Merge #22

22: Do not allocate buffer size before necessary r=gendx a=dragly

This changes the way the LZCircularBuffer is allocated. Instead of allocating the full dict size immediately, the buffer is grown on demand. This helps when parsing smaller files with large dict sizes, especially when compiling to WebAssembly and running in browser where `alloc_zeroed` is slow.

Co-authored-by: Svenn-Arne Dragly <dragly@cognite.com>

src/decode/lzbuffer.rs

@@ -139,12 +139,23 @@ where
         info!("Dict size in LZ buffer: {}", dict_size);
         Self {
             stream,
-            buf: vec![0; dict_size],
+            buf: Vec::new(),
             dict_size,
             cursor: 0,
             len: 0,
         }
     }
+
+    fn get(&self, index: usize) -> u8 {
+        *self.buf.get(index).unwrap_or(&0)
+    }
+
+    fn get_mut(&mut self, index: usize) -> &mut u8 {
+        if self.buf.len() < index + 1 {
+            self.buf.resize(index + 1, 0);
+        }
+        self.buf.get_mut(index).unwrap()
+    }
 }
 
 impl<'a, W> LZBuffer for LZCircularBuffer<'a, W>
@@ -160,7 +171,7 @@ where
         if self.len == 0 {
             lit
         } else {
-            self.buf[(self.dict_size + self.cursor - 1) % self.dict_size]
+            self.get((self.dict_size + self.cursor - 1) % self.dict_size)
         }
     }
 
@@ -180,12 +191,12 @@ where
         }
 
         let offset = (self.dict_size + self.cursor - dist) % self.dict_size;
-        Ok(self.buf[offset])
+        Ok(self.get(offset))
     }
 
     // Append a literal
     fn append_literal(&mut self, lit: u8) -> io::Result<()> {
-        self.buf[self.cursor] = lit;
+        *self.get_mut(self.cursor) = lit;
         self.cursor += 1;
         self.len += 1;
 
@@ -216,7 +227,7 @@ where
 
         let mut offset = (self.dict_size + self.cursor - dist) % self.dict_size;
         for _ in 0..len {
-            let x = self.buf[offset];
+            let x = self.get(offset);
             self.append_literal(x)?;
             offset += 1;
             if offset == self.dict_size {