模板化生成论文速记全套物料
Template-based generation for paper note cards & articles
评分卡 · 信息卡 · 封面图 · 正文模板 · 公众号排版
Score Cards · Info Cards · Cover Images · Article Templates · WeChat Formatting
如果目标是复现当前每日 paper-notes 生产流程(候选输入、选题、评分、PDF 核验、生成发布包、preflight、发布前编辑检查),请优先阅读:
pip install -r requirements.txt准备数据 → 生成图片 → 编写正文 → 完成
Prepare Data → Generate Images → Write Article → Done
复制 references/data_template.json,填入论文信息:
Copy references/data_template.json and fill in paper information:
{
"paper_title": "论文标题 / Paper Title",
"score": {
"total": 8.5,
"dimensions": [
{"label": "重要性 Impact", "value": 1.8},
{"label": "创新性 Novelty", "value": 1.7},
{"label": "可验证性 Evidence", "value": 1.6},
{"label": "产业可用性 Applicability", "value": 1.7},
{"label": "可复用性 Reusability", "value": 1.6}
]
},
"info": {
"title": "英文标题 / English Title",
"title_cn": "中文标题 / Chinese Title",
"link": "https://arxiv.org/abs/XXXX.XXXXX",
"authors": ["作者 1 / Author 1", "作者 2 / Author 2"],
"affiliations": ["机构 A / Institution A", "机构 B / Institution B"]
}
}# 评分卡 + 信息卡 / Score Card + Info Card
python3 scripts/production/generate_cards.py --data data.json --out outputs/my-paper
# 封面图(可选)/ Cover Image (optional)
python3 scripts/production/generate_cover.py --data data.json --out outputs/my-paper/cover_235.png参考 references/md_template.md 填写 A-F 六段式内容:
Refer to references/md_template.md for the A-F article structure:
| 段落 / Section | 说明 / Description |
|---|---|
| A. 研究问题 / Research Question | 一句话说明问题 / One-sentence problem statement |
| B. 核心贡献 / Core Contributions | 列出 2-3 个贡献点 / List 2-3 contributions |
| C. 方法/框架 / Method/Framework | 描述技术方法 / Describe technical approach |
| D. 关键结果 / Key Results | 指标/对比/结论 / Metrics/comparison/conclusions |
| E. 产业启示 / Industry Implications | 对行业的启发 / Implications for industry |
| F. 一句话判断 / Final Verdict | 站队结论 / One-sentence verdict |
| 文件 / File | 说明 / Description |
|---|---|
score_card.png |
五维评分卡 / 5-Dimension Score Card |
info_card.png |
论文信息卡 / Paper Info Card |
cover_235.png |
2.35:1 封面图 / 2.35:1 Cover Image |
note.md |
正文 Markdown / Article in Markdown |
article_editor_ready.html |
公众号 HTML / WeChat Article HTML |
查看完整样例 / View full examples: examples/
paper-notes/
├── references/ # 模板与说明 / Templates & Docs
├── scripts/production/ # 正式生产脚本 / Production Scripts
├── outputs/ # 最终产物 / Final Outputs
├── examples/ # 输出样例 / Output Examples
├── docs/ # 完整文档 / Full Documentation
└── assets/ # logo/字体 / Logo & Fonts
- 配色方案 / Color Scheme:编辑
references/css_main.md - 评分维度 / Score Dimensions:修改
data.json中的score.dimensions - 正文模板 / Article Template:编辑
references/md_template.md
MIT License - 自由使用,欢迎贡献 / Free to use, contributions welcome
- 公众号 "AI 系统笔记" / WeChat Official Account "AI System Notes" - 每日论文速记解读 / Daily paper note interpretations
To ensure that Google's Code Wiki parser can accurately crawl, classify, and explain this repository, the following visual schema and data flows define the automation pipeline.
graph TD
A[data.json - Input Data] --> B[scripts/production/generate_cards.py]
A --> C[scripts/production/generate_cover.py]
B --> D[outputs/score_card.png - 1080x720]
B --> E[outputs/info_card.png - 1080x720]
C --> F[outputs/cover_235.png - 2350x1000]
G[references/md_template.md] --> H[scripts/production/render_article.py]
I[references/css_main.md] --> H
H --> J[outputs/article_editor_ready.html - WeChat Rendered]
- Input Validation Module: Ingests JSON files mapping to
references/data_template.json. - Asset Generation Module: Uses PIL/Pillow to generate pixel-precise images (
score_card.png,info_card.png, andcover_235.png) utilizing the custom OTF fonts embedded inassets/. - Render Engine: Transforms standard A-F markdown structures mapped in
references/md_template.mdalong with custom CSS tokens inreferences/css_main.mdto generate WeChat-formatted HTML articles.
This tool output is optimized for high-impact social sharing and WeChat official account integration:
- 五维评分卡 / Score Card (
score_card.png):- Resolution: 1080x720px
- Design: Radar/Dimension representation of 5 indicators: Importance (Impact), Innovation (Novelty), Verifiability (Evidence), Industry Applicability (Applicability), and Reusability (Reusability).
- 论文信息卡 / Info Card (
info_card.png):- Resolution: 1080x720px
- Design: Clean layout listing metadata (Title, Link, Authors, and Affiliations) in high-contrast styling.
- 公众号封面 / Cover (
cover_235.png):- Resolution: 2350x1000px (Aspect Ratio: 2.35:1)
- Design: Minimalist post cover with title alignment.
Articles generated via scripts/production/render_article.py incorporate the following vanilla CSS framework elements:
:root {
--primary: #0052D9; /* Tencent Blue */
--secondary: #E34D59; /* Accent Red */
--bg: #F7F9FC; /* Light Grayish Blue Background */
--text: #333333; /* Standard Dark Text */
--muted: #6B7280; /* Muted Meta Info Text */
}如果这个项目对你有帮助,欢迎请我喝杯咖啡:
If this project helps you, consider buying me a coffee:
建议金额 / Suggested amounts: ¥9.9 / ¥49 / ¥199
如果你使用 PayPal,可以通过以下链接赞助:
If you prefer PayPal, you can sponsor via:
https://paypal.me/aisystemnotes
建议金额 / Suggested amounts: $1.99 / $9.99 / $49.99
