標籤: n8n

n8n 新增code節點，跳脫json字串

技術

json字串有些特殊字元，需要跳脫後才能正常使用。因此有必要在n8n加上code節點寫javascript 跳脫。

const chars = {
  "\"": "\\\"",
  "\\": "\\\\",
  "\/": "\\/",
  "\u0000": "\\u0000",
  "\u0001": "\\u0001",
  "\u0002": "\\u0002",
  "\u0003": "\\u0003",
  "\u0004": "\\u0004",
  "\u0005": "\\u0005",
  "\u0006": "\\u0006",
  "\u0007": "\\u0007",
  "\u0008": "\\u0008",
  "\u0009": "\\u0009",
  "\u000A": "\\u000A",
  "\u000B": "\\u000B",
  "\u000C": "\\u000C",
  "\u000D": "\\u000D",
  "\u000E": "\\u000E",
  "\u000F": "\\u000F",
  "\u0010": "\\u0010",
  "\u0011": "\\u0011",
  "\u0012": "\\u0012",
  "\u0013": "\\u0013",
  "\u0014": "\\u0014",
  "\u0015": "\\u0015",
  "\u0016": "\\u0016",
  "\u0017": "\\u0017",
  "\u0018": "\\u0018",
  "\u0019": "\\u0019",
  "\u001A": "\\u001A",
  "\u001B": "\\u001B",
  "\u001C": "\\u001C",
  "\u001D": "\\u001D",
  "\u001E": "\\u001E",
  "\u001F": "\\u001F",
  "\u007F": "\\u007F",
  "\u0080": "\\u0080",
  "\u0081": "\\u0081",
  "\u0082": "\\u0082",
  "\u0083": "\\u0083",
  "\u0084": "\\u0084",
  "\u0085": "\\u0085",
  "\u0086": "\\u0086",
  "\u0087": "\\u0087",
  "\u0088": "\\u0088",
  "\u0089": "\\u0089",
  "\u008A": "\\u008A",
  "\u008B": "\\u008B",
  "\u008C": "\\u008C",
  "\u008D": "\\u008D",
  "\u008E": "\\u008E",
  "\u008F": "\\u008F",
  "\u0090": "\\u0090",
  "\u0091": "\\u0091",
  "\u0092": "\\u0092",
  "\u0093": "\\u0093",
  "\u0094": "\\u0094",
  "\u0095": "\\u0095",
  "\u0096": "\\u0096",
  "\u0097": "\\u0097",
  "\u0098": "\\u0098",
  "\u0099": "\\u0099",
  "\u009A": "\\u009A",
  "\u009B": "\\u009B",
  "\u009C": "\\u009C",
  "\u009D": "\\u009D",
  "\u009E": "\\u009E",
  "\u009F": "\\u009F"
};
$json.output = $input.first().json.output.replace(/[\"\\\/\u0000-\u001F\u007F\u0080-\u009F]/g, match=>chars[match]);
return $input.all();
Leave a comment , ,

docker安裝apache tika 文件辨識系統與N8N整合

其他, 技術

現在主流文件辨識，應該都交由vision功能的AI模型處裡，辨識度高；若不想花錢使用AI，可以用apache tika頂著用。

docker 安裝 apache tika

docker run -d -p 9998:9998 --name tika-server-ocr apache/tika:latest-full
# 目前是3.1.0

安裝完畢，進入容器，安裝中文語言套件

docker exec -u root -it tika-server-ocr bash
###
apt update
apt-get install tesseract-ocr-chi-sim tesseract-ocr-chi-tra
###

測試

# Linux
# 中文圖檔 test.png
curl -T test.png http://127.0.0.1:9998/tika --header "X-Tika-OCRLanguage: eng+chi_tra+chi_sim"

N8N(nodemation)設定

前一個節點要把檔案準備好，再新增以下節點，丟給tika處理，回傳設定為text

Leave a comment , , , , ,

為了n8n能用mcp，我得改用 Nginx Proxy Manager (docker)

其他, 技術

我這種老派工程師，一直以來都用apache httpd 做反向代理，也沒用docker安裝，直到為了n8n能用mcp，果斷改用docker 安裝 nginx proxy manager。

nginx proxy manager 安裝方式請參考官方文件，安裝完畢反向對應到本地端的n8n(docker)，此時需要額外設定，避開gzip壓縮問題，才能順利搭上n8n上的mcp server。

# disable gzip and proxy buffering for both /mcp/ and /mcp-test/
location ~* ^/(?:mcp|mcp-test)/ {
    gzip                       off;
    proxy_buffering            off;
    chunked_transfer_encoding  off;

    # prevent upstream compression
    proxy_set_header           Connection "";
    proxy_set_header           Accept-Encoding "";

    # re‑declare NPM’s proxy headers using built‑in variables
    proxy_http_version         1.1;
    proxy_set_header           Host                 $host;
    proxy_set_header           X-Real-IP            $remote_addr;
    proxy_set_header           X-Forwarded-For      $proxy_add_x_forwarded_for;
    proxy_set_header           X-Forwarded-Proto    $scheme;

    # send to your upstream based on NPM’s variables
    proxy_pass                 $forward_scheme://$server:$port;
}

解法出處

Leave a comment , , , ,