{ "id": "", "meta": { "templateCredsSetupCompleted": false }, "name": "Telegram Expense Tracker \u2014 Query Bot (Template)", "tags": [], "nodes": [ { "id": "58535881-40c5-4d75-82c6-f0faa9443e1a", "name": "MSG | Telegram Inbound", "type": "n8n-nodes-base.telegramTrigger", "position": [ 2144, 1088 ], "parameters": { "updates": [ "message", "callback_query" ], "additionalFields": {} }, "credentials": { "telegramApi": { "name": "" } }, "typeVersion": 1.1 }, { "id": "69f61c14-bd15-4c5a-b0b8-ada463bff839", "name": "LLM | Parse Intent", "type": "@n8n/n8n-nodes-langchain.openAi", "position": [ 2880, 1632 ], "parameters": { "modelId": { "__rl": true, "mode": "list", "value": "gpt-4.1-nano", "cachedResultName": "GPT-4.1-NANO" }, "options": {}, "messages": { "values": [ { "role": "system", "content": "=You are a financial intent parser.\nYour ONLY task is to convert a natural language finance question into structured JSON.\nYou must:\n- Return ONLY valid JSON.\n- Do NOT include explanations.\n- Do NOT include markdown.\n- Do NOT answer the question.\n- Do NOT guess missing information.\n- Do NOT add extra fields.\n\nThe user's first name is: {{ $('MSG | Telegram Inbound').item.json.message.from.first_name }}\nThe current year is: {{ new Date().getFullYear() }}\nThe current date is: {{ new Date().toISOString().split('T')[0] }}\n\nAlways return the following JSON structure:\n{\n \"intent\": \"spending_query\",\n \"time_reference\": \"last_week, this_week, last_month, this_month, last_year, this_year, or null\",\n \"explicit_start_date\": \"YYYY-MM-DD or null\",\n \"explicit_end_date\": \"YYYY-MM-DD or null\",\n \"person\": \"string or null\",\n \"category\": [\"string\"] or null,\n \"common_only\": true, false, or null,\n \"comparison\": \"previous_period, previous_month, previous_year, or null\",\n \"group_by\": \"category, person, or null\"\n}\n\nInterpretation Rules:\n1. Date Handling (IMPORTANT):\n- If the user mentions a specific month without a year (e.g. \"february\", \"januar\"),\n assume the current year and set explicit_start_date and explicit_end_date accordingly.\n Example: \"february\" \u2192 explicit_start_date: \"2026-02-01\", explicit_end_date: \"2026-02-28\"\n- If the user refers to relative time (e.g., last week, this month, last month, this year, etc.), \n set the correct value in \"time_reference\".\n- DO NOT calculate actual dates for relative time references.\n- Leave explicit_start_date and explicit_end_date as null for relative time references.\n- If the user provides an explicit date range (e.g., 01.01.2026 - 31.01.2026),\n set explicit_start_date and explicit_end_date accordingly,\n and set time_reference to null.\n- If no time reference is mentioned \u2192 set all three fields to null.\n\n2. Person Handling:\n- If user says \"I\", \"me\", \"ich\", \"mir\" \u2192 use \"{{ $('MSG | Telegram Inbound').item.json.message.from.first_name }}\"\n- If user says \"we\", \"wir\", \"uns\", \"beide\" \u2192 null (refers to both persons, no filter needed).\n- If a specific name or nickname is mentioned that refers to another person \u2192 use that exact name as written.\n- If no person is mentioned \u2192 null.\n- NEVER invent or guess a person name.\n- The value for \"person\" must always be either the mentioned name/nickname, \"{{ $('MSG | Telegram Inbound').item.json.message.from.first_name }}\", or null.\n\n3. category:\n- If one category is mentioned, return it as a single-element array.\n- If multiple categories are mentioned, return all of them as an array.\n- Use lowercase strings.\n- If none mentioned \u2192 null.\n\n4. Common Expense Handling (VERY IMPORTANT):\nSet \"common_only\" to true if the user explicitly refers to shared expenses.\nThis includes words or phrases such as:\n- shared\n- common\n- joint\n- together\n- gemeinschaftlich\n- gemeinsam\n- gemeinsame ausgaben\n- geteilte ausgaben\n- f\u00fcr uns\n- zusammen\nIf the user explicitly says:\n- \"f\u00fcr gemeinsame Ausgaben\"\n- \"f\u00fcr gemeinsame Kosten\"\nyou MUST set \"common_only\" to true.\nSet \"common_only\" to false if the user explicitly refers to private expenses.\nThis includes:\n- private\n- pers\u00f6nlich\n- eigene ausgaben\n- nur f\u00fcr sich\n- privat\nIf shared/private is NOT explicitly mentioned \u2192 set \"common_only\" to null.\nNever assume.\n\n5. Comparison:\n- If user compares with previous month \u2192 \"previous_month\"\n- If compares with previous period \u2192 \"previous_period\"\n- If compares with previous year \u2192 \"previous_year\"\n- Otherwise \u2192 null\n\n6. Grouping:\n- If user asks for breakdown by category \u2192 \"category\"\n- If breakdown by person \u2192 \"person\"\n- Otherwise \u2192 null\n\nReturn ONLY JSON." }, { "content": "={{ $json.message.text }}" } ] } }, "credentials": { "openAiApi": { "name": "" } }, "typeVersion": 1.4 }, { "id": "1f1984d4-cc31-4730-8354-69c6ec425bbe", "name": "JS | Extract Intent JSON", "type": "n8n-nodes-base.code", "position": [ 3232, 1632 ], "parameters": { "jsCode": "const rawContent = $input.first().json.message?.content || $input.first().json.choices?.[0]?.message?.content || '{}';\nlet parsed = {};\ntry {\n const cleaned = rawContent.replace(/```json|```/g, '').trim();\n parsed = JSON.parse(cleaned);\n} catch(e) {\n parsed = { intent: 'spending_query', start_date: null, end_date: null, person: null, category: null, common_only: null, comparison: null, group_by: null };\n}\nreturn [{ json: parsed }];" }, "typeVersion": 2 }, { "id": "4e830963-ede6-4d56-b328-7f3119040143", "name": "GS | Load Expenses", "type": "n8n-nodes-base.googleSheets", "position": [ 9824, 1440 ], "parameters": { "options": {}, "sheetName": { "__rl": true, "mode": "list", "value": "gid=0", "cachedResultUrl": "https://docs.google.com/spreadsheets/d/YOUR_SPREADSHEET_ID", "cachedResultName": "Sheet1" }, "documentId": { "__rl": true, "mode": "list", "value": "YOUR_SPREADSHEET_ID", "cachedResultUrl": "https://docs.google.com/spreadsheets/d/YOUR_SPREADSHEET_ID", "cachedResultName": "expenses" } }, "credentials": { "googleSheetsOAuth2Api": { "name": "" } }, "typeVersion": 4.4 }, { "id": "4271bf57-4b8d-4f68-8896-2d55f162161e", "name": "GS | Load Categories", "type": "n8n-nodes-base.googleSheets", "position": [ 9824, 1152 ], "parameters": { "options": {}, "sheetName": { "__rl": true, "mode": "list", "value": "gid=0", "cachedResultUrl": "https://docs.google.com/spreadsheets/d/YOUR_SPREADSHEET_ID", "cachedResultName": "Sheet1" }, "documentId": { "__rl": true, "mode": "list", "value": "YOUR_SPREADSHEET_ID", "cachedResultUrl": "https://docs.google.com/spreadsheets/d/YOUR_SPREADSHEET_ID", "cachedResultName": "expense_categories" } }, "credentials": { "googleSheetsOAuth2Api": { "name": "" } }, "typeVersion": 4.4 }, { "id": "21d72a8f-e9cd-4fac-b832-f0804777d270", "name": "JS | Filter & Aggregate", "type": "n8n-nodes-base.code", "position": [ 10576, 1264 ], "parameters": { "jsCode": "// =============================\n// Date Engine\n// =============================\n\nconst now = new Date();\n\nfunction normalizeDate(date) {\n const d = new Date(date);\n if (isNaN(d)) return null;\n d.setHours(0,0,0,0);\n return d;\n}\n\nfunction getDateRange(timeReference) {\n\n switch (timeReference) {\n\n case \"this_month\":\n return {\n start: normalizeDate(new Date(now.getFullYear(), now.getMonth(), 1)),\n end: normalizeDate(now)\n };\n\n case \"last_month\":\n return {\n start: normalizeDate(new Date(now.getFullYear(), now.getMonth() - 1, 1)),\n end: normalizeDate(new Date(now.getFullYear(), now.getMonth(), 0))\n };\n\n case \"this_year\":\n return {\n start: normalizeDate(new Date(now.getFullYear(), 0, 1)),\n end: normalizeDate(now)\n };\n\n case \"last_year\":\n return {\n start: normalizeDate(new Date(now.getFullYear() - 1, 0, 1)),\n end: normalizeDate(new Date(now.getFullYear() - 1, 11, 31))\n };\n\n case \"last_week\":\n const day = now.getDay();\n const diffToMonday = (day + 6) % 7;\n\n const lastMonday = new Date(now);\n lastMonday.setDate(now.getDate() - diffToMonday - 7);\n\n const lastSunday = new Date(lastMonday);\n lastSunday.setDate(lastMonday.getDate() + 6);\n\n return {\n start: normalizeDate(lastMonday),\n end: normalizeDate(lastSunday)\n };\n\n case \"this_week\":\n const currentDay = now.getDay();\n const diff = (currentDay + 6) % 7;\n\n const monday = new Date(now);\n monday.setDate(now.getDate() - diff);\n\n return {\n start: normalizeDate(monday),\n end: normalizeDate(now)\n };\n\n default:\n return { start: null, end: null };\n }\n}\n\nfunction formatDate(date) {\n if (!date) return null;\n return date.toISOString().split('T')[0];\n}\n\n// =============================\n// Retrieve Intent\n// =============================\n\nconst intent = $('JS | Extract Intent JSON').first().json;\nconst edit3 = $('SET | Assemble Resolved Intent').first().json;\n\n// IMPORTANT: overwrite values if edited\nintent.person = edit3.person || intent.person;\nintent.category = edit3.category || intent.category;\n\n// =============================\n// Load & Normalize Data\n// =============================\n\nconst expenseItems = $('GS | Load Expenses').all().map(i => ({\n date: i.json.date,\n amount: parseFloat(i.json.amount) || 0,\n category: i.json.category?.toLowerCase() || null,\n description: i.json.description,\n common_expense: i.json.common_expense,\n person: i.json.Person?.toLowerCase() || null\n}));\n\n// =============================\n// Date Handling\n// =============================\n\nlet startDate = null;\nlet endDate = null;\n\nif (intent.time_reference) {\n const range = getDateRange(intent.time_reference);\n startDate = range.start;\n endDate = range.end;\n}\n\nif (intent.explicit_start_date) {\n startDate = normalizeDate(intent.explicit_start_date);\n}\n\nif (intent.explicit_end_date) {\n endDate = normalizeDate(intent.explicit_end_date);\n}\n\nconst formattedStart = formatDate(startDate);\nconst formattedEnd = formatDate(endDate);\n\n// =============================\n// Other Filters\n// =============================\n\nconst filterPerson = intent.person ? intent.person.toLowerCase() : null;\n\nconst filterCategories = (() => {\n const c = intent.category;\n if (!c) return [];\n if (Array.isArray(c)) {\n return c.map(x => String(x).trim().toLowerCase()).filter(Boolean);\n }\n return [String(c).trim().toLowerCase()].filter(Boolean);\n})();\n\nconst filterCommon = intent.common_only;\nconst groupBy = intent.group_by;\n\n// =============================\n// Apply Filters\n// =============================\n\nconst filtered = expenseItems.filter(row => {\n\n if (!row.date) return false;\n\n const rowDate = normalizeDate(row.date);\n if (!rowDate) return false;\n\n if (startDate && rowDate < startDate) return false;\n if (endDate && rowDate > endDate) return false;\n\n if (filterPerson && row.person !