Merge pull request #23 from maciejhirsz/0.7.1

0.7.1 performance boost, related to #6

Author: maciejhirsz

Cargo.toml

@@ -1,10 +1,8 @@
 [package]
 name = "json"
-version = "0.7.0"
+version = "0.7.1"
 authors = ["Maciej Hirsz <[email protected]>"]
 description = "JSON implementation in Rust"
 repository = "https://github.com/maciejhirsz/json-rust"
 documentation = "http://terhix.com/doc/json/"
 license = "MIT"
-
-[dependencies]

src/codegen.rs

@@ -1,122 +1,202 @@
 use JsonValue;
 
-pub struct Generator {
-    pub minify: bool,
-    code: String,
-    dent: u16,
-    spaces_per_indent: u16,
-}
+pub trait Generator {
+    fn new_line(&mut self) {}
 
-impl Generator {
-    pub fn new(minify: bool, spaces: u16) -> Self {
-        Generator {
-            minify: minify,
-            code: String::new(),
-            dent: 0,
-            spaces_per_indent: spaces
+    fn write(&mut self, slice: &[u8]);
+
+    fn write_min(&mut self, slice: &[u8], minslice: &[u8]);
+
+    fn write_char(&mut self, ch: u8);
+
+    fn indent(&mut self) {}
+
+    fn dedent(&mut self) {}
+
+    fn write_string(&mut self, string: &str) {
+        self.write_char(b'"');
+
+        for ch in string.bytes() {
+            match ch {
+                b'\\' | b'"' => {
+                    self.write_char(b'\\');
+                    self.write_char(ch);
+                },
+                b'\n' => self.write(b"\\n"),
+                b'\r' => self.write(b"\\r"),
+                b'\t' => self.write(b"\\t"),
+                0xC   => self.write(b"\\f"),
+                0x8   => self.write(b"\\b"),
+                _     => self.write_char(ch)
+            }
         }
+
+        self.write_char(b'"');
     }
 
-    pub fn new_line(&mut self) {
-        if !self.minify {
-            self.code.push('\n');
-            for _ in 0..(self.dent * self.spaces_per_indent) {
-                self.code.push(' ');
-            }
+    fn write_digits_from_u64(&mut self, mut num: u64, length: &mut u8) {
+        let digit = (num % 10) as u8;
+        num /= 10;
+        if num > 0 {
+            self.write_digits_from_u64(num, length);
         }
+        *length += 1;
+        self.write_char(digit + b'0');
     }
 
-    pub fn write_json(&mut self, json: &JsonValue) {
-        match *json {
-            JsonValue::String(ref string) => {
-                self.write_char('"');
-
-                for ch in string.chars() {
-                    match ch {
-                        '\\' | '"' => {
-                            self.write_char('\\');
-                            self.write_char(ch);
-                        },
-                        '\n'       => self.write("\\n"),
-                        '\r'       => self.write("\\r"),
-                        '\t'       => self.write("\\t"),
-                        '\u{000C}' => self.write("\\f"),
-                        '\u{0008}' => self.write("\\b"),
-                        _          => self.write_char(ch)
-                    }
-                }
+    fn write_number(&mut self, mut num: f64) {
+        let mut length = 0;
+        if num < 0.0 {
+            num = -num;
+            self.write_char(b'-');
+        }
 
