Merge pull request #618 from rsonquery/v0ldek/#613-non-borrowed-string-pattern

Added StringPattern and made Automaton no longer borrow the query

Author: V0ldek

Justfile

@@ -6,7 +6,7 @@ default: (build-all "release")
 # === INIT ===
 
 # Initialize the repository for development.
-init: check-cargo hooks-init checkout-benchmarks
+init: check-cargo hooks-init checkout-submodules
 
 # Check if cargo is installed and install it from rustup if not.
 [private]
@@ -28,10 +28,15 @@ init: check-cargo hooks-init checkout-benchmarks
 
 # Checkout and populate the benchmarks repository, excluding datasets.
 [private]
-checkout-benchmarks:
+@checkout-submodules:
     git submodule init
     git submodule update
 
+# Initialize the benchmarks crate.
+[private]
+@init-benchmarks:
+    cd crates/rsonpath-benchmarks && just init
+
 # === BUILD ===
 
 alias b := build-bin
@@ -247,7 +252,7 @@ hook-pre-commit:
     (just verify-fmt && just verify-check);
 
 [private]
-@hook-post-checkout: checkout-benchmarks
+@hook-post-checkout: checkout-submodules
 
 [private]
 assert-benchmarks-committed:

crates/rsonpath-benchmarks/Cargo.toml

@@ -33,7 +33,6 @@ libc = "0.2.168"
 lazy_static = "1.5.0"
 serde_json = "1.0.134"
 sha2 = "0.10.8"
-ouroboros = "0.18.4"
 reqwest = { version = "0.12.9", features = ["blocking"] }
 rsonpath-lib = { version = "0.9.1", default-features = false }
 rsonpath-syntax = { version = "0.3.1", default-features = false }

crates/rsonpath-benchmarks/Justfile

@@ -1,6 +1,9 @@
 [private]
 default: build-bench
 
+init:
+  cd ./src/implementations/jsurferShim && gradle wrapper
+
 # === BUILD ===
 
 alias b := build-bench

crates/rsonpath-benchmarks/README.md

@@ -12,60 +12,21 @@ Benchmark suite for [`rsonpath`](https://github.com/v0ldek/rsonpath).
 
 ## Prerequisites
 
-By default, the benches are performed against a released version of `rsonpath`.
-Usually you might want to run it against the local version to test your changes.
-To do that, pass a [patch config value] to `cargo`:
+By default, the benches are performed against the local version of `rsonpath` in the local repository.
 
-```ini
---config 'patch.crates-io.rsonpath.path = "../rsonpath"'
-```
-
-Additionally:
-
-1. An appropriate C++ compiler is required for the [`cc` crate](https://lib.rs/crates/cc) to compile the
-   JSONSki code.
-2. JDK of version at least 8 is required and your `JAVA_HOME` environment variable must be set
+1. JDK of version at least 8 is required and your `JAVA_HOME` environment variable must be set
    to its location.
+2. You need `gradle` to create the required wrapper JAR to build JSurfer. To install `gradle` it's recommended to use
+[`SDKMan`](https://sdkman.io/): `sdk install gradle 7.5`. After that is done, run `just init` in the root of `rsonpath-benchmarks`.
+This needs to be done only once.
 
 On x86_64 Ubuntu the latters can be done by installing `openjdk-17-jdk` and exporting `JAVA_HOME` as
 `/usr/lib/jvm/java-1.17.0-openjdk-amd64`.
 
 ### Download the dataset
 
-On a UNIX system with `wget` installed run the script `sh dl.sh`.
-You can also manually download the dataset and put the JSON files in the correct folder.
-
-For more information, refers to:
-
-* AST: [![DOI](https://zenodo.org/badge/DOI/10.5281/zenodo.7229269.svg)](https://doi.org/10.5281/zenodo.7229269)
-* Twitter: [![DOI](https://zenodo.org/badge/DOI/10.5281/zenodo.7229287.svg)](https://doi.org/10.5281/zenodo.7229287)
-* Crossref: [![DOI](https://zenodo.org/badge/DOI/10.5281/zenodo.7229287.svg)](https://doi.org/10.5281/zenodo.7231920)
-
-For the benchmark to work, the directory layout should be as follows:
-
-```ini
-── data
-   ├── ast
-   │   └── ast.json
-   ├── crossref
-   │   ├── crossref0.json
-   │   ├── crossref16.json
-   │   ├── crossref1.json
-   │   ├── crossref2.json
-   │   ├── crossref4.json
-   │   └── crossref8.json
-   └── twitter
-       └── twitter.json
-```
-
-The sha256sum of the JSON files, for reference:
-
-* `c3ff840d153953ee08c1d9622b20f8c1dc367ae2abcb9c85d44100c6209571af  ast/ast.json`
-* `f76da4fbd5c18889012ab9bbc222cc439b4b28f458193d297666f56fc69ec500  crossref/crossref/crossref1.json`
-* `95e0038e46ce2e94a0f9dde35ec7975280194220878f83436e320881ccd252b4  crossref/crossref/crossref2.json`
-* `f14e65d4f8df3c9144748191c1e9d46a030067af86d0cc03cc67f22149143c5d  twitter/twitter.json`
-
-TODO: checksums of other crossrefs
+Datasets are automatically downloaded on demand when an appropriate benchmark is ran. The datasets are also
+automatically checked against their known SHA256 checksum to verify integrity.
 
 ## Usage
 
@@ -83,12 +44,6 @@ cargo bench --bench <dataset> --no-default-features
 
 The folder `target/criterion` contains all the information needed to plot the experiment.
 
