Parse report_json files with serde (#18)

Replaces the hand-written `winnow`-based parser with a bunch of struct definitions along with deriving `serde::Deserialize`.
codecov · Aug 30, 2024 · 5f3091a · 5f3091a
1 parent 1b43a89
commit 5f3091a
Show file tree

Hide file tree

Showing 9 changed files with 454 additions and 1,025 deletions.
diff --git a/Cargo.lock b/Cargo.lock
diff --git a/Cargo.toml b/Cargo.toml
@@ -5,5 +5,4 @@ members = ["bindings", "core"]
 default-members = ["core"]
 
 [profile.release]
-
 debug = 1
diff --git a/core/Cargo.toml b/core/Cargo.toml
@@ -6,6 +6,7 @@ edition = "2021"
 [features]
 default = ["pyreport"]
 pyreport = []
+testing = []
 
 [dependencies]
 include_dir = "0.7.3"
@@ -14,6 +15,7 @@ rand = "0.8.5"
 rusqlite = { version = "0.31.0", features = ["bundled", "limits"] }
 rusqlite_migration = { version = "1.2.0", features = ["from-directory"] }
 seahash = "4.1.0"
+serde = { version = "1.0.204", features = ["derive"] }
 serde_json = "1.0.117"
 thiserror = "1.0.59"
 winnow = "0.5.34"
@@ -25,3 +27,4 @@ tempfile = "3.9.0"
 [[bench]]
 name = "pyreport"
 harness = false
+required-features = ["testing"]
diff --git a/core/benches/pyreport.rs b/core/benches/pyreport.rs
@@ -1,10 +1,5 @@
-use codecov_rs::{
-    error::Result,
-    parsers::{common::ReportBuilderCtx, pyreport::report_json},
-    report::{models, Report, ReportBuilder},
-};
+use codecov_rs::{parsers::pyreport::report_json, report::test::TestReportBuilder};
 use divan::Bencher;
-use winnow::Parser as _;
 
 fn main() {
     divan::main();
@@ -13,11 +8,11 @@ fn main() {
 #[divan::bench]
 fn simple_report() {
     let reports = &[
-        r#"{"files": {"src/report.rs": [0, {}, [], null]}, "sessions": {"0": {"j": "codecov-rs CI"}}}"#,
-        r#"{"files": {"src/report.rs": [0, {}, [], null], "src/report/models.rs": [1, {}, [], null]}, "sessions": {"0": {"j": "codecov-rs CI"}, "1": {"j": "codecov-rs CI 2"}}}"#,
-        r#"{"files": {}, "sessions": {"0": {"j": "codecov-rs CI"}, "1": {"j": "codecov-rs CI 2"}}}"#,
-        r#"{"files": {"src/report.rs": [0, {}, [], null], "src/report/models.rs": [1, {}, [], null]}, "sessions": {}}"#,
-        r#"{"files": {}, "sessions": {}}"#,
+        &br#"{"files": {"src/report.rs": [0, {}, [], null]}, "sessions": {"0": {"j": "codecov-rs CI"}}}"#[..],
+        &br#"{"files": {"src/report.rs": [0, {}, [], null], "src/report/models.rs": [1, {}, [], null]}, "sessions": {"0": {"j": "codecov-rs CI"}, "1": {"j": "codecov-rs CI 2"}}}"#[..],
+        &br#"{"files": {}, "sessions": {"0": {"j": "codecov-rs CI"}, "1": {"j": "codecov-rs CI 2"}}}"#[..],
+        &br#"{"files": {"src/report.rs": [0, {}, [], null], "src/report/models.rs": [1, {}, [], null]}, "sessions": {}}"#[..],
+        &br#"{"files": {}, "sessions": {}}"#[..],
     ];
 
     for input in reports {
@@ -26,175 +21,25 @@ fn simple_report() {
 }
 
 // parsing this is quite slow
-#[divan::bench(sample_count = 10)]
+#[divan::bench]
 fn complex_report(bencher: Bencher) {
     // this is a ~11M `report_json`
     let path =
         "./fixtures/pyreport/large/worker-c71ddfd4cb1753c7a540e5248c2beaa079fc3341-report_json.json";
-    let Ok(report) = std::fs::read_to_string(path) else {
+    let Ok(report) = std::fs::read(path) else {
         println!("Failed to read test report");
         return;
     };
 
