integration with hardcoded license-sim

Author: kendrickcurtis

.codacy/codacy.yaml

@@ -5,8 +5,7 @@ runtimes:
     - [email protected]
     - [email protected]
 tools:
-    - [email protected]
     - [email protected]
     - [email protected]
     - [email protected]
-    - license-sim@1.0.0
+    - license-sim@local

cmd/analyze.go

@@ -62,6 +62,8 @@ func LoadLanguageConfig() (*LanguagesConfig, error) {
 			return nil, fmt.Errorf("failed to parse YAML languages configuration file: %w", err)
 		}
 
+		// Ensure license-sim entry for PHP files
+		ensureLicenseSimEntry(&config)
 		return &config, nil
 	}
 
@@ -83,9 +85,38 @@ func LoadLanguageConfig() (*LanguagesConfig, error) {
 		return nil, fmt.Errorf("failed to parse JSON languages configuration file: %w", err)
 	}
 
+	// Ensure license-sim entry for PHP files
+	ensureLicenseSimEntry(&config)
 	return &config, nil
 }
 
+// ensureLicenseSimEntry adds or updates the license-sim entry to only support .php files
+func ensureLicenseSimEntry(config *LanguagesConfig) {
+	found := false
+	for i, tool := range config.Tools {
+		if tool.Name == "license-sim" {
+			// Overwrite to only support .php files
+			config.Tools[i].Extensions = []string{".php"}
+			config.Tools[i].Languages = []string{"PHP"}
+			found = true
+			break
+		}
+	}
+	if !found {
+		config.Tools = append(config.Tools, struct {
+			Name       string   `yaml:"name" json:"name"`
+			Languages  []string `yaml:"languages" json:"languages"`
+			Extensions []string `yaml:"extensions" json:"extensions"`
+			Files      []string `yaml:"files" json:"files"`
+		}{
+			Name:       "license-sim",
+			Languages:  []string{"PHP"},
+			Extensions: []string{".php"},
+			Files:      nil,
+		})
+	}
+}
+
 // GetFileExtension extracts the file extension from a path
 func GetFileExtension(filePath string) string {
 	return strings.ToLower(filepath.Ext(filePath))
@@ -113,14 +144,14 @@ func IsToolSupportedForFile(toolName string, filePath string, langConfig *Langua
 		if tool.Name == toolName {
 			// If tool has no extensions defined, assume it supports all files
 			if len(tool.Extensions) == 0 {
-				fmt.Printf("[DEBUG] Tool %s has no extensions defined, supports all files. File: %s\n", toolName, filePath)
+				fmt.Printf("[DEBUG1] Tool %s has no extensions defined, supports all files. File: %s\n", toolName, filePath)
 				return true
 			}
 
-			fmt.Printf("[DEBUG] Checking if tool %s supports file %s (ext: %s). Tool extensions: %v\n", toolName, filePath, fileExt, tool.Extensions)
+			//fmt.Printf("[DEBUG] Checking if tool %s supports file %s (ext: %s). Tool extensions: %v\n", toolName, filePath, fileExt, tool.Extensions)
 			for _, ext := range tool.Extensions {
 				if strings.EqualFold(ext, fileExt) {
-					fmt.Printf("[DEBUG] Tool %s supports file %s (matched ext: %s)\n", toolName, filePath, fileExt)
+					//fmt.Printf("[DEBUG2] Tool %s supports file %s (matched ext: %s)\n", toolName, filePath, fileExt)
 					return true
 				}
 			}
@@ -138,7 +169,7 @@ func IsToolSupportedForFile(toolName string, filePath string, langConfig *Langua
 	}
 
 	// If tool not found in config, assume it's supported
-	fmt.Printf("[DEBUG] Tool %s not found in language config, assuming it supports file %s\n", toolName, filePath)
+	//fmt.Printf("[DEBUG3] Tool %s not found in language config, assuming it supports file %s\n", toolName, filePath)
 	return true
 }
 

config/config.go

@@ -380,6 +380,14 @@ func (c *ConfigType) IsRuntimeInstalled(name string, runtime *plugins.RuntimeInf
 
