From d826aa899dca6a3ff53066bd9af6ed627327d5c0 Mon Sep 17 00:00:00 2001
From: ChrsMark <chrismarkou92@gmail.com>
Date: Wed, 18 Dec 2024 11:28:12 +0200
Subject: [PATCH] allow overriding container operator & disallow target path

Signed-off-by: ChrsMark <chrismarkou92@gmail.com>
---
 receiver/receivercreator/README.md         |  5 +-
 receiver/receivercreator/discovery.go      | 19 +++++-
 receiver/receivercreator/discovery_test.go | 74 ++++++++++++++++++++++
 3 files changed, 96 insertions(+), 2 deletions(-)

diff --git a/receiver/receivercreator/README.md b/receiver/receivercreator/README.md
index 915032382b6f..b378e11a98a1 100644
--- a/receiver/receivercreator/README.md
+++ b/receiver/receivercreator/README.md
@@ -558,7 +558,10 @@ io.opentelemetry.discovery.logs/config: |
 ```
 
 Note that individual settings are overridden by the configuration provided by the hints while the operators list
-is extended keeping first the `container` parser.
+is extended keeping first the `container` parser. If `container` parser is explicitly set by the user the default
+will be overridden.
+
+`include` cannot be overridden and is fixed to discovered container's log file path.
 
 
 #### Support multiple target containers
diff --git a/receiver/receivercreator/discovery.go b/receiver/receivercreator/discovery.go
index 80df7a2d8d16..752280612620 100644
--- a/receiver/receivercreator/discovery.go
+++ b/receiver/receivercreator/discovery.go
@@ -228,14 +228,31 @@ func createLogsConfig(
 		logger.Debug("could not unmarshal configuration from hint", zap.Error(err))
 	}
 
+	containerOpFound := false
 	for k, v := range userConf {
 		if k == filelogOperatorsConfigKey {
 			vlist, ok := v.([]any)
 			if !ok {
 				logger.Debug("could not parse operators configuration from hint", zap.Any("config", userConf))
 			}
-			vlist = append(cont, vlist...)
+			for _, op := range vlist {
+				operator, ok := op.(map[string]any)
+				if !ok {
+					logger.Debug("could not parse operator configuration from hint", zap.Any("operator", op))
+				}
+				if operator["type"] == "container" {
+					containerOpFound = true
+				}
+			}
+			if !containerOpFound {
+				// if no container operator found then just extend the list
+				// otherwise we just use the user provided operators as-is
+				vlist = append(cont, vlist...)
+			}
 			defaultConfMap[k] = vlist
+		} else if k == "include" {
+			// path cannot be other than the one of the target container
+			continue
 		} else {
 			defaultConfMap[k] = v
 		}
diff --git a/receiver/receivercreator/discovery_test.go b/receiver/receivercreator/discovery_test.go
index bee6fd73fa7e..90a797188ec8 100644
--- a/receiver/receivercreator/discovery_test.go
+++ b/receiver/receivercreator/discovery_test.go
@@ -239,6 +239,15 @@ operators:
 - type: add
   field: attributes.tag
   value: beta`
+	configReplaceContainer := `
+operators:
+- type: container
+  id: replaced
+- type: regex_parser
+  regex: "^(?P<time>\\d{4}-\\d{2}-\\d{2} \\d{2}:\\d{2}:\\d{2}) (?P<sev>[A-Z]*) (?P<msg>.*)$"`
+	configReplaceInclude := `
+include:
+- /var/log/pod/x/foo.log`
 
 	tests := map[string]struct {
 		inputEndpoint    observer.Endpoint
@@ -421,6 +430,71 @@ operators:
 			expectedReceiver: receiverTemplate{},
 			wantError:        true,
 			ignoreReceivers:  []string{},
+		}, `logs_pod_level_hints_replace_container_operator`: {
+			inputEndpoint: observer.Endpoint{
+				ID:     "namespace/pod-2-UID/filelog(redis)",
+				Target: "1.2.3.4:",
+				Details: &observer.PodContainer{
+					Name: "redis", Pod: observer.Pod{
+						Name:      "pod-2",
+						Namespace: "default",
+						UID:       "pod-2-UID",
+						Labels:    map[string]string{"env": "prod"},
+						Annotations: map[string]string{
+							otelLogsHints + "/enabled": "true",
+							otelLogsHints + "/config":  configReplaceContainer,
+						},
+					},
+				},
+			},
+			expectedReceiver: receiverTemplate{
+				receiverConfig: receiverConfig{
+					id: id,
+					config: userConfigMap{
+						"include":           []string{"/var/log/pods/default_pod-2_pod-2-UID/redis/*.log"},
+						"include_file_name": false,
+						"include_file_path": true,
+						"operators": []any{
+							map[string]any{"id": "replaced", "type": "container"},
+							map[string]any{"type": "regex_parser", "regex": "^(?P<time>\\d{4}-\\d{2}-\\d{2} \\d{2}:\\d{2}:\\d{2}) (?P<sev>[A-Z]*) (?P<msg>.*)$"},
+						},
+					},
+				}, signals: receiverSignals{metrics: false, logs: true, traces: false},
+			},
+			wantError:       false,
+			ignoreReceivers: []string{},
+		}, `logs_pod_level_hints_do_not_replace_path`: {
+			inputEndpoint: observer.Endpoint{
+				ID:     "namespace/pod-2-UID/filelog(redis)",
+				Target: "1.2.3.4:",
+				Details: &observer.PodContainer{
+					Name: "redis", Pod: observer.Pod{
+						Name:      "pod-2",
+						Namespace: "default",
+						UID:       "pod-2-UID",
+						Labels:    map[string]string{"env": "prod"},
+						Annotations: map[string]string{
+							otelLogsHints + "/enabled": "true",
+							otelLogsHints + "/config":  configReplaceInclude,
+						},
+					},
+				},
+			},
+			expectedReceiver: receiverTemplate{
+				receiverConfig: receiverConfig{
+					id: id,
+					config: userConfigMap{
+						"include":           []string{"/var/log/pods/default_pod-2_pod-2-UID/redis/*.log"},
+						"include_file_name": false,
+						"include_file_path": true,
+						"operators": []any{
+							map[string]any{"id": "container-parser", "type": "container"},
+						},
+					},
+				}, signals: receiverSignals{metrics: false, logs: true, traces: false},
+			},
+			wantError:       false,
+			ignoreReceivers: []string{},
 		},
 	}