企業のオンラインイメージ監視ツール

• •Googleアカウント + Gmail API認証情報
• •OpenAI API Key

{
  "nodes": [
    {
      "name": "毎朝トリガー (午前9時)",
      "type": "n8n-nodes-base.cron",
      "notes": {
        "text": "### 1. Daily Morning Trigger (9 AM)\n\nThis `Cron` node triggers the workflow automatically every **day at 9:00 AM** (based on your n8n server's local time zone). This provides a regular daily digest of activity.\n\n**To change the schedule:** Adjust the 'Hour' and 'Minute' fields to your preferred time for receiving the report.",
        "position": "right"
      },
      "position": [
        240,
        300
      ],
      "parameters": {
        "mode": "everyDay",
        "value": {
          "hour": [
            9
          ],
          "minute": [
            0
          ]
        },
        "options": {}
      },
      "typeVersion": 1,
      "id": "-9--0"
    },
    {
      "name": "企業詳細設定",
      "type": "n8n-nodes-base.set",
      "notes": {
        "text": "### 2. Set Company Details\n\nThis `Set` node defines the company you want to monitor and the keywords to search for.\n\n**Setup (CRITICAL!):**\n1.  **`companyName`:** Change `YourCompanyNameHere` to the exact name of the company you are monitoring (e.g., 'Google', 'Coca-Cola'). This is used for the email subject and AI prompts.\n2.  **`searchKeywords`:** List all relevant keywords, including common misspellings, product names, hashtags, or related terms. These will be used for searching news, Reddit, and YouTube. Example: `[\"Acme Corp\", \"AcmeCo\", \"#AcmeCorp\", \"AcmeSoftware\"]`",
        "position": "right"
      },
      "position": [
        480,
        300
      ],
      "parameters": {
        "values": [
          {
            "name": "companyName",
            "value": "YourCompanyNameHere"
          },
          {
            "name": "searchKeywords",
            "value": [
              "YourCompanyNameHere",
              "YourCompanyProduct",
              "#YourCompanyTag"
            ]
          }
        ],
        "options": {}
      },
      "typeVersion": 2,
      "id": "--1"
    },
    {
      "name": "Google ニュースRSS取得",
      "type": "n8n-nodes-base.rssFeed",
      "notes": {
        "text": "### 3.1. Fetch Google News RSS\n\nThis `RSS Feed` node fetches news articles from Google News that mention your `companyName`.\n\n**Setup:**\n1.  **URL:** It's pre-configured to search Google News using your `companyName`. You can adjust the `hl` (host language) and `gl` (geo location) parameters if needed.\n\n**Note:** Google News RSS is a good general news source, but it might not capture every single article across all sites.",
        "position": "right"
      },
      "position": [
        720,
        100
      ],
      "parameters": {
        "url": "=https://news.google.com/rss/search?q={{ encodeURIComponent($node[\"Set Company Details\"].json.companyName) }}&hl=en-US&gl=US&ceid=US:en",
        "options": {}
      },
      "typeVersion": 1,
      "id": "Google-RSS--2"
    },
    {
      "name": "ニュース統合準備",
      "type": "n8n-nodes-base.function",
      "notes": {
        "text": "### 3.2. Prepare News for Merge\n\nThis `Function` node standardizes the news articles' data into a consistent format (`source`, `title`, `text`, `link`, `publishedAt`) for later merging and AI processing.\n\n**No configuration needed.**",
        "position": "right"
      },
      "position": [
        960,
        100
      ],
      "parameters": {
        "options": {},
        "function": "const preparedItems = [];\n\nfor (const item of items) {\n  preparedItems.push({\n    json: {\n      source: 'News Article',\n      title: item.json.title || 'N/A',\n      text: item.json.contentSnippet || item.json.description || 'N/A',\n      link: item.json.link || '#',\n      publishedAt: item.json.isoDate || new Date().toISOString()\n    }\n  });\n}\n\nreturn preparedItems;"
      },
      "typeVersion": 1,
      "id": "--3"
    },
    {
      "name": "Reddit 投稿検索",
      "type": "n8n-nodes-base.reddit",
      "notes": {
        "text": "### 3.3. Search Reddit Posts\n\nThis `Reddit` node searches for posts containing your `searchKeywords`.\n\n**Setup:**\n1.  **Reddit Credential:** Click 'Credentials' and select 'New Credential'. Choose 'Reddit OAuth2 API'. Follow n8n documentation for detailed steps to create a Reddit app (type 'script') and get your client ID and secret. You'll also need your Reddit username/password for authentication.\n2.  **Query:** It's pre-configured to search using your `searchKeywords` joined by ' OR '.\n3.  **Limit & Sort:** Adjust the number of results (`limit`) and `sort` order as needed.\n\n**Note:** Reddit API has rate limits, so be mindful of how often you run this if you increase the limit significantly.",
