message-parser

TotalClaw 作者 totalclaw

将原始 WhatsApp 导出（TXT 或 JSON）解析为包含“时间戳”、“发送者”和“内容”的规范化消息对象。当用户要求在提取之前解析聊天导出、清理 WhatsApp 转储或将聊天 TXT 转换为结构化 JSON 时使用。推荐的链开始：消息解析器然后引导提取器。请勿用于线索解释、存储、总结或行动建议。

安装 / 下载方式

TotalClaw CLI推荐

totalclaw install totalclaw:totalclaw~vishalgojha-message-parser

cURL直接下载，无需登录

curl -fsSL https://skills.taituai.com/api/skills/totalclaw%3Atotalclaw~vishalgojha-message-parser/file -o vishalgojha-message-parser.md

## 概述（中文）

## 原文

# Message Parser

Convert raw chat exports into a strict array of parsed message objects.

## Quick Triggers

- Parse this WhatsApp export file.
- Convert this group dump into structured messages.
- Clean this TXT chat into timestamp/sender/content rows.

## Recommended Chain

`message-parser -> lead-extractor -> india-location-normalizer -> sentiment-priority-scorer -> summary-generator -> action-suggester -> lead-storage`

## Execute Workflow

1. Accept raw WhatsApp export input as plain text, JSON, or file contents already loaded by Supervisor.
2. Detect and parse the source format, including WhatsApp export lines like `DD/MM/YYYY, HH:MM - sender: message`.
3. Normalize each event into exactly three fields:
- `timestamp` (RFC 3339 date-time string)
- `sender` (non-empty string)
- `content` (string; allow empty message bodies)
4. Merge multiline continuation lines into the previous message when they do not start with a new timestamp marker.
5. Preserve message ordering. If timestamps collide, preserve original source order.
6. Ignore chat-system boilerplate as lead content (encryption notice, group created, member added) while preserving raw line fidelity for audit.
7. Validate output against `references/parsed-message-array.schema.json`.
8. Return only the validated array.

## Enforce Boundaries

- Never infer or extract leads.
- Never write to databases or files.
- Never generate summaries or action plans.
- Never send or schedule outbound communication.
- Never bypass Supervisor routing.

## Handle Errors

1. Return explicit parse errors for malformed entries.
2. Include line offsets when source lines are available.
3. Fail closed if output cannot satisfy the schema.