rust/src/libstd/rope.rs

1401 lines
38 KiB
Rust
Raw Normal View History

2012-03-07 20:17:30 -06:00
#[doc = "
2011-11-04 12:38:42 -05:00
High-level text containers.
Ropes are a high-level representation of text that offers
much better performance than strings for common operations,
and generally reduce memory allocations and copies, while only
entailing a small degradation of less common operations.
More precisely, where a string is represented as a memory buffer,
a rope is a tree structure whose leaves are slices of immutable
strings. Therefore, concatenation, appending, prepending, substrings,
etc. are operations that require only trivial tree manipulation,
generally without having to copy memory. In addition, the tree
structure of ropes makes them suitable as a form of index to speed-up
access to Unicode characters by index in long chunks of text.
The following operations are algorithmically faster in ropes:
2012-03-07 20:17:30 -06:00
* extracting a subrope is logarithmic (linear in strings);
* appending/prepending is near-constant time (linear in strings);
* concatenation is near-constant time (linear in strings);
* char length is constant-time (linear in strings);
* access to a character by index is logarithmic (linear in strings);
"];
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
#[doc = "The type of ropes."]
2011-11-04 12:38:42 -05:00
type rope = node::root;
/*
Section: Creating a rope
*/
2012-03-07 20:17:30 -06:00
#[doc = "Create an empty rope"]
2011-11-04 12:38:42 -05:00
fn empty() -> rope {
ret node::empty;
}
2012-03-07 20:17:30 -06:00
#[doc = "
Adopt a string as a rope.
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
# Arguments
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
* str - A valid string.
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
# Return value
2011-11-04 12:38:42 -05:00
A rope representing the same string as `str`. Depending of the length
of `str`, this rope may be empty, flat or complex.
2012-03-07 20:17:30 -06:00
# Performance notes
* this operation does not copy the string;
* the function runs in linear time.
"]
2011-11-04 12:38:42 -05:00
fn of_str(str: @str) -> rope {
2012-02-23 03:44:04 -06:00
ret of_substr(str, 0u, str::len(*str));
2011-11-04 12:38:42 -05:00
}
2012-03-07 20:17:30 -06:00
#[doc = "
2011-11-04 12:38:42 -05:00
As `of_str` but for a substring.
2012-03-07 20:17:30 -06:00
# Arguments
* byte_offset - The offset of `str` at which the rope starts.
* byte_len - The number of bytes of `str` to use.
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
# Return value
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
A rope representing the same string as `str::substr(str, byte_offset,
byte_len)`. Depending on `byte_len`, this rope may be empty, flat or
complex.
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
# Performance note
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
This operation does not copy the substring.
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
# Safety notes
* this function does _not_ check the validity of the substring;
* this function fails if `byte_offset` or `byte_len` do not match `str`.
"]
2011-11-04 12:38:42 -05:00
fn of_substr(str: @str, byte_offset: uint, byte_len: uint) -> rope {
if byte_len == 0u { ret node::empty; }
2012-02-23 03:44:04 -06:00
if byte_offset + byte_len > str::len(*str) { fail; }
2011-11-04 12:38:42 -05:00
ret node::content(node::of_substr(str, byte_offset, byte_len));
}
/*
Section: Adding things to a rope
*/
2012-03-07 20:17:30 -06:00
#[doc = "
2011-11-04 12:38:42 -05:00
Add one char to the end of the rope
2012-03-07 20:17:30 -06:00
# Performance note
* this function executes in near-constant time
"]
2011-11-04 12:38:42 -05:00
fn append_char(rope: rope, char: char) -> rope {
ret append_str(rope, @str::from_chars([char]));
}
2012-03-07 20:17:30 -06:00
#[doc = "
2011-11-04 12:38:42 -05:00
Add one string to the end of the rope
2012-03-07 20:17:30 -06:00
# Performance note
* this function executes in near-linear time
"]
2011-11-04 12:38:42 -05:00
fn append_str(rope: rope, str: @str) -> rope {
ret append_rope(rope, of_str(str))
}
2012-03-07 20:17:30 -06:00
#[doc = "
2011-11-04 12:38:42 -05:00
Add one char to the beginning of the rope
2012-03-07 20:17:30 -06:00
# Performance note
* this function executes in near-constant time
"]
2011-11-04 12:38:42 -05:00
fn prepend_char(rope: rope, char: char) -> rope {
ret prepend_str(rope, @str::from_chars([char]));
}
2012-03-07 20:17:30 -06:00
#[doc = "
2011-11-04 12:38:42 -05:00
Add one string to the beginning of the rope
2012-03-07 20:17:30 -06:00
# Performance note
* this function executes in near-linear time
"]
2011-11-04 12:38:42 -05:00
fn prepend_str(rope: rope, str: @str) -> rope {
ret append_rope(of_str(str), rope)
}
2012-03-07 20:17:30 -06:00
#[doc = "Concatenate two ropes"]
2011-11-04 12:38:42 -05:00
fn append_rope(left: rope, right: rope) -> rope {
alt(left) {
node::empty { ret right; }
2011-11-04 12:38:42 -05:00
node::content(left_content) {
alt(right) {
node::empty { ret left; }
2011-11-04 12:38:42 -05:00
node::content(right_content) {
ret node::content(node::concat2(left_content, right_content));
}
}
}
}
}
2012-03-07 20:17:30 -06:00
#[doc = "
Concatenate many ropes.
If the ropes are balanced initially and have the same height, the resulting
rope remains balanced. However, this function does not take any further
measure to ensure that the result is balanced.
