[分享] HA App通知搭配Dashboard上顯示通知歷程_update20241111

Home Assistant
1

前言

Line Notify宣佈停用之後大家開始轉進其他通訊軟體來做通知，因不想再被第三方軟體限制，加上HA app本身的定位準確度搭配第三方的整合如icloud3較為準確，而轉進研究HA本身的通知系統，HA自身的Companion App通知功能很強大，但沒辦法留存紀錄導致無法查詢舊通知，使用上有一些缺陷，依賴ChatGPT實現了在dashboard上顯示通知歷程，廢話不多說正文開始

工作原理

建立文字助手 :arrow_right:利用Shell command建立一個文字檔將通知訊息做保存 :arrow_right:建立自動化：當文字助手狀態變化，執行文字檔更新 :arrow_right:建立一個腳本將.txt內容轉成JSON格式並暫存系統:arrow_right:建立一個command sensor將文字檔訊息導出:arrow_right:建立markdown card將command sensor訊息顯示出來

設定

  1. 建立文字助手

    截圖 2024-11-12 凌晨12.00.19
    截圖 2024-11-12 凌晨12.00.19483×567 14.2 KB

  2. 利用Shell command建立歷程文字檔(保留50則)

shell_command:
  #event_input correspond to the input_text.event_input
  prepend_event_log_user: >
    /bin/bash -c "
    echo -e '{{ event_input | replace('\n', '\\n') }}\n{{ now().strftime('%Y-%m-%d %H:%M:%S') }}\n' | 
    cat - /config/www/event_log_user.txt |
    awk -v RS='' -v ORS='\n\n' 'BEGIN {count=0} {count++; if (count<=50) print}' > /config/www/event_log_user.tmp && 
    mv /config/www/event_log_user.tmp /config/www/event_log_user.txt"
  • 歷程格式為{{ now().strftime(“%Y-%m-%d %H:%M”) }} - {{ event_input }}，其中變數event_input後面會對應到{{ states(“input_text.event_input”) }}
  • /config/www/event_log_user.txt 為文字檔儲存路徑
    *count<=50 為要保存的最後50筆資料，可自行變更
  1. 建立自動化更新文字檔
alias: "Notification: Prepend Event to Log and Notify user"
description: ""
triggers:
  - entity_id:
      - input_text.event_input_for_user
    trigger: state
conditions:
  - condition: template
    value_template: "{{ states('input_text.event_input_for_user') | length > 0 }}"
actions:
  - data:
      event_input: "{{ states('input_text.event_input_for_user') }}"
    action: shell_command.prepend_event_log_user
  #此步驟是新增一個空白項目避免舊訊息反覆疊加
  - data:
      value: ""
    action: input_text.set_value
    enabled: true
    target:
      entity_id: input_text.event_input_for_user
  1. 建立腳本將txt文字轉成JSON
  • 在config目錄下建立腳本“log_to_json_user.sh”
#!/bin/bash

# Get the current date and time
current_time=$(date +"%Y-%m-%d %H:%M:%S")

# Start the JSON output with the last update timestamp
echo '{ "last_update": "'"$current_time"'", "events": ['

# Initialize a flag to check if it's the first item
first_item=true

# Read the log file line by line
while read -r line; do
    # Check if the line is a timestamp (matches YYYY-MM-DD format)
    if [[ "$line" =~ ^[0-9]{4}-[0-9]{2}-[0-9]{2} ]]; then
        # Print the message and timestamp if both are available
        if [[ -n "$message" ]]; then
            # Escape double quotes in the message
            message=$(echo "$message" | sed 's/"/\\"/g')

            # If it's not the first item, add a comma
            if [ "$first_item" = false ]; then
                echo ","
            fi

            # Print the JSON object for the event
            echo "{\"message\": \"$message\", \"timestamp\": \"$line\"}"

            # Set the first_item flag to false after printing the first event
            first_item=false
        fi
        # Reset the message for the next event
        message=""
    else
        # Append the line to the message
        if [[ -n "$message" ]]; then
            message="$message\n$line"
        else
            message="$line"
        fi
    fi
done < /config/www/event_log_user.txt

# Close the JSON array and object
echo ']}'
  • 至Terminal更改.sh檔案權限(注意檔案路徑)
chmod +x /config/www/log_to_json_user.sh
  • Command Line sensor 搭配 “log_to_json_user.sh”
command_line:
  - sensor:
      name: Event Log user
      command: "bash /config/www/log_to_json_user.sh 2>&1" # Capture script errors too
      scan_interval: 5 # Run every 5 seconds
      value_template: "{{ value_json.last_update | default('Unknown') }}" # Use 'last_update' field or default to 'Unknown'
      json_attributes:
        - events # Store the events array as attributes
  • HA sensor的state有字元數的限制，attributes的字元限制較高，要將txt的內容導入到command line sensor才可以完整顯示50則歷程，生成Entity如下
    image
    image1610×762 103 KB
  • 自動化連動通知和更新event_log_user.txt可參考進階設定的第二步驟，其中input text在輸入的時候務必確保有將每個項目換行，才會讓command line sensor的attributes變成條列式顯示，如上示意圖，這樣後面Markdown顯示出來才會是條列式的，這邊結露出來加強說明一下
      - action: input_text.set_value
        metadata: {}
        data:
          value: |
#在通知標題加上"unread"或"未讀"，然後要讓event_log_user.txt的每條項目都換行，轉換成JSON格式才會對
            **(unread){{ title }}**  
            {{ message }} 
            ![image]({{ image }})
        target:
          entity_id:
            - input_text.event_input_for_user
  1. 建立markdown card
  {% for event in state_attr('sensor.event_log_user', 'events') %}
  <ha-alert alert-type="info">{{ event.message }}
  ***{{ event.timestamp }}***</ha-alert>
  {% endfor %}
  1. 成果
    image
    image331×909 53 KB

