Ovitalmap Parcel CSV Generator

This skill processes parcel vertex coordinates — from images or text — and produces Ovitalmap-readable CSV files. It also archives every processed parcel into a structured per-country database with categorization metadata.

0) Defaults: Language, Tone, I/O

• Reply language: Chinese by default unless the user requests another language.
• Tone: non-repetitive, concise, precise, professional.
• Ambiguity handling: when inputs are unclear (e.g., parcel ID meaning, UTM zone/hemisphere, provider name), ask targeted clarifying questions via AskUserQuestion.
• CSV headers: keep original Chinese headers exactly as listed below. Do not translate them.

1) Inputs and Core Assumptions

• Input: one or more images or text blocks that contain parcel vertex coordinates.
• Coordinate parsing: auto-detect format:
  • Decimal degrees: 22.50422, 114.13472
  • DMS: 22°30'15.2"N, 114°08'05.0"E
  • UTM with zone and hemisphere
  • Other formats on a best-effort basis
• WGS84 policy:
  1. Always display the raw recognized coordinates verbatim per parcel.
  2. Always display WGS84 decimal-degree coordinates (convert if needed).
• Date token (YYMMDD):
  • Use user-provided date if given.
  • Else use today's local date.
• CSV format: UTF-8, comma-separated, no BOM.

2) Country and Parcel Codes

2.1) Country Code

• Auto-determine the country from parcel coordinates using WebSearch / coordinate lookup. Do NOT ask the user to confirm — simply state the result and proceed. The user will correct it if wrong.
• Default standard: ISO 3166-1 alpha-3 (e.g., HKG, CHN, USA, FRA). On request, allow ISO 3166-3 or other variants.
• Cross-border parcels: if coordinates happen to span borders, use the country where the majority of the parcel falls. Do not split.

2.2) Parcel Code — Always Generated

Every parcel MUST receive a parcel_code. There is no "no code" fallback.

The code follows one of two formats, with priority given to official registration IDs:

Official registration ID detection:

• Scan all text extracted from images AND any text the user provides directly.
• If a clear mining cadastre / registration / permit number appears, present it to the user for confirmation before adopting it.
• If multiple candidate IDs appear for the same parcel, pick the most specific one and ask the user to confirm.

2.3) Uniqueness Check (Mandatory Before Code Assignment)

Before finalizing any parcel_code, you MUST scan the existing archive to prevent duplicates. This is critical because:

• Users may upload multiple parcels in one session or across separate sessions on the same day.
• A new session must NOT restart SEQ from 001 — it must continue from the last used number.

Procedure:

1. Read ovitalmap_archive/{ISO3}_parcels.csv (if it exists).
2. Extract all existing parcel_code values.
3. For official registration IDs (format 1): check that {ISO3}-{OFFICIAL_ID} does NOT already exist. If it does, warn the user and ask for clarification.
4. For sequential codes (format 2): find the maximum SEQ already used for today ({ISO3}-{YYMMDD}-*). The new SEQ = max_existing + 1. If none exist for today, SEQ starts at 001.
5. If the archive file does not exist yet, SEQ starts at 001.

3) Output Files Overview

Two CSVs are produced together for the batch:

3.1) File Naming & Layout

Store generated CSVs in country subdirectories under ovitalmap_exports/:

ovitalmap_exports/{ISO3}/

Use the first parcel's code as the file base — since codes are globally unique (§2.3), this guarantees no collisions even with multiple batches on the same day:

• firstCode = the parcel_code of the first parcel in this batch.
• count = total number of parcels in this batch.
• firstCode already encodes country + date, so no redundant prefixes.

Examples:

• CHN-260609-001_N2.csv — batch of 2, first is 001.
• CHN-260609-003_N2.csv — another batch of 2 on the same day, first is 003. No collision.
• CHN-PE12345_N1.csv — single parcel with an official registration code.

4) Vertices CSV (顶点表)

Headers (exact, in order, do not translate):

文件夹,名称,经度,纬度,海拔,文本显示风格,图标样式,Comment

Field Rules

Indexing conventions

• vertex-index resets from 01 per parcel (A01, A02, …).
• Preserve original vertex order. If ordering is unclear, ask the user about clockwise reordering.

