1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
|
//@ charon-args=--opaque=utils
//@ [!borrow-check] aeneas-args=-state -split-files
//@ [coq] aeneas-args=-use-fuel
//@ [fstar] aeneas-args=-decreases-clauses -template-clauses
//@ [lean] aeneas-args=-no-gen-lib-entry
// ^ the `-no-gen-lib-entry` is because we add a custom import in the Hashmap.lean file: we do not
// want to overwrite it.
// Possible to add `--no-code-duplication` if we use the optimized MIR
// TODO: reactivate -test-trans-units
//! A hashmap implementation.
//!
//! Current limitations:
//! - all the recursive functions should be rewritten with loops, once
//! we have support for this.
//! - we will need function pointers/closures if we want to make the map
//! generic in the key type (having function pointers allows to mimic traits)
//! - for the "get" functions: we don't support borrows inside of enumerations
//! for now, so we can't return a type like `Option<&T>`. The real restriction
//! we currently have on borrows is that we forbid *nested* borrows in function
//! signatures, like in `&'a mut &'b mut T` (and the real problem comes from
//! nested *lifetimes*, not nested borrows). Getting the borrows inside of
//! enumerations mostly requires to pour some implementation time in it.
use std::vec::Vec;
pub type Key = usize; // TODO: make this generic
pub type Hash = usize;
pub enum List<T> {
Cons(Key, T, Box<List<T>>),
Nil,
}
/// A hash function for the keys.
/// Rk.: we use shared references because we anticipate on the generic
/// hash map version.
pub fn hash_key(k: &Key) -> Hash {
// Do nothing for now, we might want to implement something smarter
// in the future, or to call an external function (which will be
// abstract): we don't need to reason about the hash function.
*k
}
/// A hash map from [u64] to values
pub struct HashMap<T> {
/// The current number of entries in the table
num_entries: usize,
/// The max load factor, expressed as a fraction
max_load_factor: (usize, usize),
/// The max load factor applied to the current table length:
/// gives the threshold at which to resize the table.
max_load: usize,
/// The table itself
slots: Vec<List<T>>,
}
impl<T> HashMap<T> {
/// Allocate a vector of slots of a given size.
/// We would need a loop, but can't use loops for now...
fn allocate_slots(mut slots: Vec<List<T>>, mut n: usize) -> Vec<List<T>> {
while n > 0 {
slots.push(List::Nil);
n -= 1;
}
slots
}
/// Create a new table, with a given capacity
fn new_with_capacity(
capacity: usize,
max_load_dividend: usize,
max_load_divisor: usize,
) -> Self {
// TODO: better to use `Vec::with_capacity(capacity)` instead
// of `Vec::new()`
let slots = HashMap::allocate_slots(Vec::new(), capacity);
HashMap {
num_entries: 0,
max_load_factor: (max_load_dividend, max_load_divisor),
max_load: (capacity * max_load_dividend) / max_load_divisor,
slots,
}
}
pub fn new() -> Self {
// For now we create a table with 32 slots and a max load factor of 4/5
HashMap::new_with_capacity(32, 4, 5)
}
pub fn clear(&mut self) {
self.num_entries = 0;
let slots = &mut self.slots;
let mut i = 0;
while i < slots.len() {
slots[i] = List::Nil;
i += 1;
}
}
pub fn len(&self) -> usize {
self.num_entries
}
/// Insert in a list.
/// Return `true` if we inserted an element, `false` if we simply updated
/// a value.
fn insert_in_list(key: Key, value: T, mut ls: &mut List<T>) -> bool {
loop {
match ls {
List::Nil => {
*ls = List::Cons(key, value, Box::new(List::Nil));
return true;
}
List::Cons(ckey, cvalue, tl) => {
if *ckey == key {
*cvalue = value;
return false;
} else {
ls = tl;
}
}
}
}
}
/// Auxiliary function to insert in the hashmap without triggering a resize
fn insert_no_resize(&mut self, key: Key, value: T) {
let hash = hash_key(&key);
let hash_mod = hash % self.slots.len();
// We may want to use slots[...] instead of get_mut...
let inserted = HashMap::insert_in_list(key, value, &mut self.slots[hash_mod]);
if inserted {
self.num_entries += 1;
}
}
/// Insertion function.
