MAJOR: ignore commit hashes during checks

Author: oktalz

aspell/aspell.go

@@ -5,6 +5,7 @@ import (
 	"fmt"
 	"log"
 	"os/exec"
+	"regexp"
 	"slices"
 	"strings"
 
@@ -121,7 +122,7 @@ func (a Aspell) checkSingle(data string, allowedWords []string) error {
 	return nil
 }
 
-func (a Aspell) Check(subjects []string, commitsFull []string, content []map[string]string, junitSuite junit.Interface) error {
+func (a Aspell) Check(subjects []string, commitsFull []string, content []map[string]string, junitSuite junit.Interface, gitHashes map[string]struct{}) error {
 	var commitsFullData []string
 	for _, c := range commitsFull {
 		commit := []string{}
@@ -143,6 +144,14 @@ func (a Aspell) Check(subjects []string, commitsFull []string, content []map[str
 		commitsFullData = append(commitsFullData, strings.Join(commit, "\n"))
 	}
 
+	// Remove known git commit hashes from body portions of commit messages
+	// so they are not flagged by spell checking. Subject lines are preserved.
+	if len(gitHashes) > 0 {
+		for i, c := range commitsFullData {
+			commitsFullData[i] = removeKnownHashesFromBody(c, gitHashes)
+		}
+	}
+
 	var response strings.Builder
 	var checks []string
 	switch a.Mode {
@@ -194,6 +203,30 @@ func (a Aspell) Check(subjects []string, commitsFull []string, content []map[str
 	return nil
 }
 
+var hexStringRe = regexp.MustCompile(`[0-9a-fA-F]{7,40}`)
+
+// removeKnownHashesFromBody removes known git commit hashes from the body
+// of a commit message, leaving the subject line intact. A hex string in the
+// body is removed if it is a prefix of (or equal to) any known full hash.
+func removeKnownHashesFromBody(message string, fullHashes map[string]struct{}) string {
+	parts := strings.SplitN(message, "\n\n", 2)
+	if len(parts) < 2 {
+		return message // no body
+	}
+
+	body := hexStringRe.ReplaceAllStringFunc(parts[1], func(match string) string {
+		lower := strings.ToLower(match)
+		for hash := range fullHashes {
+			if strings.HasPrefix(hash, lower) {
+				return ""
+			}
+		}
+		return match
+	})
+
+	return parts[0] + "\n\n" + body
+}
+
 func checkWithAspellExec(subject string) (string, error) {
 	cmd := exec.Command("aspell", "--lang=en", "--list")
 	cmd.Stdin = strings.NewReader(subject)

aspell/aspell_test.go

@@ -36,6 +36,57 @@ func Test_checkWithAspell(t *testing.T) {
 	}
 }
 
+func Test_removeKnownHashesFromBody(t *testing.T) {
+	hashes := map[string]struct{}{
+		"abc1234567890abcdef1234567890abcdef12340": {},
+		"def9876543210fedcba9876543210fedcba98760": {},
+	}
+	tests := []struct {
+		name   string
+		input  string
+		expect string
+	}{
+		{
+			"no body",
+			"subject line only",
+			"subject line only",
+		},
+		{
+			"hash in body is removed",
+			"subject line\n\nThis reverts abc1234567890abcdef1234567890abcdef12340.",
+			"subject line\n\nThis reverts .",
+		},
+		{
+			"short hash in body is removed",
+			"subject line\n\nThis reverts abc1234.",
+			"subject line\n\nThis reverts .",
+		},
+		{
+			"hash in subject is preserved",
+			"revert abc1234\n\nbody text here",
+			"revert abc1234\n\nbody text here",
+		},
+		{
+			"unknown hex string preserved",
+			"subject\n\nSee fffffff for details",
+			"subject\n\nSee fffffff for details",
+		},
+		{
+			"multiple hashes in body",
+			"subject\n\nReverts abc1234 and def9876.",
+			"subject\n\nReverts  and .",
+		},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			got := removeKnownHashesFromBody(tt.input, hashes)
+			if got != tt.expect {
+				t.Errorf("removeKnownHashesFromBody() = %q, want %q", got, tt.expect)
+			}
+		})
+	}
+}
+
 func TestAspell_Check(t *testing.T) {
 	type fields struct {
 		Mode         mode
@@ -109,7 +160,7 @@ func TestAspell_Check(t *testing.T) {
 				AllowedWords: tt.fields.AllowedWords,
 				HelpText:     tt.fields.HelpText,
 			}
-			if err := a.Check(tt.args.subjects, tt.args.commitsFull, tt.args.content, &junit.JunitSuiteDummy{}); (err != nil) != tt.wantErr {
+			if err := a.Check(tt.args.subjects, tt.args.commitsFull, tt.args.content, &junit.JunitSuiteDummy{}, nil); (err != nil) != tt.wantErr {
 				t.Errorf("Aspell.Check() error = %v, wantErr %v", err, tt.wantErr)
 			}
 		})

aspell_test.go

@@ -37,7 +37,7 @@ func Test_Aspell(t *testing.T) {
 	}
 	err = aspellCheck.Check([]string{"subject"}, []string{"body"}, []map[string]string{
 		{filename: readme.String()},
-	}, &junit.JunitSuiteDummy{})
+	}, &junit.JunitSuiteDummy{}, nil)
 	if err != nil {
 		t.Errorf("checkWithAspell() error = %v", err)
 	}

check.go

@@ -575,3 +575,35 @@ func (c CommitPolicyConfig) CheckSubjectList(subjects []string, junitSuite junit
 }
 
 const requiredCmdlineArgs = 2
+
+// getGitHashes opens the git repository at repoPath and collects all commit
+// hashes (full, lowercase). These are used to ignore commit hash references
+// that appear in commit message bodies.
+func getGitHashes(repoPath string) map[string]struct{} {
+	hashes := map[string]struct{}{}
+
+	repo, err := git.PlainOpen(repoPath)
+	if err != nil {
+		log.Printf("warning: could not open git repo for hash collection: %s", err)
+		return hashes
+	}
+
+	iter, err := repo.Log(&git.LogOptions{
+		Order: git.LogOrderCommitterTime,
+		All:   true,
+	})
+	if err != nil {
+		log.Printf("warning: could not get git log for hash collection: %s", err)
+		return hashes
+	}
+
+	_ = iter.ForEach(func(c *object.Commit) error {
+		full := strings.ToLower(c.Hash.String())
+		hashes[full] = struct{}{}
+		return nil
+	})
+
+	log.Printf("collected %d git commit hashes for body hash filtering", len(hashes))
+
+	return hashes
+}

main.go

@@ -112,7 +112,9 @@ func start(junitSuite junit.Interface) {
 		return
 	}
 
-	err = aspellCheck.Check(subjects, messages, content, junitSuite)
+	gitHashes := getGitHashes(repoPath)
+
+	err = aspellCheck.Check(subjects, messages, content, junitSuite, gitHashes)
 	if err != nil {
 		log.Print("encountered one or more commit message spelling errors")
 		// log.Fatalf("%s\n", err)