2012-03-07 20:17:30 -06:00
"]
2011-11-04 12:38:42 -05:00
fn concat(v: [rope]) -> rope {
2012-03-26 20:35:18 -05:00
//Copy `v` into a mut vector
let mut len = vec::len(v);
if len == 0u { ret node::empty; }
2012-03-12 17:52:30 -05:00
let ropes = vec::to_mut(vec::from_elem(len, v[0]));
uint::range(1u, len) {|i|
ropes[i] = v[i];
}
//Merge progresively
while len > 1u {
uint::range(0u, len/2u) {|i|
ropes[i] = append_rope(ropes[2u*i], ropes[2u*i+1u]);
}
if len%2u != 0u {
ropes[len/2u] = ropes[len - 1u];
len = len/2u + 1u;
} else {
len = len/2u;
}
}
//Return final rope
ret ropes[0];
2011-11-04 12:38:42 -05:00
}
/*
Section: Keeping ropes healthy
*/
2012-03-07 20:17:30 -06:00
#[doc = "
2011-11-04 12:38:42 -05:00
Balance a rope.
2012-03-07 20:17:30 -06:00
# Return value
2011-11-04 12:38:42 -05:00
A copy of the rope in which small nodes have been grouped in memory,
and with a reduced height.
If you perform numerous rope concatenations, it is generally a good idea
to rebalance your rope at some point, before using it for other purposes.
2012-03-07 20:17:30 -06:00
"]
2011-11-04 12:38:42 -05:00
fn bal(rope:rope) -> rope {
alt(rope) {
node::empty { ret rope }
2011-11-04 12:38:42 -05:00
node::content(x) {
alt(node::bal(x)) {
option::none { rope }
2011-11-04 12:38:42 -05:00
option::some(y) { node::content(y) }
}
}
}
}
/*
Section: Transforming ropes
*/
2012-03-07 20:17:30 -06:00
#[doc = "
2011-11-04 12:38:42 -05:00
Extract a subrope from a rope.
2012-03-07 20:17:30 -06:00
# Performance note
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
* on a balanced rope, this operation takes algorithmic time;
* this operation does not involve any copying
# Safety note
* this function fails if char_offset/char_len do not represent
valid positions in rope
"]
2011-11-04 12:38:42 -05:00
fn sub_chars(rope: rope, char_offset: uint, char_len: uint) -> rope {
if char_len == 0u { ret node::empty; }
alt(rope) {
node::empty { fail }
2011-11-04 12:38:42 -05:00
node::content(node) {
if char_len > node::char_len(node) { fail }
else {
ret node::content(node::sub_chars(node, char_offset, char_len))
}
}
}
}
2012-03-07 20:17:30 -06:00
#[doc = "
2011-11-04 12:38:42 -05:00
Extract a subrope from a rope.
2012-03-07 20:17:30 -06:00
# Performance note
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
* on a balanced rope, this operation takes algorithmic time;
* this operation does not involve any copying
# Safety note
* this function fails if byte_offset/byte_len do not represent
valid positions in rope
"]
2011-11-04 12:38:42 -05:00
fn sub_bytes(rope: rope, byte_offset: uint, byte_len: uint) -> rope {
if byte_len == 0u { ret node::empty; }
alt(rope) {
node::empty { fail }
2011-11-04 12:38:42 -05:00
node::content(node) {
if byte_len > node::byte_len(node) { fail }
else {
ret node::content(node::sub_bytes(node, byte_offset, byte_len))
}
}
}
}
/*
Section: Comparing ropes
*/
2012-03-07 20:17:30 -06:00
#[doc = "
2011-11-04 12:38:42 -05:00
Compare two ropes by Unicode lexicographical order.
This function compares only the contents of the rope, not their structure.
2012-03-07 20:17:30 -06:00
# Return value
2011-11-04 12:38:42 -05:00
A negative value if `left < right`, 0 if eq(left, right) or a positive
value if `left > right`
2012-03-07 20:17:30 -06:00
"]
2011-11-04 12:38:42 -05:00
fn cmp(left: rope, right: rope) -> int {
alt((left, right)) {
(node::empty, node::empty) { ret 0; }
(node::empty, _) { ret -1;}
(_, node::empty) { ret 1;}
2011-11-04 12:38:42 -05:00
(node::content(a), node::content(b)) {
ret node::cmp(a, b);
}
}
}
2012-03-07 20:17:30 -06:00
#[doc = "
Returns `true` if both ropes have the same content (regardless of
their structure), `false` otherwise
"]
2011-11-04 12:38:42 -05:00
fn eq(left: rope, right: rope) -> bool {
ret cmp(left, right) == 0;
}
2012-03-07 20:17:30 -06:00
#[doc = "
# Arguments
2012-03-07 20:17:30 -06:00
* left - an arbitrary rope
* right - an arbitrary rope
2012-03-07 20:17:30 -06:00
# Return value
2012-03-07 20:17:30 -06:00
`true` if `left <= right` in lexicographical order (regardless of their
structure), `false` otherwise
2012-03-07 20:17:30 -06:00
"]
fn le(left: rope, right: rope) -> bool {
2011-11-04 12:38:42 -05:00
ret cmp(left, right) <= 0;
}
2012-03-07 20:17:30 -06:00
#[doc = "
# Arguments
2012-03-07 20:17:30 -06:00
* left - an arbitrary rope
* right - an arbitrary rope
2012-03-07 20:17:30 -06:00
# Return value
2012-03-07 20:17:30 -06:00
`true` if `left < right` in lexicographical order (regardless of their
structure), `false` otherwise
2012-03-07 20:17:30 -06:00
"]
2011-11-04 12:38:42 -05:00
fn lt(left: rope, right: rope) -> bool {
ret cmp(left, right) < 0;
}
2012-03-07 20:17:30 -06:00
#[doc = "
# Arguments
2012-03-07 20:17:30 -06:00
* left - an arbitrary rope
* right - an arbitrary rope
2012-03-07 20:17:30 -06:00
# Return value
`true` if `left >= right` in lexicographical order (regardless of their
structure), `false` otherwise
2012-03-07 20:17:30 -06:00
"]
fn ge(left: rope, right: rope) -> bool {
2011-11-04 12:38:42 -05:00
ret cmp(left, right) >= 0;
}
2012-03-07 20:17:30 -06:00
#[doc = "
# Arguments
2012-03-07 20:17:30 -06:00
* left - an arbitrary rope
* right - an arbitrary rope
2012-03-07 20:17:30 -06:00
# Return value
2012-03-07 20:17:30 -06:00
`true` if `left > right` in lexicographical order (regardless of their
structure), `false` otherwise
2012-03-07 20:17:30 -06:00
"]
2011-11-04 12:38:42 -05:00
fn gt(left: rope, right: rope) -> bool {
ret cmp(left, right) > 0;
}
/*
Section: Iterating
*/
2012-03-07 20:17:30 -06:00
#[doc = "
2011-11-04 12:38:42 -05:00
Loop through a rope, char by char
While other mechanisms are available, this is generally the best manner
of looping through the contents of a rope char by char. If you prefer a
loop that iterates through the contents string by string (e.g. to print
the contents of the rope or output it to the system), however,
you should rather use `traverse_components`.
