Remove unicode ans ansi color codes from ps

metacoma · Sep 15, 2023 · b5ad7f4 · b5ad7f4
1 parent 8f85da9
commit b5ad7f4
Showing 1 changed file with 17 additions and 4 deletions.
diff --git a/inventory/targets/vector.yml b/inventory/targets/vector.yml
@@ -131,7 +131,6 @@ parameters:
               .input = msg.message.input
               .output = msg.message.output
               .host = msg.host
-              .metadata = msg.message.metadata
               del(.message)
 
           kafka-linestream-to-elasticsearch:
@@ -188,6 +187,19 @@ parameters:
                 input = {}
                 ps1 = {}
               end
+              local function remove_unicode_ansi_sequences(input)
+                return input:gsub('%[%?[0-9][0-9][0-9][0-9][a-z]', ''):gsub('\27', '')
+              end
+
+              local function extract_ps(data)
+                      if (string.match(data, "^%S+@%S+:")) then
+                              print("match")
+                              return data:gsub("^([^@]+)@([^:]+).*", "%1@%2")
+                      end
+                      return data
+              end
+
+
               function process(event, emit)
                 ch = event.log.message
                 tmux_pane_id = event.log.metadata.tmux.pane_id
@@ -201,12 +213,13 @@ parameters:
                   -- we found a space character and the line matches the prompt regex
                   -- skip sent mesage if ps1 is empty, usually it's a first io-context document in the stream
                   if (data[tmux_pane_id] ~= nil and event.log.message.input ~= "" and data[tmux_pane_id] ~= {}) then
+                    event.log.metadata.ps1_start_raw = ps1_data
+                    event.log.metadata.ps1_end_raw = line[tmux_pane_id]
                     event.log.message = {
-                      ps1_start = ps1_data,
-                      ps1_end = line[tmux_pane_id],
+                      ps1_start = extract_ps((remove_unicode_ansi_sequences(ps1_data))),
+                      ps1_end = extract_ps((remove_unicode_ansi_sequences(line[tmux_pane_id]))),
                       input = input[tmux_pane_id],
                       output = table.concat(data[tmux_pane_id]),
-                      metadata = event.log.metadata
                     }
                     emit(event)
                   end