File size: 14,458 Bytes
d5bfab8 |
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 |
//! The `loda-rust similar` subcommand, identifies similar programs.
use crate::analytics::{Analytics, AnalyticsDirectory};
use crate::common::{find_asm_files_recursively, oeis_id_from_path};
use crate::common::RecordBigram;
use crate::common::SimpleLog;
use super::{Word, WordPair, WordsFromProgram};
use crate::config::Config;
use loda_rust_core::parser::ParsedProgram;
use loda_rust_core::oeis::OeisId;
use std::time::Instant;
use std::path::{Path, PathBuf};
use std::fs;
use std::collections::HashMap;
use std::collections::HashSet;
use rand::seq::SliceRandom;
use rand::SeedableRng;
use rand::rngs::StdRng;
use bit_set::BitSet;
use csv::WriterBuilder;
use serde::Serialize;
use console::Style;
use indicatif::{HumanDuration, ProgressBar};
use anyhow::Context;
const SIGNATURE_LENGTH: u8 = 30;
const IGNORE_INPUT_PROGRAM_IF_INSTRUCTION_COUNT_EXCEEDS: usize = 300;
const MAX_NUMBER_OF_ROWS_IN_OUTPUT_CSV_FILE: usize = 25;
/// Identify similar programs.
///
/// This code uses the [LSH/MinHash] algorithm.
///
/// My first attempt was slow. It was coded in Ruby and took 8 hours to compute similarity
/// for 1600 programs out of 41000 programs.
/// For this I used the unix diff tool for identifying the number of identical lines.
/// If there were more than 80% identical lines, then I added it to the csv file.
///
/// My second attempt was using [LSH/MinHash]. It was coded in Ruby.
/// All 41000 programs in the loda-programs repository can be traversed in 2 hours.
///
/// My third attempt at finding similarity between programs. This time using Rust for better preformance.
/// And using the [LSH/MinHash] algorithm.
/// All 95000 programs in the loda-programs repository can be traversed in 9 minutes.
///
/// [LSH/MinHash]: https://en.wikipedia.org/wiki/MinHash
pub struct Similar {
analytics_directory: AnalyticsDirectory,
simple_log: SimpleLog,
config: Config,
similar_programs: PathBuf,
}
impl Similar {
pub fn run() -> anyhow::Result<()> {
// Regenerate the `~/.loda-rust/analytics` directory
Analytics::oeis_run_if_expired()?;
let config = Config::load();
if !config.analytics_oeis_dir().is_dir() {
return Err(anyhow::anyhow!("Expected dir: {:?}, but it's missing.", config.analytics_oeis_dir()));
}
let analytics_directory = AnalyticsDirectory::new(
config.analytics_oeis_dir()
).with_context(||"unable to create AnalyticsDirectory instance")?;
// Ensure that the `similar-programs` dir exist
let similar_programs: PathBuf = config.similar_programs();
if !similar_programs.is_dir() {
fs::create_dir(&similar_programs)
.with_context(|| format!("Could not create dir: {:?}", similar_programs))?;
}
let log_path: PathBuf = similar_programs.join(Path::new("similar_log.txt"));
let simple_log = SimpleLog::new(&log_path)
.map_err(|e| anyhow::anyhow!("Unable to create log file at path: {:?}. error: {:?}", log_path, e))?;
let mut instance = Self {
analytics_directory,
simple_log,
config,
similar_programs,
};
instance.run_inner()
}
fn run_inner(&mut self) -> anyhow::Result<()> {
let start_time = Instant::now();
let loda_programs_oeis_dir: PathBuf = self.config.loda_programs_oeis_dir();
let instruction_bigram_csv: PathBuf = self.analytics_directory.histogram_instruction_bigram_file();
let instruction_vec: Vec<RecordBigram> = RecordBigram::parse_csv(&instruction_bigram_csv).expect("Unable to load instruction bigram csv");
let number_of_bigram_rows: usize = instruction_vec.len();
debug!("number of bigram rows: {}", number_of_bigram_rows);
let wordpair_vec = WordPair::convert(instruction_vec);
assert!(wordpair_vec.len() == number_of_bigram_rows);
let mut wordpair_to_index = HashMap::<WordPair,u16>::new();
for (index, wordpair) in wordpair_vec.iter().enumerate() {
wordpair_to_index.insert(*wordpair, index as u16);