2012-03-07 20:17:30 -06:00
# Arguments
* rope - A rope to traverse. It may be empty.
* it - A block to execute with each consecutive character of the rope.
Return `true` to continue, `false` to stop.
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
# Return value
2011-11-04 12:38:42 -05:00
`true` If execution proceeded correctly, `false` if it was interrupted,
that is if `it` returned `false` at any point.
2012-03-07 20:17:30 -06:00
"]
2012-01-23 16:59:00 -06:00
fn loop_chars(rope: rope, it: fn(char) -> bool) -> bool {
2011-11-04 12:38:42 -05:00
alt(rope) {
node::empty { ret true }
2011-11-04 12:38:42 -05:00
node::content(x) { ret node::loop_chars(x, it) }
}
}
2012-03-07 20:17:30 -06:00
#[doc = "
Loop through a rope, char by char, until the end.
2012-03-07 20:17:30 -06:00
# Arguments
* rope - A rope to traverse. It may be empty
* it - A block to execute with each consecutive character of the rope.
"]
2012-01-23 16:59:00 -06:00
fn iter_chars(rope: rope, it: fn(char)) {
2011-11-04 12:38:42 -05:00
loop_chars(rope) {|x|
it(x);
true
};
2011-11-04 12:38:42 -05:00
}
2012-03-07 20:17:30 -06:00
#[doc ="
2011-11-04 12:38:42 -05:00
Loop through a rope, string by string
While other mechanisms are available, this is generally the best manner of
looping through the contents of a rope string by string, which may be useful
e.g. to print strings as you see them (without having to copy their
contents into a new string), to send them to then network, to write them to
a file, etc.. If you prefer a loop that iterates through the contents
char by char (e.g. to search for a char), however, you should rather
use `traverse`.
2012-03-07 20:17:30 -06:00
# Arguments
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
* rope - A rope to traverse. It may be empty
* it - A block to execute with each consecutive string component of the rope.
Return `true` to continue, `false` to stop
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
# Return value
2011-11-04 12:38:42 -05:00
`true` If execution proceeded correctly, `false` if it was interrupted,
that is if `it` returned `false` at any point.
2012-03-07 20:17:30 -06:00
"]
2012-01-23 16:59:00 -06:00
fn loop_leaves(rope: rope, it: fn(node::leaf) -> bool) -> bool{
2011-11-04 12:38:42 -05:00
alt(rope) {
node::empty { ret true }
2011-11-04 12:38:42 -05:00
node::content(x) {ret node::loop_leaves(x, it)}
}
}
mod iterator {
mod leaf {
fn start(rope: rope) -> node::leaf_iterator::t {
alt(rope) {
node::empty { ret node::leaf_iterator::empty() }
2011-11-04 12:38:42 -05:00
node::content(x) { ret node::leaf_iterator::start(x) }
}
}
fn next(it: node::leaf_iterator::t) -> option<node::leaf> {
2011-11-04 12:38:42 -05:00
ret node::leaf_iterator::next(it);
}
}
mod char {
fn start(rope: rope) -> node::char_iterator::t {
alt(rope) {
node::empty { ret node::char_iterator::empty() }
2011-11-04 12:38:42 -05:00
node::content(x) { ret node::char_iterator::start(x) }
}
}
fn next(it: node::char_iterator::t) -> option<char> {
2011-11-04 12:38:42 -05:00
ret node::char_iterator::next(it)
}
}
}
/*
Section: Rope properties
*/
2012-03-07 20:17:30 -06:00
#[doc ="
Returns the height of the rope.
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
The height of the rope is a bound on the number of operations which
must be performed during a character access before finding the leaf in
which a character is contained.
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
# Performance note
Constant time.
"]
2011-11-04 12:38:42 -05:00
fn height(rope: rope) -> uint {
alt(rope) {
node::empty { ret 0u; }
2011-11-04 12:38:42 -05:00
node::content(x) { ret node::height(x); }
}
}
2012-03-07 20:17:30 -06:00
#[doc ="
The number of character in the rope
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
# Performance note
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
Constant time.
"]
2011-11-04 12:38:42 -05:00
pure fn char_len(rope: rope) -> uint {
alt(rope) {
node::empty { ret 0u; }
2011-11-04 12:38:42 -05:00
node::content(x) { ret node::char_len(x) }
}
}
2012-03-07 20:17:30 -06:00
#[doc = "
The number of bytes in the rope
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
# Performance note
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
Constant time.
"]
2011-11-04 12:38:42 -05:00
pure fn byte_len(rope: rope) -> uint {
alt(rope) {
node::empty { ret 0u; }
2011-11-04 12:38:42 -05:00
node::content(x) { ret node::byte_len(x) }
}
}
2012-03-07 20:17:30 -06:00
#[doc = "
The character at position `pos`
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
# Arguments
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
* pos - A position in the rope
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
# Safety notes
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
The function will fail if `pos` is not a valid position in the rope.
