Merge pull request #11 from jum/extract_caddy2

Extract caddy log information, second try

README.md

@@ -136,6 +136,7 @@ The following [log-opts](https://docs.docker.com/config/containers/logging/confi
 | extract-severity     | true    | Extracts the `severity` from JSON logs to set them for the log that will be sent to GCP. It will be removed from the jsonPayload section, since it is set at the root level. Currently the supported severity field names to extract are the following: `severity`, `level` |
 | extract-msg          | true    | Extracts the `msg` field from JSON logs to set the `message` field GCP expects. It will be removed from the jsonPayload section, since it is set at the root level. Fields named msg are produced for example by the golang log/slog package.                               |
 | extract-gcp          | false   | Extract trace, labels and source location fields if present and formatted for Google cloud logging. This is produced for example by the golang log/slog package with the slogdriver handler |
+| extract-caddy        | false   | Extract trace and HTTP Request from caddy if present and format for Google cloud logging.                   |
 | exclude-timestamp    | false   | Excludes timestamp fields from the final jsonPayload, since docker sends its own nanosecond precision timestamp for each log. Currently it can remove fields with the following names: `timestamp`, `time`, `ts`                                                            |
 | sleep-interval       | 500     | Milliseconds to sleep when there are no logs to send before checking again. The higher the value, the lower the CPU usage will be                                                                                                                                           |
 | credentials-file     |         | Absolute path to the GCP credentials JSON file to use when authenticating (only necessary when running the plugin outside of GCP)                                                                                                                                           |

ngcplogger.go

@@ -6,6 +6,8 @@ import (
 	"errors"
 	"fmt"
 	"log/slog"
+	"net/http"
+	"net/url"
 	"reflect"
 	"runtime"
 	"sync"
@@ -78,12 +80,14 @@ type nGCPLogger struct {
 	logger    *logging.Logger
 	instance  *instanceInfo
 	container *containerInfo
+	projectID string
 
 	extractJsonMessage bool
 	extractSeverity    bool
 	excludeTimestamp   bool
 	extractMsg         bool
 	extractGcp         bool
+	extractCaddy       bool
 }
 
 type dockerLogEntry struct {
@@ -205,11 +209,13 @@ func New(info logger.Info) (logger.Logger, error) {
 			Created:   info.ContainerCreated,
 			Metadata:  extraAttributes,
 		},
+		projectID:          project,
 		extractJsonMessage: true,
 		extractSeverity:    true,
 		excludeTimestamp:   false,
 		extractMsg:         true,
 		extractGcp:         false,
+		extractCaddy:       false,
 	}
 
 	if info.Config[logCmdKey] == "true" {
@@ -231,6 +237,9 @@ func New(info logger.Info) (logger.Logger, error) {
 	if info.Config["extract-gcp"] == "true" {
 		l.extractGcp = true
 	}
+	if info.Config["extract-caddy"] == "true" {
+		l.extractCaddy = true
+	}
 
 	if instanceResource != nil {
 		l.instance = instanceResource
@@ -292,6 +301,7 @@ func (l *nGCPLogger) Log(lMsg *logger.Message) error {
 				m["instance"] = l.instance
 				m["container"] = l.container
 				l.extractGcpFromPayload(m, &entry)
+				l.extractCaddyFromPayload(m, &entry)
 				payload = m
 			}
 		} else {
@@ -403,6 +413,63 @@ func (l *nGCPLogger) extractGcpFromPayload(m map[string]any, entry *logging.Entr
 	}
 }
 
+func (l *nGCPLogger) extractCaddyFromPayload(m map[string]any, entry *logging.Entry) {
+
+	if l.extractCaddy {
+		if val, exists := m["request"]; exists {
+			hr := logging.HTTPRequest{
+				Request: &http.Request{
+					Header: make(http.Header),
+				},
+			}
+			v := assertOrLog[map[string]any](val)
+			hr.Request.Method = assertOrLog[string](v["method"])
+			_, isTLS := v["tls"]
+			var h = "http"
+			if isTLS {
+				h = "https"
+			}
+			hr.Request.URL = &url.URL{
+				Scheme:  h,
+				Host:    assertOrLog[string](v["host"]),
+				RawPath: assertOrLog[string](v["uri"]),
+				Path:    assertOrLog[string](v["uri"]),
+			}
+			if t, ok := m["bytes_read"]; ok {
+				hr.RequestSize = int64(assertOrLog[float64](t))
+			}
+			if t, ok := m["status"]; ok {
+				hr.Status = int(assertOrLog[float64](t))
+			}
+			if t, ok := m["size"]; ok {
+				hr.ResponseSize = int64(assertOrLog[float64](t))
+			}
+			if t, ok := m["duration"]; ok {
+				hr.Latency = time.Duration(assertOrLog[float64](t) * float64(time.Second))
+			}
+			hr.Request.Proto = assertOrLog[string](v["proto"])
+			hr.RemoteIP = assertOrLog[string](v["remote_ip"]) + ":" + assertOrLog[string](v["remote_port"])
+
+			if t, ok := v["headers"]; ok {
+				headers := assertOrLog[map[string]any](t)
+				for h, v := range headers {
+					for _, s := range assertOrLog[[]any](v) {
+						hr.Request.Header.Add(h, assertOrLog[string](s))
+					}
+				}
+			}
+			entry.HTTPRequest = &hr
+			//Caddy request contains more data, don't
+			//delete.
+			//delete(m, "request")
+		}
+		if val, exists := m["traceID"]; exists {
+			entry.Trace = "projects/" + l.projectID + "/traces/" + val.(string)
+			delete(m, "traceID")
+		}
+	}
+}
+
 func (l *nGCPLogger) Close() error {
 	err := l.logger.Flush()
 	if err != nil {