Add yaml value coerce

saphyr/Cargo.toml

@@ -2,3 +2,6 @@
 name = "yaml-rust"
 version = "0.1.0"
 authors = ["Yuheng Chen <yuhengchen@sensetime.com>"]
+
+[dependencies]
+regex = "*"

saphyr/src/lib.rs

@@ -1,3 +1,5 @@
+extern crate regex;
+
 pub mod yaml;
 pub mod scanner;
 pub mod parser;

saphyr/src/parser.rs

@@ -1,4 +1,5 @@
 use scanner::*;
+use yaml::*;
 
 #[derive(Clone, Copy, PartialEq, Debug, Eq)]
 pub enum State {
@@ -67,7 +68,7 @@ impl<T: Iterator<Item=char>> Parser<T> {
             states: Vec::new(),
             state: State::StreamStart,
             marks: Vec::new(),
-            token: None
+            token: None,
         }
     }
 
@@ -104,38 +105,35 @@ impl<T: Iterator<Item=char>> Parser<T> {
         ev
     }
 
-    pub fn load(&mut self) -> Result<(), ScanError> {
+    pub fn load(&mut self) -> Result<Yaml, ScanError> {
         if !self.scanner.stream_started() {
             let ev = try!(self.parse());
             assert_eq!(ev, Event::StreamStart);
         }
 
         if self.scanner.stream_ended() {
-            return Ok(());
+            return Ok(Yaml::Null);
         }
         let ev = try!(self.parse());
         if ev == Event::StreamEnd {
-            return Ok(());
+            return Ok(Yaml::Null);
         }
-        try!(self.load_document(&ev));
-        Ok(())
+        self.load_document(&ev)
     }
 
-    fn load_document(&mut self, first_ev: &Event) -> Result<(), ScanError> {
+    fn load_document(&mut self, first_ev: &Event) -> Result<Yaml, ScanError> {
         assert_eq!(first_ev, &Event::DocumentStart);
 
         let ev = try!(self.parse());
-        try!(self.load_node(&ev));
-
-        Ok(())
+        self.load_node(&ev)
     }
 
-    fn load_node(&mut self, first_ev: &Event) -> Result<(), ScanError> {
+    fn load_node(&mut self, first_ev: &Event) -> Result<Yaml, ScanError> {
         match *first_ev {
-            Event::Scalar(_) => {
+            Event::Scalar(ref v) => {
                 // TODO scalar
                 println!("Scalar: {:?}", first_ev);
-                Ok(())
+                Ok(Yaml::String(v.clone()))
             },
             Event::SequenceStart => {
                 self.load_sequence(first_ev)
@@ -147,31 +145,36 @@ impl<T: Iterator<Item=char>> Parser<T> {
         }
     }
 
-    fn load_mapping(&mut self, first_ev: &Event) -> Result<(), ScanError> {
+    fn load_mapping(&mut self, first_ev: &Event) -> Result<Yaml, ScanError> {
         let mut ev = try!(self.parse());
+        let mut map = Hash::new();
         while ev != Event::MappingEnd {
             // key
-            try!(self.load_node(&ev));
+            let key = try!(self.load_node(&ev));
 
             // value
             ev = try!(self.parse());
-            try!(self.load_node(&ev));
+            let value = try!(self.load_node(&ev));
+
+            map.insert(key, value);
 
             // next event
             ev = try!(self.parse());
         }
-        Ok(())
+        Ok(Yaml::Hash(map))
     }
 
-    fn load_sequence(&mut self, first_ev: &Event) -> Result<(), ScanError> {
+    fn load_sequence(&mut self, first_ev: &Event) -> Result<Yaml, ScanError> {
         let mut ev = try!(self.parse());
+        let mut vec = Vec::new();
         while ev != Event::SequenceEnd {
-            try!(self.load_node(&ev));
+            let entry = try!(self.load_node(&ev));
+            vec.push(entry);
 
             // next event
             ev = try!(self.parse());
         }
-        Ok(())
+        Ok(Yaml::Array(vec))
     }
 
     fn state_machine(&mut self) -> ParseResult {
@@ -183,6 +186,10 @@ impl<T: Iterator<Item=char>> Parser<T> {
             State::DocumentStart => self.document_start(false),
             State::DocumentContent => self.document_content(),
 
