Add configurability for unicode and skipping parsers (#14)

* Add configurability for unicode and skipping parsers

- Unicode support greatly increases memory and latency and is not
generally needed.
- Compiling regex for unused parsers is just a waste of memory.

Both unicode support and device/os/user_agent parsers can now be
configured via `UserAgentParserBuilder`.

* Hide UserAgentParserBuilder behind builder method

* Update tests/bench/examples with new builder syntax

---------

Co-authored-by: David Armstrong Lewis <6754950+davidarmstronglewis@users.noreply.github.com>

Cargo.toml

@@ -15,7 +15,7 @@ keywords      = ["user", "agent", "parser", "uap", "uaparser"]
 
 [dependencies]
 lazy_static = "1.4.0"
-regex = "1.5.5"
+regex = "1.9.1"
 serde = "1.0.137"
 serde_yaml = "0.8.24"
 serde_derive = "1.0.137"

README.md

@@ -13,3 +13,13 @@ To get to the docs, clone the repo and run `cargo doc --open` to build the docum
 - `git submodule update --init` to get started
 - `cargo test`
 - `cargo test -- --nocapture` for the full results
+
+## Performance and Benchmarking
+`cargo bench` will run a criterion benchmark suite.
+
+To see memory usage of the compiled regex list you can run the examples with a tool that tracks memory usage.
+
+Example (on MacOS):
+```
+/usr/bin/time -l cargo run --examples full_parser
+```

benches/benchmark.rs

@@ -15,7 +15,11 @@ struct TestCases {
 }
 
 fn bench_os(c: &mut Criterion) {
-    let parser = UserAgentParser::from_yaml("./src/core/regexes.yaml")
+    let parser = UserAgentParser::builder()
+        .with_device(false)
+        .with_os(true)
+        .with_user_agent(false)
+        .build_from_yaml("./src/core/regexes.yaml")
         .expect("Parser creation failed");
 
     let file = File::open("./src/core/tests/test_os.yaml").unwrap();
@@ -28,10 +32,30 @@ fn bench_os(c: &mut Criterion) {
             }
         })
     });
+
+    let parser = UserAgentParser::builder()
+        .with_device(false)
+        .with_os(true)
+        .with_user_agent(false)
+        .with_unicode_support(false)
+        .build_from_yaml("./src/core/regexes.yaml")
+        .expect("Parser creation failed");
+
+    c.bench_function("parse_os unicode disabled", |b| {
+        b.iter(|| {
+            for case in &test_cases.test_cases {
+                black_box(parser.parse_os(&case.user_agent_string));
+            }
+        })
+    });
 }
 
 fn bench_device(c: &mut Criterion) {
-    let parser = UserAgentParser::from_yaml("./src/core/regexes.yaml")
+    let parser = UserAgentParser::builder()
+        .with_device(true)
+        .with_os(false)
+        .with_user_agent(false)
+        .build_from_yaml("./src/core/regexes.yaml")
         .expect("Parser creation failed");
 
     let file = File::open("./src/core/tests/test_device.yaml").unwrap();
@@ -44,13 +68,33 @@ fn bench_device(c: &mut Criterion) {
             }
         })
     });
+
+    let parser = UserAgentParser::builder()
+        .with_device(true)
+        .with_os(false)
+        .with_user_agent(false)
+        .with_unicode_support(false)
+        .build_from_yaml("./src/core/regexes.yaml")
+        .expect("Parser creation failed");
+
+    c.bench_function("parse_device unicode disabled", |b| {
+        b.iter(|| {
+            for case in &test_cases.test_cases {
+                black_box(parser.parse_device(&case.user_agent_string));
+            }
+        })
+    });
 }
 
 fn bench_ua(c: &mut Criterion) {
-    let parser = UserAgentParser::from_yaml("./src/core/regexes.yaml")
+    let parser = UserAgentParser::builder()
+        .with_device(false)
+        .with_os(false)
+        .with_user_agent(true)
+        .build_from_yaml("./src/core/regexes.yaml")
         .expect("Parser creation failed");
 
-    let file = std::fs::File::open("./src/core/tests/test_ua.yaml").unwrap();
+    let file = File::open("./src/core/tests/test_ua.yaml").unwrap();
     let test_cases: TestCases = serde_yaml::from_reader(file).unwrap();
 
     c.bench_function("parse_user_agent", |b| {
@@ -60,6 +104,22 @@ fn bench_ua(c: &mut Criterion) {
             }
         })
     });
+
+    let parser = UserAgentParser::builder()
+        .with_device(false)
+        .with_os(false)
+        .with_user_agent(true)
+        .with_unicode_support(false)
+        .build_from_yaml("./src/core/regexes.yaml")
+        .expect("Parser creation failed");
+
+    c.bench_function("parse_user_agent unicode disabled", |b| {
+        b.iter(|| {
+            for case in &test_cases.test_cases {
+                black_box(parser.parse_user_agent(&case.user_agent_string));
+            }
+        })
+    });
 }
 
 criterion_group!(

examples/full_parser.rs

@@ -0,0 +1,9 @@
+use uaparser::{Parser, UserAgentParser};
+
+fn main() {
+    let parser = UserAgentParser::builder()
+        .build_from_yaml("./src/core/regexes.yaml")
+        .expect("Parser creation failed");
+
+    println!("{:?}", parser.parse("Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/114.0.0.0 Safari/537.36"))
+}

examples/no_unicode_parser.rs

@@ -0,0 +1,10 @@
+use uaparser::{Parser, UserAgentParser};
+
+fn main() {
+    let parser = UserAgentParser::builder()
+        .with_unicode_support(false)
+        .build_from_yaml("./src/core/regexes.yaml")
+        .expect("Parser creation failed");
+
+    println!("{:?}", parser.parse("Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/114.0.0.0 Safari/537.36"))
+}

examples/os_only_no_unicode_parser.rs

@@ -0,0 +1,13 @@
+use uaparser::{Parser, UserAgentParser};
+
+fn main() {
+    let parser = UserAgentParser::builder()
+        .with_unicode_support(false)
+        .with_device(false)
+        .with_os(true)
+        .with_user_agent(false)
+        .build_from_yaml("./src/core/regexes.yaml")
+        .expect("Parser creation failed");
+
+    println!("{:?}", parser.parse("Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/114.0.0.0 Safari/537.36"))
+}

nix/shell.nix

@@ -6,10 +6,15 @@ mkShell {
 
   nativeBuildInputs = [
     (rust-bin.stable.latest.default.override {
-      extensions = [ "rust-src" ];
+      extensions = [
+        "rust-src"
+        "rust-analysis"
+        "clippy"
+      ];
     })
     cargo-criterion
     cargo-edit
+    cargo-watch
     gnuplot
   ];
 }

src/client.rs

@@ -1,4 +1,7 @@
-use super::{Deserialize, Device, Serialize, UserAgent, OS};
+use super::Device;
+use super::UserAgent;
+use super::OS;
+use super::{Deserialize, Serialize};
 
 /// Houses the `Device`, `OS`, and `UserAgent` structs, which each get parsed
 /// out from a user agent string by a `UserAgentParser`.