Add parser.GitCommit field into ParseInfo struct (#1005)

* Add parser.GitCommit field into ParseInfo struct

* Fixing styling

* Updating GitCommit to include the complete hash

Author: cristinaleonr

parser/annotation.go

@@ -83,6 +83,7 @@ func (ap *AnnotationParser) ParseAndInsert(meta map[string]bigquery.Value, testN
 			Time:       time.Now(),
 			ArchiveURL: meta["filename"].(string),
 			Filename:   testName,
+			GitCommit:  GitCommit(),
 		},
 	}
 

parser/annotation_test.go

@@ -2,11 +2,14 @@ package parser_test
 
 import (
 	"io/ioutil"
+	"strings"
 	"testing"
 
 	"cloud.google.com/go/bigquery"
 	"cloud.google.com/go/civil"
+	"github.com/go-test/deep"
 	"github.com/m-lab/etl/parser"
+	"github.com/m-lab/etl/schema"
 	"github.com/m-lab/go/rtx"
 )
 
@@ -46,6 +49,24 @@ func TestAnnotationParser_ParseAndInsert(t *testing.T) {
 			if err := n.ParseAndInsert(meta, tt.file, data); (err != nil) != tt.wantErr {
 				t.Errorf("AnnotationParser.ParseAndInsert() error = %v, wantErr %v", err, tt.wantErr)
 			}
+
+			if n.Accepted() == 1 {
+				n.Flush()
+				row := ins.data[0].(*schema.AnnotationRow)
+
+				expPI := schema.ParseInfo{
+					Version:    "https://github.com/m-lab/etl/tree/foobar",
+					Time:       row.Parser.Time,
+					ArchiveURL: "gs://mlab-test-bucket/ndt/ndt7/2020/03/18/ndt-njp6l_1585004303_00000000000170FA.json",
+					Filename:   "ndt-njp6l_1585004303_00000000000170FA.json",
+					Priority:   0,
+					GitCommit:  "12345678",
+				}
+
+				if diff := deep.Equal(row.Parser, expPI); diff != nil {
+					t.Errorf("AnnotationParser.ParseAndInsert() different summary: %s", strings.Join(diff, "\n"))
+				}
+			}
 		})
 	}
 }

parser/export_test.go

@@ -7,3 +7,7 @@ package parser
 // InitParserVersionForTest allows tests to rerun initParserVersion after initializing
 // environment variables.
 var InitParserVersionForTest = initParserVersion
+
+// InitParserGitCommitForTest allows test to rerun initParseGitCommit after initializing
+// environement variables.
+var InitParserGitCommitForTest = initParserGitCommit

parser/ndt7_result.go

@@ -81,6 +81,7 @@ func (dp *NDT7ResultParser) ParseAndInsert(meta map[string]bigquery.Value, testN
 			Time:       time.Now(),
 			ArchiveURL: meta["filename"].(string),
 			Filename:   testName,
+			GitCommit:  GitCommit(),
 		},
 	}
 

parser/ndt7_result_test.go

@@ -70,6 +70,7 @@ func TestNDT7ResultParser_ParseAndInsert(t *testing.T) {
 					ArchiveURL: "gs://mlab-test-bucket/ndt/ndt7/2020/03/18/ndt_ndt7_2020_03_18_20200318T003853.425987Z-ndt7-mlab3-syd03-ndt.tgz",
 					Filename:   "ndt7-download-20200318T000657.568382877Z.ndt-knwp4_1583603744_000000000000590E.json",
 					Priority:   0,
+					GitCommit:  "12345678",
 				}
 				if diff := deep.Equal(row.Parser, expPI); diff != nil {
 					pretty.Print(row.Parser)
@@ -108,6 +109,7 @@ func TestNDT7ResultParser_ParseAndInsert(t *testing.T) {
 					ArchiveURL: "gs://mlab-test-bucket/ndt/ndt7/2020/03/18/ndt_ndt7_2020_03_18_20200318T003853.425987Z-ndt7-mlab3-syd03-ndt.tgz",
 					Filename:   "ndt7-upload-20200318T001352.496224022Z.ndt-knwp4_1583603744_0000000000005CF2.json",
 					Priority:   0,
+					GitCommit:  "12345678",
 				}
 				if diff := deep.Equal(row.Parser, expPI); diff != nil {
 					t.Errorf("NDT7ResultParser.ParseAndInsert() different summary: %s", strings.Join(diff, "\n"))

parser/parser.go

@@ -21,31 +21,46 @@ import (
 
 func init() {
 	initParserVersion()
+	initParserGitCommit()
 }
 
-var gParserVersion = "uninitialized"
+const uninitialized = "uninitialized"
+
+var (
+	gParserVersion   = uninitialized
+	gParserGitCommit = uninitialized
+)
 
 // initParserVersion initializes the gParserVersion variable for use by all parsers.
 func initParserVersion() string {
 	release := etl.Version
 	if release != "noversion" {
 		gParserVersion = "https://github.com/m-lab/etl/tree/" + release
 	} else {
-		hash := etl.GitCommit
-		if hash != "nocommit" && len(hash) >= 8 {
-			gParserVersion = "https://github.com/m-lab/etl/tree/" + hash[0:8]
-		} else {
-			gParserVersion = "local development"
-		}
+		gParserVersion = "local development"
 	}
 	return gParserVersion
 }
 
+// initParserGitCommit initializes the gParserGitCommit variable for use by all parsers.
+func initParserGitCommit() string {
+	hash := etl.GitCommit
+	if hash != "nocommit" {
+		gParserGitCommit = hash
+	}
+	return gParserGitCommit
+}
+
 // Version returns the parser version used by parsers to annotate data rows.
 func Version() string {
 	return gParserVersion
 }
 
+// GitCommit returns the git commit hash of the build.
+func GitCommit() string {
+	return gParserGitCommit
+}
+
 // NewSinkParser creates an appropriate parser for a given data type.
 // Eventually all datatypes will use this instead of NewParser.
 func NewSinkParser(dt etl.DataType, sink row.Sink, table string, ann api.Annotator) etl.Parser {

parser/parser_test.go

@@ -19,6 +19,9 @@ import (
 func init() {
 	etl.Version = "foobar"
 	parser.InitParserVersionForTest()
+
+	etl.GitCommit = "12345678"
+	parser.InitParserGitCommitForTest()
 }
 
 // countingInserter counts the calls to InsertRows and Flush.

schema/schema.go

@@ -18,6 +18,7 @@ type ParseInfo struct {
 	ArchiveURL string
 	Filename   string
 	Priority   int64
+	GitCommit  string
 }
 
 // FindSchemaDocsFor should be used by parser row types to associate bigquery