清理城市农业类别 + 修复采集器LLM直接选题

清理:
- sustainability_cases.json移除GLO-001/CHN-001城市农业案例,替换为循环消费
- initial_cases.json移除case8东京垂直农场/case26城市屋顶农场
- strategy_topics_to_json.py移除B01/B05/D05三个种菜选题
- collector.py移除城市农业→循环消费映射,更新注释
- 删除fix_collector.py/test_image_gen.py/generate_images.py等遗留脚本
- 删除import_topics.py和automation/下旧版生成脚本

修复:
- collector.py _generate_topic_with_llm不再依赖搜索结果,无搜索时LLM直接生成
- run()始终调用LLM,不再要求web_search_results非空
- 替换sources.yaml中已失效的RSS源(澎湃/虎嗅/中新网→36氪/少数派)
This commit is contained in:
Yuzhiran Dev
2026-05-21 08:02:08 +08:00
parent c8bee712d7
commit 10996ce6ce
16 changed files with 47 additions and 980 deletions
-24
View File
@@ -1,24 +0,0 @@
#!/usr/bin/env python3
"""
在 Markdown 的 H2 标题前插入分隔线,第一个除外
"""
MD_PATH = "/root/openclaw-workspace/projects/yu-zhi-ran/content/published/2026-04-14-上海阳台种菜一年/final-article.md"
with open(MD_PATH, "r", encoding="utf-8") as f:
lines = f.readlines()
new_lines = []
first_h2_seen = False
for line in lines:
if line.startswith("## "):
if first_h2_seen:
new_lines.append("---\n\n")
else:
first_h2_seen = True
new_lines.append(line)
with open(MD_PATH, "w", encoding="utf-8") as f:
f.writelines(new_lines)
print(f"✅ 已处理 {MD_PATH}")
-39
View File
@@ -1,39 +0,0 @@
#!/usr/bin/env python3
"""
在章节标题(h2)前插入分隔线,第一个除外
"""
import re
HTML_PATH = "/root/openclaw-workspace/projects/yu-zhi-ran/content/published/2026-04-14-上海阳台种菜一年/article-optimized.html"
with open(HTML_PATH, "r", encoding="utf-8") as f:
html = f.read()
# 分隔线HTML
separator = '<div class="chapter-separator" style="margin: 40px 0 20px; border-top: 2px dashed #e0e0e0;"></div>\n'
# 找到所有 h2 标题
h2_pattern = re.compile(r'(<h2>.*?</h2>)', re.DOTALL)
matches = list(h2_pattern.finditer(html))
# 跳过第一个 h2,对其余每个插入分隔
insertions = []
for i, m in enumerate(matches[1:], start=1): # 从第二个开始
insert_pos = m.start()
insertions.append((insert_pos, separator))
# 按位置逆序插入,避免影响后续位置
insertions.sort(reverse=True, key=lambda x: x[0])
html_list = list(html)
for pos, sep in insertions:
html_list.insert(pos, sep)
new_html = ''.join(html_list)
# 写回
with open(HTML_PATH, "w", encoding="utf-8") as f:
f.write(new_html)
print(f"✅ 已插入 {len(insertions)} 个章节分隔")
print(f"📄 文件: {HTML_PATH}")
-24
View File
@@ -83,18 +83,6 @@
"credibility_rating": "⭐⭐⭐⭐",
"china_applicability": "⭐⭐⭐⭐"
},
{
"id": 8,
"title": "东京垂直农场",
"field": "可持续生活系统",
"summary": "利用高层建筑内部空间进行多层种植,实现都市粮食自给。",
"key_metrics": "单位面积产量是传统农业的10倍,节水90%",
"date": "2024",
"source": "Spread 公司",
"source_url": "https://www.spread.co.jp/",
"credibility_rating": "⭐⭐⭐⭐⭐",
"china_applicability": "⭐⭐⭐"
},
{
"id": 9,
"title": "纽约社区花园政策",
@@ -299,18 +287,6 @@
"credibility_rating": "⭐⭐⭐",
"china_applicability": "⭐⭐⭐⭐⭐"
},
{
"id": 26,
"title": "城市屋顶农场:上海社区的粮食自给实验",
"field": "可持续生活系统",
"summary": "在上海某小区屋顶建设 200㎡ 农场,一年内生产 600kg 蔬菜,减少碳足迹 1.2 吨。",
"key_metrics": "蔬菜自给率 40%, 参与家庭 50 户, 社区互动提升 300%",