== filterPerson) return false;\n\n if (filterCategories.length > 0) {\n if (!row.category) return false;\n if (!filterCategories.includes(row.category)) return false;\n }\n\n if (filterCommon !== null && filterCommon !== undefined) {\n\n const isCommon =\n String(row.common_expense).toLowerCase() === 'true' ||\n row.common_expense === true ||\n row.common_expense === 1;\n\n if (filterCommon && !isCommon) return false;\n if (!filterCommon && isCommon) return false;\n }\n\n return true;\n\n});\n\n// =============================\n// Aggregations\n// =============================\n\nconst total = filtered.reduce((sum, row) => sum + row.amount, 0);\n\n// Category breakdown\nlet categoryBreakdown = null;\n\nif (groupBy === \"category\") {\n\n const breakdown = {};\n\n filtered.forEach(row => {\n\n const cat = row.category || \"uncategorized\";\n\n breakdown[cat] = (breakdown[cat] || 0) + row.amount;\n\n });\n\n categoryBreakdown = breakdown;\n}\n\n// Person breakdown\nlet personBreakdown = null;\n\nif (groupBy === \"person\") {\n\n const breakdown = {};\n\n filtered.forEach(row => {\n\n const person = row.person || \"unknown\";\n\n breakdown[person] = (breakdown[person] || 0) + row.amount;\n\n });\n\n personBreakdown = breakdown;\n}\n\n// =============================\n// Return Result\n// =============================\n\nreturn [{\n json: {\n intent,\n total: Math.round(total * 100) / 100,\n count: filtered.length,\n categoryBreakdown,\n personBreakdown,\n period: {\n start: formattedStart,\n end: formattedEnd\n },\n filters: {\n person: intent.person,\n category: intent.category,\n common_only: intent.common_only\n }\n }\n}];" }, "typeVersion": 2 }, { "id": "b24a37e4-d82f-43e6-b343-e91f7f838d19", "name": "JS | Format Response Message", "type": "n8n-nodes-base.code", "position": [ 11056, 1264 ], "parameters": { "jsCode": "const data = $input.first().json;\nconst filters = data.filters || {};\nconst period = data.period || {};\n\nlet lines = [];\n\n// =============================\n// Header\n// =============================\n\nlines.push('\ud83d\udcca *Expense Summary*');\nlines.push('');\n\n// =============================\n// Period Display (NEW)\n// =============================\n\nif (period.start || period.end) {\n const from = period.start || 'Beginning';\n const to = period.end || 'Today';\n lines.push(`\ud83d\udcc5 Period: ${from} \u2192 ${to}`);\n} else {\n lines.push(`\ud83d\udcc5 Period: All time`);\n}\n\n// =============================\n// Applied filters\n// =============================\n\nif (filters.person) lines.push(`\ud83d\udc64 Person: ${filters.person}`);\nif (filters.category) lines.push(`\ud83c\udff7\ufe0f Category: ${filters.category}`);\n\nif (filters.common_only !== null && filters.common_only !== undefined) {\n lines.push(`\ud83d\udd17 Common expenses only: ${filters.common_only ? 'Yes' : 'No'}`);\n}\n\nlines.push('');\nlines.push(`\ud83d\udcb0 *Total: \u20ac${data.total.toFixed(2)}*`);\nlines.push(`\ud83d\udcdd Transactions: ${data.count}`);\n\n// =============================\n// Category breakdown\n// =============================\n\nif (data.categoryBreakdown) {\n lines.push('');\n lines.push('*By Category:*');\n\n const sorted = Object.entries(data.categoryBreakdown)\n .sort((a, b) => b[1] - a[1]);\n\n sorted.forEach(([cat, amt]) => {\n const pct = data.total > 0\n ? ((amt / data.total) * 100).toFixed(1)\n : 0;\n\n lines.push(` \u2022 ${cat}: $${amt.toFixed(2)} (${pct}%)`);\n });\n}\n\n// =============================\n// Person breakdown\n// =============================\n\nif (data.personBreakdown) {\n lines.push('');\n lines.push('*By Person:*');\n\n const sorted = Object.entries(data.personBreakdown)\n .sort((a, b) => b[1] - a[1]);\n\n sorted.forEach(([person, amt]) => {\n const pct = data.total > 0\n ? ((amt / data.total) * 100).toFixed(1)\n : 0;\n\n lines.push(` \u2022 ${person}: $${amt.toFixed(2)} (${pct}%)`);\n });\n}\n\n// =============================\n// Comparison placeholder\n// =============================\n\nif (data.comparison) {\n lines.push('');\n lines.push(`\ud83d\udcc8 Comparison: [Feature coming soon]`);\n}\n\n// =============================\n// Empty state\n// =============================\n\nif (data.count === 0) {\n lines.push('');\n lines.push('_No expenses found for the given filters._');\n}\n\nreturn [{ json: { message: lines.join('\\n') } }];" }, "typeVersion": 2 }, { "id": "c7dc9c2c-8bc0-4b0a-a988-27a966b3f814", "name": "TG | Send Reply", "type": "n8n-nodes-base.telegram", "position": [ 11312, 1264 ], "parameters": { "text": "={{ $json.message }}", "chatId": "={{ $('MSG | Telegram Inbound').first().json.message.chat.id }}", "additionalFields": { "parse_mode": "Markdown", "appendAttribution": false, "reply_to_message_id": "={{ $('MSG | Telegram Inbound').first().json.message.message_id }}" } }, "credentials": { "telegramApi": { "name": "" } }, "typeVersion": 1.2 }, { "id": "4876799b-af52-4218-80aa-7603258ff084", "name": "MERGE | Combine Expenses + Categories", "type": "n8n-nodes-base.merge", "position": [ 10144, 1264 ], "parameters": { "mode": "chooseBranch" }, "typeVersion": 3.2 }, { "id": "db13a254-5fde-42e4-9e82-2060a04451f0", "name": "IF | User Authorized?", "type": "n8n-nodes-base.if", "position": [ 2528, 1072 ], "parameters": { "options": {}, "conditions": { "options": { "version": 2, "leftValue": "", "caseSensitive": true, "typeValidation": "strict" }, "combinator": "or", "conditions": [ { "id": "0f85f7d5-8b7e-4bcd-ae10-d7a1785192c4", "operator": { "type": "number", "operation": "equals" }, "leftValue": "={{ $json.message.chat.id }}", "rightValue": 1000000000 }, { "id": "2d18f8cf-2910-4e20-b2f3-16a9ef04e63e", "operator": { "type": "number", "operation": "equals" }, "leftValue": "={{ $json.message.chat.id }}", "rightValue": 1000000001 } ] } }, "typeVersion": 2.2 }, { "id": "7a76b7d6-f533-4785-b822-8d7ac464d883", "name": "SPLIT | Split Categories", "type": "n8n-nodes-base.splitOut", "position": [ 4016, 768 ], "parameters": { "options": {}, "fieldToSplitOut": "category" }, "typeVersion": 1 }, { "id": "8e963faf-3079-447d-bfd9-89eff8da207e", "name": "GS | Read Category Mapping", "type": "n8n-nodes-base.googleSheets", "position": [ 4016, 1072 ], "parameters": { "options": {}, "sheetName": { "__rl": true, "mode": "list", "value": "gid=0", "cachedResultUrl": "https://docs.google.com/spreadsheets/d/YOUR_SPREADSHEET_ID", "cachedResultName": "Sheet1" }, "documentId": { "__rl": true, "mode": "list", "value": "YOUR_SPREADSHEET_ID", "cachedResultUrl": "https://docs.google.com/spreadsheets/d/YOUR_SPREADSHEET_ID", "cachedResultName": "categories_mapping" } }, "credentials": { "googleSheetsOAuth2Api": { "name": "" } }, "typeVersion": 4.7 }, { "id": "e992cd5e-ba4e-4ab5-b885-14c55546d0a3", "name": "MERGE | Join Categories with Mapping", "type": "n8n-nodes-base.merge", "position": [ 4272, 768 ], "parameters": { "mode": "combineBySql", "query": "SELECT * FROM input1 LEFT JOIN input2 ON input1.category = input2.find", "options": {} }, "typeVersion": 3.2 }, { "id": "58c889ef-e918-43d4-a7a3-1a47bf890ffc", "name": "TG | Confirm Category Suggestion", "type": "n8n-nodes-base.telegram", "position": [ 6224, 368 ], "parameters": { "chatId": "={{ $('MSG | Telegram Inbound').item.json.message.chat.id }}", "message": "=I couldn\u2019t find the category \"{{ $('IF | Category Known?').item.json.category_new }}\".\n\nDid you mean \"{{ $json.content }}\"?