        "position": "right"
      },
      "position": [
        720,
        300
      ],
      "parameters": {
        "query": "={{ $node[\"Set Company Details\"].json.searchKeywords.join(' OR ') }}",
        "options": {
          "sort": "hot",
          "limit": 20
        },
        "resource": "post",
        "operation": "search"
      },
      "credentials": {
        "redditOAuth2Api": {
          "id": "YOUR_REDDIT_CREDENTIAL_ID",
          "resolve": false
        }
      },
      "typeVersion": 1,
      "id": "Reddit--4"
    },
    {
      "name": "Reddit 統合準備",
      "type": "n8n-nodes-base.function",
      "notes": {
        "text": "### 3.4. Prepare Reddit for Merge\n\nThis `Function` node standardizes the Reddit posts' data, ensuring consistent `source`, `title`, `text`, `link`, and `publishedAt` fields for merging.\n\n**No configuration needed.**",
        "position": "right"
      },
      "position": [
        960,
        300
      ],
      "parameters": {
        "options": {},
        "function": "const preparedItems = [];\n\nfor (const item of items) {\n  preparedItems.push({\n    json: {\n      source: 'Reddit Post',\n      title: item.json.title || 'N/A',\n      text: item.json.selftext || item.json.body || 'N/A',\n      link: item.json.url || '#',\n      publishedAt: new Date(item.json.created_utc * 1000).toISOString() // Convert Unix timestamp to ISO string\n    }\n  });\n}\n\nreturn preparedItems;"
      },
      "typeVersion": 1,
      "id": "Reddit--5"
    },
    {
      "name": "YouTube 動画検索",
      "type": "n8n-nodes-base.youTube",
      "notes": {
        "text": "### 3.5. Search YouTube Videos\n\nThis `YouTube` node searches for videos mentioning your `searchKeywords`.\n\n**Setup:**\n1.  **Google Credential:** Click 'Credentials' and select 'New Credential'. Choose 'Google OAuth2 API'. You'll need to enable the YouTube Data API v3 in your Google Cloud Project and configure OAuth2 credentials (Client ID/Secret) for desktop app/other. This can be complex; search n8n docs for 'Google OAuth2' for detailed steps.\n2.  **Search:** Pre-configured to use your `searchKeywords`.\n3.  **Limit & Order:** Adjust the number of results and order as needed.\n\n**Note:** YouTube API has quotas, so excessive searches might lead to temporary limits.",
        "position": "right"
      },
      "position": [
        720,
        500
      ],
      "parameters": {
        "search": "={{ $node[\"Set Company Details\"].json.searchKeywords.join(' ') }}",
        "options": {
          "limit": 10,
          "order": "relevance"
        },
        "resource": "video",
        "operation": "list"
      },
      "credentials": {
        "googleApi": {
          "id": "YOUR_GOOGLE_API_CREDENTIAL_ID",
          "resolve": false
        }
      },
      "typeVersion": 1,
      "id": "YouTube--6"
    },
    {
      "name": "YouTube 統合準備",
      "type": "n8n-nodes-base.function",
      "notes": {
        "text": "### 3.6. Prepare YouTube for Merge\n\nThis `Function` node standardizes YouTube video data into the common format (`source`, `title`, `text`, `link`, `publishedAt`) for merging.\n\n**No configuration needed.**",
        "position": "right"
      },
      "position": [
        960,
        500
      ],
      "parameters": {
        "options": {},
        "function": "const preparedItems = [];\n\nfor (const item of items) {\n  preparedItems.push({\n    json: {\n      source: 'YouTube Video',\n      title: item.json.snippet.title || 'N/A',\n      text: item.json.snippet.description || 'N/A',\n      link: `https://www.youtube.com/watch?v=${item.json.id.videoId}` || '#',\n      publishedAt: item.json.snippet.publishedAt || new Date().toISOString()\n    }\n  });\n}\n\nreturn preparedItems;"
      },
      "typeVersion": 1,
      "id": "YouTube--7"
    },
    {
      "name": "全言及統合",
      "type": "n8n-nodes-base.itemLists",
      "notes": {
        "text": "### 4. Merge All Mentions\n\nThis `Item Lists` node combines all prepared data from Google News, Reddit, and YouTube into a single, unified list. This is the stream of all potential mentions.\n\n**No configuration needed.**",
        "position": "right"
      },
      "position": [
        1200,
        300
      ],
      "parameters": {
        "mode": "merge",
        "options": {}
      },
      "typeVersion": 1,
      "id": "--8"
    },
    {
