1
0
Fork 0
mirror of https://github.com/denoland/deno.git synced 2024-12-23 15:49:44 -05:00
denoland-deno/cli/tools/repl/session.rs

941 lines
27 KiB
Rust
Raw Normal View History

// Copyright 2018-2024 the Deno authors. All rights reserved. MIT license.
use std::sync::Arc;
use crate::args::CliOptions;
use crate::cdp;
use crate::colors;
use crate::lsp::ReplLanguageServer;
use crate::npm::CliNpmResolver;
use crate::resolver::CliResolver;
use crate::tools::test::report_tests;
use crate::tools::test::reporters::PrettyTestReporter;
use crate::tools::test::reporters::TestReporter;
use crate::tools::test::run_tests_for_worker;
use crate::tools::test::send_test_event;
use crate::tools::test::worker_has_tests;
use crate::tools::test::TestEvent;
use crate::tools::test::TestEventReceiver;
use crate::tools::test::TestFailureFormatOptions;
use deno_ast::diagnostics::Diagnostic;
use deno_ast::swc::ast as swc_ast;
use deno_ast::swc::common::comments::CommentKind;
use deno_ast::swc::visit::noop_visit_type;
use deno_ast::swc::visit::Visit;
use deno_ast::swc::visit::VisitWith;
use deno_ast::ImportsNotUsedAsValues;
use deno_ast::ModuleKind;
use deno_ast::ModuleSpecifier;
use deno_ast::ParseDiagnosticsError;
use deno_ast::ParsedSource;
use deno_ast::SourcePos;
use deno_ast::SourceRangedForSpanned;
use deno_ast::SourceTextInfo;
fix(cli): output more detailed information for steps when using JUnit reporter (#22797) This patch gets JUnit reporter to output more detailed information for test steps (subtests). ## Issue with previous implementation In the previous implementation, the test hierarchy was represented using several XML tags like the following: - `<testsuites>` corresponds to the entire test (one execution of `deno test` has exactly one `<testsuites>` tag) - `<testsuite>` corresponds to one file, such as `main_test.ts` - `<testcase>` corresponds to one `Deno.test(...)` - `<property>` corresponds to one `t.step(...)` This structure describes the test layers but one problem is that `<property>` tag is used for any use cases so some tools that can ingest a JUnit XML file might not be able to interpret `<property>` as subtests. ## How other tools address it Some of the testing frameworks in the ecosystem address this issue by fitting subtests into the `<testcase>` layer. For instance, take a look at the following Go test file: ```go package main_test import "testing" func TestMain(t *testing.T) { t.Run("child 1", func(t *testing.T) { // OK }) t.Run("child 2", func(t *testing.T) { // Error t.Fatal("error") }) } ``` Running [gotestsum], we can get the output like this: ```xml <?xml version="1.0" encoding="UTF-8"?> <testsuites tests="3" failures="2" errors="0" time="1.013694"> <testsuite tests="3" failures="2" time="0.510000" name="example/gosumtest" timestamp="2024-03-11T12:26:39+09:00"> <properties> <property name="go.version" value="go1.22.1 darwin/arm64"></property> </properties> <testcase classname="example/gosumtest" name="TestMain/child_2" time="0.000000"> <failure message="Failed" type="">=== RUN TestMain/child_2&#xA; main_test.go:12: error&#xA;--- FAIL: TestMain/child_2 (0.00s)&#xA;</failure> </testcase> <testcase classname="example/gosumtest" name="TestMain" time="0.000000"> <failure message="Failed" type="">=== RUN TestMain&#xA;--- FAIL: TestMain (0.00s)&#xA;</failure> </testcase> <testcase classname="example/gosumtest" name="TestMain/child_1" time="0.000000"></testcase> </testsuite> </testsuites> ``` This output shows that nested test cases are squashed into the `<testcase>` layer by treating them as the same layer as their parent, `TestMain`. We can still distinguish nested ones by their `name` attributes that look like `TestMain/<subtest_name>`. As described in #22795, [vitest] solves the issue in the same way as [gotestsum]. One downside of this would be that one test failure that happens in a nested test case will end up being counted multiple times, because not only the subtest but also its wrapping container(s) are considered to be failures. In fact, in the [gotestsum] output above, `TestMain/child_2` failed (which is totally expected) while its parent, `TestMain`, was also counted as failure. As https://github.com/denoland/deno/pull/20273#discussion_r1307558757 pointed out, there is a test runner that offers flexibility to prevent this, but I personally don't think the "duplicate failure count" issue is a big deal. ## How to fix the issue in this patch This patch fixes the issue with the same approach as [gotestsum] and [vitest]. More specifically, nested test cases are put into the `<testcase>` level and their names are now represented as squashed test names concatenated by `>` (e.g. `parent 2 > child 1 > grandchild 1`). This change also allows us to put a detailed error message as `<failure>` tag within the `<testcase>` tag, which should be handled nicely by third-party tools supporting JUnit XML. ## Extra fix Also, file paths embedded into XML outputs are changed from absolute path to relative path, which is helpful when running the test suites in several different environments like CI. Resolves #22795 [gotestsum]: https://github.com/gotestyourself/gotestsum [vitest]: https://vitest.dev/ --------- Co-authored-by: Bartek Iwańczuk <biwanczuk@gmail.com>
2024-03-25 11:08:46 -04:00
use deno_core::error::generic_error;
use deno_core::error::AnyError;