-                self.write_char('"');
-            },
-            JsonValue::Number(ref number) => self.write(&number.to_string()),
-            JsonValue::Boolean(ref value) => self.write(if *value { "true" } else { "false" }),
-            JsonValue::Null               => self.write("null"),
+        self.write_digits_from_u64(num as u64, &mut length);
+
+        let mut fract = num.fract();
+        if fract < 1e-10 {
+            return;
+        }
+
+        fract *= 10.0;
+        self.write_char(b'.');
+        self.write_char((fract as u8) + b'0');
+        fract = fract.fract();
+        length += 2;
+
+        while length < 17 && fract > 0.01 {
+            fract *= 10.0;
+            self.write_char((fract as u8) + b'0');
+            fract = fract.fract();
+            length += 1;
+        }
+    }
+
+    fn write_json(&mut self, json: &JsonValue) {
+        match *json {
+            JsonValue::String(ref string) => self.write_string(string),
+            JsonValue::Number(ref number) => self.write_number(*number),
+            JsonValue::Boolean(ref value) => self.write(if *value { b"true" } else { b"false" }),
+            JsonValue::Null               => self.write(b"null"),
             JsonValue::Array(ref array)   => {
-                self.write_char('[');
+                self.write_char(b'[');
                 self.indent();
                 let mut first = true;
                 for item in array {
                     if first {
                         first = false;
                         self.new_line();
                     } else {
-                        self.write(",");
+                        self.write(b",");
                         self.new_line();
                     }
                     self.write_json(item);
                 }
                 self.dedent();
                 self.new_line();
-                self.write_char(']');
+                self.write_char(b']');
             },
             JsonValue::Object(ref object) => {
-                self.write_char('{');
+                self.write_char(b'{');
                 self.indent();
                 let mut first = true;
                 for (key, value) in object.iter() {
                     if first {
                         first = false;
                         self.new_line();
                     } else {
-                        self.write(",");
+                        self.write(b",");
                         self.new_line();
                     }
-                    self.write(&format!("{:?}", key));
-                    self.write_min(": ", ":");
+                    self.write_string(key);
+                    self.write_min(b": ", b":");
                     self.write_json(value);
                 }
                 self.dedent();
                 self.new_line();
-                self.write_char('}');
+                self.write_char(b'}');
             }
         }
     }
 
-    pub fn write(&mut self, slice: &str) {
-        self.code.push_str(slice);
+    fn consume(self) -> String;
+}
+
+pub struct DumpGenerator {
+    code: Vec<u8>,
+}
+
+impl DumpGenerator {
+    pub fn new() -> Self {
+        DumpGenerator {
+            code: Vec::with_capacity(1024),
+        }
+    }
+}
+
+impl Generator for DumpGenerator {
+    fn write(&mut self, slice: &[u8]) {
+        self.code.extend_from_slice(slice);
+    }
+
+    fn write_min(&mut self, _: &[u8], minslice: &[u8]) {
+        self.code.extend_from_slice(minslice);
+    }
+
+    fn write_char(&mut self, ch: u8) {
+        self.code.push(ch);
+    }
+
+    fn consume(self) -> String {
+        String::from_utf8(self.code).unwrap()
+    }
+}
+
+pub struct PrettyGenerator {
+    code: Vec<u8>,
+    dent: u16,
+    spaces_per_indent: u16,
+}
+
+impl PrettyGenerator {
+    pub fn new(spaces: u16) -> Self {
+        PrettyGenerator {
+            code: Vec::with_capacity(1024),
+            dent: 0,
+            spaces_per_indent: spaces
+        }
     }
+}
 
-    pub fn write_min(&mut self, slice: &str, minslice: &str) {
-        if self.minify {
-            self.write(minslice);
-        } else {
-            self.write(slice);
+impl Generator for PrettyGenerator {
+    fn new_line(&mut self) {
+        self.code.push(b'\n');
+        for _ in 0..(self.dent * self.spaces_per_indent) {
+            self.code.push(b' ');
         }
     }
 
-    pub fn write_char(&mut self, ch: char) {
+    fn write(&mut self, slice: &[u8]) {
+        self.code.extend_from_slice(slice);
+    }
+
+    fn write_min(&mut self, slice: &[u8], _: &[u8]) {
+        self.code.extend_from_slice(slice);
+    }
+
+    fn write_char(&mut self, ch: u8) {
         self.code.push(ch);
     }
 
-    pub fn indent(&mut self) {
+    fn indent(&mut self) {
         self.dent += 1;
     }
 
-    pub fn dedent(&mut self) {
+    fn dedent(&mut self) {
         self.dent -= 1;
     }
 
-    pub fn consume(self) -> String {
-        self.code
+    fn consume(self) -> String {
+        String::from_utf8(self.code).unwrap()
     }
 }

src/error.rs

@@ -1,13 +1,14 @@
 use parser::Token;
 use std::error::Error;
 use std::fmt;
+use std::char;
 