"date": "2024",
"source": "城市农业网",
"source_url": "https://www.urbanfarming.org/",
"credibility_rating": "⭐⭐⭐⭐",
"china_applicability": "⭐⭐⭐⭐⭐"
},
{
"id": 27,
"title": "数字游民签证地图:2026 最新政策对比",
+11 -28
View File
@@ -1,21 +1,4 @@
[
{
"id": "GLO-001",
"country": "Japan",
"category": "城市农业",
"title": "东京垂直农场:10平米 balcony 年产蔬菜 100kg",
"core_idea": "利用多层种植架和 LED 生长灯,在狭小阳台实现全年蔬菜生产",
"data_facts": "每平米年产 10kg,较传统方式节水 90%,投资回收期 1.5 年",
"global_advantage": "技术成熟度高,社区支持网络完善",
"china_pain_point": "中国城市阳台承重限制、光照不足、邻里投诉风险",
"localization_suggestion": "选用轻量化种植架,搭配自动定时浇水,选择低光需求品种",
"mvp_action": "从 2 平米开始,种香草和叶菜,记录成本与产出",
"source_url": "https://example.com/tokyo-vertical-farm",
"credibility_rating": "⭐⭐⭐⭐⭐",
"china_applicability": "⭐⭐⭐",
"collection_date": "2026-04-19",
"status": "已验证"
},
{
"id": "GLO-002",
"country": "Sweden",
@@ -87,18 +70,18 @@
{
"id": "CHN-001",
"country": "China",
"category": "城市农业",
"title": "上海阳台种菜年省 3000 元:居民自种调查",
"core_idea": "利用阳台空间种菜,实现部分蔬菜自给,降低生活成本",
"data_facts": "20 平米阳台年产蔬菜 100kg,节省买菜支出 3000 元,投入成本 2000 元",
"global_advantage": "中国城市人口密集,阳台空间普遍存在",
"china_pain_point": "缺乏种植知识,病虫害防治困难,物业可能干涉",
"localization_suggestion": "选择易种品种(番茄、辣椒、生菜),使用有机土,与邻居共享收获",
"mvp_action": "先种 5 盆香草,成功后再扩大",
"source_url": "https://www.bilibili.com/video/BV1xx411",
"credibility_rating": "⭐⭐⭐",
"category": "循环消费",
"title": "中国二手交易平台崛起:闲鱼转转让闲置物品年交易额超5000亿",
"core_idea": "通过二手交易平台,用户可以将闲置物品变现,降低消费成本",
"data_facts": "闲鱼年交易额超5000亿,用户数超3亿,每天上架商品超200万件",
"global_advantage": "中国移动互联网普及率高,二手交易习惯逐渐养成",
"china_pain_point": "信任机制不完善,假货和退换货纠纷多",
"localization_suggestion": "选择信誉高的卖家,优先购买有质检服务的商品",
"mvp_action": "整理家中闲置物品,本月在二手平台卖出3件",
"source_url": "https://www.goofish.com/",
"credibility_rating": "⭐⭐⭐",
"china_applicability": "⭐⭐⭐⭐⭐",
"collection_date": "2026-04-19",
"collection_date": "2026-05-20",
"status": "已验证"
}
]
-138
View File
@@ -1,138 +0,0 @@
#!/usr/bin/env python3
"""
将 Markdown 文章转换为 Word 文档,嵌入图片
"""
import os
import re
from docx import Document
from docx.shared import Inches, Pt, RGBColor
from docx.enum.text import WD_ALIGN_PARAGRAPH
# 路径配置
MARKDOWN_FILE = "/root/openclaw-workspace/projects/yu-zhi-ran/content/published/2026-04-14-上海阳台种菜一年/final-article.md"
IMAGES_DIR = "/root/openclaw-workspace/projects/yu-zhi-ran/content/publishing/images"
OUTPUT_DOCX = "/root/openclaw-workspace/projects/yu-zhi-ran/content/published/2026-04-14-上海阳台种菜一年/上海阳台种菜一年_最终版.docx"
# 读取 Markdown
with open(MARKDOWN_FILE, "r", encoding="utf-8") as f:
lines = f.readlines()
doc = Document()
doc.styles['Normal'].font.name = '微软雅黑'
doc.styles['Normal'].font.size = Pt(11)
# 样式函数
def add_heading(text, level=1):
heading = doc.add_heading(text, level=level)
heading.alignment = WD_ALIGN_PARAGRAPH.LEFT
return heading
def add_paragraph(text, bold=False, italic=False):
p = doc.add_paragraph()
run = p.add_run(text)