\n\n\u2705 Yes \n\u274c No", "options": { "appendAttribution": false }, "operation": "sendAndWait", "approvalOptions": { "values": { "approvalType": "double", "approveLabel": "\u2705 Yes", "disapproveLabel": "\u274c No" } } }, "credentials": { "telegramApi": { "name": "" } }, "typeVersion": 1.2 }, { "id": "fd9c3d1c-223a-44d8-afbf-4773a11d09bd", "name": "GS | Read Allowed Categories", "type": "n8n-nodes-base.googleSheets", "position": [ 4464, 1072 ], "parameters": { "options": {}, "sheetName": { "__rl": true, "mode": "list", "value": "gid=0", "cachedResultUrl": "https://docs.google.com/spreadsheets/d/YOUR_SPREADSHEET_ID", "cachedResultName": "Sheet1" }, "documentId": { "__rl": true, "mode": "list", "value": "YOUR_SPREADSHEET_ID", "cachedResultUrl": "https://docs.google.com/spreadsheets/d/YOUR_SPREADSHEET_ID", "cachedResultName": "expense_categories" } }, "credentials": { "googleSheetsOAuth2Api": { "name": "" } }, "typeVersion": 4.7 }, { "id": "2f1f1de8-cae3-4792-920c-ba490986019e", "name": "SET | Normalize Category", "type": "n8n-nodes-base.set", "position": [ 4464, 768 ], "parameters": { "options": {}, "assignments": { "assignments": [ { "id": "8f458013-1006-4634-a1c8-aedff820a48f", "name": "category_new", "type": "string", "value": "={{ $json.replace ?? $json.category }}" } ] } }, "typeVersion": 3.4 }, { "id": "7a819214-50e5-460a-bd4d-a546eb68dc20", "name": "MERGE | Check Category against Allowed", "type": "n8n-nodes-base.merge", "position": [ 4720, 768 ], "parameters": { "mode": "combineBySql", "query": "SELECT * FROM input1\n \nLEFT JOIN input2 ON input1.category_new = input2.category", "options": {} }, "typeVersion": 3.2, "alwaysOutputData": false }, { "id": "ed0de8c5-8c55-49f2-96a0-973b49b95d6a", "name": "IF | Category Known?", "type": "n8n-nodes-base.if", "position": [ 4960, 768 ], "parameters": { "options": {}, "conditions": { "options": { "version": 3, "leftValue": "", "caseSensitive": true, "typeValidation": "strict" }, "combinator": "and", "conditions": [ { "id": "e329be08-8c46-4898-9480-5946735ad54d", "operator": { "type": "string", "operation": "notExists", "singleValue": true }, "leftValue": "={{ $json.description }}", "rightValue": "" } ] } }, "typeVersion": 2.3, "alwaysOutputData": false }, { "id": "9ecd305a-5e42-43f1-91fa-21fbb0d89e53", "name": "LLM | Classify Category", "type": "@n8n/n8n-nodes-langchain.openAi", "position": [ 5712, 368 ], "parameters": { "modelId": { "__rl": true, "mode": "list", "value": "gpt-4.1-nano", "cachedResultName": "GPT-4.1-NANO" }, "options": {}, "messages": { "values": [ { "role": "system", "content": "=You are a strict category classification system.\n\nYour task:\nMap the given input category to exactly ONE category from the allowed category list.\n\nRules:\n- You MUST select one category from the allowed list.\n- Do NOT invent new categories.\n- Do NOT return explanations.\n- Do NOT return multiple categories.\n- Output ONLY the exact category name from the allowed list.\n- Matching must be case-insensitive.\n- Choose the closest semantic match if there is no exact match.\n\nInput category:\n{{ $('IF | Category Known?').item.json.category_new }}\n\nAllowed categories:\n{{ $json.category }}" } ] } }, "credentials": { "openAiApi": { "name": "" } }, "typeVersion": 1.4 }, { "id": "73085ca0-da81-4f8f-a232-5680c7d88b88", "name": "MERGE | Combine Categories with List", "type": "n8n-nodes-base.merge", "position": [ 5216, 672 ], "parameters": { "mode": "combine", "options": {}, "combineBy": "combineAll" }, "typeVersion": 3.2 }, { "id": "19fe45dc-435b-4e70-9b7a-cd7c0bedd41e", "name": "SET | Extract LLM Category", "type": "n8n-nodes-base.set", "position": [ 5968, 368 ], "parameters": { "mode": "raw", "options": {}, "jsonOutput": "={{ $json.message }}" }, "typeVersion": 3.4 }, { "id": "20b22d0e-35bf-476f-9aba-e2abb14f0377", "name": "IF | Category Suggestion Accepted?", "type": "n8n-nodes-base.if", "position": [ 6464, 368 ], "parameters": { "options": {}, "conditions": { "options": { "version": 3, "leftValue": "", "caseSensitive": true, "typeValidation": "strict" }, "combinator": "and", "conditions": [ { "id": "4fb343c7-9eb2-4491-b017-3637fb543564", "operator": { "type": "boolean", "operation": "equals" }, "leftValue": "={{ $json.data.approved }}", "rightValue": true } ] } }, "typeVersion": 2.3 }, { "id": "1d692432-6601-47f1-9bdd-cafd50aeef5c", "name": "GS | Save Category Mapping", "type": "n8n-nodes-base.googleSheets", "position": [ 7968, 368 ], "parameters": { "columns": { "value": { "find": "={{ $json.old }}", "replace": "={{ $json.new }}" }, "schema": [ { "id": "find", "type": "string", "display": true, "required": false, "displayName": "find", "defaultMatch": false, "canBeUsedToMatch": true }, { "id": "replace", "type": "string", "display": true, "required": false, "displayName": "replace", "defaultMatch": false, "canBeUsedToMatch": true } ], "mappingMode": "defineBelow", "matchingColumns": [], "attemptToConvertTypes": false, "convertFieldsToString": false }, "options": {}, "operation": "append", "sheetName": { "__rl": true, "mode": "list", "value": "gid=0", "cachedResultUrl": "https://docs.google.com/spreadsheets/d/YOUR_SPREADSHEET_ID", "cachedResultName": "Sheet1" }, "documentId": { "__rl": true, "mode": "list", "value": "YOUR_SPREADSHEET_ID", "cachedResultUrl": "https://docs.google.com/spreadsheets/d/YOUR_SPREADSHEET_ID", "cachedResultName": "categories_mapping" } }, "credentials": { "googleSheetsOAuth2Api": { "name": "" } }, "typeVersion": 4.7 }, { "id": "750a9069-a57b-4b46-8725-42367602aad2", "name": "MERGE | Loop Entry (Category)", "type": "n8n-nodes-base.merge", "position": [ 5712, 768 ], "parameters": {}, "typeVersion": 3.2 }, { "id": "20498665-76a1-45b8-978f-e92e0fc19ed1", "name": "SET | Set Resolved Category", "type": "n8n-nodes-base.set", "position": [ 5968, 768 ], "parameters": { "options": {}, "assignments": { "assignments": [ { "id": "47a00b45-f580-4568-9913-b2359e1c1d8c", "name": "category_new", "type": "string", "value": "={{ $json.replace ?? $('IF | Category Known?').item.json.category_new}}" } ] } }, "typeVersion": 3.4 }, { "id": "e6824c4c-a839-40d0-b334-40f19aad19d0", "name": "SET | Assemble Resolved Intent", "type": "n8n-nodes-base.set", "position": [ 9472, 1264 ], "parameters": { "options": {}, "assignments": { "assignments": [ { "id": "dc8b496a-4e1a-4ba3-9180-61b77a616b99", "name": "intent", "type": "string", "value": "={{ $('JS | Extract Intent JSON').item.json.intent }}" }, { "id": "1e76280f-f49b-4225-ba90-fccc0ba3d9f2", "name": "time_reference", "type": "string", "value": "={{ $('JS | Extract Intent JSON').item.json.time_reference }}" }, { "id": "1b10b1a5-e35d-4e13-8a28-945271eb4cd6", "name": "explicit_start_date", "type": "string", "value": "={{ $('JS | Extract Intent JSON').item.json.explicit_start_date }}" }, { "id": "51c694b2-c10b-485d-8e1e-dedd797308a9", "name": "explicit_end_date", "type": "string", "value": "={{ $('JS | Extract Intent JSON').item.json.explicit_end_date }}" }, { "id": "56a2b084-547f-412f-b534-4c2a97ec52c8", "name": "person", "type": "string", "value": "={{ $json.person_new }}" }, { "id": "aa1b02a7-6908-4ba0-b8c2-a74b15516b40", "name": "category", "type": "array", "value": "={{ $json.category_new }}" }, { "id": "4bd911ba-a8e3-423c-aa2e-eab59e16a33a", "name": "common_only", "type": "string", "value": "={{ $('JS | Extract Intent JSON').item.json.common_only }}" }, { "id": "720e91ea-fc42-48b4-af78-1853736d8720", "name": "comparison", "type": "string", "value": "={{ $('JS | Extract Intent JSON').item.json.comparison }}" }, { "id": "27656405-d197-4f6e-b1d8-f90a72351f20", "name": "group_by", "type": "string", "value": "={{ $('JS | Extract Intent JSON').item.json.group_by }}" } ] } }, "typeVersion": 3.4 }, { "id": "cc31446d-2ffb-4411-afd7-64e69182b448", "name": "AGG | Aggregate Category List", "type": "n8n-nodes-base.aggregate", "position": [ 4720, 1072 ], "parameters": { "options": {}, "fieldsToAggregate": { "fieldToAggregate": [ { "fieldToAggregate": "category" } ] } }, "typeVersion": 1 }, { "id": "6f2ace60-279f-40ce-8459-53a13a603748", "name": "LOOP | Iterate Categories", "type": "n8n-nodes-base.splitInBatches", "position": [ 5472, 768 ], "parameters": { "options": {} }, "typeVersion": 3 }, { "id": "165edc36-f6bd-4231-a1e4-bd7c7cd4ecea", "name": "SET | Set New+Old Category (LLM)", "type": "n8n-nodes-base.set", "position": [ 7216, 368 ], "parameters": { "options": {}, "assignments": { "assignments": [ { "id": "5f52894c-c285-46f0-822d-44665a58555f", "name": "new", "type": "string", "value": "={{ $('SET | Extract LLM Category').item.json.content }}" }, { "id": "e02f6b7f-c27d-4620-84e4-346dcb5f26f0", "name": "old", "type": "string", "value": "={{ $('LOOP | Iterate Categories').item.json.category_new }}" } ] } }, "typeVersion": 3.4, "alwaysOutputData": true }, { "id": "8294d338-81df-4799-8d83-335699eb38d9", "name": "JS | Build Category Inline Buttons", "type": "n8n-nodes-base.code", "position": [ 6720, 176 ], "parameters": { "jsCode": "const items = $('AGG | Aggregate Category List').first().json.category;\nif (!Array.isArray(items)) {\n throw new Error('Expected category to be an array');\n}\n\nconst buttons = items.map(item => ([{\n text: String(item),\n callback_data: String(item)\n}]));\n\nreturn [{\n json: {\n inline_keyboard: buttons,\n resumeUrl: $execution.resumeUrl\n }\n}];" }, "typeVersion": 2 }, { "id": "e7a92e17-a097-4e67-8ba0-aa2073aeefb1", "name": "WAIT | Wait for Category Selection", "type": "n8n-nodes-base.wait", "position": [ 7216, 176 ], "parameters": { "resume": "webhook", "options": {}, "httpMethod": "POST" }, "typeVersion": 1.1 }, { "id": "d5c9afc3-2245-4a93-9c66-38edc4c0f4a2", "name": "HTTP | Send Category Selection Message", "type": "n8n-nodes-base.httpRequest", "position": [ 6960, 176 ], "parameters": { "url": "https://api.telegram.org/bot{{YOUR_BOT_TOKEN}}/sendMessage", "method": "POST", "options": {}, "jsonBody": "={\n \"chat_id\": \"{{ $('MSG | Telegram Inbound').item.json.message.chat.id }}\",\n \"text\": \"Bitte w\u00e4hle eine Kategorie:\u200b\",\n \"parse_mode\": \"HTML\",\n \"reply_markup\": {{ JSON.stringify({ \"inline_keyboard\": $json.inline_keyboard }) }}\n}", "sendBody": true, "specifyBody": "json" }, "typeVersion": 4.3 }, { "id": "33dba25d-9b17-4da1-8463-92dbad2c0ff3", "name": "SET | Read Callback Body (Cat.)", "type": "n8n-nodes-base.set", "position": [ 7472, 176 ], "parameters": { "mode": "raw", "options": {}, "jsonOutput": "={{ $json.body}}" }, "typeVersion": 3.4 }, { "id": "e11d27e9-66ea-4cde-a108-3c38b5fea75b", "name": "SET | Set New+Old Category (Selection)", "type": "n8n-nodes-base.set", "position": [ 7472, 368 ], "parameters": { "options": {}, "assignments": { "assignments": [ { "id": "e34ab1c9-3985-432b-9637-6a695fbf5931", "name": "new", "type": "string", "value": "={{ $json.category }}" }, { "id": "b3f6ce98-c6bc-4fea-a581-d1abcd087df4", "name": "old", "type": "string", "value": "={{ $('LOOP | Iterate Categories').item.json.category_new }}" } ] } }, "typeVersion": 3.4, "alwaysOutputData": true }, { "id": "e765fbf0-81a6-4117-9fbf-8c5c4d0f6c90", "name": "MERGE | Combine Category Mapping Entries", "type": "n8n-nodes-base.merge", "position": [ 7712, 368 ], "parameters": {}, "typeVersion": 3.2 }, { "id": "dada58e6-2abc-4cec-84c2-4bd92994e796", "name": "HTTP | Forward Category Selection", "type": "n8n-nodes-base.httpRequest", "position": [ 3312, 496 ], "parameters": { "url": "={{ $json.resumeUrl }}", "method": "POST", "options": {}, "jsonBody": "={\n \"category\": \"{{ $json.category }}\"\n}", "sendBody": true, "specifyBody": "json" }, "typeVersion": 4.3 }, { "id": "76197f7e-8f0f-46d5-84b4-c9c6bb735668", "name": "TG | Confirm Category Selection", "type": "n8n-nodes-base.telegram", "position": [ 3504, 496 ], "parameters": { "text": "=Du hast \" {{ $('MSG | Telegram Inbound').item.json.callback_query.data }}\" ausgew\u00e4hlt\"", "chatId": "={{ $('MSG | Telegram Inbound').item.json.callback_query.message.chat.id }}", "additionalFields": { "appendAttribution": false } }, "credentials": { "telegramApi": { "name": "" } }, "typeVersion": 1.2 }, { "id": "883bf86d-3fac-461a-a9b7-dcc0018eb20f", "name": "JS | Read Category Callback", "type": "n8n-nodes-base.code", "position": [ 3104, 496 ], "parameters": { "jsCode": "const callbackData = $json.callback_query.data;\nconst text = $json.callback_query.message.text;\nconst entities = $json.callback_query.message.entities;\n\n// URL aus den message entities auslesen\nconst urlEntity = entities.find(e => e.type === 'text_link');\nconst resumeUrl = urlEntity.url;\n\nreturn [{\n json: {\n category: callbackData,\n resumeUrl: resumeUrl\n }\n}];" }, "typeVersion": 2 }, { "id": "0e0c0fc1-c577-4e31-a62e-2bef9ef22815", "name": "IF | Message or Callback?", "type": "n8n-nodes-base.if", "position": [ 2352, 1088 ], "parameters": { "options": {}, "conditions": { "options": { "version": 3, "leftValue": "", "caseSensitive": true, "typeValidation": "strict" }, "combinator": "and", "conditions": [ { "id": "69655a65-f86e-4666-8f8f-d9cc2fda74b2", "operator": { "type": "object", "operation": "exists", "singleValue": true }, "leftValue": "={{ $json.message }}", "rightValue": "=" } ] } }, "typeVersion": 2.3 }, { "id": "4233a9ba-e3a9-49c7-8f07-1d36fa3fec81", "name": "IF | Category Present?", "type": "n8n-nodes-base.if", "position": [ 3824, 976 ], "parameters": { "options": {}, "conditions": { "options": { "version": 3, "leftValue": "", "caseSensitive": true, "typeValidation": "strict" }, "combinator": "and", "conditions": [ { "id": "b875ecb6-7b43-4083-8591-bec82fb6a22b", "operator": { "type": "string", "operation": "notExists", "singleValue": true }, "leftValue": "={{ $json.category[0] }}", "rightValue": "" } ] } }, "typeVersion": 2.3 }, { "id": "449ced93-3de6-4bf4-8e71-cb0acb48a43e", "name": "MERGE | Combine Category + Person", "type": "n8n-nodes-base.merge", "position": [ 8832, 1248 ], "parameters": { "numberInputs": 3 }, "typeVersion": 3.2 }, { "id": "f98b58b7-483e-495b-95c1-aa5a9fa457c4", "name": "AGG | Aggregate Resolved Categories", "type": "n8n-nodes-base.aggregate", "position": [ 6224, 768 ], "parameters": { "options": {}, "fieldsToAggregate": { "fieldToAggregate": [ { "fieldToAggregate": "category_new" } ] } }, "typeVersion": 1 }, { "id": "50f4587d-754a-4bb3-888b-b6e084ee6237", "name": "IF | Person Present?", "type": "n8n-nodes-base.if", "position": [ 3760, 2096 ], "parameters": { "options": {}, "conditions": { "options": { "version": 3, "leftValue": "", "caseSensitive": true, "typeValidation": "strict" }, "combinator": "and", "conditions": [ { "id": "b875ecb6-7b43-4083-8591-bec82fb6a22b", "operator": { "type": "string", "operation": "notExists", "singleValue": true }, "leftValue": "={{ $json.person }}", "rightValue": "" } ] } }, "typeVersion": 2.3 }, { "id": "85f203b8-bdf7-4346-aceb-30a8388dc4cf", "name": "SPLIT | Split Persons", "type": "n8n-nodes-base.splitOut", "position": [ 3968, 2064 ], "parameters": { "options": {}, "fieldToSplitOut": "person" }, "typeVersion": 1 }, { "id": "7657a339-c559-4327-bc25-5fda098783f3", "name": "MERGE | Join Persons with Mapping", "type": "n8n-nodes-base.merge", "position": [ 4224, 2064 ], "parameters": { "mode": "combineBySql", "query": "SELECT * FROM input1 LEFT JOIN input2 ON input1.category = input2.find", "options": {} }, "typeVersion": 3.2 }, { "id": "286c2311-d948-4dd0-92bb-81231cb895ea", "name": "SET | Normalize Person", "type": "n8n-nodes-base.set", "position": [ 4416, 2064 ], "parameters": { "options": {}, "assignments": { "assignments": [ { "id": "8f458013-1006-4634-a1c8-aedff820a48f", "name": "person_new", "type": "string", "value": "={{ $json.replace ?? $json.person }}" } ] } }, "typeVersion": 3.4 }, { "id": "fa32c88e-fb58-483a-8159-bc68edbeebcd", "name": "MERGE | Check Person against Allowed", "type": "n8n-nodes-base.merge", "position": [ 4704, 2048 ], "parameters": { "mode": "combineBySql", "query": "SELECT * FROM input1\n \nLEFT JOIN input2 ON input1.person_new = input2.person", "options": {} }, "typeVersion": 3.2, "alwaysOutputData": false }, { "id": "8ed005b0-aa2a-4052-868a-424d6df8d83e", "name": "IF | Person Known?", "type": "n8n-nodes-base.if", "position": [ 4944, 2048 ], "parameters": { "options": {}, "conditions": { "options": { "version": 3, "leftValue": "", "caseSensitive": true, "typeValidation": "strict" }, "combinator": "and", "conditions": [ { "id": "e329be08-8c46-4898-9480-5946735ad54d", "operator": { "type": "string", "operation": "notExists", "singleValue": true }, "leftValue": "={{ $json.description }}", "rightValue": "" } ] } }, "typeVersion": 2.3, "alwaysOutputData": false }, { "id": "2f3cf76a-d852-4479-a91a-4d93bdcb8e28", "name": "MERGE | Combine Persons with List", "type": "n8n-nodes-base.merge", "position": [ 5200, 2048 ], "parameters": { "mode": "combine", "options": {}, "combineBy": "combineAll" }, "typeVersion": 3.2 }, { "id": "9db7a679-849e-46a4-8c12-9c9296e00d4c", "name": "AGG | Aggregate Person List", "type": "n8n-nodes-base.aggregate", "position": [ 4912, 2352 ], "parameters": { "options": {}, "fieldsToAggregate": { "fieldToAggregate": [ { "fieldToAggregate": "person" } ] } }, "typeVersion": 1 }, { "id": "82d59267-2594-498f-863b-f35116802032", "name": "GS | Read Person Mapping", "type": "n8n-nodes-base.googleSheets", "position": [ 3968, 2368 ], "parameters": { "options": {}, "sheetName": { "__rl": true, "mode": "list", "value": "gid=0", "cachedResultUrl": "https://docs.google.com/spreadsheets/d/YOUR_SPREADSHEET_ID", "cachedResultName": "Sheet1" }, "documentId": { "__rl": true, "mode": "list", "value": "YOUR_SPREADSHEET_ID", "cachedResultUrl": "https://docs.google.com/spreadsheets/d/YOUR_SPREADSHEET_ID", "cachedResultName": "person_mapping" } }, "credentials": { "googleSheetsOAuth2Api": { "name": "" } }, "typeVersion": 4.7 }, { "id": "16747ce4-9d18-438d-a8df-df00508908c1", "name": "GS | Read Allowed Persons", "type": "n8n-nodes-base.googleSheets", "position": [ 4416, 2368 ], "parameters": { "options": {}, "sheetName": { "__rl": true, "mode": "list", "value": 1157427829, "cachedResultUrl": "https://docs.google.com/spreadsheets/d/YOUR_SPREADSHEET_ID", "cachedResultName": "list_persons" }, "documentId": { "__rl": true, "mode": "list", "value": "YOUR_SPREADSHEET_ID", "cachedResultUrl": "https://docs.google.com/spreadsheets/d/YOUR_SPREADSHEET_ID", "cachedResultName": "person_mapping" } }, "credentials": { "googleSheetsOAuth2Api": { "name": "" } }, "typeVersion": 4.7 }, { "id": "fa8b6b35-8747-40b8-9639-5131f279c0cb", "name": "TG | Confirm Person Suggestion", "type": "n8n-nodes-base.telegram", "position": [ 6208, 2448 ], "parameters": { "chatId": "={{ $('MSG | Telegram Inbound').item.json.message.chat.id }}", "message": "=I couldn\u2019t find the person \"{{ $('IF | Person Known?').item.json.person_new }}\".\n\nDid you mean \"{{ $json.content }}\"?\n\n\u2705 Yes \n\u274c No", "options": { "appendAttribution": false }, "operation": "sendAndWait", "approvalOptions": { "values": { "approvalType": "double", "approveLabel": "\u2705 Yes", "disapproveLabel": "\u274c No" } } }, "credentials": { "telegramApi": { "name": "" } }, "typeVersion": 1.2 }, { "id": "fa9fc8ee-c15e-4f4f-9af2-8a5dd626d788", "name": "LLM | Classify Person", "type": "@n8n/n8n-nodes-langchain.openAi", "position": [ 5696, 2448 ], "parameters": { "modelId": { "__rl": true, "mode": "list", "value": "gpt-4.1-nano", "cachedResultName": "GPT-4.1-NANO" }, "options": {}, "messages": { "values": [ { "role": "system", "content": "=You are a strict person classification system.\n\nYour task:\nMap the given input name to exactly ONE person from the allowed person list.\n\nRules:\n\n* You MUST select one person from the allowed list.\n* Do NOT invent new persons.\n* Do NOT return explanations.\n* Do NOT return multiple persons.\n* Output ONLY the exact person name from the allowed list.\n* Matching must be case-insensitive.\n* Choose the closest semantic match if there is no exact match.