Terms own their own value

src/core/codec.rs

@@ -62,7 +62,7 @@ impl Codec for SimpleCodec {
             // TODO include cause somehow
             return Err(Error::WriteError(String::from("Failed creating the term builder")));
         }
-        let mut term_buffer: String = String::new();
+        let mut term_buffer: Vec<u8> = Vec::new();
         let mut term_trie_builder = term_trie_builder_result.unwrap();
         let mut term_cursor = index.term_cursor();
         let mut offset: usize = 0;

src/core/global.rs

@@ -2,7 +2,7 @@ use std::io::{BufWriter, Write};
 use std::io;
 
 pub type DocId = usize;
-pub type FieldId = u32;
+pub type FieldId = u8;
 
 #[derive(Clone,Debug,PartialEq,PartialOrd,Eq,Hash)]
 pub struct Field(pub FieldId);

src/core/reader.rs

@@ -1,40 +1,46 @@
 use core::directory::Directory;
 use core::directory::Segment;
+use core::schema::Term;
+// use fst::stream::Streamer;
+use fst;
 
 pub struct SegmentIndexReader {
-    directory: Directory,
+    segment: Segment,
+    term_offsets: fst::Map,
 }
-
-
-// pub trait SearchableSegment {
+//
+// pub struct SegmentDocCursor<'a> {
+//     data: &'a [u8],
+// }
+//
+// struct SegmentTermCur<'a> {
+//     segment: &'a Segment,
+//     fst_streamer: fst::map::Stream<'a>,
+//     term: Term<'a>,
 //
 // }
 //
-// pub struct SimpleSearchableSegment {
-//     segment: Segment,
-// }
+// impl<'a> SegmentTermCur<'a> {
 //
-// impl SimpleSearchableSegment {
-//
-//     pub fn new(segment: &Segment) -> SimpleSearchableSegment {
-//         SimpleSearchableSegment {
-//             segment: segment.clone()
+//     fn next(&mut self,) -> Option<(Term<'a>, SegmentDocCursor<'a>)> {
+//         match self.fst_streamer.next() {
+//             Some(_) => None,
+//             None => None
 //         }
 //     }
 // }
 //
-// impl SearchableSegment for SimpleSearchableSegment {
-//
-//
-// }
-
-
 //
 // impl SegmentIndexReader {
 //
-//     pub fn open(directory: &Directory) -> IndexReader {
-// 		IndexReader {
-// 			directory: (*directory).clone(),
+//     fn term_cursor<'a>(&'a self) -> SegmentTermCur<'a> {
+//         let term: Term<'a> {
+//             self.
+//         };
+//         SegmentTermCur {
+//             segment: &self.segment,
+//             fst_streamer: self.term_offsets.stream(),
+//             term:
 //         }
 //     }
 //

src/core/schema.rs

@@ -1,5 +1,6 @@
 use core::global::*;
 use std::fmt::Write;
+use byteorder::{LittleEndian, ReadBytesExt, WriteBytesExt};
 
 #[derive(Clone,Debug,PartialEq,PartialOrd,Eq)]
 pub struct FieldValue {
@@ -8,18 +9,31 @@ pub struct FieldValue {
 }
 
 
-#[derive(Clone,Debug,PartialEq,PartialOrd,Eq,Hash)]
-pub struct Term<'a> {
-    pub field: Field,
-	pub text: &'a str,
+#[derive(Clone,PartialEq,PartialOrd,Eq,Hash)]
+pub struct Term {
+    pub data: Vec<u8>, // avoid copies
+    // pub field: Field,
+	// pub text: &'a [u8],
 }
 