      "name": "SQLite: テーブル存在確認",
      "type": "n8n-nodes-base.sqlite",
      "notes": {
        "text": "### 5.1. SQLite: Ensure Table Exists\n\nThis `SQLite` node ensures a local database table named `processed_mentions` exists. This table will store a hash of every mention seen, preventing duplicate notifications over time.\n\n**Setup:**\n* **Database:** `company_monitor` (this creates a file `company_monitor.db` in your n8n data directory).\n* **Query:** The `CREATE TABLE IF NOT EXISTS` query is pre-filled.\n\n**No further action needed**; this node runs automatically.",
        "position": "right"
      },
      "position": [
        1440,
        220
      ],
      "parameters": {
        "query": "CREATE TABLE IF NOT EXISTS processed_mentions (link_hash TEXT PRIMARY KEY, source TEXT, title TEXT, link TEXT, processed_date TEXT)",
        "database": "company_monitor"
      },
      "typeVersion": 1,
      "id": "SQLite--9"
    },
    {
      "name": "新規言及フィルター (重複排除)",
      "type": "n8n-nodes-base.function",
      "notes": {
        "text": "### 5.2. Filter New Mentions (Deduplication)\n\nThis `Function` node generates a unique hash for each mention (based on its link/title) and then checks the `processed_mentions` database to see if it's already been seen. **Only truly new mentions are passed on.**\n\n**Setup:**\n* It uses `crypto` (built-in) for hashing.\n* It executes the 'SQLite: Check If Processed' node for each item.\n\n**Important:** If you change the `SQLite: Check If Processed` node's name, update it here. Also, ensure 'Run Once Per Item' is OFF for this Function node.",
        "position": "right"
      },
      "position": [
        1440,
        380
      ],
      "parameters": {
        "options": {
          "runOncePerItem": false
        },
        "function": "const crypto = require('crypto');\n\nconst newItems = [];\n\nfor (const item of items) {\n  const link = item.json.link || item.json.title; // Use link or title as primary unique identifier\n  const linkHash = crypto.createHash('md5').update(link).digest('hex');\n\n  // Add the hash to the item for later use\n  item.json.linkHash = linkHash;\n\n  // Check if this item (based on linkHash) is already in the database\n  const result = await n8n.getNodeParameter('SQLite: Check If Processed', 'query', { json: item.json }, { execute: true });\n\n  if (result.length === 0) {\n    // Item not found in DB, it's new\n    newItems.push(item);\n  }\n}\n\nreturn newItems;"
      },
      "typeVersion": 1,
      "id": "--10"
    },
    {
      "name": "SQLite: 処理済み確認",
      "type": "n8n-nodes-base.sqlite",
      "notes": {
        "text": "### (Helper) SQLite: Check If Processed\n\nThis helper `SQLite` node is called by the 'Filter New Mentions' Function node to check if a specific `link_hash` already exists in the database.\n\n**No direct configuration needed**; it's managed by the Function node.",
        "position": "right"
      },
      "position": [
        1700,
        380
      ],
      "parameters": {
        "query": "SELECT link_hash FROM processed_mentions WHERE link_hash = '{{ $json.linkHash }}'",
        "database": "company_monitor"
      },
      "typeVersion": 1,
      "id": "SQLite--11"
    },
    {
      "name": "AI: 感情分析＆要約",
      "type": "n8n-nodes-base.openAi",
      "notes": {
        "text": "### 6. AI: Analyze Sentiment & Summarize\n\nThis `OpenAI` node processes each *new* mention to determine its sentiment and generate a brief summary.\n\n**Setup:**\n1.  **OpenAI Credential:** Select your OpenAI API Key credential.\n2.  **Model:** `gpt-3.5-turbo` is selected. For more robust analysis, consider `gpt-4o`.\n3.  **Prompt:** The system prompt guides the AI to output sentiment and summary in a structured JSON format, making it easy to parse in the next step.\n\n**Output:** Each item will have a `choices` field containing the AI's JSON output.",
        "position": "right"
      },
      "position": [
        1920,
        300
      ],
      "parameters": {
        "model": "gpt-3.5-turbo",
        "options": {},
        "messages": [
          {
            "role": "system",
            "content": "You are a sentiment analyzer and summarizer for company mentions. For the following text related to '{{ $node[\"Set Company Details\"].json.companyName }}', determine its overall sentiment (Positive, Negative, or Neutral) and provide a concise 2-sentence summary. Output in JSON format:\n\n{\n  \"sentiment\": \"[Positive/Negative/Neutral]\",\n  \"summary\": \"[Concise summary]\"\n}"
          },
          {