進階設定

若要加上訊息讀取與否這邊可以增加兩個步驟：建立數字助手 :arrow_right:通知中的標題加上未讀字樣並且讓Companion App badge的未讀數字稱加 :arrow_right:利用Shell command清除未讀字樣 :arrow_right:點開APP看完訊息後退到後台之後，消除未讀字樣

  1. 建立數字助手

    截圖 2024-11-12 凌晨12.14.54
    截圖 2024-11-12 凌晨12.14.54478×537 8.31 KB

  2. 通知自動化中的標題加上未讀字樣並且讓Companion App badge的未讀數字增加

這邊利用腳本的方式，方便做不同通知的的應用

alias: "Text: Log to Input and Notify"
sequence:
  - action: input_number.increment
    metadata: {}
    data: {}
    target:
      entity_id:
        - input_number.notification_user_badge
  - parallel:
      - action: notify.mobile_app_user
        data:
          title: "{{title}}"
          message: "{{message}}"
          data:
            image: "{{image}}"
            push:
              interruption-level: "{{level}}"
 #讓badge上疊加未讀訊息數量
              badge: "{{ states('input_number.notification_user_badge') | int }}"
              sound:
                name: "{{sound}}"
                volume: "{{volume}}"
            tag: "{{tag}}"
      - action: input_text.set_value
        metadata: {}
        data:
          value: |
#在通知標題加上"unread"或"未讀"，然後要讓event_log_user.txt的每條項目都換行，轉換成JSON格式才會對
            **(unread){{ title }}**  
            {{ message }} 
            ![image]({{ image }})
        target:
          entity_id:
            - input_text.event_input_for_user
description: ""
icon: mdi:card-text-outline
  1. 建立Shell command清除未讀字樣
shell_command:
  refresh_event_log_user: "sed -i 's/ (unread)//g; s/(unread) //g; s/(unread)//g' /config/www/event_log_user.txt"
  1. 點開APP看完訊息後消除未讀字樣，Badge歸0
    Companion App退到後台後會觸發對應的event，利用這點來達成此目的，對應event請參閱官方說明
#清除為讀字樣
alias: "Notification: Remove \"unread\" from user"
description: Mark the notification as read
triggers:
  - trigger: event
    event_type: ios.entered_background
    event_data:
      sourceDeviceID: YourDeviceID
    context:
      user_id:
        - YourUserID
conditions: []
actions:
  - action: shell_command.refresh_event_log_user
    data: {}
mode: single

#Badge歸0
alias: "Notification: Reset User Badge"
description: Badge number to "0"
triggers:
  - trigger: event
    event_type: ios.entered_background
    event_data:
      sourceDeviceID: YouDeviceID
    context:
      user_id:
        - YourUserID
conditions: []
actions:
  - action: input_number.set_value
    metadata: {}
    data:
      value: 0
    target:
      entity_id: input_number.notification_user_badge
mode: single

演示

  1. 通知action範例 - 利用步驟二的腳本完成所有步驟
alias: "Notification: Empty the water"
description: ""
triggers:
  - entity_id:
      - binary_sensor.lgdehumidifier_water_tank_full
    from: "off"
    to: "on"
    for:
      hours: 0
      minutes: 0
      seconds: 3
    trigger: state
conditions: []
actions:
  - variables:
      title: Notification
      message: |
        Empty the water tank from the dehumidifier.
      image: /local/notification/dehumidifier.jpg
      level: active
      sound: default
      volume: 0
      tag: notification
#呼叫第2步的腳本
  - action: script.text_log_to_input_and_notify
    data:
      title: "{{ title }}"
      message: "{{ message }}"
      image: "{{ image }}"
      level: "{{ level }}"
      sound: "{{ sound }}"
      volume: "{{ volume }}"
      tag: "{{ tag }}"
mode: single

.

總結

HA本身的File可以利用notify.send_message來自動生成歷程txt，但紀錄方式是最新的一筆在下方，而且是無限制的新增，應該也可以利用語法來更新這個txt，作法應該大同小異。
Companion APP的功能很多，也能直接做互動式訊息，本文是以iOS版本來撰寫，Android版本以及其餘功能應用請自行參考官方文件：https://companion.home-assistant.io/
本身非軟體專業，code或許可以在做優化，另外可以搭配自動化或Node Red流程將歷程自動更新，也可附加圖檔等markdown可支持的操作，本歷程為HA本地記錄，不限於其他通訊軟體應用，可並存。

2個讚