# Performance note
This function executes in a time proportional to the height of the
rope + the (bounded) length of the largest leaf.
"]
2011-11-04 12:38:42 -05:00
fn char_at(rope: rope, pos: uint) -> char {
alt(rope) {
node::empty { fail }
2011-11-04 12:38:42 -05:00
node::content(x) { ret node::char_at(x, pos) }
}
}
/*
Section: Implementation
*/
mod node {
2012-03-07 20:17:30 -06:00
#[doc = "Implementation of type `rope`"]
2012-01-19 17:20:57 -06:00
enum root {
2012-03-07 20:17:30 -06:00
#[doc = "An empty rope"]
2012-01-19 19:55:34 -06:00
empty,
2012-03-07 20:17:30 -06:00
#[doc = "A non-empty rope"]
2012-01-19 19:55:34 -06:00
content(@node),
2011-11-04 12:38:42 -05:00
}
2012-03-07 20:17:30 -06:00
#[doc = "
A text component in a rope.
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
This is actually a slice in a rope, so as to ensure maximal sharing.
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
# Fields
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
* byte_offset = The number of bytes skippen in `content`
* byte_len - The number of bytes of `content` to use
* char_len - The number of chars in the leaf.
* content - Contents of the leaf.
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
Note that we can have `char_len < str::char_len(content)`, if
this leaf is only a subset of the string. Also note that the
string can be shared between several ropes, e.g. for indexing
purposes.
"]
type leaf = {
byte_offset: uint,
byte_len: uint,
char_len: uint,
content: @str
};
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
#[doc = "
A node obtained from the concatenation of two other nodes
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
# Fields
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
* left - The node containing the beginning of the text.
* right - The node containing the end of the text.
* char_len - The number of chars contained in all leaves of this node.
* byte_len - The number of bytes in the subrope.
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
Used to pre-allocate the correct amount of storage for serialization.
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
* height - Height of the subrope.
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
Used for rebalancing and to allocate stacks for traversals.
"]
2011-11-04 12:38:42 -05:00
type concat = {
left: @node,//TODO: Perhaps a `vec` instead of `left`/`right`
right: @node,
char_len: uint,
byte_len: uint,
height: uint
};
2012-01-19 17:20:57 -06:00
enum node {
2012-03-07 20:17:30 -06:00
#[doc = "A leaf consisting in a `str`"]
2012-01-19 19:55:34 -06:00
leaf(leaf),
2012-03-07 20:17:30 -06:00
#[doc = "The concatenation of two ropes"]
2012-01-19 19:55:34 -06:00
concat(concat),
2011-11-04 12:38:42 -05:00
}
2012-03-07 20:17:30 -06:00
#[doc = "
2011-11-04 12:38:42 -05:00
The maximal number of chars that _should_ be permitted in a single node.
This is not a strict value
2012-03-07 20:17:30 -06:00
"]
2011-11-04 12:38:42 -05:00
const hint_max_leaf_char_len: uint = 256u;
2012-03-07 20:17:30 -06:00
#[doc = "
2011-11-04 12:38:42 -05:00
The maximal height that _should_ be permitted in a tree.
This is not a strict value
2012-03-07 20:17:30 -06:00
"]
2011-11-04 12:38:42 -05:00
const hint_max_node_height: uint = 16u;
2012-03-07 20:17:30 -06:00
#[doc = "
2011-11-04 12:38:42 -05:00
Adopt a string as a node.
If the string is longer than `max_leaf_char_len`, it is
logically split between as many leaves as necessary. Regardless,
the string itself is not copied.
Performance note: The complexity of this function is linear in
the length of `str`.
2012-03-07 20:17:30 -06:00
"]
2011-11-04 12:38:42 -05:00
fn of_str(str: @str) -> @node {
2012-02-23 03:44:04 -06:00
ret of_substr(str, 0u, str::len(*str));
2011-11-04 12:38:42 -05:00
}
2012-03-07 20:17:30 -06:00
#[doc ="
2011-11-04 12:38:42 -05:00
Adopt a slice of a string as a node.
If the slice is longer than `max_leaf_char_len`, it is logically split
between as many leaves as necessary. Regardless, the string itself
is not copied
2011-11-04 12:38:42 -05:00
2012-03-07 20:17:30 -06:00
# Arguments
2012-03-07 20:17:30 -06:00
* byte_start - The byte offset where the slice of `str` starts.
* byte_len - The number of bytes from `str` to use.
# Safety note
Behavior is undefined if `byte_start` or `byte_len` do not represent
valid positions in `str`
"]
2011-11-04 12:38:42 -05:00
fn of_substr(str: @str, byte_start: uint, byte_len: uint) -> @node {
ret of_substr_unsafer(str, byte_start, byte_len,
str::count_chars(*str, byte_start, byte_len));
}
2012-03-07 20:17:30 -06:00
#[doc = "
Adopt a slice of a string as a node.
If the slice is longer than `max_leaf_char_len`, it is logically split
between as many leaves as necessary. Regardless, the string itself
is not copied
2012-03-07 20:17:30 -06:00
# Arguments
* byte_start - The byte offset where the slice of `str` starts.
* byte_len - The number of bytes from `str` to use.