run.bold = bold
run.italic = italic
return p
# 解析 Markdown
in_code_block = False
in_table = False
table_data = []
for i, line in enumerate(lines):
line = line.rstrip('\n')
# 代码块跳过
if line.startswith('```'):
in_code_block = not in_code_block
continue
if in_code_block:
continue
# 标题
if line.startswith('# '):
add_heading(line[2:], level=1)
continue
if line.startswith('## '):
add_heading(line[3:], level=2)
continue
if line.startswith('### '):
add_heading(line[4:], level=3)
continue
# 表格处理(简化:将表格转为文本,图片位置用占位)
if line.startswith('|'):
in_table = True
table_data.append(line)
continue
if in_table and not line.startswith('|'):
in_table = False
# 可以在此转换表格,为简化直接跳过
continue
# 图片:![alt](path)
img_match = re.match(r'!\[(.*?)\]\((images/.*?)\)', line)
if img_match:
alt, path = img_match.groups()
img_full_path = os.path.join(os.path.dirname(MARKDOWN_FILE), path)
if os.path.exists(img_full_path):
try:
# 插入图片,宽度 6 英寸(约 15cm)
doc.add_picture(img_full_path, width=Inches(6))
# 居中
last_para = doc.paragraphs[-1]
last_para.alignment = WD_ALIGN_PARAGRAPH.CENTER
# 添加图片说明(可选)
if alt:
cap = doc.add_paragraph(alt)
cap.alignment = WD_ALIGN_PARAGRAPH.CENTER
cap.style = 'Caption'
except Exception as e:
doc.add_paragraph(f"[图片加载失败: {path}]")
else:
doc.add_paragraph(f"[图片缺失: {img_full_path}]")
continue
# 引用
if line.startswith('> '):
p = doc.add_paragraph(line[2:])
p.paragraph_format.left_indent = Inches(0.5)
p.italic = True
continue
# 列表
if re.match(r'^[-*] ', line):
p = doc.add_paragraph(line[2:], style='List Bullet')
continue
if re.match(r'^\d+\. ', line):
p = doc.add_paragraph(line[line.find('.')+2:], style='List Number')
continue
# 分隔线
if line.strip() == '---':
doc.add_paragraph('_' * 50)
continue
# 普通段落
if line.strip():
# 处理行内加粗、斜体
p = doc.add_paragraph()
parts = re.split(r'(\*\*[^*]+\*\*|\*[^*]+\*)', line)
for part in parts:
if part.startswith('**') and part.endswith('**'):
run = p.add_run(part[2:-2])
run.bold = True
elif part.startswith('*') and part.endswith('*'):
run = p.add_run(part[1:-1])
run.italic = True
else:
run = p.add_run(part)
else:
doc.add_paragraph() # 空行
# 保存文档
doc.save(OUTPUT_DOCX)
print(f"✅ Word 文档已生成: {OUTPUT_DOCX}")
print(f"📄 页数: {len(doc.paragraphs)} 段落")
print(f"🖼️ 图片路径: {IMAGES_DIR}")
-93
View File
@@ -1,93 +0,0 @@
#!/usr/bin/env python3
"""
生成 Word 文档,图片从发布目录的 images 文件夹读取
"""
import os
from docx import Document
from docx.shared import Inches, Pt
from docx.enum.text import WD_ALIGN_PARAGRAPH
BASE_DIR = "/root/openclaw-workspace/projects/yu-zhi-ran/content/published/2026-04-14-上海阳台种菜一年"
MD_FILE = os.path.join(BASE_DIR, "final-article.md")
IMAGES_DIR = os.path.join(BASE_DIR, "images") # 已复制的图片
OUTPUT_DOCX = os.path.join(BASE_DIR, "上海阳台种菜一年_最终版.docx")
with open(MD_FILE, "r", encoding="utf-8") as f:
lines = f.readlines()
doc = Document()
doc.styles['Normal'].font.name = '微软雅黑'