}
assert!(wordpair_to_index.len() == number_of_bigram_rows);
let indexes_array = IndexesArray::new(number_of_bigram_rows as u16, SIGNATURE_LENGTH);
let mut paths: Vec<PathBuf> = find_asm_files_recursively(&loda_programs_oeis_dir);
paths.sort();
let number_of_paths = paths.len();
if number_of_paths <= 0 {
error!("Expected 1 or more programs, but there are no programs to analyze");
return Ok(());
}
let mut program_meta_vec = Vec::<ProgramMeta>::new();
{
let content = format!("will process {} programs\n\n", number_of_paths);
self.simple_log.println(content);
let start = Instant::now();
println!("Creating signatures of programs");
let pb = ProgressBar::new(paths.len() as u64);
for path in paths {
pb.inc(1);
let program_meta = match self.analyze_program(&path, &wordpair_to_index, &indexes_array) {
Some(value) => value,
None => {
continue;
}
};
program_meta_vec.push(program_meta);
}
pb.finish_and_clear();
let green_bold = Style::new().green().bold();
println!(
"{:>12} Created signatures, in {}",
green_bold.apply_to("Finished"),
HumanDuration(start.elapsed())
);
let content = format!("\n\nnumber of program_meta items: {}", program_meta_vec.len());
self.simple_log.println(content);
}
{
let start = Instant::now();
println!("Finding similarities between NxN signatures");
let mut bitset = BitSet::with_capacity(number_of_bigram_rows);
let mut comparison_results = Vec::<ComparisonResult>::new();
comparison_results.reserve(program_meta_vec.len());
let pb = ProgressBar::new(program_meta_vec.len() as u64);
for (index0, program0) in program_meta_vec.iter().enumerate() {
pb.inc(1);
comparison_results.clear();
for (index1, program1) in program_meta_vec.iter().enumerate() {
if index0 == index1 {
continue;
}
bitset.clone_from(&program0.signature);
bitset.intersect_with(&program1.signature);
let overlap_count: u16 = bitset.len() as u16;
if overlap_count == 0 {
continue;
}
let comparison_result = ComparisonResult::new(program1.program_id, overlap_count);
comparison_results.push(comparison_result);
}
comparison_results.sort_by(|a, b| {
a.overlap_count.cmp(&b.overlap_count).reverse()
.then(a.program_id.cmp(&b.program_id))
});
comparison_results.truncate(MAX_NUMBER_OF_ROWS_IN_OUTPUT_CSV_FILE);
OutputManager::create_csv_file(&comparison_results, program0.program_id, &self.similar_programs)
.with_context(|| format!("Failed to create csv file for program id: {:?}", program0.program_id))?;
}
pb.finish_and_clear();
let green_bold = Style::new().green().bold();
println!(
"{:>12} Found similarities, in {}",
green_bold.apply_to("Finished"),
HumanDuration(start.elapsed())
);
}
let content = format!("similar end, elapsed: {:?} ms", start_time.elapsed().as_millis());
self.simple_log.println(content);
Ok(())
}
fn analyze_program(
&mut self,
path: &Path,
wordpair_to_index: &HashMap<WordPair,u16>,
indexes_array: &IndexesArray
) -> Option<ProgramMeta> {
let oeis_id: OeisId = match oeis_id_from_path(path) {
Some(value) => value,
None => {
return None;
}
};
let program_id: u32 = oeis_id.raw();
let parsed_program: ParsedProgram = match Self::load_program(path) {
Ok(value) => value,
Err(error) => {
let content = format!("Skipped program: {} path: {:?} error: {:?}", oeis_id.a_number(), path, error);
self.simple_log.println(&content);
return None;
}
};
let line_count_raw: usize = parsed_program.instruction_vec.len();
if line_count_raw > IGNORE_INPUT_PROGRAM_IF_INSTRUCTION_COUNT_EXCEEDS {
error!("Skipped a program that is too long. path: {:?}", path);
let content = format!("Skipped a program that is too long. path: {:?}", path);
self.simple_log.println(&content);
return None;
}
let line_count = line_count_raw as u16;
let words: Vec<Word> = parsed_program.as_words();
let n = words.len();
if n < 2 {
return None;
}
let mut match_set = HashSet::<u16>::new();
for i in 1..n {
let word0: Word = words[i-1];