 // IsToolInstalled checks if a tool is already installed
 func (c *ConfigType) IsToolInstalled(name string, tool *plugins.ToolInfo) bool {
+	// Special case for license-sim: treat as installed if duncan.py exists in a sibling directory
+	if name == "license-sim" {
+		repoParent := filepath.Dir(c.repositoryDirectory)
+		scriptPath := filepath.Join(repoParent, "license-sim", "duncan.py")
+		if _, err := os.Stat(scriptPath); err == nil {
+			return true
+		}
+	}
 	// If there are no binaries, check the install directory
 	if len(tool.Binaries) == 0 {
 		_, err := os.Stat(tool.InstallDir)

config/tools-installer.go

@@ -92,9 +92,9 @@ func InstallTool(name string, toolInfo *plugins.ToolInfo, registry string) error
 		isRuntimeInstalled = true
 	}
 
-	// Skip installation for local binary tools (no runtime, no download URL)
-	if toolInfo.Runtime == "" && toolInfo.DownloadURL == "" {
-		logger.Info("Skipping installation for local binary tool", logrus.Fields{
+	// Skip installation for local binary tools (no runtime, no download URL, and version == "local")
+	if toolInfo.Version == "local" {
+		logger.Info("Skipping installation for local binary tool (version=local)", logrus.Fields{
 			"tool":    name,
 			"version": toolInfo.Version,
 		})

domain/tool.go

@@ -31,6 +31,7 @@ const (
 	Semgrep      string = "6792c561-236d-41b7-ba5e-9d6bee0d548b"
 	Lizard       string = "76348462-84b3-409a-90d3-955e90abfb87"
 	Revive       string = "bd81d1f4-1406-402d-9181-1274ee09f1aa"
+	LicenseSim   string = "b7e1c2a4-5f3d-4e2a-9c8b-1a2b3c4d5e6f" // generated UUID
 )
 
 type ToolInfo struct {
@@ -50,4 +51,5 @@ var SupportedToolsMetadata = map[string]ToolInfo{
 	Lizard:       {Name: "lizard", Priority: 0},
 	Semgrep:      {Name: "semgrep", Priority: 0},
 	Revive:       {Name: "revive", Priority: 0},
+	LicenseSim:   {Name: "license-sim", Priority: 0},
 }

tools/licenseSimRunner.go

@@ -32,6 +32,9 @@ func RunLicenseSim(workDirectory string, binary string, files []string, outputFi
 	parts := strings.Split(binary, " ")
 	cmdArgs := append(parts[1:], "search", "-f", fileToCheck, "-e", ext)
 
+	// Always add --json for machine-readable output
+	cmdArgs = append(cmdArgs, "--json")
+
 	if outputFormat == "sarif" {
 		tempFile, err := os.CreateTemp("", "license-sim-*.json")
 		if err != nil {
@@ -41,7 +44,6 @@ func RunLicenseSim(workDirectory string, binary string, files []string, outputFi
 		tempFile.Close()
 		defer os.Remove(tempFilePath)
 
-		cmdArgs = append(cmdArgs, "--json")
 		cmd := exec.Command(parts[0], cmdArgs...)
 		cmd.Dir = workDirectory
 		cmd.Env = append(os.Environ(), "KMP_DUPLICATE_LIB_OK=TRUE")

utils/sarif.go

@@ -342,16 +342,18 @@ func ConvertLicenseSimToSarifWithFile(licenseSimOutput []byte, scannedFile strin
 	}
 
 	for _, issue := range issues {
-		referenceFile := issue.FilePath
 		ruleId := issue.License
 		if ruleId == "" {
 			ruleId = "license-sim-match"
 		}
+		// Compose message with similarity score and reference file if available
+		percent := int(issue.Similarity * 100)
 		msg := issue.Message
 		if msg == "" {
-			msg = "code similar to licensed code in " + referenceFile
-		} else {
-			msg = msg + " (reference: " + referenceFile + ")"
+			msg = fmt.Sprintf("code similar to licensed code (%d%%)", percent)
+			if issue.FilePath != "" {
+				msg += " in " + issue.FilePath
+			}
 		}
 		startLine := issue.Line
 		if startLine <= 0 {