo, corrections) in main {
         let value = itertools::join(corrections.iter().map(|s| format!("{:?}", s)), ", ");
         let value = format!("&[{}]", value);
-        builder.entry(unicase::UniCase::new(typo), &value);
+
+        let key = format!("{:?}", typo);
+        writeln!(file, "  ({}, {}),", key, &value).unwrap();
     }
-    let codegenned = builder.build();
+    writeln!(file, "];").unwrap();
-    writeln!(file, "{}", codegenned).unwrap();
-    writeln!(file, ";").unwrap();
     writeln!(file).unwrap();
 
     writeln!(
         file,
-        "pub static AMERICAN_DICTIONARY: phf::Map<unicase::UniCase<&'static str>, &[&'static str]> = ",
+        "pub static AMERICAN_DICTIONARY: &[(&str, &[&str])] = &["
     )
     .unwrap();
-    let mut builder = phf_codegen::Map::new();
+    for (typo, corrections) in american.into_iter() {
-    for (typo, corrections) in american {
         let value = itertools::join(corrections.iter().map(|s| format!("{:?}", s)), ", ");
         let value = format!("&[{}]", value);
-        builder.entry(unicase::UniCase::new(typo), &value);
+
+        let key = format!("{:?}", typo);
+        writeln!(file, "  ({}, {}),", key, &value).unwrap();
     }
-    let codegenned = builder.build();
+    writeln!(file, "];").unwrap();
-    writeln!(file, "{}", codegenned).unwrap();
-    writeln!(file, ";").unwrap();
     writeln!(file).unwrap();
 
     writeln!(
         file,
-        "pub static BRITISH_DICTIONARY: phf::Map<unicase::UniCase<&'static str>, &[&'static str]> = ",
+        "pub static BRITISH_DICTIONARY: &[(&str, &[&str])] = &["
     )
     .unwrap();
-    let mut builder = phf_codegen::Map::new();
+    for (typo, corrections) in british.into_iter() {
-    for (typo, corrections) in british {
         let value = itertools::join(corrections.iter().map(|s| format!("{:?}", s)), ", ");
         let value = format!("&[{}]", value);
-        builder.entry(unicase::UniCase::new(typo), &value);
+
+        let key = format!("{:?}", typo);
+        writeln!(file, "  ({}, {}),", key, &value).unwrap();
     }
-    let codegenned = builder.build();
+    writeln!(file, "];").unwrap();
-    writeln!(file, "{}", codegenned).unwrap();
-    writeln!(file, ";").unwrap();
 }
 
 #[derive(Debug, StructOpt)]

crates/wikipedia-dict/Cargo.toml

@@ -15,7 +15,9 @@ publish = false
 azure-devops = { project = "crate-ci", pipeline = "typos" }
 codecov = { repository = "crate-ci/typos" }
 
+[package.metadata.release]
+disable-release = true
+
 [dependencies]
-phf = { version = "0.8", features = ["unicase"] }
 unicase = "2.5"
 log = "0.4"

crates/wikipedia-dict/codegen/Cargo.toml

@@ -14,9 +14,10 @@ publish = false
 azure-devops = { project = "crate-ci", pipeline = "typos" }
 codecov = { repository = "crate-ci/typos" }
 
+[package.metadata.release]
+disable-release = true
+
 [dependencies]
-phf = { version = "0.8", features = ["unicase"] }
-phf_codegen = "0.8"
 unicase = "2.5"
 itertools = "0.10"
 codegenrs = "1.0"

crates/wikipedia-dict/codegen/src/main.rs

@@ -32,24 +32,18 @@ fn generate<W: std::io::Write>(file: &mut W) {
     .unwrap();
     writeln!(file, "#![allow(clippy::unreadable_literal)]",).unwrap();
     writeln!(file).unwrap();
-    writeln!(file, "use unicase::UniCase;").unwrap();
 
     let dict = parse_dict(DICT);
 
-    writeln!(
+    writeln!(file, "pub static WORD_DICTIONARY: &[(&str, &[&str])] = &[").unwrap();
-        file,
-        "pub static WORD_DICTIONARY: phf::Map<unicase::UniCase<&'static str>, &[&'static str]> = ",
-    )
-    .unwrap();
-    let mut builder = phf_codegen::Map::new();
     for (typo, corrections) in dict {
         let value = itertools::join(corrections.iter().map(|s| format!("{:?}", s)), ", ");
         let value = format!("&[{}]", value);
-        builder.entry(unicase::UniCase::new(typo), &value);
+
+        let key = format!("{:?}", typo);
+        writeln!(file, "  ({}, {}),", key, &value).unwrap();
     }
-    let codegenned = builder.build();
+    writeln!(file, "];").unwrap();
-    writeln!(file, "{}", codegenned).unwrap();
-    writeln!(file, ";").unwrap();
 }
 
 #[derive(Debug, StructOpt)]