\n\nInput name:\n{{ $('IF | Person Known?').item.json.person_new }}\n\nAllowed persons:\n{{ $json.person }}\n" } ] } }, "credentials": { "openAiApi": { "name": "" } }, "typeVersion": 1.4 }, { "id": "ef710337-260f-495c-97aa-7ccb5420b7dc", "name": "SET | Extract LLM Person", "type": "n8n-nodes-base.set", "position": [ 5952, 2448 ], "parameters": { "mode": "raw", "options": {}, "jsonOutput": "={{ $json.message }}" }, "typeVersion": 3.4 }, { "id": "fd5713f9-9501-420f-9d12-bb8b84fda620", "name": "IF | Person Suggestion Accepted?", "type": "n8n-nodes-base.if", "position": [ 6448, 2448 ], "parameters": { "options": {}, "conditions": { "options": { "version": 3, "leftValue": "", "caseSensitive": true, "typeValidation": "strict" }, "combinator": "and", "conditions": [ { "id": "4fb343c7-9eb2-4491-b017-3637fb543564", "operator": { "type": "boolean", "operation": "equals" }, "leftValue": "={{ $json.data.approved }}", "rightValue": true } ] } }, "typeVersion": 2.3 }, { "id": "262227e8-5564-4caf-a833-edc83bd36894", "name": "GS | Save Person Mapping", "type": "n8n-nodes-base.googleSheets", "position": [ 8176, 1904 ], "parameters": { "columns": { "value": { "find": "={{ $json.old }}", "replace": "={{ $json.new }}" }, "schema": [ { "id": "find", "type": "string", "display": true, "required": false, "displayName": "find", "defaultMatch": false, "canBeUsedToMatch": true }, { "id": "replace", "type": "string", "display": true, "required": false, "displayName": "replace", "defaultMatch": false, "canBeUsedToMatch": true } ], "mappingMode": "defineBelow", "matchingColumns": [], "attemptToConvertTypes": false, "convertFieldsToString": false }, "options": {}, "operation": "append", "sheetName": { "__rl": true, "mode": "list", "value": "gid=0", "cachedResultUrl": "https://docs.google.com/spreadsheets/d/YOUR_SPREADSHEET_ID", "cachedResultName": "Sheet1" }, "documentId": { "__rl": true, "mode": "list", "value": "YOUR_SPREADSHEET_ID", "cachedResultUrl": "https://docs.google.com/spreadsheets/d/YOUR_SPREADSHEET_ID", "cachedResultName": "person_mapping" } }, "credentials": { "googleSheetsOAuth2Api": { "name": "" } }, "typeVersion": 4.7 }, { "id": "e6850bcd-ec85-4a14-ae06-73878206a278", "name": "LOOP | Iterate Persons", "type": "n8n-nodes-base.splitInBatches", "position": [ 5456, 2048 ], "parameters": { "options": {} }, "typeVersion": 3 }, { "id": "741e158c-14ed-4ff1-aba3-27b0f648ab20", "name": "SET | Set New+Old Person (LLM)", "type": "n8n-nodes-base.set", "position": [ 7664, 2304 ], "parameters": { "options": {}, "assignments": { "assignments": [ { "id": "5f52894c-c285-46f0-822d-44665a58555f", "name": "new", "type": "string", "value": "={{ $('SET | Extract LLM Person').item.json.content }}" }, { "id": "e02f6b7f-c27d-4620-84e4-346dcb5f26f0", "name": "old", "type": "string", "value": "={{ $('LOOP | Iterate Persons').item.json.person_new }}" } ] } }, "typeVersion": 3.4, "alwaysOutputData": true }, { "id": "41280f1e-9534-4947-908a-41a7caef3397", "name": "JS | Build Person Inline Buttons", "type": "n8n-nodes-base.code", "position": [ 6672, 1712 ], "parameters": { "jsCode": "const items = $('AGG | Aggregate Person List').first().json.person;\n\nif (!Array.isArray(items)) {\n throw new Error('Expected person to be an array');\n}\n\nconst buttons = items.map(item => ([{\n text: String(item),\n callback_data: String(item)\n}]));\n\nreturn [{\n json: {\n inline_keyboard: buttons,\n resumeUrl: $execution.resumeUrl\n }\n}];" }, "typeVersion": 2 }, { "id": "5f9c0563-ee99-4147-9081-f499af2bb507", "name": "WAIT | Wait for Person Selection", "type": "n8n-nodes-base.wait", "position": [ 7168, 1712 ], "parameters": { "resume": "webhook", "options": {}, "httpMethod": "POST" }, "typeVersion": 1.1 }, { "id": "5b86154e-57b0-4b2d-a7d7-1e11cc878b67", "name": "HTTP | Send Person Selection Message", "type": "n8n-nodes-base.httpRequest", "position": [ 6912, 1712 ], "parameters": { "url": "https://api.telegram.org/bot{{YOUR_BOT_TOKEN}}/sendMessage", "method": "POST", "options": {}, "jsonBody": "={\n \"chat_id\": \"{{ $('MSG | Telegram Inbound').item.json.message.chat.id }}\",\n \"text\": \"Bitte w\u00e4hle eine Person:\u200b\",\n \"parse_mode\": \"HTML\",\n \"reply_markup\": {{ JSON.stringify({ \"inline_keyboard\": $json.inline_keyboard }) }}\n}", "sendBody": true, "specifyBody": "json" }, "typeVersion": 4.3 }, { "id": "b879f1c5-ffb3-45b6-b7f5-1b772b431643", "name": "SET | Read Callback Body (Person)", "type": "n8n-nodes-base.set", "position": [ 7424, 1712 ], "parameters": { "mode": "raw", "options": {}, "jsonOutput": "={{ $json.body}}" }, "typeVersion": 3.4 }, { "id": "1a9aa8de-b20f-4cac-9fde-9d889caee19e", "name": "SET | Set New+Old Person (Selection)", "type": "n8n-nodes-base.set", "position": [ 7664, 1712 ], "parameters": { "options": {}, "assignments": { "assignments": [ { "id": "e34ab1c9-3985-432b-9637-6a695fbf5931", "name": "new", "type": "string", "value": "={{ $json.person }}" }, { "id": "b3f6ce98-c6bc-4fea-a581-d1abcd087df4", "name": "old", "type": "string", "value": "={{ $('LOOP | Iterate Persons').item.json.person_new }}" } ] } }, "typeVersion": 3.4, "alwaysOutputData": true }, { "id": "205a6d3a-9108-4b8b-9698-5ba8db844f88", "name": "MERGE | Combine Person Mapping Entries", "type": "n8n-nodes-base.merge", "position": [ 7920, 1904 ], "parameters": {}, "typeVersion": 3.2 }, { "id": "eaee6930-3da6-40a9-ba0f-84ab8543b451", "name": "MERGE | Loop Entry (Person)", "type": "n8n-nodes-base.merge", "position": [ 5696, 2256 ], "parameters": {}, "typeVersion": 3.2 }, { "id": "e2241bb1-eae6-4311-909c-a67bb2bd52fe", "name": "SET | Set Resolved Person", "type": "n8n-nodes-base.set", "position": [ 6208, 2256 ], "parameters": { "options": {}, "assignments": { "assignments": [ { "id": "47a00b45-f580-4568-9913-b2359e1c1d8c", "name": "person_new", "type": "string", "value": "={{ $json.replace ?? $('IF | Person Known?').item.json.person_new}}" } ] } }, "typeVersion": 3.4 }, { "id": "9a37179d-2d3c-48e3-9ce1-1624b626b6fd", "name": "MERGE | Combine Person Mapping", "type": "n8n-nodes-base.merge", "position": [ 6448, 2048 ], "parameters": {}, "typeVersion": 3.2 }, { "id": "183d2e80-d18b-403e-a3f0-86d41fd0c0b1", "name": "IF | Category or Person Callback?", "type": "n8n-nodes-base.if", "position": [ 2816, 576 ], "parameters": { "options": {}, "conditions": { "options": { "version": 3, "leftValue": "", "caseSensitive": true, "typeValidation": "strict" }, "combinator": "and", "conditions": [ { "id": "a6650d78-63a0-4bc4-ac46-e076eba14def", "operator": { "type": "string", "operation": "contains" }, "leftValue": "{{json.message.text}}", "rightValue": "Kategorie" } ] } }, "typeVersion": 2.3 }, { "id": "0ac9e7b3-a794-4a42-b48b-1d42ba317bbc", "name": "HTTP | Forward Person Selection", "type": "n8n-nodes-base.httpRequest", "position": [ 3312, 688 ], "parameters": { "url": "={{ $json.resumeUrl }}", "method": "POST", "options": {}, "jsonBody": "={\n \"person\": \"{{ $json.person }}\"\n}", "sendBody": true, "specifyBody": "json" }, "typeVersion": 4.3 }, { "id": "36db9da8-b32c-4b48-b7d4-6871eed49cb0", "name": "TG | Confirm Person Selection", "type": "n8n-nodes-base.telegram", "position": [ 3504, 688 ], "parameters": { "text": "=Du hast \" {{ $('MSG | Telegram Inbound').item.json.callback_query.data }}\" ausgew\u00e4hlt\"", "chatId": "={{ $('MSG | Telegram Inbound').item.json.callback_query.message.chat.id }}", "additionalFields": { "appendAttribution": false } }, "credentials": { "telegramApi": { "name": "" } }, "typeVersion": 1.2 }, { "id": "17fceb64-ac93-42b7-af34-fc5f1d0cb3d2", "name": "JS | Read Person Callback", "type": "n8n-nodes-base.code", "position": [ 3104, 688 ], "parameters": { "jsCode": "const callbackData = $json.callback_query.data;\nconst text = $json.callback_query.message.text;\nconst entities = $json.callback_query.message.entities;\n\n// URL aus den message entities auslesen\nconst urlEntity = entities.find(e => e.type === 'text_link');\nconst resumeUrl = urlEntity.url;\n\nreturn [{\n json: {\n person: callbackData,\n resumeUrl: resumeUrl\n }\n}];" }, "typeVersion": 2 }, { "id": "a742d127-062b-4f7d-8c4d-0b089439508e", "name": "JS | Merge Intent Fields", "type": "n8n-nodes-base.code", "position": [ 9168, 1264 ], "parameters": { "jsCode": "const items = $input.all();\n\nconst result = {};\n\nfor (const item of items) {\n for (const [key, value] of Object.entries(item.json)) {\n if (\n value !== undefined &&\n value !== null &&\n value !== \"\" &&\n value !== \"undefined\"\n ) {\n result[key] = value;\n }\n }\n}\n\nreturn [{\n json: result,\n pairedItem: { item: 0 }\n}];" }, "typeVersion": 2 }, { "id": "f4353b36-c5f3-4e13-baf8-a257e81fbf43", "name": "StickyNote_Layer1", "type": "n8n-nodes-base.stickyNote", "position": [ 1968, 0 ], "parameters": { "color": 5, "width": 1740, "height": 1288, "content": "## \ud83d\udfe2 LAYER 1 \u2014 INPUT\nReceives incoming Telegram messages & callbacks.\nAuthorizes users and routes by message type.\n\n---\n\n\u2699\ufe0f **ACTION REQUIRED**\n\n**1. Connect Telegram Credential**\nOpen the `MSG | Telegram Inbound` node and connect your Telegram Bot credential.\nDo the same for all other `TG |` nodes in this layer.\n\u2192 *Nodes: MSG | Telegram Inbound, TG | Confirm Category Selection, TG | Confirm Person Selection*\n\n**2. Add authorized Chat IDs**\nOpen `IF | User Authorized?` and replace the placeholder values (`1000000000`, `1000000001`) with the real Telegram Chat IDs of all users who should have access.\nEach person can find their Chat ID by messaging @userinfobot on Telegram.\n" }, "typeVersion": 1 }, { "id": "b6684ba8-b5ab-41fe-8c1c-9ec20255983d", "name": "StickyNote_Layer2", "type": "n8n-nodes-base.stickyNote", "position": [ 1968, 1312 ], "parameters": { "width": 1730, "height": 1284, "content": "## \ud83d\udd35 LAYER 2 \u2014 INTENT PARSING\n\n---\n\n\u2699\ufe0f **ACTION REQUIRED**\n\n**1. Connect OpenAI Credential**\nOpen the `LLM | Parse Intent` node and connect your OpenAI API credential." }, "typeVersion": 1 }, { "id": "c6ddda36-1b63-428b-932d-2d9c857aa12d", "name": "StickyNote_Layer3", "type": "n8n-nodes-base.stickyNote", "position": [ 3712, 0 ], "parameters": { "color": 3, "width": 4970, "height": 1296, "content": "## \ud83d\udfe1 LAYER 3a \u2014 ENTITY RESOLUTION: CATEGORIES\nResolves raw category strings to canonical names via mapping sheets.