use deno_core::futures::channel::mpsc::UnboundedReceiver;
use deno_core::futures::FutureExt;
use deno_core::futures::StreamExt;
use deno_core::serde_json;
use deno_core::serde_json::Value;
use deno_core::unsync::spawn;
fix(cli): output more detailed information for steps when using JUnit reporter (#22797) This patch gets JUnit reporter to output more detailed information for test steps (subtests). ## Issue with previous implementation In the previous implementation, the test hierarchy was represented using several XML tags like the following: - `<testsuites>` corresponds to the entire test (one execution of `deno test` has exactly one `<testsuites>` tag) - `<testsuite>` corresponds to one file, such as `main_test.ts` - `<testcase>` corresponds to one `Deno.test(...)` - `<property>` corresponds to one `t.step(...)` This structure describes the test layers but one problem is that `<property>` tag is used for any use cases so some tools that can ingest a JUnit XML file might not be able to interpret `<property>` as subtests. ## How other tools address it Some of the testing frameworks in the ecosystem address this issue by fitting subtests into the `<testcase>` layer. For instance, take a look at the following Go test file: ```go package main_test import "testing" func TestMain(t *testing.T) { t.Run("child 1", func(t *testing.T) { // OK }) t.Run("child 2", func(t *testing.T) { // Error t.Fatal("error") }) } ``` Running [gotestsum], we can get the output like this: ```xml <?xml version="1.0" encoding="UTF-8"?> <testsuites tests="3" failures="2" errors="0" time="1.013694"> <testsuite tests="3" failures="2" time="0.510000" name="example/gosumtest" timestamp="2024-03-11T12:26:39+09:00"> <properties> <property name="go.version" value="go1.22.1 darwin/arm64"></property> </properties> <testcase classname="example/gosumtest" name="TestMain/child_2" time="0.000000"> <failure message="Failed" type="">=== RUN TestMain/child_2&#xA; main_test.go:12: error&#xA;--- FAIL: TestMain/child_2 (0.00s)&#xA;</failure> </testcase> <testcase classname="example/gosumtest" name="TestMain" time="0.000000"> <failure message="Failed" type="">=== RUN TestMain&#xA;--- FAIL: TestMain (0.00s)&#xA;</failure> </testcase> <testcase classname="example/gosumtest" name="TestMain/child_1" time="0.000000"></testcase> </testsuite> </testsuites> ``` This output shows that nested test cases are squashed into the `<testcase>` layer by treating them as the same layer as their parent, `TestMain`. We can still distinguish nested ones by their `name` attributes that look like `TestMain/<subtest_name>`. As described in #22795, [vitest] solves the issue in the same way as [gotestsum]. One downside of this would be that one test failure that happens in a nested test case will end up being counted multiple times, because not only the subtest but also its wrapping container(s) are considered to be failures. In fact, in the [gotestsum] output above, `TestMain/child_2` failed (which is totally expected) while its parent, `TestMain`, was also counted as failure. As https://github.com/denoland/deno/pull/20273#discussion_r1307558757 pointed out, there is a test runner that offers flexibility to prevent this, but I personally don't think the "duplicate failure count" issue is a big deal. ## How to fix the issue in this patch This patch fixes the issue with the same approach as [gotestsum] and [vitest]. More specifically, nested test cases are put into the `<testcase>` level and their names are now represented as squashed test names concatenated by `>` (e.g. `parent 2 > child 1 > grandchild 1`). This change also allows us to put a detailed error message as `<failure>` tag within the `<testcase>` tag, which should be handled nicely by third-party tools supporting JUnit XML. ## Extra fix Also, file paths embedded into XML outputs are changed from absolute path to relative path, which is helpful when running the test suites in several different environments like CI. Resolves #22795 [gotestsum]: https://github.com/gotestyourself/gotestsum [vitest]: https://vitest.dev/ --------- Co-authored-by: Bartek Iwańczuk <biwanczuk@gmail.com>
2024-03-25 11:08:46 -04:00
use deno_core::url::Url;
use deno_core::LocalInspectorSession;
use deno_core::PollEventLoopOptions;
use deno_graph::Position;
use deno_graph::PositionRange;
use deno_graph::SpecifierWithRange;
use deno_runtime::worker::MainWorker;
use deno_semver::npm::NpmPackageReqReference;
use node_resolver::NodeResolutionKind;
use node_resolver::ResolutionMode;
use once_cell::sync::Lazy;
use regex::Match;
use regex::Regex;
use tokio::sync::Mutex;
fn comment_source_to_position_range(
comment_start: SourcePos,
m: &Match,
text_info: &SourceTextInfo,
is_jsx_import_source: bool,
) -> PositionRange {
// the comment text starts after the double slash or slash star, so add 2
let comment_start = comment_start + 2;
// -1 and +1 to include the quotes, but not for jsx import sources because
// they don't have quotes
let padding = if is_jsx_import_source { 0 } else { 1 };
PositionRange {
start: Position::from_source_pos(
comment_start + m.start() - padding,
text_info,
),
end: Position::from_source_pos(
comment_start + m.end() + padding,
text_info,
),
}
}
/// We store functions used in the repl on this object because
/// the user might modify the `Deno` global or delete it outright.
