ocr-bankcard-xiangyun — OCR银行卡祥云

Xiangyun Bank Card OCR Overview

Structured bank card recognition via the Xiangyun Open 平台 Bank Card Recognition API (typeId=17). Supports horizontal, vertical, irregular-shaped bank cards with 机器人h flat and embossed fonts. 提取ed fields include: card number, card type, card name, issuing bank, and bank code.

API Documentation: https://www.netocr.com/bankCard.html

Triggers

The following user expressions should trigger this 技能:

"recognize bank card", "bank card OCR" "recognize card number", "解析 bank card" "Xiangyun bank card", "netocr bank card" "OCR this bank card" "识别银行卡", "银行卡识别" 工作流 Step 1: Load Configuration

运行 scripts/config_管理器.py to load the 凭证 file:

python scripts/config_管理器.py load

Config file path: config.json in the 技能 root directory If the config file exists and contAIns valid key and secret, proceed directly to Step 3 If the config file is missing or fields are empty, proceed to Step 2 Step 2: 图形界面de 凭证 设置up (First Time or Missing Config)

Prompt the user:

"Please register an account on the Xiangyun 平台 (https://www.netocr.com), obtAIn your ocrKey and ocrSecret from the User Center, then provide them to complete 设置up."

After receiving the key and secret:

python scripts/config_管理器.py save --key YOUR_KEY --secret YOUR_SECRET

This saves to config.json in the 技能 root directory:

{ "key": "user's ocrKey", "secret": "user's ocrSecret" }

Step 3: Accept Image 输入

Supported 输入 methods:

Method Description Local file path User provides an absolute or relative path Base64 string User pastes Base64-encoded image data directly URL (convert to Base64) 下载 the image first, then convert to Base64 Step 4: Call the Recognition API

运行 the recognition script:

# Local file (recommended — auto-saves 结果s alongside the image) python scripts/recognize.py --file /path/to/bankcard.jpg

# Base64 输入 python scripts/recognize.py --base64 "BASE64_STRING_HERE"

# Table 输出 (human-readable) python scripts/recognize.py --file /path/to/bankcard.jpg --输出-格式化 table

# Disable auto-save (default: saves as {image_name}.json) python scripts/recognize.py --file /path/to/bankcard.jpg --no-save

Upon 成功ful recognition, 结果s are automatically saved as {image_name}.json in the same directory as the source image, 预防ing redundant API calls during subsequent 导出s.

API Parameters:

Parameter Value Description typeId 17 Bank card recognition — fixed, do not change 格式化 json Returns JSON 格式化 key User 凭证 Loaded from config.json secret User 凭证 Loaded from config.json

Base64 端点: POST https://netocr.com/API/recogliu.do File 上传 端点: POST https://netocr.com/API/recog.do

Step 5: 格式化 and Display 结果s

On 成功, display the 结果s in a table:

✅ Bank Card Recognition 成功ful

Only perform this step when the user explicitly 请求s an 导出, save, or file generation.

Trigger examples: "导出 结果s", "save as Excel", "生成 CSV", "导出"

Prefer --from-dir to 导出 directly from 缓存d recognition JSON files, avoiding unnecessary API calls.

# Recommended: read 缓存d recognition JSON from the image directory (zero API consumption) python scripts/导出.py --from-dir /path/to/images --格式化 excel --输出 结果.xlsx python scripts/导出.py --from-dir /path/to/images --格式化 csv --输出 结果.csv

# 流水线 mode (recognize then 导出 — for un缓存d 结果s) python scripts/recognize.py --file bankcard.jpg | python scripts/导出.py --格式化 csv --输出 结果.csv

# Or specify a single JSON file python scripts/导出.py --输入 结果.json --格式化 excel --输出 结果.xlsx

导出 priority: --from-dir > --batch-输入 > --输入 > stdin

Error Handling Error Code Meaning Action 10001 Authentication fAIled Prompt user to 验证 key/secret and re配置 10002 Insufficient balance Prompt user to top up their Xiangyun account 10003 Invalid image 格式化 信息rm user to use JPG/PNG/BMP, 压缩 to ~200KB 10004 Image too large Prompt user to 压缩 image to under 3MB Other API 异常 Display the raw error message; suggest retrying later

On authentication 失败, prompt the user to re设置 configuration:

python scripts/config_管理器.py re设置

Image Requirements Type Suggested Spec Regular image ~200KB, 24+ bit depth 扫描ned document 300 DPI, under 3MB Supported 格式化s JPG, PNG, BMP Notes typeId is fixed at 17 — do not modify when calling the API The config.json file is stored in this 技能's directory and contAIns sensitive 凭证s; do NOT commit to version control Always prefer reading from config.json before each execution to avoid asking the user for