* char_len - The number of chars in `str` in the interval
[byte_start, byte_start+byte_len(
# Safety notes
* Behavior is undefined if `byte_start` or `byte_len` do not represent
valid positions in `str`
* Behavior is undefined if `char_len` does not accurately represent the
number of chars between byte_start and byte_start+byte_len
"]
fn of_substr_unsafer(str: @str, byte_start: uint, byte_len: uint,
char_len: uint) -> @node {
2012-02-23 03:44:04 -06:00
assert(byte_start + byte_len <= str::len(*str));
2011-11-04 12:38:42 -05:00
let candidate = @leaf({
byte_offset: byte_start,
byte_len: byte_len,
char_len: char_len,
content: str});
if char_len <= hint_max_leaf_char_len {
ret candidate;
} else {
//Firstly, split `str` in slices of hint_max_leaf_char_len
let mut leaves = uint::div_ceil(char_len, hint_max_leaf_char_len);
2011-11-04 12:38:42 -05:00
//Number of leaves
2012-03-12 17:52:30 -05:00
let nodes = vec::to_mut(vec::from_elem(leaves, candidate));
2011-11-04 12:38:42 -05:00
let mut i = 0u;
let mut offset = byte_start;
2011-11-04 12:38:42 -05:00
let first_leaf_char_len =
if char_len%hint_max_leaf_char_len == 0u {
hint_max_leaf_char_len
} else {
char_len%hint_max_leaf_char_len
};
while i < leaves {
let chunk_char_len: uint =
if i == 0u { first_leaf_char_len }
else { hint_max_leaf_char_len };
let chunk_byte_len =
str::count_bytes(*str, offset, chunk_char_len);
2011-11-04 12:38:42 -05:00
nodes[i] = @leaf({
byte_offset: offset,
byte_len: chunk_byte_len,
char_len: chunk_char_len,
content: str
});
offset += chunk_byte_len;
i += 1u;
}
//Then, build a tree from these slices by collapsing them
while leaves > 1u {
i = 0u;
while i < leaves - 1u {//Concat nodes 0 with 1, 2 with 3 etc.
nodes[i/2u] = concat2(nodes[i], nodes[i + 1u]);
i += 2u;
}
if i == leaves - 1u {
//And don't forget the last node if it is in even position
nodes[i/2u] = nodes[i];
}
leaves = uint::div_ceil(leaves, 2u);
}
ret nodes[0u];
}
}
pure fn byte_len(node: @node) -> uint {
alt(*node) {//TODO: Could we do this without the pattern-matching?
leaf(y) { ret y.byte_len; }
concat(y){ ret y.byte_len; }
}
}
pure fn char_len(node: @node) -> uint {
alt(*node) {
leaf(y) { ret y.char_len; }
concat(y) { ret y.char_len; }
}
}
2012-03-07 20:17:30 -06:00
#[doc ="
Concatenate a forest of nodes into one tree.
2012-03-07 20:17:30 -06:00
# Arguments
* forest - The forest. This vector is progressively rewritten during
execution and should be discarded as meaningless afterwards.
"]
2012-03-26 20:35:18 -05:00
fn tree_from_forest_destructive(forest: [mut @node]) -> @node {
let mut i = 0u;
let mut len = vec::len(forest);
2011-11-04 12:38:42 -05:00
while len > 1u {
i = 0u;
while i < len - 1u {//Concat nodes 0 with 1, 2 with 3 etc.
let mut left = forest[i];
let mut right = forest[i+1u];
2011-11-04 12:38:42 -05:00
let left_len = char_len(left);
let right_len= char_len(right);
let mut left_height= height(left);
let mut right_height=height(right);
if left_len + right_len > hint_max_leaf_char_len {
2011-11-04 12:38:42 -05:00
if left_len <= hint_max_leaf_char_len {
left = flatten(left);
left_height = height(left);
2011-11-04 12:38:42 -05:00
}
if right_len <= hint_max_leaf_char_len {
right = flatten(right);
right_height = height(right);
2011-11-04 12:38:42 -05:00
}
}
if left_height >= hint_max_node_height {
left = of_substr_unsafer(@serialize_node(left),
0u,byte_len(left),
left_len);
}
if right_height >= hint_max_node_height {
right = of_substr_unsafer(@serialize_node(right),
0u,byte_len(right),
right_len);
}
2011-11-04 12:38:42 -05:00
forest[i/2u] = concat2(left, right);
i += 2u;
}
if i == len - 1u {
//And don't forget the last node if it is in even position
forest[i/2u] = forest[i];
}
len = uint::div_ceil(len, 2u);
}
ret forest[0];
}
fn serialize_node(node: @node) -> str unsafe {
let mut buf = vec::to_mut(vec::from_elem(byte_len(node), 0u8));
let mut offset = 0u;//Current position in the buffer
2011-11-04 12:38:42 -05:00
let it = leaf_iterator::start(node);
loop {
2011-11-04 12:38:42 -05:00
alt(leaf_iterator::next(it)) {
option::none { break; }
2011-11-04 12:38:42 -05:00
option::some(x) {
//TODO: Replace with memcpy or something similar
let mut local_buf: [u8] =
unsafe::reinterpret_cast(*x.content);
let mut i = x.byte_offset;
2011-11-04 12:38:42 -05:00
while i < x.byte_len {
buf[offset] = local_buf[i];
offset += 1u;
i += 1u;
}
unsafe::forget(local_buf);
2011-11-04 12:38:42 -05:00
}
}
}
let str : str = unsafe::reinterpret_cast(buf);
unsafe::forget(buf);//TODO: Check if this is correct
2011-11-04 12:38:42 -05:00
ret str;
}
2012-03-07 20:17:30 -06:00
#[doc ="
2011-11-04 12:38:42 -05:00
Replace a subtree by a single leaf with the same contents.
2012-03-07 20:17:30 -06:00
* Performance note
This function executes in linear time.
"]
2011-11-04 12:38:42 -05:00
fn flatten(node: @node) -> @node unsafe {
alt(*node) {
leaf(_) { ret node }
concat(x) {
ret @leaf({
byte_offset: 0u,
byte_len: x.byte_len,
char_len: x.char_len,
content: @serialize_node(node)
})
}
}
}
2012-03-07 20:17:30 -06:00
#[doc ="
Balance a node.