pub static REPL_INTERNALS_NAME: Lazy<String> = Lazy::new(|| {
let now = std::time::SystemTime::now();
let seconds = now
.duration_since(std::time::SystemTime::UNIX_EPOCH)
.unwrap()
.as_secs();
// use a changing variable name to make it hard to depend on this
format!("__DENO_REPL_INTERNALS_{seconds}__")
});
fn get_prelude() -> String {
format!(
r#"
Object.defineProperty(globalThis, "{0}", {{
enumerable: false,
writable: false,
value: {{
lastEvalResult: undefined,
lastThrownError: undefined,
inspectArgs: Deno[Deno.internal].inspectArgs,
noColor: Deno.noColor,
get closed() {{
return typeof globalThis.closed === 'undefined' ? false : globalThis.closed;
}}
}},
}});
Object.defineProperty(globalThis, "_", {{
configurable: true,
get: () => {0}.lastEvalResult,
set: (value) => {{
Object.defineProperty(globalThis, "_", {{
value: value,
writable: true,
enumerable: true,
configurable: true,
}});
console.log("Last evaluation result is no longer saved to _.");
}},
}});
Object.defineProperty(globalThis, "_error", {{
configurable: true,
get: () => {0}.lastThrownError,
set: (value) => {{
Object.defineProperty(globalThis, "_error", {{
value: value,
writable: true,
enumerable: true,
configurable: true,
}});
console.log("Last thrown error is no longer saved to _error.");
}},
}});
globalThis.clear = console.clear.bind(console);
"#,
*REPL_INTERNALS_NAME
)
}
pub enum EvaluationOutput {
Value(String),
Error(String),
}
impl std::fmt::Display for EvaluationOutput {
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
match self {
EvaluationOutput::Value(value) => f.write_str(value),
EvaluationOutput::Error(value) => f.write_str(value),
}
}
}
pub fn result_to_evaluation_output(
r: Result<EvaluationOutput, AnyError>,
) -> EvaluationOutput {
match r {
Ok(value) => value,
Err(err) => {
EvaluationOutput::Error(format!("{} {:#}", colors::red("error:"), err))
}
}
}
#[derive(Debug)]
pub struct TsEvaluateResponse {
pub ts_code: String,
pub value: cdp::EvaluateResponse,
}
struct ReplJsxState {
factory: String,
frag_factory: String,
import_source: Option<String>,
}
pub struct ReplSession {
npm_resolver: Arc<dyn CliNpmResolver>,
resolver: Arc<CliResolver>,
pub worker: MainWorker,
session: LocalInspectorSession,
pub context_id: u64,
pub language_server: ReplLanguageServer,
pub notifications: Arc<Mutex<UnboundedReceiver<Value>>>,
referrer: ModuleSpecifier,
main_module: ModuleSpecifier,
test_reporter_factory: Box<dyn Fn() -> Box<dyn TestReporter>>,
/// This is only optional because it's temporarily taken when evaluating.
test_event_receiver: Option<TestEventReceiver>,
jsx: ReplJsxState,
experimental_decorators: bool,
}
impl ReplSession {
pub async fn initialize(
cli_options: &CliOptions,
npm_resolver: Arc<dyn CliNpmResolver>,
resolver: Arc<CliResolver>,
mut worker: MainWorker,
main_module: ModuleSpecifier,
test_event_receiver: TestEventReceiver,
) -> Result<Self, AnyError> {
let language_server = ReplLanguageServer::new_initialized().await?;
let mut session = worker.create_inspector_session();
worker
.js_runtime
.with_event_loop_future(
session
.post_message::<()>("Runtime.enable", None)
.boxed_local(),
PollEventLoopOptions::default(),
)
.await?;
// Enabling the runtime domain will always send trigger one executionContextCreated for each
// context the inspector knows about so we grab the execution context from that since
// our inspector does not support a default context (0 is an invalid context id).
let context_id: u64;
let mut notification_rx = session.take_notification_rx();
loop {
let notification = notification_rx.next().await.unwrap();
let notification =
serde_json::from_value::<cdp::Notification>(notification)?;
if notification.method == "Runtime.executionContextCreated" {
let execution_context_created = serde_json::from_value::<
cdp::ExecutionContextCreated,
>(notification.params)?;
assert!(execution_context_created
.context
.aux_data
.get("isDefault")
.unwrap()
.as_bool()
.unwrap());
context_id = execution_context_created.context.id;
break;
}
}
assert_ne!(context_id, 0);
let referrer =
deno_core::resolve_path("./$deno$repl.mts", cli_options.initial_cwd())
.unwrap();
fix(cli): output more detailed information for steps when using JUnit reporter (#22797) This patch gets JUnit reporter to output more detailed information for test steps (subtests). ## Issue with previous implementation In the previous implementation, the test hierarchy was represented using several XML tags like the following: - `<testsuites>` corresponds to the entire test (one execution of `deno test` has exactly one `<testsuites>` tag) - `<testsuite>` corresponds to one file, such as `main_test.ts` - `<testcase>` corresponds to one `Deno.test(...)` - `<property>` corresponds to one `t.step(...)` This structure describes the test layers but one problem is that `<property>` tag is used for any use cases so some tools that can ingest a JUnit XML file might not be able to interpret `<property>` as subtests. ## How other tools address it Some of the testing frameworks in the ecosystem address this issue by fitting subtests into the `<testcase>` layer. For instance, take a look at the following Go test file: ```go package main_test import "testing" func TestMain(t *testing.T) { t.Run("child 1", func(t *testing.T) { // OK }) t.Run("child 2", func(t *testing.T) { // Error t.Fatal("error") }) } ``` Running [gotestsum], we can get the output like this: ```xml <?xml version="1.0" encoding="UTF-8"?