Add more scientific research tool detectors

detect/detect.go

@@ -597,7 +597,7 @@ func (e *Engine) recursiveGlob(pattern string) bool {
 	suffix := parts[1] // e.g. "*.py"
 
 	// Use the cached extension set for simple "**/*.ext" patterns
-	if strings.HasPrefix(suffix, "*.") {
+	if strings.HasPrefix(suffix, "*.") && strings.Count(suffix, ".") == 1 {
 		ext := suffix[1:] // ".py"
 		e.loadFileExts()
 		return e.fileExts[ext] > 0

detect/research_tools_test.go

@@ -0,0 +1,149 @@
+package detect
+
+import "testing"
+
+func TestResearchToolsProject(t *testing.T) {
+	report := runOn(t, "../testdata/research-tools-project")
+
+	assertToolDetected(t, report, "test", "nf-test")
+	assertToolDetected(t, report, "build", "nf-core")
+	assertToolDetected(t, report, "docs", "MultiQC")
+	assertToolDetected(t, report, "infrastructure", "Dockstore")
+	assertToolDetected(t, report, "infrastructure", "DVC")
+	assertToolDetected(t, report, "build", "cibuildwheel")
+	assertToolDetected(t, report, "docs", "MyST-Parser")
+	assertToolDetected(t, report, "test", "BenchmarkTools.jl")
+	assertToolDetected(t, report, "lint", "lintr")
+	assertToolDetected(t, report, "format", "styler")
+	assertToolDetected(t, report, "build", "targets")
+	assertToolDetected(t, report, "test", "vdiffr")
+	assertToolDetected(t, report, "test", "tinytest")
+}
+
+func TestResearchToolSignals(t *testing.T) {
+	tests := []struct {
+		name     string
+		files    map[string]string
+		category string
+		tool     string
+	}{
+		{
+			name: "nf-test test file",
+			files: map[string]string{
+				"nextflow.config":                   "nextflow.enable.dsl=2\n",
+				"tests/modules/fastqc/main.nf.test": "nextflow_process { script \"main.nf\" }\n",
+			},
+			category: "test",
+			tool:     "nf-test",
+		},
+		{
+			name: "nf-core metadata",
+			files: map[string]string{
+				".nf-core.yml": "repository_type: pipeline\n",
+			},
+			category: "build",
+			tool:     "nf-core",
+		},
+		{
+			name: "nested MultiQC config",
+			files: map[string]string{
+				"assets/multiqc_config.yml": "title: Example analysis\n",
+			},
+			category: "docs",
+			tool:     "MultiQC",
+		},
+		{
+			name: "Dockstore GitHub descriptor",
+			files: map[string]string{
+				".github/.dockstore.yml": "version: 1.2\nworkflows:\n  - subclass: NFL\n    primaryDescriptorPath: /nextflow.config\n",
+			},
+			category: "infrastructure",
+			tool:     "Dockstore",
+		},
+		{
+			name: "DVC project config",
+			files: map[string]string{
+				".dvc/config": "[core]\n    remote = storage\n",
+			},
+			category: "infrastructure",
+			tool:     "DVC",
+		},
+		{
+			name: "cibuildwheel pyproject config",
+			files: map[string]string{
+				"pyproject.toml": "[project]\nname = \"native-example\"\nversion = \"1.0.0\"\n\n[tool.cibuildwheel]\ntest-command = \"pytest {project}/tests\"\n",
+			},
+			category: "build",
+			tool:     "cibuildwheel",
+		},
+		{
+			name: "MyST-Parser dependency",
+			files: map[string]string{
+				"pyproject.toml": "[project]\nname = \"research-docs\"\nversion = \"1.0.0\"\ndependencies = [\"myst-parser>=4\"]\n",
+			},
+			category: "docs",
+			tool:     "MyST-Parser",
+		},
+		{
+			name: "BenchmarkTools Julia dependency",
+			files: map[string]string{
+				"Project.toml": "name = \"NumericalModel\"\nuuid = \"12345678-1234-1234-1234-123456789abc\"\nversion = \"0.1.0\"\n\n[deps]\nBenchmarkTools = \"6e4b80f9-dd63-53aa-95a3-0cdb28fa8baf\"\n",
+			},
+			category: "test",
+			tool:     "BenchmarkTools.jl",
+		},
+		{
+			name: "lintr R dependency",
+			files: map[string]string{
+				"DESCRIPTION": "Package: modeltools\nVersion: 1.0.0\nSuggests: lintr\n",
+			},
+			category: "lint",
+			tool:     "lintr",
+		},
+		{
+			name: "styler R dependency",
+			files: map[string]string{
+				"DESCRIPTION": "Package: modeltools\nVersion: 1.0.0\nSuggests: styler\n",
+			},
+			category: "format",
+			tool:     "styler",
+		},
+		{
+			name: "targets pipeline",
+			files: map[string]string{
+				"_targets.R": "library(targets)\nlist(tar_target(result, run_model()))\n",
+			},
+			category: "build",
+			tool:     "targets",
+		},
+		{
+			name: "vdiffr R dependency",
+			files: map[string]string{
+				"DESCRIPTION": "Package: plottools\nVersion: 1.0.0\nSuggests: vdiffr\n",
+			},
+			category: "test",
+			tool:     "vdiffr",
+		},
+		{
+			name: "tinytest package runner",
+			files: map[string]string{
+				"DESCRIPTION":      "Package: modeltools\nVersion: 1.0.0\nSuggests: tinytest\n",
+				"tests/tinytest.R": "if (requireNamespace(\"tinytest\")) tinytest::test_package(\"modeltools\")\n",
+			},
+			category: "test",
+			tool:     "tinytest",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			dir := t.TempDir()
+			for path, content := range tt.files {
+				writeProjectFile(t, dir, path, content)
+			}
+
+			report := runOn(t, dir)
+			assertToolDetected(t, report, tt.category, tt.tool)
+		})
+	}
+}

