feat: add Byte_buf, a byte buffer.

This commit is contained in:
Simon Cruanes 2022-02-03 16:53:31 -05:00
parent bc8b7b168b
commit 5f064dbbbf
No known key found for this signature in database
GPG key ID: EBFFF6F283F3A2B4
4 changed files with 182 additions and 0 deletions

114
src/core/CCByte_buffer.ml Normal file
View file

@ -0,0 +1,114 @@
type 'a iter = ('a -> unit) -> unit
type t = {
mutable bytes: bytes;
mutable sz: int;
}
let create ?(size=0) () : t =
let bytes =
if size=0 then Bytes.unsafe_of_string "" else Bytes.create size in
{ sz=size; bytes }
let[@inline] capacity self : int = Bytes.length self.bytes
let[@inline] bytes self = self.bytes
let[@inline] length self = self.sz
let[@inline] is_empty self = self.sz = 0
let[@inline] clear self = self.sz <- 0
let grow_cap_ self =
min Sys.max_string_length
(let n = capacity self in n + n lsl 1 + 5)
let grow_to_ self newcap =
if newcap = capacity self then (
invalid_arg "byte_buf: cannot grow further";
);
let newbytes = Bytes.create newcap in
Bytes.blit self.bytes 0 newbytes 0 self.sz;
self.bytes <- newbytes
let[@inline never] grow_ self =
let newcap = grow_cap_ self in
grow_to_ self newcap
let ensure_cap self n =
if n>capacity self then (
let newcap = max n (grow_cap_ self) in
grow_to_ self newcap
)
let shrink_to self n =
if self.sz > n then self.sz <- n
let append_buf (self:t) buf : unit =
let n = Buffer.length buf in
ensure_cap self (length self + n);
Buffer.blit buf 0 self.bytes self.sz n;
self.sz <- self.sz + n
let append_subbytes self b off len =
ensure_cap self (length self + len);
Bytes.blit b off self.bytes self.sz len;
self.sz <- self.sz + len
let append_bytes self b = append_subbytes self b 0 (Bytes.length b)
let append_string self s = append_bytes self (Bytes.unsafe_of_string s)
let append_substring self s off len = append_subbytes self (Bytes.unsafe_of_string s) off len
let[@inline] add_char_unsafe_ self c =
Bytes.unsafe_set self.bytes self.sz c;
self.sz <- self.sz + 1
let[@inline] add_char self c =
if self.sz = capacity self then grow_ self;
add_char_unsafe_ self c
let[@inline] unsafe_get self i = Bytes.unsafe_get self.bytes i
let[@inline] unsafe_set self i c = Bytes.unsafe_set self.bytes i c
let[@inline] get self i =
if i < 0 || i >= self.sz then invalid_arg "Byte_buf.get";
unsafe_get self i
let[@inline] set self i c =
if i < 0 || i >= self.sz then invalid_arg "Byte_buf.set";
unsafe_set self i c
let[@inline] append_iter self i = i (add_char self)
let[@inline] append_seq self seq = Seq.iter (add_char self) seq
let fold_left f acc self =
let {bytes; sz} = self in (* capture current content *)
let acc = ref acc in
for i=0 to sz do
acc := f !acc (Bytes.unsafe_get bytes i)
done;
!acc
let iter f self =
let {bytes; sz} = self in (* capture current content *)
for i=0 to sz do
f (Bytes.unsafe_get bytes i)
done
let of_seq seq =
let self = create ~size:32 () in
append_seq self seq;
self
let of_iter iter =
let self = create ~size:32 () in
append_iter self iter;
self
let to_iter self yield = iter yield self
let to_seq self =
let {bytes;sz} = self in
let rec s i () =
if i= sz then Seq.Nil
else Seq.Cons (Bytes.unsafe_get bytes i, s (i+1))
in
s 0

View file

@ -0,0 +1,66 @@
(** Byte buffer.
A dynamic vector of bytes.
@since NEXT_RELEASE
*)
type t
type 'a iter = ('a -> unit) -> unit
val create : ?size:int -> unit -> t
val length : t -> int
(** Current length. *)
val capacity : t -> int
(** Current capacity (size of the array returned by {!bytes}) *)
val bytes : t -> bytes
(** Access the underlying byte buffer. This buffer can change after
operations that affect the capacity (e.g. {!add_char}). *)
val clear : t -> unit
val ensure_cap : t -> int -> unit
(** [ensure_cap self n] ensures that [capacity self >= n]. *)
val shrink_to : t -> int -> unit
(** [shrink_to buf n] reduces [length buf] to at most [n].
Does nothing if the length is already <= n. *)
val add_char : t -> char -> unit
(** Push a character at the end. *)
val append_bytes : t -> bytes -> unit
val append_subbytes : t -> bytes -> int -> int -> unit
val append_string : t -> string -> unit
val append_substring : t -> string -> int -> int -> unit
val append_buf : t -> Buffer.t -> unit
val append_iter : t -> char iter -> unit
val append_seq : t -> char Seq.t -> unit
val get : t -> int -> char
val unsafe_get : t -> int -> char
val set : t -> int -> char -> unit
val unsafe_set : t -> int -> char -> unit
val iter : (char -> unit) -> t -> unit
val fold_left : ('a -> char -> 'a) -> 'a -> t -> 'a
val of_iter : char iter -> t
val of_seq : char Seq.t -> t
val to_iter : t -> char iter
val to_seq : t -> char Seq.t

View file

@ -5,6 +5,7 @@
module Array = CCArray
module Bool = CCBool
module Byte_buf = CCByte_buf
module Char = CCChar
module Equal = CCEqual
module Either = CCEither

View file

@ -5,6 +5,7 @@
module Array = CCArrayLabels
module Bool = CCBool
module Byte_buf = CCByte_buf
module Char = CCChar
module Equal = CCEqualLabels
module Either = CCEither