> <testsuites tests="3" failures="2" errors="0" time="1.013694"> <testsuite tests="3" failures="2" time="0.510000" name="example/gosumtest" timestamp="2024-03-11T12:26:39+09:00"> <properties> <property name="go.version" value="go1.22.1 darwin/arm64"></property> </properties> <testcase classname="example/gosumtest" name="TestMain/child_2" time="0.000000"> <failure message="Failed" type="">=== RUN TestMain/child_2&#xA; main_test.go:12: error&#xA;--- FAIL: TestMain/child_2 (0.00s)&#xA;</failure> </testcase> <testcase classname="example/gosumtest" name="TestMain" time="0.000000"> <failure message="Failed" type="">=== RUN TestMain&#xA;--- FAIL: TestMain (0.00s)&#xA;</failure> </testcase> <testcase classname="example/gosumtest" name="TestMain/child_1" time="0.000000"></testcase> </testsuite> </testsuites> ``` This output shows that nested test cases are squashed into the `<testcase>` layer by treating them as the same layer as their parent, `TestMain`. We can still distinguish nested ones by their `name` attributes that look like `TestMain/<subtest_name>`. As described in #22795, [vitest] solves the issue in the same way as [gotestsum]. One downside of this would be that one test failure that happens in a nested test case will end up being counted multiple times, because not only the subtest but also its wrapping container(s) are considered to be failures. In fact, in the [gotestsum] output above, `TestMain/child_2` failed (which is totally expected) while its parent, `TestMain`, was also counted as failure. As https://github.com/denoland/deno/pull/20273#discussion_r1307558757 pointed out, there is a test runner that offers flexibility to prevent this, but I personally don't think the "duplicate failure count" issue is a big deal. ## How to fix the issue in this patch This patch fixes the issue with the same approach as [gotestsum] and [vitest]. More specifically, nested test cases are put into the `<testcase>` level and their names are now represented as squashed test names concatenated by `>` (e.g. `parent 2 > child 1 > grandchild 1`). This change also allows us to put a detailed error message as `<failure>` tag within the `<testcase>` tag, which should be handled nicely by third-party tools supporting JUnit XML. ## Extra fix Also, file paths embedded into XML outputs are changed from absolute path to relative path, which is helpful when running the test suites in several different environments like CI. Resolves #22795 [gotestsum]: https://github.com/gotestyourself/gotestsum [vitest]: https://vitest.dev/ --------- Co-authored-by: Bartek Iwańczuk <biwanczuk@gmail.com>
2024-03-25 11:08:46 -04:00
let cwd_url =
Url::from_directory_path(cli_options.initial_cwd()).map_err(|_| {
generic_error(format!(
"Unable to construct URL from the path of cwd: {}",
cli_options.initial_cwd().to_string_lossy(),
))
})?;
let ts_config_for_emit = cli_options
.resolve_ts_config_for_emit(deno_config::deno_json::TsConfigType::Emit)?;
let (transpile_options, _) =
crate::args::ts_config_to_transpile_and_emit_options(
ts_config_for_emit.ts_config,
)?;
let experimental_decorators = transpile_options.use_ts_decorators;
let mut repl_session = ReplSession {
npm_resolver,
resolver,
worker,
session,
context_id,
language_server,
referrer,
notifications: Arc::new(Mutex::new(notification_rx)),
fix(cli): output more detailed information for steps when using JUnit reporter (#22797) This patch gets JUnit reporter to output more detailed information for test steps (subtests). ## Issue with previous implementation In the previous implementation, the test hierarchy was represented using several XML tags like the following: - `<testsuites>` corresponds to the entire test (one execution of `deno test` has exactly one `<testsuites>` tag) - `<testsuite>` corresponds to one file, such as `main_test.ts` - `<testcase>` corresponds to one `Deno.test(...)` - `<property>` corresponds to one `t.step(...)` This structure describes the test layers but one problem is that `<property>` tag is used for any use cases so some tools that can ingest a JUnit XML file might not be able to interpret `<property>` as subtests. ## How other tools address it Some of the testing frameworks in the ecosystem address this issue by fitting subtests into the `<testcase>` layer. For instance, take a look at the following Go test file: ```go package main_test import "testing" func TestMain(t *testing.T) { t.Run("child 1", func(t *testing.T) { // OK }) t.Run("child 2", func(t *testing.T) { // Error t.Fatal("error") }) } ``` Running [gotestsum], we can get the output like this: ```xml <?xml version="1.0" encoding="UTF-8"?> <testsuites tests="3" failures="2" errors="0" time="1.013694"> <testsuite tests="3" failures="2" time="0.510000" name="example/gosumtest" timestamp="2024-03-11T12:26:39+09:00"> <properties> <property name="go.version" value="go1.22.1 darwin/arm64"></property> </properties> <testcase classname="example/gosumtest" name="TestMain/child_2" time="0.000000"> <failure message="Failed" type="">=== RUN TestMain/child_2&#xA; main_test.go:12: error&#xA;--- FAIL: TestMain/child_2 (0.00s)&#xA;</failure> </testcase> <testcase classname="example/gosumtest" name="TestMain" time="0.000000"> <failure message="Failed" type="">=== RUN TestMain&#xA;--- FAIL: TestMain (0.00s)&#xA;</failure> </testcase> <testcase classname="example/gosumtest" name="TestMain/child_1" time="0.000000"></testcase> </testsuite> </testsuites> ``` This output shows that nested test cases are squashed into the `<testcase>` layer by treating them as the same layer as their parent, `TestMain`. We can still distinguish nested ones by their `name` attributes that look like `TestMain/<subtest_name>`. As described in #22795, [vitest] solves the issue in the same way as [gotestsum]. One downside of this would be that one test failure that happens in a nested test case will end up being counted multiple times, because not only the subtest but also its wrapping container(s) are considered to be failures. In fact, in the [gotestsum] output above, `TestMain/child_2` failed (which is totally expected) while its parent, `TestMain`, was also counted as failure. As https://github.com/denoland/deno/pull/20273#discussion_r1307558757 