stonewareslord/this_algorithm
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Update to use four digits and 4k wordlist size

Browse Source
This commit is contained in:
Austen Adler 2023-03-06 22:57:07 -05:00
parent 9aa6a68de0
commit db83a68923
7 changed files with 54 additions and 48 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

16
docs/ALGORITHM.adoc
View File

@ -19,14 +19,14 @@ If you want to see the steps to get to this definition, go to link:DESIGN.html[D
[source,title='xpin and S2 CellID Format'] [source,title='xpin and S2 CellID Format']
---- ----
=== xpin Format === === xpin Format ===
WORD2 (13 bits) : vvvvvvvvvvvvv WORD2 (12 bits) [52, 63] : vvvvvvvvvvvv
WORD1 (13 bits) : | |vvv vvvvvvvvvv WORD1 (12 bits) [40, 51] : | |vvvv vvvvvvvv
WORD0 (13 bits) : | | |vvvvvv vvvvvvv WORD0 (12 bits) [28, 39] : | | |vvvvvvvv vvvv
0000 (10 bits) : | | | |vvvvvvvvv v 0000 (13 bits) [15, 27] : | | | |vvvvvvvvvvvv v
Not represented : | | | | | Not represented : | | | | |
: | | | | | : | | | | |
Bit : 63 51 48 38 32 25 16| 0 Bit : 63 52 48 40 32 28 16| 0
: | | | | | | | | | : | | | | | | | | |
: 0100101110101000 1011100010010011 1001001100100100 1100000000000000 : 0100101110101000 1011100010010011 1001001100100100 1100000000000000
=== S2 CellID Format === | | || | === S2 CellID Format === | | || |
Face number : ^^^ || | Face number : ^^^ || |

17
justfile
View File

@ -1,6 +1,6 @@
build: fmt rust-build build-wasm js-build build: fmt rust-build wasm-build js-build
all: clean fmt build build-docs # rust-test all: clean fmt build docs-build # rust-test
rsync -ha ./web-frontend/build/ build/ rsync -ha ./web-frontend/build/ build/
du -shc build/* | sort -h du -shc build/* | sort -h
@ -11,16 +11,21 @@ rust-test:
rust-build: rust-build:
cargo build --all cargo build --all
build-wasm: wasm-build:
wasm-pack build --target web xpin-wasm wasm-pack build --target web xpin-wasm
js-build: build-wasm wordlist-build:
. wordlist/venv/bin/activate && cd wordlist && for i in *.py; do "./${i}"; done
js-build: wasm-build docs-build
# mkdir -p ./web-frontend/static/docs/
# rsync -ha ./build/docs/ ./web-frontend/static/docs/
yarn --cwd ./web-frontend/ build yarn --cwd ./web-frontend/ build
clean: clean:
rm -vrf build rm -vrf build
build-docs: docs-build:
earthly +docs earthly +docs
cargo doc --all cargo doc --all
@ -43,3 +48,5 @@ init:
yarn --cwd ./web-frontend/ yarn --cwd ./web-frontend/
cargo fetch cargo fetch
cargo install wasm-pack cargo install wasm-pack
if [ ! -d "wordlist/venv" ]; then python3 -m virtualenv wordlist/venv -p "$(which python3)"; fi; export OSTYPE=linux-gnu && . wordlist/venv/bin/activate
export OSTYPE=linux-gnu && . wordlist/venv/bin/activate && pip install -r wordlist/requirements.txt && python -c 'import nltk; nltk.download('wordnet')'

3
src/conversions.rs
View File

@ -2,6 +2,9 @@ use std::ops::RangeInclusive;
use s2::{cellid::CellID, latlng::LatLng}; use s2::{cellid::CellID, latlng::LatLng};
pub(crate) const TWELVE_BITS: u64 = 0b1111_1111_1111;
pub(crate) const TEN_BITS: u64 = 0b1111_1111_11;
use crate::Error; use crate::Error;
pub fn lat_lon_to_cellid(lat: f64, lon: f64) -> Result<CellID, Error> { pub fn lat_lon_to_cellid(lat: f64, lon: f64) -> Result<CellID, Error> {

