what it says on the tin; an implementation of about half of cs50's problem sets in rust. includes hello world, hello, it's me, mario_less, mario_more, cash, credit, scrabble, readability, caesar, substitution, plurality, runoff, tideman, volume, recover, and speller.
all of these solutions come with in-file tests that duplicate a sufficiently large subset of check50's functionality such that i can be confident these are all correct. [perhaps with the exception of recover, but that problem sets' functionality is so specific that evaluating correctness is pretty self-explanatory; i settled with seeing if it spits out the 50 jpgs from card.raw, which i'm happy to report it does!] note that starting with the pset2 solutions i began to use a module i created called rust50 that duplicates the functionality of the cs50.h library; from then on, i don't test the stdin/stdout interfacing as much as i did for pset1's problems.
to check my work / run it yourself, clone the repo, cd into any of the problem directories (e.g. pset1/cash, pset2/readability) and run cargo run, or cargo run --release if you're in a hurry. to check the tests, run cargo test.
there are some implementations here that are definitely more interesting than others; tideman, for instance, shows off a number of cool things about rust (the dfs() function ended up being really pleasant to look at). speller, though, is the only problem set in cs50 that takes a nontrivial amount of time to execute, and so ends up being a place where rust can show off its potential for, uh, performantness.
this binary matches cs50's speller50 binary's functionality exactly, which was a bit of a pain, since speller50's precise functionality is often a consequence of its very C-ish implementation (i.e., a string like rust50's is 'ignored' since it includes digits, but the spellchecker stops ignoring characters when it hits the apostrophe, so the s after the apostrophe has to be rescued and spellchecked; weird, right?) the tests for the file just confirm that its output is identical to that of speller50.
one snag; the provided test files xueqin2.txt and holmes.txt aren't valid UTF-8! to work with them, i would've had to rewrite this to deal with byte arrays ([u8]) instead of strings (String), since rust strings are necessarily UTF-8. hence, the tests exclude them; kick up the times below by like ~20% when comparing against the times-to-spellcheck-all-files for speller50 in C.
so, extremely loose benchmarks: going to pset5/speller and running the following [the -- --nocapture flag just prints out each text as the program spellchecks it]
cargo test test_spellcheck_load_once --release -- --nocapture
completes successfully in ~0.91s on my computer. this test loads the dictionary once and then fully spellchecks the 26 test files that are valid UTF-8, from cat.txt ("A cat is not a caterpillar.") to shakespeare.txt (his complete works), comparing this program's outputs against the expected outputs from speller50. running
cargo test test_spellcheck_load_every_time --release -- --nocapture
completes successfully in ~1.34s on my computer. this test does the same things as the previous one, but reloads the dictionary for every text it spellchecks; this is more in line with how check50 runs your code. given the best score on the big board for speller is something around ~5s, i'm happy with this result!
on nushell on my computer, going to pset5/speller and running
timeit ./target/release/speller dictionaries/large texts/shakespeare.txt
returns in about ~.31s. doing the same for speller50 on my computer after recompiling it myself;
timeit ./speller50 dictionaries/large texts/shakespeare.txt
returns in about ~1.65s.
- some of these implementations could benefit from a second pass, of course; an ancillary purpose of this was to learn a bit more about rust, so you can see my understanding of it develop as i go along. also, of course of course, i'm constrained for time!
- at some point i ought to publish a
rust50crate proper! pset4/filterisn't here yet; my sincere apologies.