pointed out, there is a test runner that offers flexibility to prevent this, but I personally don't think the "duplicate failure count" issue is a big deal. ## How to fix the issue in this patch This patch fixes the issue with the same approach as [gotestsum] and [vitest]. More specifically, nested test cases are put into the `<testcase>` level and their names are now represented as squashed test names concatenated by `>` (e.g. `parent 2 > child 1 > grandchild 1`). This change also allows us to put a detailed error message as `<failure>` tag within the `<testcase>` tag, which should be handled nicely by third-party tools supporting JUnit XML. ## Extra fix Also, file paths embedded into XML outputs are changed from absolute path to relative path, which is helpful when running the test suites in several different environments like CI. Resolves #22795 [gotestsum]: https://github.com/gotestyourself/gotestsum [vitest]: https://vitest.dev/ --------- Co-authored-by: Bartek Iwańczuk <biwanczuk@gmail.com>
2024-03-25 11:08:46 -04:00
test_reporter_factory: Box::new(move || {
Box::new(PrettyTestReporter::new(
false,
true,
false,
true,
cwd_url.clone(),
TestFailureFormatOptions::default(),
fix(cli): output more detailed information for steps when using JUnit reporter (#22797) This patch gets JUnit reporter to output more detailed information for test steps (subtests). ## Issue with previous implementation In the previous implementation, the test hierarchy was represented using several XML tags like the following: - `<testsuites>` corresponds to the entire test (one execution of `deno test` has exactly one `<testsuites>` tag) - `<testsuite>` corresponds to one file, such as `main_test.ts` - `<testcase>` corresponds to one `Deno.test(...)` - `<property>` corresponds to one `t.step(...)` This structure describes the test layers but one problem is that `<property>` tag is used for any use cases so some tools that can ingest a JUnit XML file might not be able to interpret `<property>` as subtests. ## How other tools address it Some of the testing frameworks in the ecosystem address this issue by fitting subtests into the `<testcase>` layer. For instance, take a look at the following Go test file: ```go package main_test import "testing" func TestMain(t *testing.T) { t.Run("child 1", func(t *testing.T) { // OK }) t.Run("child 2", func(t *testing.T) { // Error t.Fatal("error") }) } ``` Running [gotestsum], we can get the output like this: ```xml <?xml version="1.0" encoding="UTF-8"?> <testsuites tests="3" failures="2" errors="0" time="1.013694"> <testsuite tests="3" failures="2" time="0.510000" name="example/gosumtest" timestamp="2024-03-11T12:26:39+09:00"> <properties> <property name="go.version" value="go1.22.1 darwin/arm64"></property> </properties> <testcase classname="example/gosumtest" name="TestMain/child_2" time="0.000000"> <failure message="Failed" type="">=== RUN TestMain/child_2&#xA; main_test.go:12: error&#xA;--- FAIL: TestMain/child_2 (0.00s)&#xA;</failure> </testcase> <testcase classname="example/gosumtest" name="TestMain" time="0.000000"> <failure message="Failed" type="">=== RUN TestMain&#xA;--- FAIL: TestMain (0.00s)&#xA;</failure> </testcase> <testcase classname="example/gosumtest" name="TestMain/child_1" time="0.000000"></testcase> </testsuite> </testsuites> ``` This output shows that nested test cases are squashed into the `<testcase>` layer by treating them as the same layer as their parent, `TestMain`. We can still distinguish nested ones by their `name` attributes that look like `TestMain/<subtest_name>`. As described in #22795, [vitest] solves the issue in the same way as [gotestsum]. One downside of this would be that one test failure that happens in a nested test case will end up being counted multiple times, because not only the subtest but also its wrapping container(s) are considered to be failures. In fact, in the [gotestsum] output above, `TestMain/child_2` failed (which is totally expected) while its parent, `TestMain`, was also counted as failure. As https://github.com/denoland/deno/pull/20273#discussion_r1307558757 pointed out, there is a test runner that offers flexibility to prevent this, but I personally don't think the "duplicate failure count" issue is a big deal. ## How to fix the issue in this patch This patch fixes the issue with the same approach as [gotestsum] and [vitest]. More specifically, nested test cases are put into the `<testcase>` level and their names are now represented as squashed test names concatenated by `>` (e.g. `parent 2 > child 1 > grandchild 1`). This change also allows us to put a detailed error message as `<failure>` tag within the `<testcase>` tag, which should be handled nicely by third-party tools supporting JUnit XML. ## Extra fix Also, file paths embedded into XML outputs are changed from absolute path to relative path, which is helpful when running the test suites in several different environments like CI. Resolves #22795 [gotestsum]: https://github.com/gotestyourself/gotestsum [vitest]: https://vitest.dev/ --------- Co-authored-by: Bartek Iwańczuk <biwanczuk@gmail.com>
2024-03-25 11:08:46 -04:00
))
}),
main_module,
test_event_receiver: Some(test_event_receiver),
jsx: ReplJsxState {
factory: "React.createElement".to_string(),
frag_factory: "React.Fragment".to_string(),
import_source: None,
},
experimental_decorators,
};
// inject prelude
repl_session.evaluate_expression(&get_prelude()).await?;
Ok(repl_session)
}
pub fn set_test_reporter_factory(
&mut self,
f: Box<dyn Fn() -> Box<dyn TestReporter>>,
) {
self.test_reporter_factory = f;
}
pub async fn closing(&mut self) -> Result<bool, AnyError> {
let expression = format!(r#"{}.closed"#, *REPL_INTERNALS_NAME);
let closed = self
.evaluate_expression(&expression)
.await?
