initial commit

internal/parser/access.go

@@ -0,0 +1,70 @@
+package parser
+
+import (
+	"fmt"
+	"regexp"
+	"strconv"
+	"time"
+)
+
+// AccessEntry holds a parsed nginx combined-format access log entry.
+type AccessEntry struct {
+	RemoteAddr string
+	RemoteUser string
+	Time       time.Time
+	Method     string
+	Path       string
+	Protocol   string
+	Status     int
+	BytesSent  int64
+	Referer    string
+	UserAgent  string
+}
+
+// accessLogRe matches nginx combined log format.
+// Groups: remoteAddr, remoteUser, time, method, path, protocol, status, bytes, referer, userAgent
+var accessLogRe = regexp.MustCompile(
+	`^(\S+) - (\S+) \[([^\]]+)\] "(\S+) (\S+) (\S+)" (\d+) (\d+|-) "([^"]*)" "([^"]*)"$`,
+)
+
+const accessTimeLayout = "02/Jan/2006:15:04:05 -0700"
+
+// ParseAccessLine parses a single nginx access log line.
+// Returns an error for malformed lines.
+func ParseAccessLine(line string) (AccessEntry, error) {
+	m := accessLogRe.FindStringSubmatch(line)
+	if m == nil {
+		return AccessEntry{}, fmt.Errorf("access: no match: %q", line)
+	}
+
+	t, err := time.Parse(accessTimeLayout, m[3])
+	if err != nil {
+		return AccessEntry{}, fmt.Errorf("access: parse time %q: %w", m[3], err)
+	}
+
+	status, err := strconv.Atoi(m[7])
+	if err != nil {
+		return AccessEntry{}, fmt.Errorf("access: parse status %q: %w", m[7], err)
+	}
+
+	var bytes int64
+	if m[8] != "-" {
+		bytes, err = strconv.ParseInt(m[8], 10, 64)
+		if err != nil {
+			return AccessEntry{}, fmt.Errorf("access: parse bytes %q: %w", m[8], err)
+		}
+	}
+
+	return AccessEntry{
+		RemoteAddr: m[1],
+		RemoteUser: m[2],
+		Time:       t,
+		Method:     m[4],
+		Path:       m[5],
+		Protocol:   m[6],
+		Status:     status,
+		BytesSent:  bytes,
+		Referer:    m[9],
+		UserAgent:  m[10],
+	}, nil
+}