\nUnknown entities \u2192 LLM classification \u2192 user confirmation \u2192 alias saved.\n\n---\n\n\u2699\ufe0f **ACTION REQUIRED**\n\n**1. Connect your `categories_mapping` Google Sheet**\nOpen `GS | Read Category Mapping` and `GS | Save Category Mapping`.\nReplace `YOUR_SPREADSHEET_ID` with the ID of your `categories_mapping` sheet.\nRequired columns: `find` \u00b7 `replace`\n\n**2. Connect your `expense_categories` Google Sheet**\nOpen `GS | Read Allowed Categories`.\nReplace `YOUR_SPREADSHEET_ID` with the ID of your `expense_categories` sheet.\nRequired columns: `category` \u00b7 `description` \u00b7 `examples`\n\n**3. Connect OpenAI Credential**\nOpen `LLM | Classify Category` and connect your OpenAI API credential.\n\n**4. Connect Telegram Credential**\nOpen `TG | Confirm Category Suggestion` and connect your Telegram Bot credential.\n\n**5. Set your Bot Token in HTTP nodes**\nOpen `HTTP | Send Category Selection Message`.\nReplace `{{YOUR_BOT_TOKEN}}` in the URL with your actual Telegram Bot Token." }, "typeVersion": 1 }, { "id": "90172522-bfdb-4fa7-8ef3-48c2ee914994", "name": "StickyNote_Layer4", "type": "n8n-nodes-base.stickyNote", "position": [ 8688, 0 ], "parameters": { "color": 4, "width": 1712, "height": 2592, "content": "## \ud83d\udfe0 LAYER 4 \u2014 QUERY ENGINE\nLoads expense data from Google Sheets.\nMerges resolved intent with raw data for filtering.\n\n---\n\n\u2699\ufe0f **ACTION REQUIRED**\n\n**1. Connect your `expenses` Google Sheet**\nOpen `GS | Load Expenses` and replace `YOUR_SPREADSHEET_ID` with your expenses sheet ID.\nRequired columns: `date` \u00b7 `amount` \u00b7 `category` \u00b7 `description` \u00b7 `common_expense` \u00b7 `Person`\n\n**2. Connect your `expense_categories` Google Sheet**\nOpen `GS | Load Categories` and replace `YOUR_SPREADSHEET_ID` with your categories sheet ID.\nRequired columns: `category` \u00b7 `description` \u00b7 `examples`" }, "typeVersion": 1 }, { "id": "4e94aa93-bca3-4087-b52c-de55b3f7be23", "name": "StickyNote_Layer5", "type": "n8n-nodes-base.stickyNote", "position": [ 10400, 0 ], "parameters": { "color": 2, "width": 564, "height": 2592, "content": "## \ud83d\udd34 LAYER 5 \u2014 ANALYTICS & AGGREGATION\nApplies date, person, category and common_only filters.\nComputes totals, category breakdown and person breakdown.\n\n---\n\n\u2699\ufe0f **ACTION REQUIRED**\n\nNo setup required for this layer.\nThe `JS | Filter & Aggregate` node works automatically based on the resolved intent from the previous layers.\nIf you add custom columns to your expenses sheet, update the filter logic in this node accordingly." }, "typeVersion": 1 }, { "id": "a566ae23-972c-4c15-aa4d-b2ea0e14c8f4", "name": "StickyNote_Layer6", "type": "n8n-nodes-base.stickyNote", "position": [ 10976, 0 ], "parameters": { "color": 6, "width": 560, "height": 2592, "content": "## \ud83d\udfe3 LAYER 6 \u2014 RESPONSE\nFormats the final message and sends it back via Telegram.\n\n---\n\n\u2699\ufe0f **ACTION REQUIRED**\n\n**1. Connect Telegram Credential**\nOpen `TG | Send Reply` and connect your Telegram Bot credential.\n\n**2. Adjust the message format (optional)**\nOpen `JS | Format Response Message` to customise the output format, currency symbol, or language of the reply message." }, "typeVersion": 1 }, { "id": "ed4de5df-0c7f-49e6-9580-9bcec3a29756", "name": "StickyNote_Layer", "type": "n8n-nodes-base.stickyNote", "position": [ 3712, 1296 ], "parameters": { "color": 6, "width": 4970, "height": 1296, "content": "## \ud83d\udfe1 LAYER 3b \u2014 ENTITY RESOLUTION: PERSONS\n\n\n---\n\n\u2699\ufe0f **ACTION REQUIRED**\n\n**1. Connect your `person_mapping` Google Sheet**\nOpen `GS | Read Person Mapping` and `GS | Save Person Mapping`.\nReplace `YOUR_SPREADSHEET_ID` with the ID of your `person_mapping` sheet.\nRequired columns: `find` \u00b7 `replace`\n\n**2. Connect your `list_persons` sheet**\nOpen `GS | Read Allowed Persons`.\nThis reads from the `list_persons` tab of the same `person_mapping` spreadsheet.\nRequired column: `person`\n\n**3. Connect OpenAI Credential**\nOpen `LLM | Classify Person` and connect your OpenAI API credential.\n\n**4. Connect Telegram Credential**\nOpen `TG | Confirm Person Suggestion` and connect your Telegram Bot credential.\n\n**5. Set your Bot Token in HTTP nodes**\nOpen `HTTP | Send Person Selection Message`.\nReplace `{{YOUR_BOT_TOKEN}}` in the URL with your actual Telegram Bot Token." }, "typeVersion": 1 }, { "id": "453a9f82-a297-4fb6-adc1-08f2e5619986", "name": "DOC | Layer 1 \u2014 Input", "type": "n8n-nodes-base.stickyNote", "position": [ 1968, 464 ], "parameters": { "color": 5, "width": 540, "height": 500, "content": "\n\n**Purpose:** Entry point of the workflow. Receives all incoming Telegram events and routes them to the correct processing path.\n\n**What happens here:**\n- `MSG | Telegram Inbound` \u2014 Listens for incoming Telegram messages and callback queries (inline button taps).\n- `IF | Message or Callback?` \u2014 Splits the flow: regular text messages go to intent parsing, callback queries (button responses) are handled separately.\n- `IF | User Authorized?` \u2014 Checks whether the sender's chat ID is on the allowlist. Unauthorized users are silently dropped.\n- `IF | Category or Person Callback?` \u2014 Determines whether an incoming callback is a category or person selection and routes accordingly.\n- `JS | Read Category Callback` / `JS | Read Person Callback` \u2014 Extracts the selected value and the resume URL from the callback payload.\n- `HTTP | Forward Category/Person Selection` \u2014 Resumes the waiting workflow branch by POSTing the selected value to the internal webhook.\n- `TG | Confirm Category/Person Selection` \u2014 Sends a short confirmation message to the user acknowledging their selection." }, "typeVersion": 1 }, { "id": "a1283c4a-bad8-4116-a97c-0d2bbaa0f5ef", "name": "DOC | Layer 2 \u2014 Intent Parsing", "type": "n8n-nodes-base.stickyNote", "position": [ 1984, 1664 ], "parameters": { "width": 412, "height": 392, "content": "\n**Purpose:** Converts the user's free-text message into a structured JSON object that all downstream layers can work with.\n\n**What happens here:**\n- `LLM | Parse Intent` \u2014 Sends the raw message to GPT-4.1-nano with a strict system prompt. Returns a JSON object containing: intent, time_reference, explicit dates, person, category, common_only flag, comparison type and group_by dimension.\n- `JS | Extract Intent JSON` \u2014 Cleans the raw LLM response (strips markdown fences if present) and parses it into a proper JSON node output. Falls back to a safe empty-intent object if parsing fails." }, "typeVersion": 1 }, { "id": "f593b6d4-1dce-475b-928d-7c9850bda508", "name": "DOC | Layer 3a \u2014 Entity Resolution: Categories", "type": "n8n-nodes-base.stickyNote", "position": [ 4384, 128 ], "parameters": { "color": 3, "width": 540, "height": 644, "content": "\n\n**Purpose:** Resolves raw category strings from the intent into valid canonical category names as defined in the categories master sheet.\n\n**What happens here:**\n- `IF | Category Present?` \u2014 Skips the entire branch if no category was mentioned.\n- `SPLIT | Split Categories` \u2014 Splits multi-category intents into individual items for per-item processing.\n- `GS | Read Category Mapping` \u2014 Loads the alias\u2192canonical mapping sheet (e.g. \"essen\" \u2192 \"food\").\n- `MERGE | Join Categories with Mapping` + `SET | Normalize Category` \u2014 Applies alias replacement where available, keeps original otherwise.\n- `GS | Read Allowed Categories` + `MERGE | Check Category against Allowed` \u2014 Validates the normalised category against the master list.\n- `IF | Category Known?` \u2014 Known categories pass through directly; unknown ones enter the LLM resolution loop.\n- `LOOP | Iterate Categories` \u2014 Processes unresolved categories one by one.\n- `LLM | Classify Category` \u2014 GPT maps the unknown input to the closest canonical category from the allowed list.\n- `TG | Confirm Category Suggestion` \u2014 Sends the LLM suggestion to the user as a Yes/No confirmation.\n- `IF | Category Suggestion Accepted?` \u2014 On confirmation the mapping is saved; otherwise an inline-button picker is shown (`JS | Build Category Inline Buttons` \u2192 `HTTP | Send Category Selection Message` \u2192 `WAIT | Wait for Category Selection`).\n- `GS | Save Category Mapping` \u2014 Persists the newly learned alias for future queries." }, "typeVersion": 1 }, { "id": "1b9d5cb8-c8b9-4afd-919f-ef51ffbd2109", "name": "DOC | Layer 3b \u2014 Entity Resolution: Persons", "type": "n8n-nodes-base.stickyNote", "position": [ 4448, 1376 ], "parameters": { "color": 6, "width": 572, "height": 556, "content": "**Purpose:** Mirrors Layer 3a but for person names \u2014 resolves nicknames and unknown names to canonical person records.\n\n**What happens here:**\n- `IF | Person Present?` \u2014 Skips the branch entirely when no person was specified.\n- `SPLIT | Split Persons` \u2014 Splits the person field into individual items.\n- `GS | Read Person Mapping` \u2014 Loads the alias\u2192canonical person mapping (e.g. \"nickname\" \u2192 \"Full Name\").\n- `MERGE | Join Persons with Mapping` + `SET | Normalize Person` \u2014 Applies alias replacement where available.