2012-03-07 20:17:30 -06:00
# Algorithm
* if the node height is smaller than `hint_max_node_height`, do nothing
* otherwise, gather all leaves as a forest, rebuild a balanced node,
concatenating small leaves along the way
2012-03-07 20:17:30 -06:00
# Return value
* `option::none` if no transformation happened
* `option::some(x)` otherwise, in which case `x` has the same contents
as `node` bot lower height and/or fragmentation.
2012-03-07 20:17:30 -06:00
"]
fn bal(node: @node) -> option<@node> {
if height(node) < hint_max_node_height { ret option::none; }
//1. Gather all leaves as a forest
2012-03-26 20:35:18 -05:00
let mut forest = [mut];
let it = leaf_iterator::start(node);
loop {
alt (leaf_iterator::next(it)) {
option::none { break; }
2012-03-26 20:35:18 -05:00
option::some(x) { forest += [mut @leaf(x)]; }
2011-11-04 12:38:42 -05:00
}
}
//2. Rebuild tree from forest
let root = @*tree_from_forest_destructive(forest);
ret option::some(root);
2011-11-04 12:38:42 -05:00
}
2012-03-07 20:17:30 -06:00
#[doc ="
Compute the subnode of a node.
2012-03-07 20:17:30 -06:00
# Arguments
* node - A node
* byte_offset - A byte offset in `node`
* byte_len - The number of bytes to return
# Performance notes
2012-03-07 20:17:30 -06:00
* this function performs no copying;
* this function executes in a time proportional to the height of `node`.
2012-03-07 20:17:30 -06:00
# Safety notes
This function fails if `byte_offset` or `byte_len` do not represent
valid positions in `node`.
2012-03-07 20:17:30 -06:00
"]
2011-11-04 12:38:42 -05:00
fn sub_bytes(node: @node, byte_offset: uint, byte_len: uint) -> @node {
let mut node = node;
let mut byte_offset = byte_offset;
loop {
2011-11-04 12:38:42 -05:00
if byte_offset == 0u && byte_len == node::byte_len(node) {
ret node;
2011-11-04 12:38:42 -05:00
}
alt(*node) {
node::leaf(x) {
let char_len =
str::count_chars(*x.content, byte_offset, byte_len);
ret @leaf({byte_offset: byte_offset,
2011-11-04 12:38:42 -05:00
byte_len: byte_len,
char_len: char_len,
content: x.content});
}
node::concat(x) {
let left_len: uint = node::byte_len(x.left);
if byte_offset <= left_len {
if byte_offset + byte_len <= left_len {
//Case 1: Everything fits in x.left, tail-call
2011-11-04 12:38:42 -05:00
node = x.left;
} else {
//Case 2: A (non-empty, possibly full) suffix
//of x.left and a (non-empty, possibly full) prefix
//of x.right
let left_result =
sub_bytes(x.left, byte_offset, left_len);
let right_result =
sub_bytes(x.right, 0u, left_len - byte_offset);
ret concat2(left_result, right_result);
2011-11-04 12:38:42 -05:00
}
} else {
//Case 3: Everything fits in x.right
byte_offset -= left_len;
node = x.right;
}
}
}
};
2011-11-04 12:38:42 -05:00
}
2012-03-07 20:17:30 -06:00
#[doc ="
Compute the subnode of a node.
2012-03-07 20:17:30 -06:00
# Arguments
* node - A node
* char_offset - A char offset in `node`
* char_len - The number of chars to return
# Performance notes
2012-03-07 20:17:30 -06:00
* this function performs no copying;
* this function executes in a time proportional to the height of `node`.
2012-03-07 20:17:30 -06:00
# Safety notes
This function fails if `char_offset` or `char_len` do not represent
valid positions in `node`.
2012-03-07 20:17:30 -06:00
"]
2011-11-04 12:38:42 -05:00
fn sub_chars(node: @node, char_offset: uint, char_len: uint) -> @node {
let mut node = node;
let mut char_offset = char_offset;
loop {
alt(*node) {
node::leaf(x) {
if char_offset == 0u && char_len == x.char_len {
ret node;
}
let byte_offset =
str::count_bytes(*x.content, 0u, char_offset);
let byte_len =
str::count_bytes(*x.content, byte_offset, char_len);
ret @leaf({byte_offset: byte_offset,
byte_len: byte_len,
char_len: char_len,
content: x.content});
}
node::concat(x) {
if char_offset == 0u && char_len == x.char_len {ret node;}
let left_len : uint = node::char_len(x.left);
if char_offset <= left_len {
if char_offset + char_len <= left_len {
//Case 1: Everything fits in x.left, tail call
node = x.left;
} else {
//Case 2: A (non-empty, possibly full) suffix
//of x.left and a (non-empty, possibly full) prefix
//of x.right
let left_result =
sub_chars(x.left, char_offset, left_len);
let right_result =
sub_chars(x.right, 0u, left_len - char_offset);
ret concat2(left_result, right_result);
}
2011-11-04 12:38:42 -05:00
} else {
//Case 3: Everything fits in x.right, tail call
node = x.right;
char_offset -= left_len;
2011-11-04 12:38:42 -05:00
}
}
2011-11-04 12:38:42 -05:00
}
};
2011-11-04 12:38:42 -05:00
}
fn concat2(left: @node, right: @node) -> @node {
ret @concat({left : left,
right : right,
char_len: char_len(left) + char_len(right),
byte_len: byte_len(left) + byte_len(right),
height: uint::max(height(left), height(right)) + 1u
2011-11-04 12:38:42 -05:00
})
}
fn height(node: @node) -> uint {
alt(*node) {
leaf(_) { ret 0u; }
concat(x) { ret x.height; }
}
}
fn cmp(a: @node, b: @node) -> int {
let ita = char_iterator::start(a);
let itb = char_iterator::start(b);
let mut result = 0;
let mut pos = 0u;
2011-11-04 12:38:42 -05:00
while result == 0 {
alt((char_iterator::next(ita), char_iterator::next(itb))) {
(option::none, option::none) {
2011-11-04 12:38:42 -05:00
break;
}
(option::some(chara), option::some(charb)) {
result = char::cmp(chara, charb);
}
(option::some(_), _) {
result = 1;
}
(_, option::some(_)) {
result = -1;
}
}
pos += 1u;
}
ret result;
}
2012-01-23 16:59:00 -06:00
fn loop_chars(node: @node, it: fn(char) -> bool) -> bool {
2011-11-04 12:38:42 -05:00
ret loop_leaves(node, {|leaf|
str::all_between(*leaf.content,
leaf.byte_offset,
leaf.byte_len, it)
2011-11-04 12:38:42 -05:00
})
}
2012-03-07 20:17:30 -06:00
#[doc ="
Loop through a node, leaf by leaf
2012-03-07 20:17:30 -06:00
# Arguments
2012-03-07 20:17:30 -06:00
* rope - A node to traverse.