knowledge/_shared/dockstore.toml

@@ -0,0 +1,20 @@
+[tool]
+name = "Dockstore"
+category = "infrastructure"
+homepage = "https://dockstore.org"
+docs = "https://docs.dockstore.org"
+repo = "https://github.com/dockstore/dockstore"
+description = "Registry for scientific tools and workflows"
+
+[detect]
+files = [".dockstore.yml", ".dockstore.yaml", ".github/.dockstore.yml", ".github/.dockstore.yaml"]
+
+[config]
+files = [".dockstore.yml", ".dockstore.yaml", ".github/.dockstore.yml", ".github/.dockstore.yaml"]
+
+[taxonomy]
+role = ["cli-tool"]
+layer = ["infrastructure"]
+function = ["deployment"]
+domain = ["scientific-computing"]
+audience = ["researcher"]

knowledge/_shared/dvc.toml

@@ -0,0 +1,25 @@
+[tool]
+name = "DVC"
+category = "infrastructure"
+homepage = "https://dvc.org"
+docs = "https://dvc.org/doc"
+repo = "https://github.com/iterative/dvc"
+description = "Version control for data and machine learning projects"
+
+[detect]
+files = ["dvc.yaml", "dvc.lock", ".dvc/config", ".dvcignore", "*.dvc", "**/*.dvc"]
+dependencies = ["dvc"]
+
+[commands]
+run = "dvc pull"
+alternatives = ["dvc repro", "dvc status"]
+
+[config]
+files = ["dvc.yaml", "dvc.lock", ".dvc/config", ".dvcignore"]
+lockfile = "dvc.lock"
+
+[taxonomy]
+role = ["cli-tool"]
+function = ["file-management"]
+domain = ["data-science"]
+audience = ["data-scientist", "researcher"]

knowledge/_shared/multiqc.toml

@@ -0,0 +1,23 @@
+[tool]
+name = "MultiQC"
+category = "docs"
+homepage = "https://multiqc.info"
+docs = "https://docs.seqera.io/multiqc"
+repo = "https://github.com/MultiQC/MultiQC"
+description = "Aggregate reports for bioinformatics analyses"
+
+[detect]
+files = ["multiqc_config.yml", "multiqc_config.yaml", ".multiqc_config.yml", ".multiqc_config.yaml", "**/multiqc_config.yml", "**/multiqc_config.yaml"]
+dependencies = ["multiqc"]
+
+[commands]
+run = "multiqc ."
+
+[config]
+files = ["multiqc_config.yml", "multiqc_config.yaml", ".multiqc_config.yml", ".multiqc_config.yaml", "**/multiqc_config.yml", "**/multiqc_config.yaml"]
+
+[taxonomy]
+role = ["cli-tool"]
+function = ["visualization"]
+domain = ["scientific-computing"]
+audience = ["researcher"]

knowledge/_shared/nf-core.toml

@@ -0,0 +1,24 @@
+[tool]
+name = "nf-core"
+category = "build"
+homepage = "https://nf-co.re"
+docs = "https://nf-co.re/docs/nf-core-tools"
+repo = "https://github.com/nf-core/tools"
+description = "Tools for creating and maintaining nf-core Nextflow pipelines"
+
+[detect]
+files = [".nf-core.yml", ".nf-core.yaml"]
+dependencies = ["nf-core"]
+
+[commands]
+run = "nf-core pipelines lint"
+alternatives = ["nf-core pipelines schema lint", "nf-core pipelines create"]
+
+[config]
+files = [".nf-core.yml", ".nf-core.yaml"]
+
+[taxonomy]
+role = ["build-tool"]
+function = ["automation"]
+domain = ["scientific-computing"]
+audience = ["researcher"]