-impl<'a> Term<'a> {
-    pub fn write_into(&self, term_str: &mut String) {
-        term_str.clear();
-        let Field(field_idx) = self.field;
-        // TODO avoid writing the field idx.
-        term_str.write_fmt(format_args!("{}:{}", field_idx, self.text));
+impl Term {
+
+    // TODO avoid all these copies.
+
+    pub fn from_field_text(field: Field, text: &str) -> Term {
+        let mut buffer = Vec::with_capacity(1 + text.len());
+        let Field(field_idx) = field;
+        buffer.clear();
+        buffer.push(field_idx);
+        buffer.extend(text.as_bytes());
+        Term {
+            data: buffer,
+        }
+    }
+
+    pub fn write_into(&self, buf: &mut Vec<u8>) {
+        buf.clear();
+        buf.extend(&self.data);
     }
 }
 

src/core/serial.rs

@@ -3,7 +3,7 @@ use core::schema::*;
 
 // Trait sufficient to serialize a segment.
 pub trait SerializableSegment<'a> {
-    type TermCur: TermCursor<'a>; // TODO rename TermCursorImpl
+    type TermCur: TermCursor; // TODO rename TermCursorImpl
     fn term_cursor(&'a self) -> Self::TermCur;
 }
 
@@ -14,7 +14,8 @@ pub trait DocCursor: Iterator<Item=DocId> {
 
 // TODO make iteration over Fields somehow sorted
 
-pub trait TermCursor<'a>  {
+pub trait TermCursor  {
     type DocCur: DocCursor;
-    fn next(&mut self,) -> Option<(Term<'a>, Self::DocCur)>;
+    // fn next(&mut self,) -> Option<(Term<'a>, Self::DocCur)>;
+    fn next(&mut self,) -> Option<(Term, Self::DocCur)>;
 }

src/core/writer.rs

@@ -15,6 +15,8 @@ use byteorder::{NativeEndian, ReadBytesExt, WriteBytesExt};
 use std::iter::Peekable;
 use core::serial::*;
 use core::error::*;
+use std::cell::RefCell;
+use std::borrow::BorrowMut;
 
 pub struct SimplePostingsWriter {
 	doc_ids: Vec<DocId>,
@@ -166,11 +168,12 @@ impl<'a> CIWTermCursor<'a> {
 		}
 	}
 
-	fn get_term(&self) -> Term<'a> {
-		Term {
-			field: self.field.clone(),
-			text: self.current_form_postings.as_ref().unwrap().form,
-		}
+	fn get_term(&self) -> Term {
+		Term::from_field_text(self.field.clone(), self.current_form_postings.as_ref().unwrap().form)
+		// Term {
+		// 	field: self.field.clone(),
+		// 	text: self.current_form_postings.as_ref().unwrap().form,
+		// }
 	}
 
 	fn doc_cursor(&self,) -> CIWDocCursor<'a> {
@@ -218,11 +221,11 @@ impl<'a> CIWTermCursor<'a> {
 }
 
 
-impl<'a> TermCursor<'a> for CIWTermCursor<'a> {
+impl<'a> TermCursor for CIWTermCursor<'a> {
 
 	type DocCur = CIWDocCursor<'a>;
 
-	fn next(&mut self,) -> Option<(Term<'a>, CIWDocCursor<'a>)> {
+	fn next(&mut self,) -> Option<(Term, CIWDocCursor<'a>)> {
 		if self.advance() {
 			Some((self.get_term(), self.doc_cursor()))
 		}
@@ -242,7 +245,7 @@ impl<'a> SerializableSegment<'a> for IndexWriter {
 	fn term_cursor(&'a self) -> CIWTermCursor<'a> {
 		let mut field_it: hash_map::Iter<'a, Field, FieldWriter> = self.term_writers.iter();
 		let (field, field_writer) = field_it.next().unwrap(); // TODO handle no field
-		let term_cursor = CIWTermCursor {
+		CIWTermCursor {
 			field_it: field_it,
 			form_it: CIWFormCursor {
 				term_it: field_writer.term_index.iter(),
@@ -250,9 +253,8 @@ impl<'a> SerializableSegment<'a> for IndexWriter {
 			},
 			field: field,
 			current_form_postings: None,
-		};
+		}
 		// TODO handle having no fields at all
-		term_cursor
 	}
 }