Refactor fname matching

dataset/dataset.go

@@ -207,7 +207,10 @@ func (t *Target) SetupHaddock3Scenario(wd string, s input.Scenario) (runner.Job,
 	}
 
 	// Generate the run.toml file - it will handle the restraints
-	_, _ = t.WriteRunToml(sPath, s.Parameters.General, s.Parameters.Modules)
+	_, err := t.WriteRunToml(sPath, s.Parameters.General, s.Parameters.Modules)
+	if err != nil {
+		return runner.Job{}, err
+	}
 
 	j := runner.Job{
 		ID:   t.ID + "_" + s.Name,
@@ -252,6 +255,12 @@ func (t *Target) WriteRunToml(projectDir string, general map[string]interface{},
 	}
 	runTomlString += "]\n\n"
 
+	fnameArray := [][]string{
+		t.Restraints,
+		t.Toppar,
+		t.MiscPDB,
+	}
+
 	// NOTE: THE ORDER OF THE MODULES IS IMPORTANT!!
 	// Range over the modules in the order they are defined
 	v := reflect.ValueOf(mod)
@@ -282,27 +291,19 @@ func (t *Target) WriteRunToml(projectDir string, general map[string]interface{},
 				}
 				runTomlString += "[" + m + "]\n"
 				for k, v := range field.Interface().(map[string]interface{}) {
+
+					// Find the file to be used as fname
 					if strings.Contains(k, "_fname") {
-						// Identify the fname parameters based on the pattern, considering
-						//  word boundaries
-						pattern := regexp.MustCompile(`\b` + v.(string) + `\b`)
-
-						// Find the restraints that match the pattern
-						for _, r := range t.Restraints {
-							if pattern.MatchString(r) {
-								runTomlString += k + " = \"../data/" + filepath.Base(r) + "\"\n"
-							}
-						}
-						// Find the Toppar that matches the pattern
-						for _, r := range t.Toppar {
-							if pattern.MatchString(r) {
-								runTomlString += k + " = \"../data/" + filepath.Base(r) + "\"\n"
+						pattern := regexp.MustCompile(v.(string))
+
+						for _, fArr := range fnameArray {
+							fname, err := utils.FindFname(fArr, pattern)
+							if err != nil {
+								return "", err
 							}
-						}
-						// Find if a MiscPDB that matches the pattern
-						for _, r := range t.MiscPDB {
-							if pattern.MatchString(r) {
-								runTomlString += k + " = \"../data/" + filepath.Base(r) + "\"\n"
+							// If fname is not empty, add it to the TOML file
+							if fname != "" {
+								runTomlString += k + " = \"../data/" + filepath.Base(fname) + "\"\n"
 							}
 						}
 

dataset/dataset_test.go

@@ -526,7 +526,7 @@ func TestSetupHaddock3Scenario(t *testing.T) {
 						"some-param": "some-value",
 					},
 					Rigidbody: map[string]interface{}{
-						"ambig_fname": "_ti.tbl",
+						"ambig_fname": "_ti",
 					},
 				},
 			},
@@ -576,6 +576,22 @@ func TestSetupHaddock3Scenario(t *testing.T) {
 		t.Errorf("Scenario was not written to disk")
 	}
 
+	// Fail to setup a scenario in which multiple patterns match
+	target = Target{
+		ID:           "1abc",
+		Receptor:     []string{"dummy.pdb", "dummy.pdb"},
+		ReceptorList: "pdb-files.txt",
+		Ligand:       []string{"dummy.pdb", "dummy.pdb"},
+		LigandList:   "pdb-files.txt",
+		Restraints:   []string{"1abc_ti.tbl", "1abc_ti5.tbl"},
+		Toppar:       []string{"custom.top", "custom.param"},
+	}
+
+	_, err = target.SetupHaddock3Scenario(wd, s)
+	if err == nil {
+		t.Errorf("Failed to detect wrong scenario")
+	}
+
 }
 
 func TestTarget_SetupHaddock24Scenario(t *testing.T) {

example/example_haddock30.yml

@@ -1,11 +1,11 @@
 general:
-  executable: /trinity/login/rodrigo/repos/haddock-runner/example/haddock3.sh
+  executable: /home/rodrigo/repos/haddock-runner/example/haddock3.sh
   max_concurrent: 4
-  haddock_dir: /trinity/login/rodrigo/repos/haddock3
+  haddock_dir: /home/rodrigo/repos/haddock3
   receptor_suffix: _r_u
   ligand_suffix: _l_u
-  input_list: /trinity/login/rodrigo/repos/haddock-runner/example/input_list.txt
-  work_dir: /trinity/login/rodrigo/repos/haddock-runner/bm-goes-here
+  input_list: /home/rodrigo/repos/haddock-runner/example/input_list.txt
+  work_dir: /home/rodrigo/repos/haddock-runner/bm-goes-here
 
 scenarios:
   - name: true-interface
@@ -20,20 +20,20 @@ scenarios:
           autohis: true
         rigidbody:
           sampling: 10
-          ambig_fname: _ti
-          unambig_fname: _unambig
-          ligand_top_fname: _custom_top
-          ligand_param_fname: _custom_param
+          ambig_fname: _ti.tbl
+          unambig_fname: _unambig.tbl
+          ligand_top_fname: _ligand.top
+          ligand_param_fname: _ligand.param
         seletop:
           select: 2
         flexref:
-          ambig_fname: _ti
-          unambig_fname: _unambig
-          ligand_top_fname: _custom_top
-          ligand_param_fname: _custom_param
+          ambig_fname: _ti.tbl
+          unambig_fname: _unambig.tbl
+          ligand_top_fname: _ligand.top
+          ligand_param_fname: _ligand.param
         emref:
         caprieval:
-          reference_fname: _ref
+          reference_fname: _ref.pdb
 
   - name: center-of-mass
     parameters:

utils/utils.go

@@ -201,3 +201,20 @@ func ContainsCG(s string) bool {
 	s = strings.ToLower(s)
 	return regexp.MustCompile(`cg`).MatchString(s)
 }
+
+// FindFname checks the array of strings for a pattern, returns an error if multiple files match
+func FindFname(arr []string, pattern *regexp.Regexp) (string, error) {
+
+	var fname string
+	for _, f := range arr {
+		if pattern.MatchString(f) {
+			if fname != "" {
+				err := errors.New("multiple files match the pattern: `" + pattern.String() + "` please use a more specific pattern")
+				return "", err
+			}
+			fname = f
+		}
+	}
+
+	return fname, nil
+}

utils/utils_test.go

@@ -4,6 +4,7 @@ import (
 	"flag"
 	"os"
 	"reflect"
+	"regexp"
 	"testing"
 )
 
@@ -314,3 +315,20 @@ func TestContainsCG(t *testing.T) {
 		t.Errorf("Failed to detect cg")
 	}
 }
+
+func TestFindFname(t *testing.T) {
+
+	testArr := []string{"abc", "bcd", "cde"}
+	pattern := regexp.MustCompile("abc")
+	_, err := FindFname(testArr, pattern)
+	if err != nil {
+		t.Errorf("Failed to find name: %s", err)
+	}
+
+	pattern = regexp.MustCompile("cd")
+	_, err = FindFname(testArr, pattern)
+	if err == nil {
+		t.Errorf("Failed to detect multiple files")
+	}
+
+}