+            State::BlockNode => self.parse_node(true, false),
+            State::BlockNodeOrIndentlessSequence => self.parse_node(true, true),
+            State::FlowNode => self.parse_node(false, false),
+
             State::BlockMappingFirstKey => self.block_mapping_key(true),
             State::BlockMappingKey => self.block_mapping_key(false),
             State::BlockMappingValue => self.block_mapping_value(),
@@ -240,7 +247,6 @@ impl<T: Iterator<Item=char>> Parser<T> {
             _ if implicit => {
                 self.push_state(State::DocumentEnd);
                 self.state = State::BlockNode;
-                self.skip();
                 Ok(Event::DocumentStart)
             },
             _ => {
@@ -460,7 +466,7 @@ mod test {
     use super::*;
     #[test]
     fn test_parser() {
-        let s: String = "---
+        let s: String = "
 # comment
 a0 bb: val
 a1:
@@ -474,8 +480,8 @@ a4:
     - 2
 ".to_string();
         let mut parser = Parser::new(s.chars());
-        parser.load().unwrap();
-
+        let out = parser.load().unwrap();
+        println!("DOC {:?}", out);
     }
 }
 

saphyr/src/scanner.rs

@@ -540,7 +540,7 @@ impl<T: Iterator<Item=char>> Scanner<T> {
                 // indicators ends a plain scalar
                 match self.ch() {
                     ':' if is_blankz(self.buffer[1]) => break,
-                    ',' | ':' | '?' | '[' | ']' |'{' |'}' => break,
+                    ',' | ':' | '?' | '[' | ']' |'{' |'}' if self.flow_level > 0 => break,
                     _ => {}
                 }
 
@@ -718,7 +718,7 @@ a4:
 ".to_string();
         let p = Scanner::new(s.chars());
         for t in p {
-            // println!("{:?}", t);
+            println!("{:?}", t);
         }
     }
 }

saphyr/src/yaml.rs

@@ -1,11 +1,12 @@
 use std::collections::{HashMap, BTreeMap};
 use std::string;
+use regex::Regex;
 
-#[derive(Clone, PartialEq, PartialOrd, Debug)]
+#[derive(Clone, PartialEq, PartialOrd, Debug, Eq, Ord)]
 pub enum Yaml {
     I64(i64),
-    U64(u64),
-    F64(f64),
+    //U64(u64),
+    //F64(f64),
     String(string::String),
     Boolean(bool),
     Array(self::Array),
@@ -14,7 +15,7 @@ pub enum Yaml {
 }
 
 pub type Array = Vec<Yaml>;
-pub type Hash = BTreeMap<string::String, Yaml>;
+pub type Hash = BTreeMap<Yaml, Yaml>;
 
 /// The errors that can arise while parsing a YAML stream.
 #[derive(Clone, Copy, PartialEq, Debug)]
@@ -37,3 +38,74 @@ pub enum ErrorCode {
     NotFourDigit,
     NotUtf8,
 }
+
+macro_rules! define_as (
+    ($name:ident, $t:ident, $yt:ident) => (
+pub fn $name(&self) -> Option<$t> {
+    match *self {
+        Yaml::$yt(v) => Some(v),
+        _ => None
+    }
+}
+    );
+);
+
+macro_rules! define_as_ref (
+    ($name:ident, $t:ty, $yt:ident) => (
+pub fn $name(&self) -> Option<$t> {
+    match *self {
+        Yaml::$yt(ref v) => Some(v),
+        _ => None
+    }
+}
+    );
+);
+
+// these regex are from libyaml-rust project
+macro_rules! regex(
+    ($s:expr) => (Regex::new($s).unwrap());
+);
+impl Yaml {
+    define_as!(as_i64, i64, I64);
+    define_as!(as_bool, bool, Boolean);
+
+    define_as_ref!(as_str, &str, String);
+    define_as_ref!(as_hash, &Hash, Hash);
+    define_as_ref!(as_vec, &Array, Array);
+
+    pub fn is_null(&self) -> bool {
+        match *self {
+            Yaml::Null => true,
+            _ => false
+        }
+    }
+
+    pub fn as_f64(&self) -> Option<f64> {
+        // XXX(chenyh) precompile me
+        let float_pattern = regex!(r"^([-+]?)(\.[0-9]+|[0-9]+(\.[0-9]*)?([eE][-+]?[0-9]+)?)$");
+        match *self {
+            Yaml::String(ref v) if float_pattern.is_match(v) => {
+                v.parse::<f64>().ok()
+            },
+            _ => None
+        }
+    }
+}
+
+#[cfg(test)]
+mod test {
+    use parser::Parser;
+    use yaml::Yaml;
+    // #[test]
+    fn test_coerce() {
+        let s = "---
+a: 1
+b: 2.2
+c: [1, 2]
+";
+        let mut parser = Parser::new(s.chars());
+        let out = parser.load().unwrap();
+        //assert_eq!(out.as_hash().unwrap()[&Yaml::String("a".to_string())].as_i64().unwrap(), 1i64);
+    }
+}
+