doc.styles['Normal'].font.size = Pt(11)
def add_heading(text, level=1):
heading = doc.add_heading(text, level=level)
heading.alignment = WD_ALIGN_PARAGRAPH.LEFT
return heading
for line in lines:
line = line.rstrip('\n')
if line.startswith('# '):
add_heading(line[2:], level=1)
continue
if line.startswith('## '):
add_heading(line[3:], level=2)
continue
if line.startswith('### '):
add_heading(line[4:], level=3)
continue
if line.startswith('---'):
doc.add_paragraph('_' * 60)
continue
# 图片
if line.startswith('!['):
import re
m = re.match(r'!\[(.*?)\]\((images/.*?)\)', line)
if m:
alt, fname = m.groups()
img_path = os.path.join(IMAGES_DIR, os.path.basename(fname))
if os.path.exists(img_path):
try:
doc.add_picture(img_path, width=Inches(6))
last_para = doc.paragraphs[-1]
last_para.alignment = WD_ALIGN_PARAGRAPH.CENTER
except Exception as e:
doc.add_paragraph(f"[图片错误: {fname}]")
else:
doc.add_paragraph(f"[缺失图片: {fname}]")
continue
# 空行
if not line.strip():
doc.add_paragraph()
continue
# 普通段落,处理粗体斜体
p = doc.add_paragraph()
parts = []
tmp = line
while '**' in tmp:
parts.append(tmp[:tmp.find('**')])
tmp = tmp[tmp.find('**')+2:]
if '**' in tmp:
parts.append(('bold', tmp[:tmp.find('**')]))
tmp = tmp[tmp.find('**')+2:]
else:
parts.append(('bold', tmp))
break
if not parts:
parts = [line]
for part in parts:
if isinstance(part, tuple):
style, text = part
run = p.add_run(text)
run.bold = (style == 'bold')
else:
p.add_run(part)
doc.save(OUTPUT_DOCX)
print(f"✅ Word 已生成: {OUTPUT_DOCX}")
print(f"📄 段落数: {len(doc.paragraphs)}")
-74
View File
@@ -1,74 +0,0 @@
#!/usr/bin/env python3
"""
生成 HTML,图片使用相对路径 'images/xxx.png'(确保图片在发布目录的 images 子文件夹中)
"""
import os
import re
BASE_DIR = "/root/openclaw-workspace/projects/yu-zhi-ran/content/published/2026-04-14-上海阳台种菜一年"
MD_FILE = os.path.join(BASE_DIR, "final-article.md")
OUT_HTML = os.path.join(BASE_DIR, "上海阳台种菜一年_可复制.html")
with open(MD_FILE, "r", encoding="utf-8") as f:
content = f.read()
# 替换图片为 HTML img 标签,保持相对路径
def replace_img(match):
alt, path = match.groups()
return f'<img src="{path}" alt="{alt}" style="max-width:100%; margin:20px 0; display:block;">'
content = re.sub(r'!\[(.*?)\]\((images/.*?)\)', replace_img, content)
# 转换 Markdown 为 HTML
html_lines = []
for line in content.split('\n'):
if line.startswith('# '):
html_lines.append(f'<h1>{line[2:]}</h1>')
elif line.startswith('## '):
html_lines.append(f'<h2>{line[3:]}</h2>')
elif line.startswith('### '):
html_lines.append(f'<h3>{line[4:]}</h3>')
elif line.startswith('---'):
html_lines.append('<hr style="border:none;border-top:2px dashed #ddd;margin:40px 0;">')
elif line.startswith('> '):
html_lines.append(f'<blockquote style="border-left:4px solid #4CAF50;background:#f9f9f9;padding:10px 20px;margin:20px 0;color:#666;">{line[2:]}</blockquote>')
elif re.match(r'^[-*] ', line):