            "role": "user",
            "content": "Source: {{ $json.source }}\\nTitle: {{ $json.title }}\\nText: {{ $json.text }}"
          }
        ]
      },
      "credentials": {
        "openAiApi": {
          "id": "YOUR_OPENAI_CREDENTIAL_ID",
          "resolve": false
        }
      },
      "typeVersion": 1,
      "id": "AI--12"
    },
    {
      "name": "AI結果処理＆分類",
      "type": "n8n-nodes-base.function",
      "notes": {
        "text": "### 7. Process AI Results & Categorize\n\nThis `Function` node parses the AI's JSON output and categorizes all *new* mentions by their sentiment (Positive, Negative, Neutral). It also handles potential AI parsing errors.\n\n**Output:** A single item containing categorized lists of new mentions (e.g., `positive: [...]`, `negative: [...]`).\n\n**No configuration needed.**",
        "position": "right"
      },
      "position": [
        2160,
        300
      ],
      "parameters": {
        "options": {},
        "function": "const results = {\n  positive: [],\n  neutral: [],\n  negative: [],\n  noAnalysis: []\n};\n\nfor (const item of items) {\n  try {\n    const aiOutput = JSON.parse(item.json.choices[0].message.content);\n    const sentiment = aiOutput.sentiment ? aiOutput.sentiment.toLowerCase() : 'unknown';\n    const summary = aiOutput.summary || 'No summary provided.';\n\n    const data = {\n      source: item.json.source,\n      title: item.json.title,\n      link: item.json.link,\n      summary: summary,\n      publishedAt: item.json.publishedAt,\n      linkHash: item.json.linkHash // Pass the hash for database insertion\n    };\n\n    if (sentiment === 'positive') {\n      results.positive.push(data);\n    } else if (sentiment === 'negative') {\n      results.negative.push(data);\n    } else if (sentiment === 'neutral') {\n      results.neutral.push(data);\n    } else {\n      results.noAnalysis.push(data);\n    }\n  } catch (e) {\n    results.noAnalysis.push({ ...item.json, summary: `AI analysis failed: ${e.message}` });\n  }\n}\n\nreturn [{ json: results }];"
      },
      "typeVersion": 1,
      "id": "AI--13"
    },
    {
      "name": "SQLite: 処理済み言及記録",
      "type": "n8n-nodes-base.sqlite",
      "notes": {
        "text": "### 8. SQLite: Record Processed Mentions\n\nThis `SQLite` node records all the *newly processed* mentions into your `processed_mentions` database. This is crucial for the deduplication step in future runs.\n\n**Setup:**\n* **Database:** `company_monitor`\n* **Query:** The `INSERT INTO` query is pre-filled, storing the unique hash and other details of the processed mention.\n\n**No further action needed**; it automatically stores the data.",
        "position": "right"
      },
      "position": [
        2400,
        220
      ],
      "parameters": {
        "query": "INSERT INTO processed_mentions (link_hash, source, title, link, processed_date) VALUES ('{{ $json.linkHash }}', '{{ $json.source }}', '{{ $json.title }}', '{{ $json.link }}', '{{ new Date().toISOString() }}')",
        "database": "company_monitor"
      },
      "typeVersion": 1,
      "id": "SQLite--14"
    },
    {
      "name": "レポートメールフォーマット",
      "type": "n8n-nodes-base.function",
      "notes": {
        "text": "### 9. Format Report Email\n\nThis `Function` node constructs the final email body. It groups mentions by sentiment and provides summaries and links for each. It also handles the case where no *new* mentions were found.\n\n**Customization:**\n* You can adjust the headings, introduction, or the level of detail for each mention.\n* The email body uses Markdown for formatting (bolding, lists).\n\n**No configuration needed.**",
        "position": "right"
      },
      "position": [
        2400,
        380
      ],
      "parameters": {
        "options": {},
        "function": "const data = items[0].json;\nconst companyName = $node[\"Set Company Details\"].json.companyName;\n\nlet emailBody = `Good morning! Here's your daily online presence report for **${companyName}**.\\n\\n`;\n\n// Helper to format a list of mentions\nconst formatMentions = (mentions, heading) => {\n  if (mentions.length === 0) return '';\n\n  let formatted = `### ${heading} (${mentions.length} mentions)\\n`;\n  mentions.forEach(m => {\n    formatted += `* **[${m.source}]** ${m.title}\\n` +\n                 `    Summary: ${m.summary}\\n` +\n                 `    Link: ${m.link}\\n` +\n                 `    (Published: ${new Date(m.publishedAt).toLocaleDateString()})\\n\\n`;\n  });\n  return formatted;\n};\n\nemailBody += formatMentions(data.positive, 'Positive Mentions');\nemailBody += formatMentions(data.neutral, 'Neutral Mentions');\nemailBody += formatMentions(data.negative, 'Negative Mentions');\nemailBody += formatMentions(data.noAnalysis, 'Mentions with No Analysis / Error');\n\n\n// Check if any new mentions were processed\nif (data.positive.length + data.neutral.length + data.negative.length + data.noAnalysis.length === 0) {\n  emailBody = `Good morning! No *new* online mentions found for **${companyName}** today.`;\n}\n\nreturn [{ json: { emailSubject: `📊 Daily Online Presence Report for ${companyName}`, emailBody: emailBody } }];"