* it - A block to execute with each consecutive leaf of the node.
Return `true` to continue, `false` to stop
2012-03-07 20:17:30 -06:00
# Arguments
`true` If execution proceeded correctly, `false` if it was interrupted,
that is if `it` returned `false` at any point.
2012-03-07 20:17:30 -06:00
"]
2012-01-23 16:59:00 -06:00
fn loop_leaves(node: @node, it: fn(leaf) -> bool) -> bool{
let mut current = node;
loop {
2011-11-04 12:38:42 -05:00
alt(*current) {
leaf(x) {
ret it(x);
2011-11-04 12:38:42 -05:00
}
concat(x) {
if loop_leaves(x.left, it) { //non tail call
current = x.right; //tail call
} else {
ret false;
2011-11-04 12:38:42 -05:00
}
}
}
};
2011-11-04 12:38:42 -05:00
}
2012-03-07 20:17:30 -06:00
#[doc ="
# Arguments
* pos - A position in the rope
# Return value
2012-03-07 20:17:30 -06:00
The character at position `pos`
2012-03-07 20:17:30 -06:00
# Safety notes
2012-03-07 20:17:30 -06:00
The function will fail if `pos` is not a valid position in the rope.
Performance note: This function executes in a time
proportional to the height of the rope + the (bounded)
length of the largest leaf.
2012-03-07 20:17:30 -06:00
"]
fn char_at(node: @node, pos: uint) -> char {
let mut node = node;
let mut pos = pos;
loop {
alt *node {
leaf(x) {
ret str::char_at(*x.content, pos);
}
concat({left, right, _}) {
let left_len = char_len(left);
node = if left_len > pos { left }
else { pos -= left_len; right };
}
}
};
}
2011-11-04 12:38:42 -05:00
mod leaf_iterator {
type t = {
2012-03-26 20:35:18 -05:00
stack: [mut @node],
mut stackpos: int
2011-11-04 12:38:42 -05:00
};
fn empty() -> t {
2012-03-26 20:35:18 -05:00
let stack : [mut @node] = [mut];
ret {stack: stack, mut stackpos: -1}
2011-11-04 12:38:42 -05:00
}
fn start(node: @node) -> t {
2012-03-12 17:52:30 -05:00
let stack = vec::to_mut(vec::from_elem(height(node)+1u, node));
2011-11-04 12:38:42 -05:00
ret {
stack: stack,
2012-03-26 20:35:18 -05:00
mut stackpos: 0
2011-11-04 12:38:42 -05:00
}
}
fn next(it: t) -> option<leaf> {
2011-11-04 12:38:42 -05:00
if it.stackpos < 0 { ret option::none; }
loop {
2011-11-04 12:38:42 -05:00
let current = it.stack[it.stackpos];
it.stackpos -= 1;
alt(*current) {
concat(x) {
it.stackpos += 1;
it.stack[it.stackpos] = x.right;
it.stackpos += 1;
it.stack[it.stackpos] = x.left;
}
leaf(x) {
ret option::some(x);
2011-11-04 12:38:42 -05:00
}
}
};
2011-11-04 12:38:42 -05:00
}
}
mod char_iterator {
type t = {
leaf_iterator: leaf_iterator::t,
2012-03-26 20:35:18 -05:00
mut leaf: option<leaf>,
mut leaf_byte_pos: uint
2011-11-04 12:38:42 -05:00
};
fn start(node: @node) -> t {
ret {
leaf_iterator: leaf_iterator::start(node),
2012-03-26 20:35:18 -05:00
mut leaf: option::none,
mut leaf_byte_pos: 0u
2011-11-04 12:38:42 -05:00
}
}
fn empty() -> t {
ret {
leaf_iterator: leaf_iterator::empty(),
2012-03-26 20:35:18 -05:00
mut leaf: option::none,
mut leaf_byte_pos: 0u
2011-11-04 12:38:42 -05:00
}
}
fn next(it: t) -> option<char> {
loop {
2011-11-04 12:38:42 -05:00
alt(get_current_or_next_leaf(it)) {
option::none { ret option::none; }
option::some(_) {
2011-11-04 12:38:42 -05:00
let next_char = get_next_char_in_leaf(it);
alt(next_char) {
option::none {
2011-11-04 12:38:42 -05:00
cont;
}
option::some(_) {
ret next_char;
2011-11-04 12:38:42 -05:00
}
}
}
}
};
2011-11-04 12:38:42 -05:00
}
fn get_current_or_next_leaf(it: t) -> option<leaf> {
2011-11-04 12:38:42 -05:00
alt(it.leaf) {
option::some(_) { ret it.leaf }
option::none {
2011-11-04 12:38:42 -05:00
let next = leaf_iterator::next(it.leaf_iterator);
alt(next) {
option::none { ret option::none }
option::some(_) {
2011-11-04 12:38:42 -05:00
it.leaf = next;
it.leaf_byte_pos = 0u;
ret next;
}
}
}
}
}
fn get_next_char_in_leaf(it: t) -> option<char> {
2011-11-04 12:38:42 -05:00
alt(it.leaf) {
option::none { ret option::none }
option::some(aleaf) {
if it.leaf_byte_pos >= aleaf.byte_len {
2011-11-04 12:38:42 -05:00
//We are actually past the end of the leaf
