Avocado example (#315)

# Rationale for this change

More examples in the repo provide better documentation as well as
provide some regression testing on top of other tests.

# What changes are included in this PR?

A simple example is added.

# Are these changes tested?
Yes

.github/workflows/lint-and-test.yml

@@ -124,6 +124,8 @@ jobs:
         run: cargo run --example books
       - name: Run brands example
         run: cargo run --example brands
+      - name: Run avocado-prices example
+        run: cargo run --example avocado-prices
       - name: Run plastics example
         run: cargo run --example plastics
       - name: Run posql_db example (With Blitzar)
@@ -240,4 +242,4 @@ jobs:
       - name: Install solhint
         run: npm install -g solhint
       - name: Run tests
-        run: solhint -c 'crates/proof-of-sql/.solhint.json' 'crates/proof-of-sql/**/*.sol' -w 0
+        run: solhint -c 'crates/proof-of-sql/.solhint.json' 'crates/proof-of-sql/**/*.sol' -w 0

crates/proof-of-sql/Cargo.toml

@@ -119,6 +119,10 @@ required-features = [ "arrow" ]
 name = "plastics"
 required-features = [ "arrow" ]
 
+[[example]]
+name = "avocado-prices"
+required-features = [ "arrow" ]
+
 [[bench]]
 name = "posql_benches"
 harness = false

crates/proof-of-sql/examples/avocado-prices/avocado-prices.csv

@@ -0,0 +1,37 @@
+Year,Price
+1990,96
+1991,100
+1992,269
+1993,149
+1994,127
+1995,153
+1996,232
+1997,127
+1998,249
+1999,240
+2000,241
+2001,90
+2002,91
+2003,169
+2004,167
+2005,56
+2006,230
+2007,174
+2008,124
+2009,92
+2010,201
+2011,167
+2012,125
+2013,147
+2014,285
+2015,154
+2016,106
+2017,223
+2018,85
+2019,145
+2020,147
+2021,68
+2022,142
+2023,281
+2024,164
+

crates/proof-of-sql/examples/avocado-prices/main.rs

@@ -0,0 +1,124 @@
+//! Example to use Proof of SQL with datasets
+//! To run, use `cargo run --example avocado-prices`.
+//!
+//! NOTE: If this doesn't work because you do not have the appropriate GPU drivers installed,
+//! you can run `cargo run --release --example avocado-prices --no-default-features --features="arrow cpu-perf"` instead. It will be slower for proof generation.
+use arrow::datatypes::SchemaRef;
+use arrow_csv::{infer_schema_from_files, ReaderBuilder};
+use proof_of_sql::{
+    base::database::{OwnedTable, OwnedTableTestAccessor},
+    proof_primitive::dory::{
+        DynamicDoryCommitment, DynamicDoryEvaluationProof, ProverSetup, PublicParameters,
+        VerifierSetup,
+    },
+    sql::{parse::QueryExpr, postprocessing::apply_postprocessing_steps, proof::QueryProof},
+};
+use rand::{rngs::StdRng, SeedableRng};
+use std::{fs::File, time::Instant};
+
+// We generate the public parameters and the setups used by the prover and verifier for the Dory PCS.
+// The `max_nu` should be set such that the maximum table size is less than `2^(2*max_nu-1)`.
+// For a sampling:
+// max_nu = 3 => max table size is 32 rows
+// max_nu = 4 => max table size is 128 rows
+// max_nu = 8 => max table size is 32768 rows
+// max_nu = 10 => max table size is 0.5 million rows
+// max_nu = 15 => max table size is 0.5 billion rows
+// max_nu = 20 => max table size is 0.5 trillion rows
+// Note: we will eventually load these from a file.
+const DORY_SETUP_MAX_NU: usize = 8;
+// This should be a "nothing-up-my-sleeve" phrase or number.
+const DORY_SEED: [u8; 32] = *b"len 32 rng seed - Space and Time";
+
+/// # Panics
+/// Will panic if the query does not parse or the proof fails to verify.
+fn prove_and_verify_query(
+    sql: &str,
+    accessor: &OwnedTableTestAccessor<DynamicDoryEvaluationProof>,
+    prover_setup: &ProverSetup,
+    verifier_setup: &VerifierSetup,
+) {
+    // Parse the query:
+    println!("Parsing the query: {sql}...");
+    let now = Instant::now();
+    let query_plan = QueryExpr::<DynamicDoryCommitment>::try_new(
+        sql.parse().unwrap(),
+        "avocado".parse().unwrap(),
+        accessor,
+    )
+    .unwrap();
+    println!("Done in {} ms.", now.elapsed().as_secs_f64() * 1000.);
+
+    // Generate the proof and result:
+    print!("Generating proof...");
+    let now = Instant::now();
+    let (proof, provable_result) = QueryProof::<DynamicDoryEvaluationProof>::new(
+        query_plan.proof_expr(),
+        accessor,
+        &prover_setup,
+    );
+    println!("Done in {} ms.", now.elapsed().as_secs_f64() * 1000.);
+
+    // Verify the result with the proof:
+    print!("Verifying proof...");
+    let now = Instant::now();
+    let result = proof
+        .verify(
+            query_plan.proof_expr(),
+            accessor,
+            &provable_result,
+            &verifier_setup,
+        )
+        .unwrap();
+    let result = apply_postprocessing_steps(result.table, query_plan.postprocessing());
+    println!("Verified in {} ms.", now.elapsed().as_secs_f64() * 1000.);
+
+    // Display the result
+    println!("Query Result:");
+    println!("{result:?}");
+}
+
+fn main() {
+    let mut rng = StdRng::from_seed(DORY_SEED);
+    let public_parameters = PublicParameters::rand(DORY_SETUP_MAX_NU, &mut rng);
+    let prover_setup = ProverSetup::from(&public_parameters);
+    let verifier_setup = VerifierSetup::from(&public_parameters);
+
+    let filename = "./crates/proof-of-sql/examples/avocado-prices/avocado-prices.csv";
+    let data_batch = ReaderBuilder::new(SchemaRef::new(
+        infer_schema_from_files(&[filename.to_string()], b',', None, true).unwrap(),
+    ))
+    .with_header(true)
+    .build(File::open(filename).unwrap())
+    .unwrap()
+    .next()
+    .unwrap()
+    .unwrap();
+
+    // Load the table into an "Accessor" so that the prover and verifier can access the data/commitments.
+    let accessor = OwnedTableTestAccessor::<DynamicDoryEvaluationProof>::new_from_table(
+        "avocado.prices".parse().unwrap(),
+        OwnedTable::try_from(data_batch).unwrap(),
+        0,
+        &prover_setup,
+    );
+
+    prove_and_verify_query(
+        "SELECT COUNT(*) AS total FROM prices",
+        &accessor,
+        &prover_setup,
+        &verifier_setup,
+    );
+    prove_and_verify_query(
+        "SELECT Price, COUNT(*) AS total FROM prices GROUP BY Price ORDER BY total",
+        &accessor,
+        &prover_setup,
+        &verifier_setup,
+    );
+    prove_and_verify_query(
+        "SELECT Year, COUNT(*) AS total FROM prices WHERE Price > 100 GROUP BY Year ORDER BY total DESC LIMIT 5",
+        &accessor,
+        &prover_setup,
+        &verifier_setup,
+    );
+}