Merge "Bugfix for unicode characters in commit message"

tools/repo_diff/service/repodiff/controllers/differential.go

@@ -199,7 +199,7 @@ func readCSVFiles(projectCSVFile, commitCSVFile string) ([]ent.DiffRow, []ent.Co
 	if err != nil {
 		return nil, nil, errors.Wrap(err, "Error converting CSV file to entities")
 	}
-	commitRows, err := csvFileToCommitRows(commitCSVFile)
+	commitRows, err := CSVFileToCommitRows(commitCSVFile)
 	if err != nil {
 		return nil, nil, errors.Wrap(err, "Error converting CSV file to entities")
 	}
@@ -252,7 +252,7 @@ func toDiffRows(entities []interface{}) ([]ent.DiffRow, error) {
 	return diffRows, nil
 }
 
-func csvFileToCommitRows(csvFile string) ([]ent.CommitRow, error) {
+func CSVFileToCommitRows(csvFile string) ([]ent.CommitRow, error) {
 	entities, err := filesystem.CSVFileToEntities(
 		csvFile,
 		func(cols []string) (interface{}, error) {

tools/repo_diff/service/repodiff/controllers/regression_test.go

@@ -0,0 +1,31 @@
+package controllers
+
+import (
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+
+	cst "repodiff/constants"
+	ent "repodiff/entities"
+	"repodiff/repositories"
+)
+
+func TestRegressionIncorrectStringValue(t *testing.T) {
+	commitRows, _ := CSVFileToCommitRows("testdata/commit.csv")
+	analyzed := make([]ent.AnalyzedCommitRow, len(commitRows))
+	for i, row := range commitRows {
+		analyzed[i] = ent.AnalyzedCommitRow{
+			CommitRow: row,
+			Type:      cst.Empty,
+		}
+	}
+
+	c, _ := repositories.NewCommitRepository(
+		ent.MappedDiffTarget{
+			UpstreamTarget:   1,
+			DownstreamTarget: 2,
+		},
+	)
+	err := c.InsertCommitRows(analyzed)
+	assert.Equal(t, nil, err, "Error should be nil")
+}

tools/repo_diff/service/repodiff/controllers/testdata/commit.csv

@@ -0,0 +1,2 @@
+Date,Commit,Downstream Project,Author,Subject
+2018/04/28,f3bc9021add1f9cb458223dd374c58c69f53e207,platform/frameworks/support,aurimas@google.com,Move to AGP 3.0.0 stable 😁

tools/repo_diff/service/repodiff/interactors/manifest_test.go

@@ -5,19 +5,19 @@ import (
 
 	"github.com/stretchr/testify/assert"
 
-	c "repodiff/constants"
+	cst "repodiff/constants"
-	e "repodiff/entities"
+	ent "repodiff/entities"
 	"repodiff/persistence/filesystem"
 )
 
 func TestProjectNamesToType(t *testing.T) {
-	var common, downstream, upstream e.ManifestFile
+	var common, downstream, upstream ent.ManifestFile
 	filesystem.ReadXMLAsEntity("testdata/common_manifest.xml", &common)
 	filesystem.ReadXMLAsEntity("testdata/downstream_manifest.xml", &downstream)
 	filesystem.ReadXMLAsEntity("testdata/upstream_manifest.xml", &upstream)
 
 	nameToType := ProjectNamesToType(
-		e.ManifestFileGroup{
+		&ent.ManifestFileGroup{
 			Common:     common,
 			Upstream:   upstream,
 			Downstream: downstream,
@@ -27,7 +27,7 @@ func TestProjectNamesToType(t *testing.T) {
 
 	distinctCount := 0
 	for _, projectType := range nameToType {
-		if projectType == c.DifferentialSpecific {
+		if projectType == cst.DifferentialSpecific {
 			distinctCount++
 		}
 	}

tools/repo_diff/service/repodiff/interactors/strings.go

@@ -1,11 +1,15 @@
 package interactors
 
 import (
+	"regexp"
 	"sort"
+	"strings"
 )
 
 type simpleSet map[string]bool
 
+var unicode = regexp.MustCompile("[^\x00-\x7F]+")
+
 func (s simpleSet) Contains(other string) bool {
 	enabled, exists := s[other]
 	return exists && enabled
@@ -45,7 +49,7 @@ func SetSubtract(add, negate []string) []string {
 }
 
 func SetUnion(slice1, slice2 []string) []string {
-	return allKeys(
+	union := allKeys(
 		sliceToSimpleSet(
 			append(
 				slice1,
@@ -53,6 +57,8 @@ func SetUnion(slice1, slice2 []string) []string {
 			),
 		),
 	)
+	sort.Strings(union)
+	return union
 }
 
 func sliceToSimpleSet(s []string) simpleSet {
@@ -87,3 +93,23 @@ func allKeys(sets ...simpleSet) []string {
 	}
 	return keys
 }
+
+func FilterNoUnicode(s string) string {
+	badCharacters := sliceToSimpleSet(
+		unicode.FindAllString(s, -1),
+	)
+	if len(badCharacters) == 0 {
+		return s
+	}
+	validCharacters := make([]string, 0, len(s))
+	for _, rune_ := range s {
+		char := string(rune_)
+		if !badCharacters.Contains(char) {
+			validCharacters = append(validCharacters, char)
+		}
+	}
+	return strings.Join(
+		validCharacters,
+		"",
+	)
+}

tools/repo_diff/service/repodiff/interactors/strings_test.go

@@ -90,3 +90,23 @@ func TestSetUnion(t *testing.T) {
 	union := SetUnion(s1, s2)
 	assert.Equal(t, expected, union, "Union of s2 and s1")
 }
+
+func TestFilterNoUnicodeWithUnicode(t *testing.T) {
+	regressionStr := "Move to AGP 3.0.0 stable 😁"
+	assert.Equal(
+		t,
+		"Move to AGP 3.0.0 stable ",
+		FilterNoUnicode(regressionStr),
+		"Function should filter out unicode characters",
+	)
+}
+
+func TestFilterNoUnicodeWithNoUnicode(t *testing.T) {
+	validStr := "I'm a regular string with no whacky unicode chars"
+	assert.Equal(
+		t,
+		validStr,
+		FilterNoUnicode(validStr),
+		"No change should occur",
+	)
+}

tools/repo_diff/service/repodiff/mappers/mappers.go

@@ -12,6 +12,7 @@ import (
 
 	"repodiff/constants"
 	e "repodiff/entities"
+	"repodiff/interactors"
 	"repodiff/utils"
 )
 
@@ -123,7 +124,7 @@ func commitRowToPersistCols(c e.AnalyzedCommitRow, uuidBytes string, timestamp i
 		c.Commit,
 		c.DownstreamProject,
 		c.Author,
-		c.Subject,
+		interactors.FilterNoUnicode(c.Subject),
 		c.Type,
 	}
 }