23
src/lib.rs
View File

@ -23,9 +23,9 @@ pub type Number = u32;
pub const CELLID_LEVEL: u64 = 23; pub const CELLID_LEVEL: u64 = 23;
/// The maximum number value for V0 /// The maximum number value for V0
pub const V0_MAX_NUMBER: u32 = 1024; pub const V0_MAX_NUMBER: u32 = 9120;
/// The minimum number value for V0 /// The minimum number value for V0
pub const V0_MIN_NUMBER: u32 = 1; pub const V0_MIN_NUMBER: u32 = 1024;
// Any encoding or decoding error // Any encoding or decoding error
#[derive(Error, Debug, Eq, PartialEq)] #[derive(Error, Debug, Eq, PartialEq)]
@ -186,9 +186,10 @@ impl Address<'_> {
} }
fn from_components(number: Number, other_components: &[&str]) -> Result<Self, Error> { fn from_components(number: Number, other_components: &[&str]) -> Result<Self, Error> {
match extract_version(number)? { // match extract_version(number)? {
Version::V0 => Self::parse_v0(number, other_components), // Version::V0 => Self::parse_v0(number, other_components),
} // }
Self::parse_v0(number, other_components)
} }
/// Get the address as a [`CellID`] value /// Get the address as a [`CellID`] value
@ -211,12 +212,12 @@ impl Address<'_> {
/// ///
/// The version number is set by the two bits 11 and 12 /// The version number is set by the two bits 11 and 12
// TODO: impl TryFrom ? // TODO: impl TryFrom ?
const fn extract_version(number: Number) -> Result<Version, Error> { // const fn extract_version(number: Number) -> Result<Version, Error> {
match ((number >> 10) & 0b11) as u8 { // match ((number >> 10) & 0b11) as u8 {
0 => Ok(Version::V0), // 0 => Ok(Version::V0),
v => Err(Error::UnimplementedVersion(v)), // v => Err(Error::UnimplementedVersion(v)),
} // }
} // }
impl Display for Address<'_> { impl Display for Address<'_> {
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result { fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {

32
src/v0.rs
View File

@ -2,6 +2,8 @@ use crate::{conversions, Address, Version, CELLID_LEVEL};
use std::ops::{Add, RangeInclusive}; use std::ops::{Add, RangeInclusive};
use words::NUMBER_TO_WORDS; use words::NUMBER_TO_WORDS;
use conversions::TEN_BITS;
use conversions::TWELVE_BITS;
use s2::{cell::Cell, cellid::CellID}; use s2::{cell::Cell, cellid::CellID};
pub struct UnpackedCellID { pub struct UnpackedCellID {
@ -21,10 +23,10 @@ impl From<CellID> for UnpackedCellID {
impl From<u64> for UnpackedCellID { impl From<u64> for UnpackedCellID {
fn from(cellid: u64) -> Self { fn from(cellid: u64) -> Self {
Self { Self {
number_bits: conversions::extract_binary(cellid, 15..=24) as u16, number_bits: conversions::extract_binary(cellid, 15..=27) as u16,
word0_bits: conversions::extract_binary(cellid, 25..=37) as u16, word0_bits: conversions::extract_binary(cellid, 28..=39) as u16,
word1_bits: conversions::extract_binary(cellid, 38..=50) as u16, word1_bits: conversions::extract_binary(cellid, 40..=51) as u16,
word2_bits: conversions::extract_binary(cellid, 51..=63) as u16, word2_bits: conversions::extract_binary(cellid, 52..=63) as u16,
} }
} }
} }
@ -36,7 +38,7 @@ impl From<UnpackedCellID> for Address<'_> {
let word2 = words::NUMBER_TO_WORDS[unpacked_cellid.word2_bits as usize][0]; let word2 = words::NUMBER_TO_WORDS[unpacked_cellid.word2_bits as usize][0];
Self { Self {
number: u32::from(unpacked_cellid.number_bits), number: u32::from(unpacked_cellid.number_bits) + 1024,
words: [word0, word1, word2], words: [word0, word1, word2],
version: Version::V0, version: Version::V0,
} }
@ -46,7 +48,7 @@ impl From<UnpackedCellID> for Address<'_> {
impl From<&Address<'_>> for UnpackedCellID { impl From<&Address<'_>> for UnpackedCellID {
fn from(value: &Address) -> Self { fn from(value: &Address) -> Self {
Self { Self {
number_bits: value.number as u16, number_bits: value.number as u16 - 1024,
word0_bits: value.words[0].number, word0_bits: value.words[0].number,
word1_bits: value.words[1].number, word1_bits: value.words[1].number,
word2_bits: value.words[2].number, word2_bits: value.words[2].number,
@ -57,17 +59,15 @@ impl From<&Address<'_>> for UnpackedCellID {
#[allow(clippy::use_self)] #[allow(clippy::use_self)]
impl From<UnpackedCellID> for u64 { impl From<UnpackedCellID> for u64 {
fn from(value: UnpackedCellID) -> Self { fn from(value: UnpackedCellID) -> Self {
let ten_bits = 0b1111_1111_11;
let thirteen_bits = 0b1_1111_1111_1111;
let mut ret = 0b0; let mut ret = 0b0;
// Add words in reverse order // Add words in reverse order
ret = (ret << 13) | (u64::from(value.word2_bits) & thirteen_bits); ret = (ret << 12) | (u64::from(value.word2_bits) & TWELVE_BITS);
ret = (ret << 13) | (u64::from(value.word1_bits) & thirteen_bits); ret = (ret << 12) | (u64::from(value.word1_bits) & TWELVE_BITS);
ret = (ret << 13) | (u64::from(value.word0_bits) & thirteen_bits); ret = (ret << 12) | (u64::from(value.word0_bits) & TWELVE_BITS);
// Add the number // Add the number
ret = (ret << 10) | (u64::from(value.number_bits) & ten_bits); ret = (ret << 13) | (u64::from(value.number_bits) & TEN_BITS);
// Add the final bit // Add the final bit
ret = (ret << 1) | 0b1; ret = (ret << 1) | 0b1;
@ -84,11 +84,3 @@ impl From<UnpackedCellID> for CellID {
Self(value.into()) Self(value.into())
} }
} }
// impl<A: AsRef<Address<'a>>> From<A> for UnpackedCellID {
// fn from(addr: A) -> Self {
// let number_bits = addr.
// CellID(ret)
// }
// }