.result
.value
.unwrap()
.as_bool()
.unwrap();
Ok(closed)
}
pub async fn post_message_with_event_loop<T: serde::Serialize>(
&mut self,
method: &str,
params: Option<T>,
) -> Result<Value, AnyError> {
self
.worker
.js_runtime
.with_event_loop_future(
self.session.post_message(method, params).boxed_local(),
PollEventLoopOptions {
// NOTE(bartlomieju): this is an important bit; we don't want to pump V8
// message loop here, so that GC won't run. Otherwise, the resulting
// object might be GC'ed before we have a chance to inspect it.
pump_v8_message_loop: false,
..Default::default()
},
)
.await
}
pub async fn run_event_loop(&mut self) -> Result<(), AnyError> {
self.worker.run_event_loop(true).await
}
pub async fn evaluate_line_and_get_output(
&mut self,
line: &str,
) -> EvaluationOutput {
fn format_diagnostic(diagnostic: &deno_ast::ParseDiagnostic) -> String {
let display_position = diagnostic.display_position();
format!(
"{}: {} at {}:{}",
colors::red("parse error"),
diagnostic.message(),
display_position.line_number,
display_position.column_number,
)
}
async fn inner(
session: &mut ReplSession,
line: &str,
) -> Result<EvaluationOutput, AnyError> {
match session.evaluate_line_with_object_wrapping(line).await {
Ok(evaluate_response) => {
let cdp::EvaluateResponse {
result,
exception_details,
} = evaluate_response.value;
Ok(if let Some(exception_details) = exception_details {
session.set_last_thrown_error(&result).await?;
let description = match exception_details.exception {
Some(exception) => {
if let Some(description) = exception.description {
description
} else if let Some(value) = exception.value {
value.to_string()
} else {
"undefined".to_string()
}
}
None => "Unknown exception".to_string(),
};
EvaluationOutput::Error(format!(
"{} {}",
exception_details.text, description
))
} else {
session
.language_server
.commit_text(&evaluate_response.ts_code)
.await;
session.set_last_eval_result(&result).await?;
let value = session.get_eval_value(&result).await?;
EvaluationOutput::Value(value)
})
}
Err(err) => {
// handle a parsing diagnostic
match err.downcast_ref::<deno_ast::ParseDiagnostic>() {
Some(diagnostic) => {
Ok(EvaluationOutput::Error(format_diagnostic(diagnostic)))
}
None => match err.downcast_ref::<ParseDiagnosticsError>() {
Some(diagnostics) => Ok(EvaluationOutput::Error(
diagnostics
.0
.iter()
.map(format_diagnostic)
.collect::<Vec<_>>()
.join("\n\n"),
)),
None => Err(err),
},
}
}
}
}
let result = inner(self, line).await;
result_to_evaluation_output(result)
}
pub async fn evaluate_line_with_object_wrapping(
&mut self,
line: &str,
) -> Result<TsEvaluateResponse, AnyError> {
// Expressions like { "foo": "bar" } are interpreted as block expressions at the
// statement level rather than an object literal so we interpret it as an expression statement
// to match the behavior found in a typical prompt including browser developer tools.
let wrapped_line = if line.trim_start().starts_with('{')
&& !line.trim_end().ends_with(';')
{
format!("({})", &line)
} else {
line.to_string()
};
let evaluate_response = self.evaluate_ts_expression(&wrapped_line).await;
// If that fails, we retry it without wrapping in parens letting the error bubble up to the
// user if it is still an error.
let result = if wrapped_line != line
&& (evaluate_response.is_err()
|| evaluate_response
.as_ref()
.unwrap()
.value
.exception_details
.is_some())
{
self.evaluate_ts_expression(line).await
} else {
evaluate_response
};
if worker_has_tests(&mut self.worker) {
let report_tests_handle = spawn(report_tests(
self.test_event_receiver.take().unwrap(),
(self.test_reporter_factory)(),
));
run_tests_for_worker(
&mut self.worker,
&self.main_module,
&Default::default(),
&Default::default(),
)
.await
.unwrap();
send_test_event(
&self.worker.js_runtime.op_state(),
TestEvent::ForceEndReport,
)
.unwrap();
self.test_event_receiver = Some(report_tests_handle.await.unwrap().1);
}
result
}
async fn set_last_thrown_error(
&mut self,
error: &cdp::RemoteObject,
) -> Result<(), AnyError> {
self
.post_message_with_event_loop(
"Runtime.callFunctionOn",
Some(cdp::CallFunctionOnArgs {
function_declaration: format!(
r#"function (object) {{ {}.lastThrownError = object; }}"#,
*REPL_INTERNALS_NAME
),
object_id: None,
arguments: Some(vec![error.into()]),
silent: None,
return_by_value: None,
generate_preview: None,
user_gesture: None,
await_promise: None,
execution_context_id: Some(self.context_id),
object_group: None,
throw_on_side_effect: None,
}),
)
.await?;
Ok(())
}
async fn set_last_eval_result(
&mut self,
evaluate_result: &cdp::RemoteObject,
) -> Result<(), AnyError> {
self
.post_message_with_event_loop(
"Runtime.callFunctionOn",
Some(cdp::CallFunctionOnArgs {
function_declaration: format!(
r#"function (object) {{ {}.lastEvalResult = object; }}"#,
*REPL_INTERNALS_NAME
),
object_id: None,
arguments: Some(vec![evaluate_result.into()]),
silent: None,
return_by_value: None,
generate_preview: None,
user_gesture: None,
await_promise: None,
execution_context_id: Some(self.context_id),
object_group: None,
throw_on_side_effect: None,
}),
)
.await?;
Ok(())
}
pub async fn call_function_on_args(
&mut self,
function_declaration: String,
args: &[cdp::RemoteObject],
) -> Result<cdp::CallFunctionOnResponse, AnyError> {
let arguments: Option<Vec<cdp::CallArgument>> = if args.is_empty() {
None
} else {
Some(args.iter().map(|a| a.into()).collect())
};
let inspect_response = self
.post_message_with_event_loop(
"Runtime.callFunctionOn",
Some(cdp::CallFunctionOnArgs {
function_declaration,
object_id: None,
arguments,
silent: None,
return_by_value: None,
generate_preview: None,
user_gesture: None,
await_promise: None,
execution_context_id: Some(self.context_id),
object_group: None,
throw_on_side_effect: None,
}),
)
.await?;
let response: cdp::CallFunctionOnResponse =
serde_json::from_value(inspect_response)?;
Ok(response)
}
pub async fn get_eval_value(
&mut self,
evaluate_result: &cdp::RemoteObject,
) -> Result<String, AnyError> {
// TODO(caspervonb) we should investigate using previews here but to keep things
// consistent with the previous implementation we just get the preview result from
// Deno.inspectArgs.