5) Boundary CSV (边界表)

Headers (exact, in order, do not translate):

文件夹,名称,经纬度[经度+纬度],线条宽度,线条颜色,线条不透明度,闭合,线型,轨迹风格,Comment

Field Rules

6) Quality Checks and Special Cases

1. Range checks: longitude ∈ [-180, 180], latitude ∈ [-90, 90]. Flag any violations explicitly.
2. Duplicate vertices: keep first occurrence, mark later duplicates in the output and ask the user for confirmation.
3. Cross-border parcels: use the country where the majority area falls (§2.1). Do not split.
4. Missing altitude: leave the field empty; never infer or guess.
5. Naming collisions: if 文件夹 + 名称 duplicates occur, append -1, -2, … to the 名称 field and report the collision to the user.
6. Coordinate precision: preserve all decimal digits from the conversion. Do not round.

7) Archiving and Categorization (Database)

After generating the CSVs for a parcel, you MUST archive the parcel into the workspace database.

7.1) Archive Location

All archives live flat in ovitalmap_archive/ at the workspace root. One file per country:

ovitalmap_archive/{ISO3}_parcels.csv

If the file does not exist yet, create it with headers.

7.2) Archive CSV Headers

parcel_code,country,nearest_city,provider_name,archive_date,boundary_coords,provider_notes,cadastre_code

7.3) Field Rules for Archiving

7.4) Provider Name Workflow

1. At Step 1 (before any code assignment or file writing), ask the user: "Who provided this parcel? (请提供此地块的提供者姓名)"
2. After the user provides a name, immediately scan all existing providers from both:
   • ovitalmap_archive/{ISO3}_parcels.csv (per-country archive)
   • ovitalmap_archive/master.csv (cross-country archive)
3. Proactive fuzzy deduplication — compare the user-supplied name against every existing provider. Flag a match if any of the following hold:
   • Exact match: identical string (case-insensitive).
   • Whitespace / punctuation variation: same characters but different spacing, hyphens, or dots (e.g., "Zhang San" ≈ "Zhang-San" ≈ "ZhangSan").
   • Chinese ↔ Pinyin overlap: one is Chinese characters and the other is the corresponding pinyin (e.g., "张三" ≈ "Zhang San").
   • Partial / suffix / prefix overlap: one name is fully contained within another OR they differ only by a common suffix/prefix (e.g., "李", "总", "先生", "经理", "老板", "哥", "姐"). Examples:
     • "中非李" ≈ "中非李总" (suffix variation).
     • "三一李" ≈ "三一李总" (suffix variation).
     • "李总" ⊂ "中非李总" and ⊂ "三一李总" (short name matches multiple).
4. Handle matches based on count and type:
   • Single candidate match: ask the user explicitly:

     "检测到提供者 '{input_name}' 与已有记录 '{existing_name}' 高度相似。是否为同一个人？"

     • If the user confirms same person → reuse the existing provider_name spelling.
     • If the user says different person → treat as a new provider.
     • If the user is unsure → keep the new name but add a note in provider_notes: Possible duplicate of '{existing_name}'.
   • Multiple candidates match (e.g., user says "李总", archive has "中非李总" and "三一李总"): list ALL candidates and ask the user to specify:

     "'{input_name}' 在已有记录中匹配到多位提供者：① {candidate1} ② {candidate2}。请问对应的是哪一位？或都不是(新建)？"

     • If the user picks one → reuse that existing provider_name.
     • If the user says none of them → treat as a new provider.
5. If no similar match exists, add the name as a new provider.
6. If the user declines to provide a name, use "Unknown" and note it.
7. Do not proceed to code assignment until provider name is confirmed.

8) Master Spreadsheet

After archiving to ovitalmap_archive/{ISO3}_parcels.csv, also append the same row(s) to a single consolidated file:

ovitalmap_archive/master.csv

This is a single spreadsheet containing every parcel from every country — the complete cross-country view.

Headers (one extra column vs per-country parcels.csv):

ISO3,parcel_code,country,nearest_city,provider_name,archive_date,boundary_coords,provider_notes,cadastre_code

Append every newly archived parcel row here immediately after writing to the per-country file. If master.csv does not exist, create it with headers.

9) Interaction Flow

Guard Rule

Do NOT write or modify any file until user confirms BOTH coordinates AND provider name. Step 1 is a hard gate — all subsequent steps (2–4) are blocked until the user explicitly confirms the parsed coordinates AND the provider(s).

When invoked, follow this sequence:

1. Parse, Verify & Confirm Provider

• Show raw coordinates and WGS84 conversion per parcel.
• Ask for the provider name(s) (§7.4). Identify each parcel's provider.
• Display ⚠️ "请核实以下识别和转换后的坐标是否与原始数据一致，并确认提供者姓名". → STOP HERE. Wait for user to confirm or correct coordinates AND provider. If user provides corrections, re-parse/re-confirm until both are confirmed.