it.leaf = option::none;
ret option::none
} else {
let {ch, next} =
str::char_range_at(*aleaf.content,
it.leaf_byte_pos + aleaf.byte_offset);
it.leaf_byte_pos = next - aleaf.byte_offset;
2011-11-04 12:38:42 -05:00
ret option::some(ch)
}
}
}
}
}
}
2012-01-17 21:05:07 -06:00
#[cfg(test)]
mod tests {
//Utility function, used for sanity check
fn rope_to_string(r: rope) -> str {
alt(r) {
node::empty { ret "" }
2012-01-17 21:05:07 -06:00
node::content(x) {
2012-03-26 20:35:18 -05:00
let str = @mut "";
fn aux(str: @mut str, node: @node::node) unsafe {
2012-01-17 21:05:07 -06:00
alt(*node) {
node::leaf(x) {
*str += str::slice(
*x.content, x.byte_offset,
x.byte_offset + x.byte_len);
2012-01-17 21:05:07 -06:00
}
node::concat(x) {
aux(str, x.left);
aux(str, x.right);
}
}
}
aux(str, x);
ret *str
}
}
}
#[test]
fn trivial() {
assert char_len(empty()) == 0u;
assert byte_len(empty()) == 0u;
}
#[test]
fn of_string1() {
let sample = @"0123456789ABCDE";
let r = of_str(sample);
assert char_len(r) == str::char_len(*sample);
2012-01-17 21:05:07 -06:00
assert rope_to_string(r) == *sample;
}
#[test]
fn of_string2() {
2012-03-26 20:35:18 -05:00
let buf = @ mut "1234567890";
let mut i = 0;
2012-01-17 21:05:07 -06:00
while i < 10 { *buf = *buf + *buf; i+=1;}
let sample = @*buf;
let r = of_str(sample);
assert char_len(r) == str::char_len(*sample);
2012-01-17 21:05:07 -06:00
assert rope_to_string(r) == *sample;
let mut string_iter = 0u;
2012-02-23 03:44:04 -06:00
let string_len = str::len(*sample);
2012-01-17 21:05:07 -06:00
let rope_iter = iterator::char::start(r);
let mut equal = true;
let mut pos = 0u;
2012-01-17 21:05:07 -06:00
while equal {
alt(node::char_iterator::next(rope_iter)) {
option::none {
2012-01-17 21:05:07 -06:00
if string_iter < string_len {
equal = false;
} break; }
option::some(c) {
let {ch, next} = str::char_range_at(*sample, string_iter);
string_iter = next;
if ch != c { equal = false; break; }
}
}
pos += 1u;
}
assert equal;
}
#[test]
fn iter1() {
2012-03-26 20:35:18 -05:00
let buf = @ mut "1234567890";
let mut i = 0;
2012-01-17 21:05:07 -06:00
while i < 10 { *buf = *buf + *buf; i+=1;}
let sample = @*buf;
let r = of_str(sample);
let mut len = 0u;
2012-01-17 21:05:07 -06:00
let it = iterator::char::start(r);
loop {
2012-01-17 21:05:07 -06:00
alt(node::char_iterator::next(it)) {
option::none { break; }
2012-01-17 21:05:07 -06:00
option::some(_) { len += 1u; }
}
}
assert len == str::char_len(*sample);
2012-01-17 21:05:07 -06:00
}
#[test]
fn bal1() {
let init = @ "1234567890";
2012-03-26 20:35:18 -05:00
let buf = @ mut * init;
let mut i = 0;
2012-01-17 21:05:07 -06:00
while i < 8 { *buf = *buf + *buf; i+=1;}
let sample = @*buf;
let r1 = of_str(sample);
let mut r2 = of_str(init);
2012-01-17 21:05:07 -06:00
i = 0;
while i < 8 { r2 = append_rope(r2, r2); i+= 1;}
assert eq(r1, r2);
let r3 = bal(r2);
assert char_len(r1) == char_len(r3);
assert eq(r1, r3);
}
#[test]
fn char_at1() {
//Generate a large rope
let mut r = of_str(@ "123456789");
2012-01-17 21:05:07 -06:00
uint::range(0u, 10u){|_i|
r = append_rope(r, r);
}
//Copy it in the slowest possible way
let mut r2 = empty();
2012-01-17 21:05:07 -06:00
uint::range(0u, char_len(r)){|i|
r2 = append_char(r2, char_at(r, i));
}
assert eq(r, r2);
let mut r3 = empty();
2012-01-17 21:05:07 -06:00
uint::range(0u, char_len(r)){|i|
r3 = prepend_char(r3, char_at(r, char_len(r) - i - 1u));
}
assert eq(r, r3);
//Additional sanity checks
let balr = bal(r);
let bal2 = bal(r2);
let bal3 = bal(r3);
assert eq(r, balr);
assert eq(r, bal2);
assert eq(r, bal3);
assert eq(r2, r3);
assert eq(bal2, bal3);
}
#[test]
fn concat1() {
//Generate a reasonable rope
let chunk = of_str(@ "123456789");
let mut r = empty();
2012-01-17 21:05:07 -06:00
uint::range(0u, 10u){|_i|
r = append_rope(r, chunk);
}
//Same rope, obtained with rope::concat
2012-03-12 17:52:30 -05:00
let r2 = concat(vec::from_elem(10u, chunk));
2012-01-17 21:05:07 -06:00
assert eq(r, r2);
}
}