\n- `GS | Read Allowed Persons` + `MERGE | Check Person against Allowed` \u2014 Validates the resolved name against the known persons list.\n- `IF | Person Known?` \u2014 Known persons pass through; unknown ones enter the LLM resolution loop.\n- `LOOP | Iterate Persons` \u2014 Processes unresolved person names one at a time.\n- `LLM | Classify Person` \u2014 GPT maps the unknown name to the nearest known person from the allowed list.\n- `TG | Confirm Person Suggestion` \u2014 Asks the user to confirm the suggestion via Yes/No button.\n- `IF | Person Suggestion Accepted?` \u2014 On confirmation the mapping is saved; otherwise a button picker is shown (`JS | Build Person Inline Buttons` \u2192 `HTTP | Send Person Selection Message` \u2192 `WAIT | Wait for Person Selection`).\n- `GS | Save Person Mapping` \u2014 Stores the new alias for future use." }, "typeVersion": 1 }, { "id": "c7f0bb66-e951-4962-ac01-999e805c887c", "name": "DOC | Layer 4 \u2014 Query Engine", "type": "n8n-nodes-base.stickyNote", "position": [ 8704, 416 ], "parameters": { "color": 4, "width": 524, "height": 428, "content": "\n**Purpose:** Assembles the fully resolved intent and loads raw expense data from Google Sheets, ready for filtering.\n\n**What happens here:**\n- `SET | Assemble Resolved Intent` \u2014 Merges canonical category and person values back into the original intent object, producing one clean query descriptor.\n- `GS | Load Expenses` \u2014 Reads all rows from the expenses Google Sheet (date, amount, category, person, common_expense flag).\n- `GS | Load Categories` \u2014 Loads the categories reference sheet in parallel (used for validation during filtering).\n- `MERGE | Combine Expenses + Categories` \u2014 Combines both data sources into a single output so the next layer has everything it needs." }, "typeVersion": 1 }, { "id": "4f17b844-0e95-469e-a6b3-b33edf5b96f7", "name": "DOC | Layer 5 \u2014 Analytics & Aggregation", "type": "n8n-nodes-base.stickyNote", "position": [ 10448, 480 ], "parameters": { "color": 2, "width": 348, "height": 340, "content": "\n**Purpose:** Applies all filters from the resolved intent to the expense data and computes the requested aggregations.\n\n**What happens here:**\n- `JS | Filter & Aggregate` \u2014 Single code node that does everything: resolves relative date references (this_week, last_month, \u2026) into concrete ranges, applies person / category / common_only filters row by row, then computes total amount, transaction count, and optional breakdowns by category or person (driven by the group_by field in the intent)." }, "typeVersion": 1 }, { "id": "6d8d1b8c-93db-4374-9240-b01174d354d5", "name": "DOC | Layer 6 \u2014 Response", "type": "n8n-nodes-base.stickyNote", "position": [ 11008, 128 ], "parameters": { "color": 6, "width": 460, "height": 308, "content": "**Purpose:** Formats the aggregation result into a human-readable Telegram message and delivers it back to the user.\n\n**What happens here:**\n- `JS | Format Response Message` \u2014 Builds the final message string: period header, applied filters, total amount, transaction count, and \u2014 if requested \u2014 a sorted category or person breakdown with percentage shares. Handles the empty-result case gracefully.\n- `TG | Send Reply` \u2014 Sends the formatted message as a Markdown reply to the original chat, referencing the triggering message ID so it appears as a thread reply." }, "typeVersion": 1 }, { "id": "9c55a9fc-45aa-4ac4-a94f-7736abee4e73", "name": "Sticky Note", "type": "n8n-nodes-base.stickyNote", "position": [ 1008, 0 ], "parameters": { "color": 7, "width": 912, "height": 2576, "content": "# \ud83d\udd04 WORKFLOW OVERVIEW\n\n## \ud83d\udfe9 Input & Security (Layer 1)\n\nEvery incoming Telegram message or button tap is received by the trigger.\n\nThe workflow checks whether the sender's **Chat ID** is on the approved list \u2014 unauthorized users are silently ignored.\n\nCallback responses (inline button taps for category or person disambiguation) are also handled here and routed back into the correct waiting branch.\n\n---\n\n## \ud83d\udfe6 Intent Parsing (Layer 2)\n\nThe raw message is sent to **GPT-4.1-nano** with a strict system prompt.\n\nThe model returns a structured JSON object containing:\n\n- `intent`\n- `time_reference`\n- `explicit_date_range`\n- `person`\n- `category`\n- `common_only`\n- `comparison_type`\n- `group_by`\n\nA fallback **Code Node** cleans the response and handles parsing errors gracefully.\n\n---\n\n## \ud83d\udfe8 Entity Resolution\n\n### Layer 3a \u2014 Categories \n### Layer 3b \u2014 Persons\n\nRaw category strings and person names from the intent are resolved to canonical values using **Google Sheets mapping tables**.\n\nIf a value is unknown:\n\n1. GPT classifies it against the allowed list \n2. The user is asked to confirm via a **Yes/No Telegram button**\n\nIf the user rejects the suggestion, an **inline button picker** is shown instead.\n\nEvery confirmed match is saved back to the mapping sheet so future queries resolve automatically.\n\n**The system learns over time.**\n\n---\n\n## \ud83d\udfe7 Query Engine (Layer 4)\n\nThe fully resolved intent is assembled into one clean **query descriptor**.\n\nAll expense rows and the category reference list are loaded from **Google Sheets in parallel** and merged into a single data object.\n\n---\n\n## \ud83d\udfe5 Analytics & Aggregation (Layer 5)\n\nA single **Code Node** applies all filters:\n\n- date range \n- relative time references (last month, this week, etc.) \n- person \n- category \n- `common_only` flag \n\nIt then computes:\n\n- **total amount**\n- **transaction count**\n- optional **breakdowns**\n\nBreakdowns are calculated depending on the `group_by` field:\n\n- by **category**\n- by **person**\n\n---\n\n## \ud83d\udfea Response (Layer 6)\n\nThe result is formatted into a clean **Markdown message** including:\n\n- period \n- applied filters \n- total \n- count \n- optional breakdowns with percentage shares \n\nThe reply is sent directly back to the user's **Telegram chat** as a thread reply to the original message." }, "typeVersion": 1 }, { "id": "14b52ff0-e0db-42be-8f44-066315ba7642", "name": "Sticky Note1", "type": "n8n-nodes-base.stickyNote", "position": [ -16, 0 ], "parameters": { "color": 7, "width": 1024, "height": 2576, "content": "# HOW IT WORKS \u2014 AI TELEGRAM EXPENSE TRACKER (QUERY MODE)\n\nThis workflow answers natural language expense questions from Telegram using AI.\n\nInstead of opening spreadsheets, simply send a message like:\n\n- *How much did I spend on food last month?*\n- *What did we spend together this week?*\n- *Show me a breakdown by category for January.*\n\nThe workflow understands your question, resolves ambiguous names or categories, queries your expense data, and replies with a formatted summary \u2014 automatically.\n\n\n\n# \ud83d\udccb WHAT YOU NEED\n\nFour **Google Sheets** are required before using this workflow.\n\n## \ud83d\udccb WHAT YOU NEED\n\nFive Google Sheets are required before using this workflow.\n\n---\n\n**expenses**\n\nColumns:\n- date (date of expense)\n- amount (amount of expense)\n- category (expense category - e.g. supermarket)\n- description (description of the expense)\n- common_expense (true if it was an expense for all people - e.g. in a household)\n- Person - (Who made the expense)\n\nPurpose:\nMain expense table containing all recorded transactions.\n\n---\n\n**expense_categories**\n\nColumns:\n- category\n- description\n- examples\n\nPurpose:\nReference list of allowed expense categories used for classification.\n\n---\n\n**categories_mapping**\n\nColumns:\n- find\n- replace\n\nPurpose:\nMaps user aliases to canonical categories.\n\nExample:\n- sushi \u2192 food\n\n---\n\n**person_mapping**\n\nColumns:\n- find\n- replace\n\nPurpose:\nMaps nicknames or aliases to canonical person names.\n\nExample:\n- Johnny -> John\n\n---\n\n**list_persons**\n\nColumns:\n- person\n- description\n\nPurpose:\nReference list of all persons that can appear in the expense data.\n\n---\n\n# \ud83d\udca1 KEY FEATURES\n\n- Natural language queries in **German & English**\n- AI-powered intent parsing with **GPT-4.1-nano**\n- **Self-learning category & person resolution system**\n- Interactive disambiguation via **Telegram inline buttons**\n- Relative date support: this week, last month, this year\n- Group-by breakdowns: by category or by person\n- `common_only` filter for shared expense queries\n\n---\n\n# \u2b50 MULTI-USER SETUP\n\nThis workflow supports **multiple users simultaneously**.\n\nAdd all approved **Chat IDs** in the node:\n\nIF | User Authorized?\n\nEach expense query is automatically scoped to the sender unless:\n\n- a specific **person** is mentioned\n- or **\"we\"** is used.\n\nShared expenses are identified by the column:\n\ncommon_expense (boolean true / false)\n\nQueries like:\n\n*\"what did we spend together?