Okay, this is my first ever Power Automate flow, so its not pretty. I think "Compose 2" and "Compose 3" are not required. If there are any suggestions on how to eliminate the need for 2x "Parse JSON" steps I am all ears. If there is a way to bypass the JSON steps altogether, even better.
I hope my experience, can help someone else.
Power Automate – Trigger
When keywords are mentioned
Message type: Channel
Keywords to search for: “[my keyword]”
Team: [my channel]
Channel – 1: General
Power Automate – Actions
Initialize variable – msg email
Name: msg_email
Type: String
Value: [empty]
Initialize variable – msg caseNumber
Name: msg_caseNumber
Type: String
Value: [empty]
Initialize variable – msg nextAction
Name: msg_nextAction
Type: String
Value: [empty]
Apply to each
Select an output from previous steps: Value = triggerOutputs()?[‘body/value’]
Get message details
Message: Message ID = items(‘Apply_to_each’)?[messageId’]
Message type: Channel
Team: Team ID = items(‘Apply_to_each’)?[‘teamId’]
Channel: Channel ID = items(‘Apply_to_each’)?[‘channelId’]
Parent message ID = [empty]
Parse JSON
Content: Body = body(‘Get_message_details’)
Schema: [I took the output of the message and generated a schema]
{
"type": "object",
"properties": {
"@@odata.context": {
"type": "string"
},
"id": {
"type": "string"
},
"replyToId": {},
"etag": {
"type": "string"
},
"messageType": {
"type": "string"
},
"createdDateTime": {
"type": "string"
},
"lastModifiedDateTime": {
"type": "string"
},
"lastEditedDateTime": {},
"deletedDateTime": {},
"subject": {},
"summary": {
"type": "string"
},
"chatId": {},
"importance": {
"type": "string"
},
"locale": {
"type": "string"
},
"webUrl": {
"type": "string"
},
"policyViolation": {},
"eventDetail": {},
"from": {
"type": "object",
"properties": {
"device": {},
"user": {},
"application": {
"type": "object",
"properties": {
"id": {
"type": "string"
},
"displayName": {
"type": "string"
},
"applicationIdentityType": {
"type": "string"
}
}
}
}
},
"body": {
"type": "object",
"properties": {
"contentType": {
"type": "string"
},
"content": {
"type": "string"
},
"plainTextContent": {
"type": "string"
}
}
},
"channelIdentity": {
"type": "object",
"properties": {
"teamId": {
"type": "string"
},
"channelId": {
"type": "string"
}
}
},
"attachments": {
"type": "array",
"items": {
"type": "object",
"properties": {
"id": {
"type": "string"
},
"contentType": {
"type": "string"
},
"contentUrl": {},
"content": {
"type": "string"
},
"name": {},
"thumbnailUrl": {}
},
"required": [
"id",
"contentType",
"contentUrl",
"content",
"name",
"thumbnailUrl"
]
}
},
"mentions": {
"type": "array"
},
"reactions": {
"type": "array"
},
"messageLink": {
"type": "string"
},
"threadType": {
"type": "string"
},
"teamId": {
"type": "string"
},
"channelId": {
"type": "string"
}
}
}
Apply to each 3
Select an output from previous steps: attachments = body(‘Parse_JSON’)?[‘attachments’]
Compose 2
Inputs: content x = items(‘Apply_to_each_3’)?[‘content’]
Parse JSON 2
Content: Outputs x = outputs(‘Compose_2’)
Schema: [I took the output of the attachment item and generated a schema]
{
"type": "object",
"properties": {
"summary": {
"type": "string"
},
"title": {
"type": "string"
},
"themeColor": {
"type": "string"
},
"sections": {
"type": "array",
"items": {
"type": "object",
"properties": {
"text": {
"type": "string"
},
"title": {
"type": "string"
},
"activityTitle": {
"type": "string"
},
"activitySubtitle": {
"type": "string"
},
"activityText": {
"type": "string"
},
"markdown": {
"type": "boolean"
},
"startGroup": {
"type": "boolean"
}
},
"required": [
"markdown",
"startGroup"
]
}
},
"potentialAction": {
"type": "array"
}
}
}
Compose 3
Inputs: bodu(…) x = body(‘Parse_JSON_2’)[‘sections’][0][‘text’]
Note: after Compose 3 there are 3 parallel branches to create 3 variables which then converge back at Get an @mention token for a user.
1. Branch for variable msg_email
Find text position “(“
Text: Outputs x = outputs(‘Compose_3’)
Search Text: (
Substring email
Text: Outputs x = outputs(‘Compose_3’)
Starting Position: add(…) x = add(int(outputs('Find_text_position_"("')?['body']),1)
Length: sub(…) x = sub(sub(int(outputs('Find_text_position_")"')?['body']),int(outputs('Find_text_position_"("')?['body'])),1)
Set variable msg email
Name: msg_email
Value: Body x = body(‘Substring_email’)
2. Branch for variable msg_caseNumber
Find text position “[case prefix]“
Text: Outputs x = outputs(‘Compose_3’)
Search Text: [case prefix]
Substring caseNumber
Text: Outputs x = outputs(‘Compose_3’)
Starting Position: Body x = body(‘Find_text_position_”[case prefix]”’)
Length: 11
Set variable msg caseNumber
Name: msg_caseNumber
Value: Body x = body(‘Substring_caseNumber’)
3. Branch for variable msg_nextAction
Find text position “colour:blue“
Text: Outputs x = outputs(‘Compose_3’)
Search Text: color:blue
Substring nextAction+end tag
Text: Outputs x = outputs(‘Compose_3’)
Starting Position: add(…) x = add(12,int(outputs('Find_text_position_"color:blue"')?['body']))
Length: [empty]
Substring nextAction
Text: Body x = body(‘Substring_nextAction+end_tag’)
Starting Position: 0
Length: sub(…) x = sub(length(outputs('Substring_nextAction+end_tag')?['body']),7)
Set variable msg nextAction
Name: msg_nextAction
Value: Body = body(‘Substring_nextAction’)
Get an @mention token for a user
User: variables(‘msg_email’)
Reply with a message in a channel
Post as: Flow bot
Post in: Channel
Message ID: outputs(‘Get_message_deatils’)?[‘body/id’]
Channel: Channel ID = outputs(‘Get_message_details’)?[‘body/channelId’]
Message:
<p>@{outputs('Get_an_@mention_token_for_a_user')?['body/atMention']} Case @{variables('msg_caseNumber')} has an <span style="color: red"><strong>overdue Next Action</strong></span>. The stated action was:.<br>
<br>
<span style="color: blue"></span><span style="color: blue">@{variables('msg_nextAction')}</span><span style="color: blue"></span></p>