html_lines.append(f'<li>{line[2:]}</li>')
elif re.match(r'^\d+\. ', line):
html_lines.append(f'<li>{line[line.find(". ")+2:]}</li>')
elif line.strip() == '':
html_lines.append('<br>')
else:
# 处理行内粗体斜体
tmp = re.sub(r'\*\*(.*?)\*\*', r'<strong>\1</strong>', line)
tmp = re.sub(r'\*(.*?)\*', r'<em>\1</em>', tmp)
html_lines.append(f'<p>{tmp}</p>')
html = f'''<!DOCTYPE html>
<html>
<head>
<meta charset="UTF-8">
<title>上海阳台种菜一年</title>
<style>
body {{ font-family: "Microsoft YaHei", sans-serif; max-width: 800px; margin: 0 auto; padding: 20px; line-height: 1.8; }}
h1 {{ font-size: 28px; border-bottom: 2px solid #4CAF50; padding-bottom: 10px; }}
h2 {{ font-size: 24px; margin-top: 40px; border-left: 4px solid #4CAF50; padding-left: 10px; }}
h3 {{ font-size: 20px; margin-top: 30px; color: #666; }}
img {{ max-width: 100%; height: auto; border-radius: 4px; margin: 20px 0; display: block; margin-left: auto; margin-right: auto; }}
blockquote {{ border-left: 4px solid #4CAF50; background: #f9f9f9; padding: 10px 20px; margin: 20px 0; color: #666; }}
li {{ margin-bottom: 8px; }}
p {{ margin-bottom: 16px; }}
</style>
</head>
<body>
{chr(10).join(html_lines)}
</body>
</html>'''
with open(OUT_HTML, "w", encoding="utf-8") as f:
f.write(html)
print(f"✅ HTML 已生成: {OUT_HTML}")
print(f"📊 字符数: {len(content)}")
print(f"🖼️ 图片路径: images/ (需与 HTML 同目录的 images 文件夹)")
-90
View File
@@ -1,90 +0,0 @@
#!/usr/bin/env python3
"""
生成图文混排的 HTML(图片内联为 base64),方便直接复制
"""
import os
import re
import base64
BASE_DIR = "/root/openclaw-workspace/projects/yu-zhi-ran/content/published/2026-04-14-上海阳台种菜一年"
MD_FILE = os.path.join(BASE_DIR, "final-article.md")
IMAGES_DIR = os.path.join(BASE_DIR, "images") # 使用发布目录内的 images
OUT_HTML = os.path.join(BASE_DIR, "上海阳台种菜一年_内联.html")
# 读取 Markdown
with open(MD_FILE, "r", encoding="utf-8") as f:
content = f.read()
# 预加载图片并转为 base64
image_cache = {}
for fname in os.listdir(IMAGES_DIR):
if fname.endswith('.png'):
path = os.path.join(IMAGES_DIR, fname)
with open(path, "rb") as imgf:
b64 = base64.b64encode(imgf.read()).decode('utf-8')
image_cache[fname] = b64
# 替换图片
def replace_img(match):
alt = match.group(1)
fname = match.group(2)
key = os.path.basename(fname)
if key in image_cache:
return f'<img src="data:image/png;base64,{image_cache[key]}" alt="{alt}" style="max-width:100%; margin:20px 0; display:block;">'
else:
return f'<p>[图片缺失: {fname}]</p>'
content = re.sub(r'!\[(.*?)\]\((images/.*?)\)', replace_img, content)
# Markdown 转 HTML
html_lines = []
for line in content.split('\n'):
if line.startswith('# '):
html_lines.append(f'<h1>{line[2:]}</h1>')
elif line.startswith('## '):
html_lines.append(f'<h2>{line[3:]}</h2>')
elif line.startswith('### '):
html_lines.append(f'<h3>{line[4:]}</h3>')
elif line.startswith('---'):
html_lines.append('<hr>')
elif line.startswith('> '):
html_lines.append(f'<blockquote>{line[2:]}</blockquote>')
