Examples

Each file runs on its own from the repository root, no arguments unless noted:

arturo examples/json.art

File	What it is	Why read it
json.art	JSON in nine rules	The proposal's headline claim, and the shortest complete grammar here. Ends by showing what scanError says about {"a": 01}.
csv.art	RFC 4180 CSV in seven rules	Quoting rules, doubled quotes inside quoted fields, and why to/thru are not always the answer.
arturo-scan.art	Arturo reading one file of its own source	Every top-level function definition in carpintero.art, name and parameter list, extracted without running any of it.
arturo-corpus.art	Arturo reading all of its own source	Takes the path to an Arturo checkout. The differential test that validates stripComments against a megabyte of real source, plus a recursive into walk that finds every definition at any nesting depth. See below.
safeload.art	loadSafe on a file that breaks the lexer	Runs hangs-the-lexer.art, a file that hangs arturo outright, by stripping its comments with a Carpintero grammar before the lexer sees them.
bench.art	The Phase 3 numbers	scan against the native regex engine, memoization against a deliberately exponential grammar, and cost per kilobyte as the input doubles.

Running the corpus example

arturo-corpus.art is the only one that needs an argument, a checkout of Arturo itself:

git clone --depth 1 https://github.com/arturo-lang/arturo
arturo examples/arturo-corpus.art ./arturo

Budget about half a minute for the megabyte, roughly a third of it in stripComments and the rest in the tree walk. Both were several times slower until the grammars were tuned. The write-up has the before-and-after and the two rules of thumb behind it.

The run reports four things: how many files the stripper rewrote without changing the program the lexer builds, how many never lexed in either form (Arturo keeps deliberate syntax errors in its test suite, and those are not stripper failures), how many compared unequal, and how many function definitions the block grammar found.

The results, and the interpreter bug the "compared unequal" line turned up, are written up in the manual.