knowledge/_shared/nf-test.toml

@@ -0,0 +1,23 @@
+[tool]
+name = "nf-test"
+category = "test"
+homepage = "https://www.nf-test.com"
+docs = "https://www.nf-test.com/docs/getting-started/"
+repo = "https://github.com/askimed/nf-test"
+description = "Testing tool for Nextflow pipelines"
+
+[detect]
+files = ["nf-test.config", "**/*.nf.test", "**/*.nf.test.snap"]
+
+[commands]
+run = "nf-test test"
+alternatives = ["nf-test list", "nf-test status"]
+
+[config]
+files = ["nf-test.config"]
+
+[taxonomy]
+role = ["testing-framework"]
+function = ["testing"]
+domain = ["scientific-computing"]
+audience = ["researcher"]

knowledge/julia/benchmarktools.toml

@@ -0,0 +1,21 @@
+[tool]
+name = "BenchmarkTools.jl"
+category = "test"
+homepage = "https://juliaci.github.io/BenchmarkTools.jl"
+docs = "https://juliaci.github.io/BenchmarkTools.jl/stable/manual/"
+repo = "https://github.com/JuliaCI/BenchmarkTools.jl"
+description = "Benchmarking and performance tracking for Julia"
+
+[detect]
+dependencies = ["BenchmarkTools"]
+dev_dependencies = ["BenchmarkTools"]
+ecosystems = ["julia"]
+
+[detect.key_exists]
+"Project.toml" = ["deps.BenchmarkTools", "extras.BenchmarkTools", "weakdeps.BenchmarkTools"]
+
+[taxonomy]
+role = ["testing-framework"]
+function = ["benchmarking"]
+domain = ["scientific-computing"]
+audience = ["researcher"]

knowledge/python/cibuildwheel.toml

@@ -0,0 +1,25 @@
+[tool]
+name = "cibuildwheel"
+category = "build"
+homepage = "https://cibuildwheel.pypa.io"
+docs = "https://cibuildwheel.pypa.io/en/stable/"
+repo = "https://github.com/pypa/cibuildwheel"
+description = "Build and test Python wheels across supported platforms"
+
+[detect]
+dependencies = ["cibuildwheel"]
+dev_dependencies = ["cibuildwheel"]
+ecosystems = ["python"]
+
+[detect.key_exists]
+"pyproject.toml" = ["tool.cibuildwheel"]
+
+[commands]
+run = "cibuildwheel"
+alternatives = ["python -m cibuildwheel"]
+
+[config]
+files = ["pyproject.toml"]
+
+[taxonomy]
+role = ["build-tool"]

knowledge/python/myst-parser.toml

@@ -0,0 +1,20 @@
+[tool]
+name = "MyST-Parser"
+category = "docs"
+homepage = "https://myst-parser.readthedocs.io"
+docs = "https://myst-parser.readthedocs.io/en/latest/"
+repo = "https://github.com/executablebooks/MyST-Parser"
+description = "MyST Markdown parser and Sphinx extension"
+
+[detect]
+dependencies = ["myst-parser"]
+dev_dependencies = ["myst-parser"]
+ecosystems = ["python"]
+
+[detect.file_contains]
+"conf.py" = ["myst_parser"]
+"**/conf.py" = ["myst_parser"]
+
+[taxonomy]
+role = ["library"]
+function = ["documentation"]

knowledge/r/lintr.toml

@@ -0,0 +1,23 @@
+[tool]
+name = "lintr"
+category = "lint"
+homepage = "https://lintr.r-lib.org"
+docs = "https://lintr.r-lib.org/reference/"
+repo = "https://github.com/r-lib/lintr"
+description = "Static analysis and style checks for R"
+
+[detect]
+files = [".lintr", ".lintr.R"]
+dependencies = ["lintr"]
+dev_dependencies = ["lintr"]
+ecosystems = ["r"]
+
+[commands]
+run = "Rscript -e 'lintr::lint_dir()'"
+alternatives = ["Rscript -e 'lintr::lint_package()'"]
+
+[config]
+files = [".lintr", ".lintr.R"]
+
+[taxonomy]
+role = ["linter"]

knowledge/r/styler.toml

@@ -0,0 +1,19 @@
+[tool]
+name = "styler"
+category = "format"
+homepage = "https://styler.r-lib.org"
+docs = "https://styler.r-lib.org/reference/"
+repo = "https://github.com/r-lib/styler"
+description = "Source code formatter for R"
+
+[detect]
+dependencies = ["styler"]
+dev_dependencies = ["styler"]
+ecosystems = ["r"]
+
+[commands]
+run = "Rscript -e 'styler::style_pkg()'"
+alternatives = ["Rscript -e 'styler::style_dir()'"]
+
+[taxonomy]
+role = ["formatter"]