test: add OpenSeek corpus to CI

.github/workflows/nightly_check.yml

@@ -52,3 +52,21 @@ jobs:
       - name: moon test
         run: |
           moon test --target all
+
+      - name: checkout openseek corpus
+        uses: actions/checkout@v4
+        with:
+          repository: moonbitlang/openseek
+          ref: main
+          path: .openseek
+          fetch-depth: 1
+          submodules: recursive
+          persist-credentials: false
+
+      - name: test openseek corpus
+        env:
+          OPENSEEK_ROOT: ${{ github.workspace }}/.openseek
+        run: |
+          echo "OpenSeek commit: $(git -C "$OPENSEEK_ROOT" rev-parse HEAD)"
+          git -C "$OPENSEEK_ROOT" submodule status --recursive
+          moon test test/openseek_test/openseek_test.mbt --target native --include-skipped

.github/workflows/stable_check.yml

@@ -49,3 +49,21 @@ jobs:
       - name: moon test
         run: |
           moon test --target all
+
+      - name: checkout openseek corpus
+        uses: actions/checkout@v4
+        with:
+          repository: moonbitlang/openseek
+          ref: main
+          path: .openseek
+          fetch-depth: 1
+          submodules: recursive
+          persist-credentials: false
+
+      - name: test openseek corpus
+        env:
+          OPENSEEK_ROOT: ${{ github.workspace }}/.openseek
+        run: |
+          echo "OpenSeek commit: $(git -C "$OPENSEEK_ROOT" rev-parse HEAD)"
+          git -C "$OPENSEEK_ROOT" submodule status --recursive
+          moon test test/openseek_test/openseek_test.mbt --target native --include-skipped

test/openseek_test/moon.pkg

@@ -0,0 +1,7 @@
+import {
+  "moonbitlang/parser",
+  "moonbitlang/parser/untyped_cst",
+  "moonbitlang/lexer/basic",
+  "moonbitlang/x/fs",
+  "moonbitlang/core/env",
+} for "test"

test/openseek_test/openseek_test.mbt

@@ -0,0 +1,115 @@
+///|
+fn join_path(dir : String, entry : String) -> String {
+  if dir.has_suffix("/") {
+    dir + entry
+  } else {
+    dir + "/" + entry
+  }
+}
+
+///|
+fn collect_mbt_files(dir : String, files : Array[String]) -> Unit raise Error {
+  let entries = @fs.read_dir(dir) catch {
+    err => fail("failed to read directory \{dir}: \{Repr(err)}")
+  }
+  for entry in entries {
+    if entry != ".git" {
+      let path = join_path(dir, entry)
+      if !@fs.path_exists(path) {
+        if entry.has_suffix(".mbt") {
+          files.push(path)
+        }
+        continue
+      }
+      let is_dir = @fs.is_dir(path) catch {
+        err => fail("failed to inspect \{path}: \{Repr(err)}")
+      }
+      if !is_dir && entry.has_suffix(".mbt") {
+        files.push(path)
+      } else if is_dir {
+        collect_mbt_files(path, files)
+      }
+    }
+  }
+}
+
+///|
+fn reports_from_view(view : ArrayView[@basic.Report]) -> Array[@basic.Report] {
+  let reports : Array[@basic.Report] = []
+  for report in view {
+    reports.push(report)
+  }
+  reports
+}
+
+///|
+fn record_diagnostics(
+  failures : Array[String],
+  path : String,
+  parser : String,
+  reports : Array[@basic.Report],
+) -> Unit {
+  if !reports.is_empty() {
+    let rendered = @basic.show_loc.protect(String, fn() {
+      reports.to_json().stringify(indent=2)
+    })
+    failures.push("\{path}\nparser: \{parser}\ndiagnostics:\n\{rendered}")
+  }
+}
+
+///|
+#skip("requires an external openseek checkout")
+test "all OpenSeek MoonBit sources parse without diagnostics" {
+  let root = @env.get_env_var("OPENSEEK_ROOT").unwrap_or_else(() => {
+    fail("OPENSEEK_ROOT must point to an OpenSeek checkout")
+  })
+  if root.is_empty() {
+    fail("OPENSEEK_ROOT must not be empty")
+  }
+  let root_is_dir = @fs.is_dir(root) catch {
+    err => fail("failed to inspect OPENSEEK_ROOT \{root}: \{Repr(err)}")
+  }
+  if !root_is_dir {
+    fail("OPENSEEK_ROOT is not a directory: \{root}")
+  }
+  let files : Array[String] = []
+  collect_mbt_files(root, files)
+  files.sort()
+  if files.is_empty() {
+    fail("no .mbt files found under OPENSEEK_ROOT: \{root}")
+  }
+  let failures : Array[String] = []
+  for path in files {
+    let source = @fs.read_file_to_string(path) catch {
+      err => fail("failed to read \{path}: \{Repr(err)}")
+    }
+    let (_, handrolled_reports) = @parser.parse_string(
+      source,
+      name=path,
+      parser=Handrolled,
+    )
+    record_diagnostics(failures, path, "Handrolled", handrolled_reports)
+    let (_, moonyacc_reports) = @parser.parse_string(
+      source,
+      name=path,
+      parser=MoonYacc,
+    )
+    record_diagnostics(failures, path, "MoonYacc", moonyacc_reports)
+    let cst = @untyped_cst.parse_structure(source, name=path)
+    record_diagnostics(
+      failures,
+      path,
+      "untyped CST",
+      reports_from_view(cst.diagnostics_view()),
+    )
+    let (_, lowering_reports) = cst.to_impls()
+    record_diagnostics(failures, path, "untyped CST to_impls", lowering_reports)
+  }
+  if !failures.is_empty() {
+    fail(
+      "OpenSeek corpus parsing failed for \{failures.length()} parser/file checks:\n\n" +
+      failures.join("\n\n"),
+    )
+  }
+  println("parsed \{files.length()} OpenSeek .mbt files with all three parsers")
+}

test/openseek_test/pkg.generated.mbti

@@ -0,0 +1,12 @@
+// Generated using `moon info`, DON'T EDIT IT
+package "moonbitlang/parser/test/openseek_test"
+
+// Values
+
+// Errors
+
+// Types and methods
+
+// Type aliases
+
+// Traits