      },
      "typeVersion": 1,
      "id": "--15"
    },
    {
      "name": "レポートメール送信",
      "type": "n8n-nodes-base.gmail",
      "notes": {
        "text": "### 10. Send Report Email\n\nThis `Gmail` node sends the compiled daily online presence report to your mailbox.\n\n**Setup:**\n1.  **Gmail Credential:** Select your Gmail API credential.\n2.  **From Email:** Enter your Gmail address (must match the authenticated account).\n3.  **To Email:** **IMPORTANT: Change `YOUR_RECIPIENT_EMAIL@example.com` to your actual email address!**\n4.  **Subject & Text:** These fields pull the formatted subject and body from the previous 'Format Report Email' node.\n\n**Test this node by running the workflow** to ensure you receive the email.",
        "position": "right"
      },
      "position": [
        2640,
        300
      ],
      "parameters": {
        "text": "={{ $json.emailBody }}",
        "options": {},
        "subject": "={{ $json.emailSubject }}",
        "toEmail": "YOUR_RECIPIENT_EMAIL@example.com",
        "fromEmail": "YOUR_GMAIL_EMAIL@gmail.com"
      },
      "credentials": {
        "gmailApi": {
          "id": "YOUR_GMAIL_CREDENTIAL_ID",
          "resolve": false
        }
      },
      "typeVersion": 2,
      "id": "--16"
    }
  ],
  "pinData": {},
  "version": 1,
  "connections": {
    "--8": {
      "main": [
        [
          {
            "node": "--10",
            "type": "main"
          }
        ]
      ]
    },
    "--15": {
      "main": [
        [
          {
            "node": "--16",
            "type": "main"
          }
        ]
      ]
    },
    "Reddit--4": {
      "main": [
        [
          {
            "node": "Reddit--5",
            "type": "main"
          }
        ]
      ]
    },
    "--1": {
      "main": [
        [
          {
            "node": "Google-RSS--2",
            "type": "main"
          },
          {
            "node": "Reddit--4",
            "type": "main"
          },
          {
            "node": "YouTube--6",
            "type": "main"
          }
        ]
      ]
    },
    "Google-RSS--2": {
      "main": [
        [
          {
            "node": "--3",
            "type": "main"
          }
        ]
      ]
    },
    "YouTube--6": {
      "main": [
        [
          {
            "node": "YouTube--7",
            "type": "main"
          }
        ]
      ]
    },
    "--3": {
      "main": [
        [
          {
            "node": "--8",
            "type": "main"
          }
        ]
      ]
    },
    "Reddit--5": {
      "main": [
        [
          {
            "node": "--8",
            "type": "main"
          }
        ]
      ]
    },
    "YouTube--7": {
      "main": [
        [
          {
            "node": "--8",
            "type": "main"
          }
        ]
      ]
    },
    "SQLite--9": {
      "main": [
        [
          {
            "node": "--10",
            "type": "main"
          }
        ]
      ]
    },
    "-9--0": {
      "main": [
        [
          {
            "node": "--1",
            "type": "main"
          }
        ]
      ]
    },
    "AI--13": {
      "main": [
        [
          {
            "node": "SQLite--14",
            "type": "main"
          },
          {
            "node": "--15",
            "type": "main"
          }
        ]
      ]
    },
    "AI--12": {
      "main": [
        [
          {
            "node": "AI--13",
            "type": "main"
          }
        ]
      ]
    },
    "SQLite--14": {
      "main": [
        []
      ]
    },
    "--10": {
      "main": [
        [
          {
            "node": "AI--12",
            "type": "main"
          }
        ]
      ],
      "output": [
        {
          "type": "item",
          "toIndex": 0,
          "fromIndex": 0,
          "destination": [
            {
              "node": "SQLite: 処理済み確認",
              "input": "input1"
            }
          ]
        }
      ]
    }
  }
}