let response = self
.call_function_on_args(
format!(
r#"function (object) {{
try {{
return {0}.inspectArgs(["%o", object], {{ colors: !{0}.noColor }});
}} catch (err) {{
return {0}.inspectArgs(["%o", err]);
}}
}}"#,
*REPL_INTERNALS_NAME
),
&[evaluate_result.clone()],
)
.await?;
let value = response.result.value.unwrap();
let s = value.as_str().unwrap();
Ok(s.to_string())
}
async fn evaluate_ts_expression(
&mut self,
expression: &str,
) -> Result<TsEvaluateResponse, AnyError> {
let parsed_source =
match parse_source_as(expression.to_string(), deno_ast::MediaType::Tsx) {
Ok(parsed) => parsed,
Err(err) => {
if let Ok(parsed) = parse_source_as(
expression.to_string(),
deno_ast::MediaType::TypeScript,
) {
parsed
} else {
return Err(err);
}
}
};
self
.check_for_npm_or_node_imports(&parsed_source.program())
.await?;
self.analyze_and_handle_jsx(&parsed_source);
let transpiled_src = parsed_source
.transpile(
&deno_ast::TranspileOptions {
use_ts_decorators: self.experimental_decorators,
use_decorators_proposal: !self.experimental_decorators,
emit_metadata: false,
imports_not_used_as_values: ImportsNotUsedAsValues::Preserve,
transform_jsx: true,
precompile_jsx: false,
precompile_jsx_skip_elements: None,
precompile_jsx_dynamic_props: None,
jsx_automatic: self.jsx.import_source.is_some(),
jsx_development: false,
jsx_factory: self.jsx.factory.clone(),
jsx_fragment_factory: self.jsx.frag_factory.clone(),
jsx_import_source: self.jsx.import_source.clone(),
var_decl_imports: true,
verbatim_module_syntax: false,
},
&deno_ast::TranspileModuleOptions {
module_kind: Some(ModuleKind::Esm),
},
&deno_ast::EmitOptions {
source_map: deno_ast::SourceMapOption::None,
source_map_base: None,
source_map_file: None,
inline_sources: false,
remove_comments: false,
},
)?
.into_source()
.text;
let value = self
.evaluate_expression(&format!("'use strict'; void 0;{transpiled_src}"))
.await?;
Ok(TsEvaluateResponse {
ts_code: expression.to_string(),
value,
})
}
fn analyze_and_handle_jsx(&mut self, parsed_source: &ParsedSource) {
let Some(analyzed_pragmas) = analyze_jsx_pragmas(parsed_source) else {
return;
};
if !analyzed_pragmas.has_any() {
return;
}
if let Some(jsx) = analyzed_pragmas.jsx {
self.jsx.factory = jsx.text;
self.jsx.import_source = None;
}
if let Some(jsx_frag) = analyzed_pragmas.jsx_fragment {
self.jsx.frag_factory = jsx_frag.text;
self.jsx.import_source = None;
}
if let Some(jsx_import_source) = analyzed_pragmas.jsx_import_source {
self.jsx.import_source = Some(jsx_import_source.text);
}
}
async fn check_for_npm_or_node_imports(
&mut self,
program: &swc_ast::Program,
) -> Result<(), AnyError> {
let Some(npm_resolver) = self.npm_resolver.as_managed() else {
return Ok(()); // don't auto-install for byonm
};
let mut collector = ImportCollector::new();
program.visit_with(&mut collector);
let resolved_imports = collector
.imports
.iter()
.flat_map(|i| {
self
.resolver
.resolve(
i,
&self.referrer,
deno_graph::Position::zeroed(),
ResolutionMode::Import,
NodeResolutionKind::Execution,
)
.ok()
.or_else(|| ModuleSpecifier::parse(i).ok())
})
.collect::<Vec<_>>();
let npm_imports = resolved_imports
.iter()
.flat_map(|url| NpmPackageReqReference::from_specifier(url).ok())
.map(|r| r.into_inner().req)
.collect::<Vec<_>>();
let has_node_specifier =
resolved_imports.iter().any(|url| url.scheme() == "node");
if !npm_imports.is_empty() || has_node_specifier {
npm_resolver
.add_and_cache_package_reqs(&npm_imports)
.await?;
// prevent messages in the repl about @types/node not being cached
if has_node_specifier {
npm_resolver.inject_synthetic_types_node_package().await?;
}
}
Ok(())
}
async fn evaluate_expression(
&mut self,
expression: &str,
) -> Result<cdp::EvaluateResponse, AnyError> {
self
.post_message_with_event_loop(
"Runtime.evaluate",
Some(cdp::EvaluateArgs {
expression: expression.to_string(),
object_group: None,
include_command_line_api: None,
silent: None,
context_id: Some(self.context_id),
return_by_value: None,
generate_preview: None,
user_gesture: None,
await_promise: None,
throw_on_side_effect: None,
timeout: None,
disable_breaks: None,
repl_mode: Some(true),
allow_unsafe_eval_blocked_by_csp: None,
unique_context_id: None,
}),
)
.await
.and_then(|res| serde_json::from_value(res).map_err(|e| e.into()))
}
}
/// Walk an AST and get all import specifiers for analysis if any of them is
/// an npm specifier.