2. Assign Codes (only after step 1 confirmed)

• Auto-determine country from coordinates, derive ISO3. State it, move on.
• Scan the archive for existing codes (§2.3). Detect any official registration IDs from image text or user input.
• Generate and display all parcel_code values. The user will correct if wrong.

3. Build CSVs

Build both Vertices CSV (§4) and Boundary CSV (§5) together. Run quality checks (§6). State the generated filenames:

CSV 文件已生成:

• 顶点表: {firstCode}_N{count}.csv
• 边界表: {firstCode}_N{count}_boundary.csv

4. Archive

• Factual city lookup from coordinates (§7.3).
• Append to ovitalmap_archive/{ISO3}_parcels.csv and ovitalmap_archive/master.csv.
• Summarize: country, nearest city, provider, codes, date.

5. Wrap Up

Note any assumptions, edge cases, or corrections.

9b) Post-Archive Coordinate Correction

Users may correct a parcel's coordinates after it has been archived. When this happens:

Procedure

1. Identify the parcel by its parcel_code in ovitalmap_archive/{ISO3}_parcels.csv and ovitalmap_archive/master.csv.

2. Backup before modifying — copy the current archive files to a backup:

   ovitalmap_backups/{ISO3}_parcels_{YYMMDD_HHMM}.csv
   ovitalmap_backups/master_{YYMMDD_HHMM}.csv
   

   Create ovitalmap_backups/ if it does not exist.

3. Update the row in the per-country archive and master spreadsheet:

   • Replace boundary_coords with the corrected boundary string.
   • Update archive_date to the current date (marking the correction date).
   • Append a note to provider_notes: "[{date}] Coordinates corrected. Original backup: {backup_file}".

4. Regenerate the export CSVs for the corrected parcel in ovitalmap_exports/{ISO3}/.

5. Summarize what was changed, where backups are, and ask the user to verify.

10) Example Output Snippet

⚠️ 请核实以下识别和转换后的坐标是否与原始数据一致，并确认提供者姓名

原始坐标:
• 地块 1: 22°30'15.2"N, 114°08'05.0"E; 22°30'14.8"N, 114°08'08.3"E; …
• 地块 2: 22.50422, 114.13472; 22.50418, 114.13480; …

WGS84:
• 地块 1: 114.13472, 22.50422; 114.13564, 22.50411; …
• 地块 2: 114.13472, 22.50422; 114.13500, 22.50416; …

请提供此地块的提供者姓名？

确认坐标无误并确认提供者后继续处理？(如有修正请提供)

→ 用户确认坐标 + 提供者: 张三

所属国家: 中国 (CHN)

正在扫描 CHN 存档…
  现有编码: CHN-260609-001, CHN-260609-002
  → 地块 1 → CHN-260609-003
  → 地块 2 → CHN-260609-004

图像文本检测到注册号 PE12345 → 确认为地块 2 的官方登记号 → CHN-PE12345

最终编码:
• 地块 1: CHN-260609-003
• 地块 2: CHN-PE12345

CSV 文件已生成:
• 顶点表: CHN-260609-003_N2.csv
• 边界表: CHN-260609-003_N2_boundary.csv

--- 归档 ---
查询最近城市中…
✓ 深圳市 (Shenzhen), 中国

归档完成:
• 国家: 中国 (CHN)    • 最近城市: 深圳市    • 提供者: 张三
• 地块编号: CHN-260609-003, CHN-PE12345
• 归档日期: 2026-06-09    • Cadastre: (待后续技能)

11) Future Extensions (Placeholders)

• Cadastre minier lookup: another skill will be developed to query official mining cadastre registries and populate the cadastre_code field.
• Email dispatch: the summary spreadsheet will be periodically sent to a configured email address. The email configuration (recipient, SMTP, schedule) is not yet implemented.

12) File Structure

workspace/
├── ovitalmap_exports/             # Generated CSVs (by country subdir)
│   ├── CHN/
│   │   └── CHN-260609-001_N2.csv
│   ├── HKG/
│   │   └── HKG-260609-001_N1.csv
│   └── …
├── ovitalmap_archive/             # Persistent database (flat)
│   ├── master.csv                 # All parcels, all countries
│   ├── CHN_parcels.csv
│   ├── HKG_parcels.csv
│   └── …
├── ovitalmap_backups/             # Pre-correction snapshots (§9b)
│   ├── CHN_parcels_260609_1430.csv
│   └── master_260609_1430.csv
└── .trae/skills/ovitalmap-parcel-csv/
    └── SKILL.md