Implement load_from_bytes

Closes #155

Also helps in some cases with #142, when the BOM is at the beginning of the file (common),
but not in corner case where the BOM is at the start of a document which is not the first one.

Author: mkmik

Cargo.toml

@@ -12,6 +12,7 @@ edition = "2018"
 
 [dependencies]
 linked-hash-map = "0.5.3"
+encoding = "0.2"
 
 [dev-dependencies]
 quickcheck = "0.9"

src/yaml.rs

@@ -157,6 +157,19 @@ impl MarkedEventReceiver for YamlLoader {
     }
 }
 
+#[derive(Debug)]
+pub enum LoadError {
+    IO(std::io::Error),
+    Scan(ScanError),
+    Decode(std::borrow::Cow<'static, str>),
+}
+
+impl From<std::io::Error> for LoadError {
+    fn from(error: std::io::Error) -> Self {
+        LoadError::IO(error)
+    }
+}
+
 impl YamlLoader {
     fn insert_new_node(&mut self, node: (Yaml, usize)) {
         // valid anchor id starts from 1
@@ -197,6 +210,42 @@ impl YamlLoader {
         parser.load(&mut loader, true)?;
         Ok(loader.docs)
     }
+
+    pub fn load_from_bytes(mut source: impl std::io::Read) -> Result<Vec<Yaml>, LoadError> {
+        let mut buffer = Vec::new();
+        source.read_to_end(&mut buffer)?;
+
+        // Decodes the input buffer using either UTF-8, UTF-16LE or UTF-16BE depending on the BOM codepoint.
+        // If the buffer doesn't start with a BOM codepoint, it will use a fallback encoding obtained by
+        // detect_utf16_endianness.
+        let (res, _) = encoding::types::decode(
+            &buffer,
+            encoding::DecoderTrap::Strict,
+            detect_utf16_endianness(&buffer),
+        );
+        let s = res.map_err(LoadError::Decode)?;
+        YamlLoader::load_from_str(&s).map_err(LoadError::Scan)
+    }
+}
+
+/// The encoding crate knows how to tell apart UTF-8 from UTF-16LE and utf-16BE, when the
+/// bytestream starts with BOM codepoint.
+/// However, it doesn't even attempt to guess the UTF-16 endianness of the input bytestream since
+/// in the general case the bytestream could start with a codepoint that uses both bytes.
+///
+/// The YAML-1.2 spec mandates that the first character of a YAML document is an ASCII character.
+/// This allows the encoding to be deduced by the pattern of null (#x00) characters.
+//
+/// See spec at https://yaml.org/spec/1.2/spec.html#id2771184
+fn detect_utf16_endianness(b: &[u8]) -> encoding::types::EncodingRef {
+    if b.len() > 1 && (b[0] != b[1]) {
+        if b[0] == 0 {
+            return encoding::all::UTF_16BE;
+        } else if b[1] == 0 {
+            return encoding::all::UTF_16LE;
+        }
+    }
+    encoding::all::UTF_8
 }
 
 macro_rules! define_as (
@@ -736,4 +785,67 @@ subcommands3:
         let s = "[".repeat(10_000) + &"]".repeat(10_000);
         assert!(YamlLoader::load_from_str(&s).is_err());
     }
+
+    #[test]
+    fn test_read_bom() {
+        let s = b"\xef\xbb\xbf---
+a: 1
+b: 2.2
+c: [1, 2]
+";
+        let out = YamlLoader::load_from_bytes(s as &[u8]).unwrap();
+        let doc = &out[0];
+        assert_eq!(doc["a"].as_i64().unwrap(), 1i64);
+        assert_eq!(doc["b"].as_f64().unwrap(), 2.2f64);
+        assert_eq!(doc["c"][1].as_i64().unwrap(), 2i64);
+        assert!(doc["d"][0].is_badvalue());
+    }
+
+    #[test]
+    fn test_read_utf16le() {
+        let s = b"\xff\xfe-\x00-\x00-\x00
+\x00a\x00:\x00 \x001\x00
+\x00b\x00:\x00 \x002\x00.\x002\x00
+\x00c\x00:\x00 \x00[\x001\x00,\x00 \x002\x00]\x00
+\x00";
+        let out = YamlLoader::load_from_bytes(s as &[u8]).unwrap();
+        let doc = &out[0];
+        println!("GOT: {:?}", doc);
+        assert_eq!(doc["a"].as_i64().unwrap(), 1i64);
+        assert_eq!(doc["b"].as_f64().unwrap(), 2.2f64);
+        assert_eq!(doc["c"][1].as_i64().unwrap(), 2i64);
+        assert!(doc["d"][0].is_badvalue());
+    }
+
+    #[test]
+    fn test_read_utf16be() {
+        let s = b"\xfe\xff\x00-\x00-\x00-\x00
+\x00a\x00:\x00 \x001\x00
+\x00b\x00:\x00 \x002\x00.\x002\x00
+\x00c\x00:\x00 \x00[\x001\x00,\x00 \x002\x00]\x00
+";
+        let out = YamlLoader::load_from_bytes(s as &[u8]).unwrap();
+        let doc = &out[0];
+        println!("GOT: {:?}", doc);
+        assert_eq!(doc["a"].as_i64().unwrap(), 1i64);
+        assert_eq!(doc["b"].as_f64().unwrap(), 2.2f64);
+        assert_eq!(doc["c"][1].as_i64().unwrap(), 2i64);
+        assert!(doc["d"][0].is_badvalue());
+    }
+
+    #[test]
+    fn test_read_utf16le_nobom() {
+        let s = b"-\x00-\x00-\x00
+\x00a\x00:\x00 \x001\x00
+\x00b\x00:\x00 \x002\x00.\x002\x00
+\x00c\x00:\x00 \x00[\x001\x00,\x00 \x002\x00]\x00
+\x00";
+        let out = YamlLoader::load_from_bytes(s as &[u8]).unwrap();
+        let doc = &out[0];
+        println!("GOT: {:?}", doc);
+        assert_eq!(doc["a"].as_i64().unwrap(), 1i64);
+        assert_eq!(doc["b"].as_f64().unwrap(), 2.2f64);
+        assert_eq!(doc["c"][1].as_i64().unwrap(), 2i64);
+        assert!(doc["d"][0].is_badvalue());
+    }
 }