9
wordlist/03-exclude.csv
View File

@ -79,6 +79,7 @@ ALBERTA
ALBION ALBION
ALBUMIN ALBUMIN
ALBUQUERQUE ALBUQUERQUE
ALCOHOL
ALCOHOLISM ALCOHOLISM
ALDER ALDER
ALDERMAN ALDERMAN
@ -382,6 +383,8 @@ BLEED
BLEEDING BLEEDING
BLEST BLEST
BLEW BLEW
BLOWING
BLUE
BLUNT BLUNT
BO BO
BOB BOB
@ -880,6 +883,7 @@ DENMARK
DENNIS DENNIS
DENT DENT
DENVER DENVER
DEPRESSION
DEPRESSIVE DEPRESSIVE
DERBY DERBY
DERBYSHIRE DERBYSHIRE
@ -1120,6 +1124,7 @@ FAHRENHEIT
FAIL FAIL
FAILING FAILING
FAILURE FAILURE
FAINT
FAIRBANKS FAIRBANKS
FAIRFAX FAIRFAX
FAKE FAKE
@ -3333,7 +3338,3 @@ ZOOLOGICAL
ZOOLOGY ZOOLOGY
ZULU ZULU
ZURICH ZURICH
ALCOHOL
DEPRESSION
FAINT
BLOWING

1 WORD
79 ALBION
80 ALBUMIN
81 ALBUQUERQUE
82 ALCOHOL
83 ALCOHOLISM
84 ALDER
85 ALDERMAN
383 BLEEDING
384 BLEST
385 BLEW
386 BLOWING
387 BLUE
388 BLUNT
389 BO
390 BOB
883 DENNIS
884 DENT
885 DENVER
886 DEPRESSION
887 DEPRESSIVE
888 DERBY
889 DERBYSHIRE
1124 FAIL
1125 FAILING
1126 FAILURE
1127 FAINT
1128 FAIRBANKS
1129 FAIRFAX
1130 FAKE
3338 ZOOLOGY
3339 ZULU
3340 ZURICH
ALCOHOL
DEPRESSION
FAINT
BLOWING

2
words/build.rs
View File

@ -98,10 +98,12 @@ pub static NUMBER_TO_WORDS: &[&[&Word]] = &["#
#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)] #[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
pub struct Word { pub struct Word {
/// The word itself /// The word itself
#[serde(rename = "WORD")]
pub word: String, pub word: String,
/// The binary representation of this number /// The binary representation of this number
/// ///
/// The words are responsible for 13 bits of data, so this is fine to fit in a u16 /// The words are responsible for 13 bits of data, so this is fine to fit in a u16
#[serde(rename = "NUMBER")]
pub number: u16, pub number: u16,
} }