\"*\n\nautomatically filter for **shared entries only**." }, "typeVersion": 1 } ], "active": false, "settings": { "executionOrder": "v1" }, "versionId": "", "connections": { "GS | Load Expenses": { "main": [ [ { "node": "MERGE | Combine Expenses + Categories", "type": "main", "index": 0 } ] ] }, "IF | Person Known?": { "main": [ [ { "node": "MERGE | Combine Persons with List", "type": "main", "index": 0 } ], [ { "node": "MERGE | Loop Entry (Person)", "type": "main", "index": 1 } ] ] }, "LLM | Parse Intent": { "main": [ [ { "node": "JS | Extract Intent JSON", "type": "main", "index": 0 } ] ] }, "GS | Load Categories": { "main": [ [ { "node": "MERGE | Combine Expenses + Categories", "type": "main", "index": 1 } ] ] }, "IF | Category Known?": { "main": [ [ { "node": "MERGE | Combine Categories with List", "type": "main", "index": 0 } ], [ { "node": "MERGE | Loop Entry (Category)", "type": "main", "index": 1 } ] ] }, "IF | Person Present?": { "main": [ [ { "node": "MERGE | Combine Person Mapping", "type": "main", "index": 0 } ], [ { "node": "SPLIT | Split Persons", "type": "main", "index": 0 }, { "node": "GS | Read Person Mapping", "type": "main", "index": 0 }, { "node": "GS | Read Allowed Persons", "type": "main", "index": 0 } ] ] }, "IF | User Authorized?": { "main": [ [ { "node": "LLM | Parse Intent", "type": "main", "index": 0 } ] ] }, "LLM | Classify Person": { "main": [ [ { "node": "SET | Extract LLM Person", "type": "main", "index": 0 } ] ] }, "SPLIT | Split Persons": { "main": [ [ { "node": "MERGE | Join Persons with Mapping", "type": "main", "index": 0 } ] ] }, "IF | Category Present?": { "main": [ [ { "node": "MERGE | Combine Category + Person", "type": "main", "index": 0 } ], [ { "node": "SPLIT | Split Categories", "type": "main", "index": 0 }, { "node": "GS | Read Category Mapping", "type": "main", "index": 0 }, { "node": "GS | Read Allowed Categories", "type": "main", "index": 0 } ] ] }, "LOOP | Iterate Persons": { "main": [ [ { "node": "MERGE | Loop Entry (Person)", "type": "main", "index": 0 } ], [ { "node": "LLM | Classify Person", "type": "main", "index": 0 } ] ] }, "MSG | Telegram Inbound": { "main": [ [ { "node": "IF | Message or Callback?", "type": "main", "index": 0 } ] ] }, "SET | Normalize Person": { "main": [ [ { "node": "MERGE | Check Person against Allowed", "type": "main", "index": 0 } ] ] }, "JS | Filter & Aggregate": { "main": [ [ { "node": "JS | Format Response Message", "type": "main", "index": 0 } ] ] }, "LLM | Classify Category": { "main": [ [ { "node": "SET | Extract LLM Category", "type": "main", "index": 0 } ] ] }, "GS | Read Person Mapping": { "main": [ [ { "node": "MERGE | Join Persons with Mapping", "type": "main", "index": 1 } ] ] }, "GS | Save Person Mapping": { "main": [ [ { "node": "LOOP | Iterate Persons", "type": "main", "index": 0 } ] ] }, "JS | Extract Intent JSON": { "main": [ [ { "node": "IF | Category Present?", "type": "main", "index": 0 }, { "node": "IF | Person Present?", "type": "main", "index": 0 } ] ] }, "JS | Merge Intent Fields": { "main": [ [ { "node": "SET | Assemble Resolved Intent", "type": "main", "index": 0 } ] ] }, "SET | Extract LLM Person": { "main": [ [ { "node": "TG | Confirm Person Suggestion", "type": "main", "index": 0 } ] ] }, "SET | Normalize Category": { "main": [ [ { "node": "MERGE | Check Category against Allowed", "type": "main", "index": 0 } ] ] }, "SPLIT | Split Categories": { "main": [ [ { "node": "MERGE | Join Categories with Mapping", "type": "main", "index": 0 } ] ] }, "GS | Read Allowed Persons": { "main": [ [ { "node": "MERGE | Check Person against Allowed", "type": "main", "index": 1 }, { "node": "AGG | Aggregate Person List", "type": "main", "index": 0 } ] ] }, "IF | Message or Callback?": { "main": [ [ { "node": "IF | User Authorized?", "type": "main", "index": 0 } ], [ { "node": "IF | Category or Person Callback?", "type": "main", "index": 0 } ] ] }, "JS | Read Person Callback": { "main": [ [ { "node": "HTTP | Forward Person Selection", "type": "main", "index": 0 } ] ] }, "LOOP | Iterate Categories": { "main": [ [ { "node": "MERGE | Loop Entry (Category)", "type": "main", "index": 0 } ], [ { "node": "LLM | Classify Category", "type": "main", "index": 0 } ] ] }, "SET | Set Resolved Person": { "main": [ [ { "node": "MERGE | Combine Person Mapping", "type": "main", "index": 1 } ] ] }, "GS | Read Category Mapping": { "main": [ [ { "node": "MERGE | Join Categories with Mapping", "type": "main", "index": 1 } ] ] }, "GS | Save Category Mapping": { "main": [ [ { "node": "LOOP | Iterate Categories", "type": "main", "index": 0 } ] ] }, "SET | Extract LLM Category": { "main": [ [ { "node": "TG | Confirm Category Suggestion", "type": "main", "index": 0 } ] ] }, "AGG | Aggregate Person List": { "main": [ [ { "node": "MERGE | Combine Persons with List", "type": "main", "index": 1 } ] ] }, "JS | Read Category Callback": { "main": [ [ { "node": "HTTP | Forward Category Selection", "type": "main", "index": 0 } ] ] }, "MERGE | Loop Entry (Person)": { "main": [ [ { "node": "SET | Set Resolved Person", "type": "main", "index": 0 } ] ] }, "SET | Set Resolved Category": { "main": [ [ { "node": "AGG | Aggregate Resolved Categories", "type": "main", "index": 0 } ] ] }, "GS | Read Allowed Categories": { "main": [ [ { "node": "MERGE | Check Category against Allowed", "type": "main", "index": 1 }, { "node": "AGG | Aggregate Category List", "type": "main", "index": 0 } ] ] }, "JS | Format Response Message": { "main": [ [ { "node": "TG | Send Reply", "type": "main", "index": 0 } ] ] }, "AGG | Aggregate Category List": { "main": [ [ { "node": "MERGE | Combine Categories with List", "type": "main", "index": 1 } ] ] }, "MERGE | Loop Entry (Category)": { "main": [ [ { "node": "SET | Set Resolved Category", "type": "main", "index": 0 } ] ] }, "MERGE | Combine Person Mapping": { "main": [ [ { "node": "MERGE | Combine Category + Person", "type": "main", "index": 2 } ] ] }, "SET | Assemble Resolved Intent": { "main": [ [ { "node": "GS | Load Expenses", "type": "main", "index": 0 }, { "node": "GS | Load Categories", "type": "main", "index": 0 } ] ] }, "SET | Set New+Old Person (LLM)": { "main": [ [ { "node": "MERGE | Combine Person Mapping Entries", "type": "main", "index": 1 } ] ] }, "TG | Confirm Person Suggestion": { "main": [ [ { "node": "IF | Person Suggestion Accepted?", "type": "main", "index": 0 } ] ] }, "HTTP | Forward Person Selection": { "main": [ [ { "node": "TG | Confirm Person Selection", "type": "main", "index": 0 } ] ] }, "SET | Read Callback Body (Cat.)": { "main": [ [ { "node": "SET | Set New+Old Category (Selection)", "type": "main", "index": 0 } ] ] }, "IF | Person Suggestion Accepted?": { "main": [ [ { "node": "SET | Set New+Old Person (LLM)", "type": "main", "index": 0 } ], [ { "node": "JS | Build Person Inline Buttons", "type": "main", "index": 0 } ] ] }, "JS | Build Person Inline Buttons": { "main": [ [ { "node": "HTTP | Send Person Selection Message", "type": "main", "index": 0 } ] ] }, "SET | Set New+Old Category (LLM)": { "main": [ [ { "node": "MERGE | Combine Category Mapping Entries", "type": "main", "index": 1 } ] ] }, "TG | Confirm Category Suggestion": { "main": [ [ { "node": "IF | Category Suggestion Accepted?", "type": "main", "index": 0 } ] ] }, "WAIT | Wait for Person Selection": { "main": [ [ { "node": "SET | Read Callback Body (Person)", "type": "main", "index": 0 } ] ] }, "HTTP | Forward Category Selection": { "main": [ [ { "node": "TG | Confirm Category Selection", "type": "main", "index": 0 } ] ] }, "IF | Category or Person Callback?": { "main": [ [ { "node": "JS | Read Category Callback", "type": "main", "index": 0 } ], [ { "node": "JS | Read Person Callback", "type": "main", "index": 0 } ] ] }, "MERGE | Combine Category + Person": { "main": [ [ { "node": "JS | Merge Intent Fields", "type": "main", "index": 0 } ] ] }, "MERGE | Combine Persons with List": { "main": [ [ { "node": "LOOP | Iterate Persons", "type": "main", "index": 0 } ] ] }, "MERGE | Join Persons with Mapping": { "main": [ [ { "node": "SET | Normalize Person", "type": "main", "index": 0 } ] ] }, "SET | Read Callback Body (Person)": { "main": [ [ { "node": "SET | Set New+Old Person (Selection)", "type": "main", "index": 0 } ] ] }, "IF | Category Suggestion Accepted?": { "main": [ [ { "node": "SET | Set New+Old Category (LLM)", "type": "main", "index": 0 } ], [ { "node": "JS | Build Category Inline Buttons", "type": "main", "index": 0 } ] ] }, "JS | Build Category Inline Buttons": { "main": [ [ { "node": "HTTP | Send Category Selection Message", "type": "main", "index": 0 } ] ] }, "WAIT | Wait for Category Selection": { "main": [ [ { "node": "SET | Read Callback Body (Cat.)", "type": "main", "index": 0 } ] ] }, "AGG | Aggregate Resolved Categories": { "main": [ [ { "node": "MERGE | Combine Category + Person", "type": "main", "index": 1 } ] ] }, "HTTP | Send Person Selection Message": { "main": [ [ { "node": "WAIT | Wait for Person Selection", "type": "main", "index": 0 } ] ] }, "MERGE | Check Person against Allowed": { "main": [ [ { "node": "IF | Person Known?", "type": "main", "index": 0 } ] ] }, "MERGE | Combine Categories with List": { "main": [ [ { "node": "LOOP | Iterate Categories", "type": "main", "index": 0 } ] ] }, "MERGE | Join Categories with Mapping": { "main": [ [ { "node": "SET | Normalize Category", "type": "main", "index": 0 } ] ] }, "SET | Set New+Old Person (Selection)": { "main": [ [ { "node": "MERGE | Combine Person Mapping Entries", "type": "main", "index": 0 } ] ] }, "MERGE | Combine Expenses + Categories": { "main": [ [ { "node": "JS | Filter & Aggregate", "type": "main", "index": 0 } ] ] }, "HTTP | Send Category Selection Message": { "main": [ [ { "node": "WAIT | Wait for Category Selection", "type": "main", "index": 0 } ] ] }, "MERGE | Check Category against Allowed": { "main": [ [ { "node": "IF | Category Known?", "type": "main", "index": 0 } ] ] }, "MERGE | Combine Person Mapping Entries": { "main": [ [ { "node": "GS | Save Person Mapping", "type": "main", "index": 0 } ] ] }, "SET | Set New+Old Category (Selection)": { "main": [ [ { "node": "MERGE | Combine Category Mapping Entries", "type": "main", "index": 0 } ] ] }, "MERGE | Combine Category Mapping Entries": { "main": [ [ { "node": "GS | Save Category Mapping", "type": "main", "index": 0 } ] ] } } }