/// May trigger a resize of the hash table.
pub fn insert(&mut self, key: Key, value: T) {
// Insert
self.insert_no_resize(key, value);
// Resize if necessary
if self.len() > self.max_load {
self.try_resize()
}
}
/// The resize function, called if we need to resize the table after
/// an insertion.
fn try_resize(&mut self) {
// Check that we can resize: we need to check that there are no overflows.
// Note that we are conservative about the usize::MAX.
// Also note that `as usize` is a trait, but we apply it to a constant
// here, which gets compiled by the MIR interpreter (so we don't see
// the conversion, actually).
// Rk.: this is a hit heavy...
let max_usize = u32::MAX as usize;
let capacity = self.slots.len();
// Checking that there won't be overflows by using the fact that, if m > 0:
// n * m <= p <==> n <= p / m
let n1 = max_usize / 2;
if capacity <= n1 / self.max_load_factor.0 {
// Create a new table with a higher capacity
let mut ntable = HashMap::new_with_capacity(
capacity * 2,
self.max_load_factor.0,
self.max_load_factor.1,
);
// Move the elements to the new table
HashMap::move_elements(&mut ntable, &mut self.slots, 0);
// Replace the current table with the new table
self.slots = ntable.slots;
self.max_load = ntable.max_load;
}
}
/// Auxiliary function called by [try_resize] to move all the elements
/// from the table to a new table
fn move_elements<'a>(ntable: &'a mut HashMap<T>, slots: &'a mut Vec<List<T>>, mut i: usize) {
while i < slots.len() {
// Move the elements out of the slot i
let ls = std::mem::replace(&mut slots[i], List::Nil);
// Move all those elements to the new table
HashMap::move_elements_from_list(ntable, ls);
// Do the same for slot i+1
i += 1;
}
}
/// Auxiliary function.
fn move_elements_from_list(ntable: &mut HashMap<T>, mut ls: List<T>) {
// As long as there are elements in the list, move them
loop {
match ls {
List::Nil => return, // We're done
List::Cons(k, v, tl) => {
// Insert the element in the new table
ntable.insert_no_resize(k, v);
// Move the elements out of the tail
ls = *tl;
}
}
}
}
/// Returns `true` if the map contains a value for the specified key.
pub fn contains_key(&self, key: &Key) -> bool {
let hash = hash_key(key);
let hash_mod = hash % self.slots.len();
HashMap::contains_key_in_list(key, &self.slots[hash_mod])
}
/// Returns `true` if the list contains a value for the specified key.
pub fn contains_key_in_list(key: &Key, mut ls: &List<T>) -> bool {
loop {
match ls {
List::Nil => return false,
List::Cons(ckey, _, tl) => {
if *ckey == *key {
return true;
} else {
ls = tl;
}
}
}
}
}
/// We don't support borrows inside of enumerations for now, so we
/// can't return an option...
/// TODO: add support for that
fn get_in_list<'a, 'k>(key: &'k Key, mut ls: &'a List<T>) -> &'a T {
loop {
match ls {
List::Nil => panic!(),
List::Cons(ckey, cvalue, tl) => {
if *ckey == *key {
return cvalue;
} else {
ls = tl;
}
}
}
}
}
pub fn get<'a, 'k>(&'a self, key: &'k Key) -> &'a T {
let hash = hash_key(key);
let hash_mod = hash % self.slots.len();
HashMap::get_in_list(key, &self.slots[hash_mod])
}
pub fn get_mut_in_list<'a, 'k>(mut ls: &'a mut List<T>, key: &'k Key) -> &'a mut T {
while let List::Cons(ckey, cvalue, tl) = ls {
if *ckey == *key {
return cvalue;
} else {
ls = tl;
}
}
panic!()
}
/// Same remark as for [get].
pub fn get_mut<'a, 'k>(&'a mut self, key: &'k Key) -> &'a mut T {
let hash = hash_key(key);
let hash_mod = hash % self.slots.len();
HashMap::get_mut_in_list(&mut self.slots[hash_mod], key)
}
/// Remove an element from the list.