let word1: Word = words[i];
let wordpair = WordPair { word0: word0, word1: word1 };
let index: u16 = match wordpair_to_index.get(&wordpair) {
Some(value) => *value,
None => {
error!("Unrecognized bigram, not found in vocabulary. Skipping. {:?}", wordpair);
continue;
}
};
match_set.insert(index);
}
let signature: BitSet = indexes_array.compute_signature(&match_set);
let program_meta = ProgramMeta::new(
program_id,
PathBuf::from(path),
line_count,
signature
);
Some(program_meta)
}
fn load_program(path: &Path) -> anyhow::Result<ParsedProgram> {
let contents: String = match fs::read_to_string(path) {
Ok(value) => value,
Err(error) => {
return Err(anyhow::anyhow!("load program, error: {:?} path: {:?}", error, path));
}
};
let parsed_program: ParsedProgram = match ParsedProgram::parse_program(&contents) {
Ok(value) => value,
Err(error) => {
return Err(anyhow::anyhow!("load program, error: {:?} path: {:?}", error, path));
}
};
Ok(parsed_program)
}
}
struct ProgramMeta {
program_id: u32,
#[allow(dead_code)]
path_input: PathBuf,
#[allow(dead_code)]
line_count: u16,
signature: BitSet,
}
impl ProgramMeta {
fn new(program_id: u32, path_input: PathBuf, line_count: u16, signature: BitSet) -> Self {
Self {
program_id: program_id,
path_input: path_input,
line_count: line_count,
signature: signature
}
}
}
struct IndexesArray {
indexes_array: Vec<Vec<u16>>
}
impl IndexesArray {
fn new(vocabulary_size: u16, signature_length: u8) -> Self {
// Create permutations of the numbers between (0 .. vocabulary_size-1)
let mut indexes_array: Vec<Vec<u16>> = vec!();
let original_indexes: Vec<u16> = (0..vocabulary_size).collect();
for i in 0..signature_length {
let mut rng = StdRng::seed_from_u64(i as u64);
let mut indexes: Vec<u16> = original_indexes.clone();
indexes.shuffle(&mut rng);
indexes_array.push(indexes);
}
Self {
indexes_array: indexes_array
}
}
fn compute_signature(&self, match_set: &HashSet<u16>) -> BitSet {
let mut result = BitSet::new();
for indexes in &self.indexes_array {
for (key, value) in indexes.iter().enumerate() {
if match_set.contains(value) {
result.insert(key);
break;
}
}
}
result
}
}
#[derive(Serialize)]
struct ComparisonResult {
#[serde(rename = "program id")]
program_id: u32,
#[serde(rename = "overlap")]
overlap_count: u16,
}
impl ComparisonResult {
fn new(program_id: u32, overlap_count: u16) -> Self {
Self {
program_id: program_id,
overlap_count: overlap_count,
}
}
}
struct OutputManager {}
impl OutputManager {
fn create_csv_file(records: &Vec<ComparisonResult>, program_id: u32, output_rootdir: &Path) -> anyhow::Result<()> {
let (dirname_string, filename_string) = Self::output_dir_and_file(program_id);
let dirname = Path::new(&dirname_string);
let filename = Path::new(&filename_string);
let path_output_dir: PathBuf = output_rootdir.join(dirname);
let path_output_file: PathBuf = path_output_dir.join(filename);
Self::create_csv_file_inner(records, &path_output_dir, &path_output_file)
.with_context(|| format!("Save csv file at: {:?}", path_output_file))?;
Ok(())
}
// Used for construct a path like: "/absolute/path/123/A123456_similarity_lsh.csv"
fn output_dir_and_file(program_id: u32) -> (String, String) {
let dir_index: u32 = program_id / 1000;
let dir_index_string: String = format!("{:0>3}", dir_index);
let filename_string: String = format!("A{:0>6}_similarity_lsh.csv", program_id);
(dir_index_string, filename_string)
}
fn create_csv_file_inner(records: &Vec<ComparisonResult>, output_path_dir: &Path, output_path_file: &Path) -> anyhow::Result<()> {
if !output_path_dir.is_dir() {
fs::create_dir(output_path_dir)
.with_context(|| format!("Could not create dir: {:?}", output_path_dir))?;
}
let mut wtr = WriterBuilder::new()
.has_headers(true)
.delimiter(b';')
.from_path(output_path_file)?;
for record in records {
wtr.serialize(record)?;
}
wtr.flush()?;
Ok(())
}
}
|