-    if report.starts_with("version https://git-lfs.github.com/spec/v1\n") {
+    if report.starts_with(b"version https://git-lfs.github.com/spec/v1\n") {
         println!("Sample report has not been pulled from Git LFS");
         return;
     }
 
     bencher.bench(|| run_parsing(&report));
 }
 
-fn run_parsing(input: &str) {
-    let report_builder = TestReport::default();
-    let mut stream = report_json::ReportOutputStream::<&str, TestReport, TestReport> {
-        input,
-        state: ReportBuilderCtx::new(report_builder),
-    };
-    report_json::parse_report_json
-        .parse_next(&mut stream)
-        .unwrap();
-}
-
-#[derive(Debug, Default)]
-struct TestReport {
-    files: Vec<models::SourceFile>,
-    uploads: Vec<models::RawUpload>,
-}
-
-impl Report for TestReport {
-    fn list_files(&self) -> Result<Vec<models::SourceFile>> {
-        todo!()
-    }
-
-    fn list_contexts(&self) -> Result<Vec<models::Context>> {
-        todo!()
-    }
-
-    fn list_coverage_samples(&self) -> Result<Vec<models::CoverageSample>> {
-        todo!()
-    }
-
-    fn list_branches_for_sample(
-        &self,
-        _sample: &models::CoverageSample,
-    ) -> Result<Vec<models::BranchesData>> {
-        todo!()
-    }
-
-    fn get_method_for_sample(
-        &self,
-        _sample: &models::CoverageSample,
-    ) -> Result<Option<models::MethodData>> {
-        todo!()
-    }
-
-    fn list_spans_for_sample(
-        &self,
-        _sample: &models::CoverageSample,
-    ) -> Result<Vec<models::SpanData>> {
-        todo!()
-    }
-
-    fn list_contexts_for_sample(
-        &self,
-        _sample: &models::CoverageSample,
-    ) -> Result<Vec<models::Context>> {
-        todo!()
-    }
-
-    fn list_samples_for_file(
-        &self,
-        _file: &models::SourceFile,
-    ) -> Result<Vec<models::CoverageSample>> {
-        todo!()
-    }
-
-    fn list_raw_uploads(&self) -> Result<Vec<models::RawUpload>> {
-        todo!()
-    }
-
-    fn merge(&mut self, _other: &Self) -> Result<()> {
-        todo!()
-    }
-
-    fn totals(&self) -> Result<models::ReportTotals> {
-        todo!()
-    }
-}
-
-impl ReportBuilder<TestReport> for TestReport {
-    fn insert_file(&mut self, path: &str) -> Result<models::SourceFile> {
-        let file = models::SourceFile::new(path);
-        self.files.push(file.clone());
-        Ok(file)
-    }
-
-    fn insert_raw_upload(
-        &mut self,
-        mut upload_details: models::RawUpload,
-    ) -> Result<models::RawUpload> {
-        upload_details.id = self.uploads.len() as i64;
-        self.uploads.push(upload_details.clone());
-        Ok(upload_details)
-    }
-
-    fn insert_context(&mut self, _name: &str) -> Result<models::Context> {
-        todo!()
-    }
-
-    fn insert_coverage_sample(
-        &mut self,
-        _sample: models::CoverageSample,
-    ) -> Result<models::CoverageSample> {
-        todo!()
-    }
-
-    fn multi_insert_coverage_sample(
-        &mut self,
-        _samples: Vec<&mut models::CoverageSample>,
-    ) -> Result<()> {
-        todo!()
-    }
-
-    fn insert_branches_data(
-        &mut self,
-        _branch: models::BranchesData,
-    ) -> Result<models::BranchesData> {
-        todo!()
-    }
-
-    fn multi_insert_branches_data(
-        &mut self,
-        _branches: Vec<&mut models::BranchesData>,
-    ) -> Result<()> {
-        todo!()
-    }
-
-    fn insert_method_data(&mut self, _method: models::MethodData) -> Result<models::MethodData> {
-        todo!()
-    }
-
-    fn multi_insert_method_data(&mut self, _methods: Vec<&mut models::MethodData>) -> Result<()> {
-        todo!()
-    }
-
-    fn insert_span_data(&mut self, _span: models::SpanData) -> Result<models::SpanData> {
-        todo!()
-    }
-
-    fn multi_insert_span_data(&mut self, _spans: Vec<&mut models::SpanData>) -> Result<()> {
-        todo!()
-    }
-
-    fn associate_context(&mut self, _assoc: models::ContextAssoc) -> Result<models::ContextAssoc> {
-        todo!()
-    }
-
-    fn multi_associate_context(&mut self, _assocs: Vec<&mut models::ContextAssoc>) -> Result<()> {
-        todo!()
-    }
-
-    fn build(self) -> Result<Self> {
-        Ok(self)
-    }
+fn run_parsing(input: &[u8]) {
+    let mut report_builder = TestReportBuilder::default();
+    report_json::parse_report_json(input, &mut report_builder).unwrap();
 }