/// Return the removed element.
fn remove_from_list(key: &Key, mut ls: &mut List<T>) -> Option<T> {
loop {
match ls {
List::Nil => return None,
// We have to use a guard and split the Cons case into two
// branches, otherwise the borrow checker is not happy.
List::Cons(ckey, _, _) if *ckey == *key => {
// We have to move under borrows, so we need to use
// [std::mem::replace] in several steps.
// Retrieve the tail
let mv_ls = std::mem::replace(ls, List::Nil);
match mv_ls {
List::Nil => unreachable!(),
List::Cons(_, cvalue, tl) => {
// Make the list equal to its tail
*ls = *tl;
// Return the dropped value
return Some(cvalue);
}
}
}
List::Cons(_, _, tl) => {
ls = tl;
}
}
}
}
/// Same remark as for [get].
pub fn remove(&mut self, key: &Key) -> Option<T> {
let hash = hash_key(key);
let hash_mod = hash % self.slots.len();
let x = HashMap::remove_from_list(key, &mut self.slots[hash_mod]);
match x {
Option::None => Option::None,
Option::Some(x) => {
self.num_entries -= 1;
Option::Some(x)
}
}
}
}
// This is a module so we can tell charon to leave it opaque
mod utils {
use crate::*;
/// Serialize a hash map - we don't have traits, so we fix the type of the
/// values (this is not the interesting part anyway)
pub(crate) fn serialize(_hm: HashMap<u64>) {
unimplemented!();
}
/// Deserialize a hash map - we don't have traits, so we fix the type of the
/// values (this is not the interesting part anyway)
pub(crate) fn deserialize() -> HashMap<u64> {
unimplemented!();
}
}
pub fn insert_on_disk(key: Key, value: u64) {
// Deserialize
let mut hm = utils::deserialize();
// Update
hm.insert(key, value);
// Serialize
utils::serialize(hm);
}
/// I currently can't retrieve functions marked with the attribute #[test],
/// while I want to extract the unit tests and use the normalize on them,
/// so I have to define the test functions somewhere and call them from
/// a test function.
/// TODO: find a way to do that.
#[allow(dead_code)]
fn test1() {
let mut hm: HashMap<u64> = HashMap::new();
hm.insert(0, 42);
hm.insert(128, 18);
hm.insert(1024, 138);
hm.insert(1056, 256);
// Rk.: `&128` introduces a ref constant value
// TODO: add support for this
// Rk.: this only happens if we query the MIR too late (for instance,
// the optimized MIR). It is not a problem if we query, say, the
// "built" MIR.
let k = 128;
assert!(*hm.get(&k) == 18);
let k = 1024;
let x = hm.get_mut(&k);
*x = 56;
assert!(*hm.get(&k) == 56);
let x = hm.remove(&k);
// If we write `x == Option::Some(56)` rust introduces
// a call to `core::cmp::PartialEq::eq`, which is a trait
// I don't support for now.
// Also, I haven't implemented support for `unwrap` yet...
match x {
Option::None => panic!(),
Option::Some(x) => assert!(x == 56),
};
let k = 0;
assert!(*hm.get(&k) == 42);
let k = 128;
assert!(*hm.get(&k) == 18);
let k = 1056;
assert!(*hm.get(&k) == 256);
}
#[test]
fn tests() {
test1();
}
|