 #[derive(Debug)]
 pub enum JsonError {
     UnexpectedToken(String),
     UnexpectedCharacter(char),
     UnexpectedEndOfJson,
-    CantCastCodepointToCharacter(u32),
+    FailedUtf8Parsing,
     ArrayIndexOutOfBounds,
     WrongType(String),
     UndefinedField(String),
@@ -18,6 +19,12 @@ impl JsonError {
         JsonError::UnexpectedToken(format!("{:?}", token))
     }
 
+    pub fn unexpected_character(byte: u8) -> Self {
+        JsonError::UnexpectedCharacter(
+            char::from_u32(byte as u32).unwrap_or('?')
+        )
+    }
+
     pub fn wrong_type(expected: &str) -> Self {
         JsonError::WrongType(expected.into())
     }
@@ -32,13 +39,13 @@ impl fmt::Display for JsonError {
         use JsonError::*;
 
         match *self {
-            UnexpectedToken(ref s) => write!(f, "Unexpected token: {}", s),
-            UnexpectedCharacter(c) => write!(f, "Unexpected character: {}", c),
-            UnexpectedEndOfJson => write!(f, "Unexpected end of JSON"),
-            CantCastCodepointToCharacter(i) => write!(f, "Cannot cast this codepoint to a character: {}", i),
-            ArrayIndexOutOfBounds => write!(f, "Array index out of bounds!"),
-            WrongType(ref s) => write!(f, "Wrong type, expected: {}", s),
-            UndefinedField(ref s) => write!(f, "Undefined field: {}", s)
+            UnexpectedToken(ref s)  => write!(f, "Unexpected token: {}", s),
+            UnexpectedCharacter(ch) => write!(f, "Unexpected character: {}", ch),
+            UnexpectedEndOfJson     => write!(f, "Unexpected end of JSON"),
+            FailedUtf8Parsing       => write!(f, "Failed to parse UTF-8 bytes"),
+            ArrayIndexOutOfBounds   => write!(f, "Array index out of bounds!"),
+            WrongType(ref s)        => write!(f, "Wrong type, expected: {}", s),
+            UndefinedField(ref s)   => write!(f, "Undefined field: {}", s)
         }
     }
 }
@@ -47,13 +54,13 @@ impl Error for JsonError {
     fn description(&self) -> &str {
         use JsonError::*;
         match *self {
-            UnexpectedToken(_) => "Unexpected token",
+            UnexpectedToken(_)     => "Unexpected token",
             UnexpectedCharacter(_) => "Unexpected character",
-            UnexpectedEndOfJson => "Unexpected end of JSON",
-            CantCastCodepointToCharacter(_) => "Cannot cast this codepoint to a character",
-            ArrayIndexOutOfBounds => "Array index out of bounds!",
-            WrongType(_) => "Wrong type",
-            UndefinedField(_) => "Undefined field",
+            UnexpectedEndOfJson    => "Unexpected end of JSON",
+            FailedUtf8Parsing      => "Failed to read bytes as UTF-8 from JSON",
+            ArrayIndexOutOfBounds  => "Array index out of bounds!",
+            WrongType(_)           => "Wrong type",
+            UndefinedField(_)      => "Undefined field",
         }
     }
 }

src/lib.rs

@@ -194,7 +194,7 @@ pub use value::JsonValue::Null;
 pub type JsonResult<T> = Result<T, JsonError>;
 
 pub use parser::parse;
-use codegen::Generator;
+use codegen::{ Generator, PrettyGenerator, DumpGenerator };
 
 use std::collections::HashMap;
 use std::collections::BTreeMap;
@@ -206,15 +206,15 @@ pub type Object = BTreeMap<String, JsonValue>;
 impl JsonValue {
     /// Prints out the value as JSON string.
     pub fn dump(&self) -> String {
-        let mut gen = Generator::new(true, 0);
+        let mut gen = DumpGenerator::new();
         gen.write_json(self);
         gen.consume()
     }
 
     /// Pretty prints out the value as JSON string. Takes an argument that's
     /// number of spaces to indent new blocks with.
     pub fn pretty(&self, spaces: u16) -> String {
-        let mut gen = Generator::new(false, spaces);
+        let mut gen = PrettyGenerator::new(spaces);
         gen.write_json(self);
         gen.consume()
     }