struct ImportCollector {
pub imports: Vec<String>,
}
impl ImportCollector {
pub fn new() -> Self {
Self { imports: vec![] }
}
}
impl Visit for ImportCollector {
noop_visit_type!();
fn visit_call_expr(&mut self, call_expr: &swc_ast::CallExpr) {
if !matches!(call_expr.callee, swc_ast::Callee::Import(_)) {
return;
}
if !call_expr.args.is_empty() {
let arg = &call_expr.args[0];
if let swc_ast::Expr::Lit(swc_ast::Lit::Str(str_lit)) = &*arg.expr {
self.imports.push(str_lit.value.to_string());
}
}
}
fn visit_module_decl(&mut self, module_decl: &swc_ast::ModuleDecl) {
use deno_ast::swc::ast::*;
match module_decl {
ModuleDecl::Import(import_decl) => {
if import_decl.type_only {
return;
}
self.imports.push(import_decl.src.value.to_string());
}
ModuleDecl::ExportAll(export_all) => {
self.imports.push(export_all.src.value.to_string());
}
ModuleDecl::ExportNamed(export_named) => {
if let Some(src) = &export_named.src {
self.imports.push(src.value.to_string());
}
}
_ => {}
}
}
}
fn parse_source_as(
source: String,
media_type: deno_ast::MediaType,
) -> Result<deno_ast::ParsedSource, AnyError> {
let specifier = if media_type == deno_ast::MediaType::Tsx {
ModuleSpecifier::parse("file:///repl.tsx").unwrap()
} else {
ModuleSpecifier::parse("file:///repl.ts").unwrap()
};
let parsed = deno_ast::parse_module(deno_ast::ParseParams {
specifier,
text: source.into(),
media_type,
capture_tokens: true,
maybe_syntax: None,
scope_analysis: false,
})?;
Ok(parsed)
}
// TODO(bartlomieju): remove these and use regexes from `deno_graph`
/// Matches the `@jsxImportSource` pragma.
static JSX_IMPORT_SOURCE_RE: Lazy<Regex> =
Lazy::new(|| Regex::new(r"(?i)^[\s*]*@jsxImportSource\s+(\S+)").unwrap());
/// Matches the `@jsx` pragma.
static JSX_RE: Lazy<Regex> =
Lazy::new(|| Regex::new(r"(?i)^[\s*]*@jsx\s+(\S+)").unwrap());
/// Matches the `@jsxFrag` pragma.
static JSX_FRAG_RE: Lazy<Regex> =
Lazy::new(|| Regex::new(r"(?i)^[\s*]*@jsxFrag\s+(\S+)").unwrap());
#[derive(Default, Debug)]
struct AnalyzedJsxPragmas {
/// Information about `@jsxImportSource` pragma.
jsx_import_source: Option<SpecifierWithRange>,
/// Matches the `@jsx` pragma.
jsx: Option<SpecifierWithRange>,
/// Matches the `@jsxFrag` pragma.
jsx_fragment: Option<SpecifierWithRange>,
}
impl AnalyzedJsxPragmas {
fn has_any(&self) -> bool {
self.jsx_import_source.is_some()
|| self.jsx.is_some()
|| self.jsx_fragment.is_some()
}
}
/// Analyze provided source and return information about carious pragmas
/// used to configure the JSX transforms.
fn analyze_jsx_pragmas(
parsed_source: &ParsedSource,
) -> Option<AnalyzedJsxPragmas> {
if !matches!(
parsed_source.media_type(),
deno_ast::MediaType::Jsx | deno_ast::MediaType::Tsx
) {
return None;
}
let mut analyzed_pragmas = AnalyzedJsxPragmas::default();
for c in parsed_source.get_leading_comments()?.iter() {
if c.kind != CommentKind::Block {
continue; // invalid
}
if let Some(captures) = JSX_IMPORT_SOURCE_RE.captures(&c.text) {
if let Some(m) = captures.get(1) {
analyzed_pragmas.jsx_import_source = Some(SpecifierWithRange {
text: m.as_str().to_string(),
range: comment_source_to_position_range(
c.start(),
&m,
parsed_source.text_info_lazy(),
true,
),
});
}
}
if let Some(captures) = JSX_RE.captures(&c.text) {
if let Some(m) = captures.get(1) {
analyzed_pragmas.jsx = Some(SpecifierWithRange {
text: m.as_str().to_string(),
range: comment_source_to_position_range(
c.start(),
&m,
parsed_source.text_info_lazy(),
false,
),
});
}
}
if let Some(captures) = JSX_FRAG_RE.captures(&c.text) {
if let Some(m) = captures.get(1) {
analyzed_pragmas.jsx_fragment = Some(SpecifierWithRange {
text: m.as_str().to_string(),
range: comment_source_to_position_range(
c.start(),
&m,
parsed_source.text_info_lazy(),
false,
),
});
}
}
}
Some(analyzed_pragmas)
}