-As a reminder, to test against local changes instead of a crates.io version:
-
-```bash
-cargo bench --bench <dataset> --config 'patch.crates-io.rsonpath.path = "../rsonpath"'
-```
-
 ## Plotting
 
 To plot the result once the is bench done:

crates/rsonpath-benchmarks/src/implementations/rsonpath.rs

@@ -1,15 +1,10 @@
 use crate::framework::implementation::Implementation;
-use ouroboros::self_referencing;
+use rsonpath::{engine::Compiler, input::MmapInput};
 use rsonpath::{
-    engine::main::MainEngine,
+    engine::{main::MainEngine, Engine},
     input::OwnedBytes,
     result::{Match, Sink},
 };
-use rsonpath::{
-    engine::{Compiler, Engine},
-    input::MmapInput,
-};
-use rsonpath_syntax::JsonPathQuery;
 use std::{convert::Infallible, fmt::Display, fs, io};
 use thiserror::Error;
 
@@ -18,16 +13,8 @@ pub struct RsonpathCount {}
 pub struct RsonpathMmap {}
 pub struct RsonpathMmapCount {}
 
-#[self_referencing()]
-pub struct RsonpathQuery {
-    query: JsonPathQuery,
-    #[borrows(query)]
-    #[not_covariant]
-    engine: MainEngine<'this>,
-}
-
 impl Implementation for Rsonpath {
-    type Query = RsonpathQuery;
+    type Query = MainEngine;
 
     type File = OwnedBytes<Vec<u8>>;
 
@@ -52,25 +39,20 @@ impl Implementation for Rsonpath {
 
     fn compile_query(&self, query: &str) -> Result<Self::Query, Self::Error> {
         let query = rsonpath_syntax::parse(query).unwrap();
+        let engine = MainEngine::compile_query(&query).map_err(RsonpathError::CompilerError)?;
 
-        let rsonpath = RsonpathQuery::try_new(query, |query| {
-            MainEngine::compile_query(query).map_err(RsonpathError::CompilerError)
-        })?;
-
-        Ok(rsonpath)
+        Ok(engine)
     }
 
     fn run(&self, query: &Self::Query, file: &Self::File) -> Result<Self::Result<'_>, Self::Error> {
-        query
-            .with_engine(|engine| engine.matches(file, &mut VoidSink))
-            .map_err(RsonpathError::EngineError)?;
+        query.matches(file, &mut VoidSink).map_err(RsonpathError::EngineError)?;
 
         Ok("[not collected]")
     }
 }
 
 impl Implementation for RsonpathCount {
-    type Query = RsonpathQuery;
+    type Query = MainEngine;
 
     type File = OwnedBytes<Vec<u8>>;
 
@@ -95,25 +77,20 @@ impl Implementation for RsonpathCount {
 
     fn compile_query(&self, query: &str) -> Result<Self::Query, Self::Error> {
         let query = rsonpath_syntax::parse(query).unwrap();
+        let engine = MainEngine::compile_query(&query).map_err(RsonpathError::CompilerError)?;
 
-        let rsonpath = RsonpathQuery::try_new(query, |query| {
-            MainEngine::compile_query(query).map_err(RsonpathError::CompilerError)
-        })?;
-
-        Ok(rsonpath)
+        Ok(engine)
     }
 
     fn run(&self, query: &Self::Query, file: &Self::File) -> Result<Self::Result<'_>, Self::Error> {
-        query
-            .with_engine(|engine| engine.count(file))
-            .map_err(RsonpathError::EngineError)?;
+        query.count(file).map_err(RsonpathError::EngineError)?;
 
         Ok("[not collected]")
     }
 }
 
 impl Implementation for RsonpathMmap {
-    type Query = RsonpathQuery;
+    type Query = MainEngine;
 
     type File = MmapInput;
 
@@ -138,25 +115,20 @@ impl Implementation for RsonpathMmap {
 
     fn compile_query(&self, query: &str) -> Result<Self::Query, Self::Error> {
         let query = rsonpath_syntax::parse(query).unwrap();
+        let engine = MainEngine::compile_query(&query).map_err(RsonpathError::CompilerError)?;
 
-        let rsonpath = RsonpathQuery::try_new(query, |query| {
-            MainEngine::compile_query(query).map_err(RsonpathError::CompilerError)
-        })?;
-
-        Ok(rsonpath)
+        Ok(engine)
     }
 
     fn run(&self, query: &Self::Query, file: &Self::File) -> Result<Self::Result<'_>, Self::Error> {
-        query
-            .with_engine(|engine| engine.matches(file, &mut VoidSink))
-            .map_err(RsonpathError::EngineError)?;
+        query.matches(file, &mut VoidSink).map_err(RsonpathError::EngineError)?;
 
         Ok("[not collected]")
     }
 }
 
 impl Implementation for RsonpathMmapCount {
-    type Query = RsonpathQuery;
+    type Query = MainEngine;
 
     type File = MmapInput;
 
@@ -181,18 +153,13 @@ impl Implementation for RsonpathMmapCount {
 
     fn compile_query(&self, query: &str) -> Result<Self::Query, Self::Error> {
         let query = rsonpath_syntax::parse(query).unwrap();
+        let engine = MainEngine::compile_query(&query).map_err(RsonpathError::CompilerError)?;
 
-        let rsonpath = RsonpathQuery::try_new(query, |query| {
-            MainEngine::compile_query(query).map_err(RsonpathError::CompilerError)
-        })?;
-
-        Ok(rsonpath)
+        Ok(engine)
     }
 
     fn run(&self, query: &Self::Query, file: &Self::File) -> Result<Self::Result<'_>, Self::Error> {
-        query
-            .with_engine(|engine| engine.count(file))
-            .map_err(RsonpathError::EngineError)?;
+        query.count(file).map_err(RsonpathError::EngineError)?;
 
         Ok("[not collected]")
     }