elif re.match(r'^[-*] ', line):
html_lines.append(f'<li>{line[2:]}</li>')
elif re.match(r'^\d+\. ', line):
html_lines.append(f'<li>{line[line.find(". ")+2:]}</li>')
elif line.strip() == '':
html_lines.append('<br>')
else:
tmp = re.sub(r'\*\*(.*?)\*\*', r'<strong>\1</strong>', line)
tmp = re.sub(r'\*(.*?)\*', r'<em>\1</em>', tmp)
html_lines.append(f'<p>{tmp}</p>')
html = f'''<!DOCTYPE html>
<html>
<head>
<meta charset="UTF-8">
<title>上海阳台种菜一年</title>
<style>
body {{ font-family: "Microsoft YaHei", sans-serif; max-width: 800px; margin: 0 auto; padding: 20px; line-height: 1.8; }}
h1 {{ font-size: 28px; border-bottom: 2px solid #4CAF50; padding-bottom: 10px; }}
h2 {{ font-size: 24px; margin-top: 40px; border-left: 4px solid #4CAF50; padding-left: 10px; }}
h3 {{ font-size: 20px; margin-top: 30px; color: #666; }}
img {{ max-width: 100%; height: auto; border-radius: 4px; margin: 20px 0; display: block; margin-left: auto; margin-right: auto; }}
blockquote {{ border-left: 4px solid #4CAF50; background: #f9f9f9; padding: 10px 20px; margin: 20px 0; color: #666; }}
li {{ margin-bottom: 8px; }}
p {{ margin-bottom: 16px; }}
</style>
</head>
<body>
{chr(10).join(html_lines)}
</body>
</html>'''
with open(OUT_HTML, "w", encoding="utf-8") as f:
f.write(html)
print(f"✅ HTML 已生成: {OUT_HTML}")
print(f"📊 字符数: {len(content)}")
print(f"🖼️ 内嵌图片: {len(image_cache)}")
-50
View File
@@ -1,50 +0,0 @@
#!/usr/bin/env python3
"""
对 article.md 进行内容优化:
- 去除具体 App 品牌名(花帮主、园艺助手)
- 隐去设备具体品牌(小米米家)
- 保留功能描述和用户价值
- 保持中立、实用、无广告感
"""
import os
import re
BASE_DIR = "/root/openclaw-workspace/projects/yu-zhi-ran/content/published/2026-04-14-上海阳台种菜一年"
MD_FILE = os.path.join(BASE_DIR, "final-article.md")
OUT_MD = os.path.join(BASE_DIR, "final-article-optimized.md")
with open(MD_FILE, "r", encoding="utf-8") as f:
content = f.read()
# 1. 替换具体 App 名称 -> 通用描述
content = re.sub(r'花帮主', '一些第三方种植App', content)
content = re.sub(r'园艺助手', '另一些生活助手类App', content)
content = re.sub(r'(\*\*)花帮主(\*\*)AI识别病虫害,准确率85%', '**一些第三方种植App**,可以通过 AI 识别病虫害,准确率在 80% 以上', content)
# 2. 替换设备品牌 -> 通用描述
content = re.sub(r'小米米家灌溉套装', '智能灌溉套装', content)
content = re.sub(r'LED补光灯', 'LED 植物补光灯', content)
# 3. 移除可能带有广告嫌疑的表述(如“效果最好”、“推荐”等),改为中性描述
content = re.sub(r'强烈推荐(易种)', '适合新手(易种)', content)
content = re.sub(r'强烈推荐', '推荐', content)
# 4. 图片描述调整(不影响图片本身,只调整 alt 文本和图片说明)
# 图片文件保留不变,只调整 Markdown 中的说明文字
content = re.sub(r'!\[App截图\]', '[App功能截图]', content)
content = re.sub(r'App截图', 'App功能界面示意', content)
# 5. 增加免责声明(在文末)
if "声明:" not in content:
content = content.rstrip() + "\n\n---\n\n> **声明**:本文提及的工具和设备仅为个人使用经验分享,不构成商业推荐。读者可根据自身需求选择类似产品。\n"
with open(OUT_MD, "w", encoding="utf-8") as f:
f.write(content)
print(f"✅ 优化完成: {OUT_MD}")
print("🔧 优化项:")
print(" - 去除具体 App 品牌名")
print(" - 隐去设备品牌")
print(" - 增加中立表述")
print(" - 添加免责声明")