export_platform_statistics_xlsx.py 4.4 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191
  1. #!/usr/bin/env python3
  2. # -*- coding: utf-8 -*-
  3. """
  4. 从 stdin 读取 JSON(平台统计列表),生成 xlsx 并输出到 stdout(二进制)。
  5. 输入 JSON 格式:
  6. {
  7. "platforms": [
  8. {
  9. "platform": "baijiahao",
  10. "viewsCount": 5,
  11. "commentsCount": 1,
  12. "likesCount": 0,
  13. "fansIncrease": 1,
  14. "updateTime": "2026-01-27 14:30:00"
  15. }
  16. ]
  17. }
  18. """
  19. import json
  20. import sys
  21. from io import BytesIO
  22. from datetime import datetime
  23. # Ensure stdin is read as UTF-8 (important on Windows when Node passes UTF-8 JSON)
  24. if sys.platform == "win32":
  25. import io as _io
  26. if hasattr(sys.stdin, "buffer"):
  27. sys.stdin = _io.TextIOWrapper(sys.stdin.buffer, encoding="utf-8", errors="replace")
  28. try:
  29. from openpyxl import Workbook
  30. from openpyxl.styles import Font, Alignment, PatternFill, Border, Side
  31. from openpyxl.utils import get_column_letter
  32. except Exception as e: # pragma: no cover
  33. sys.stderr.write(
  34. "Missing dependency: openpyxl. Please install it in your python env.\n"
  35. "Example: pip install -r server/python/requirements.txt\n"
  36. f"Detail: {e}\n"
  37. )
  38. sys.exit(3)
  39. HEADERS = [
  40. "平台",
  41. "阅读(播放量)",
  42. "评论量",
  43. "点赞量",
  44. "涨粉量",
  45. "更新时间",
  46. ]
  47. COL_WIDTHS = [12, 16, 12, 12, 10, 18]
  48. def _safe_int(v):
  49. try:
  50. if v is None or v == "":
  51. return None
  52. return int(float(v))
  53. except Exception:
  54. return None
  55. def _safe_str(v) -> str:
  56. """
  57. 将任意值转换为字符串,并过滤掉 Excel 不支持的非法 Unicode 代理字符。
  58. """
  59. if v is None:
  60. return ""
  61. try:
  62. s = str(v)
  63. except Exception:
  64. s = repr(v)
  65. return "".join(ch for ch in s if not (0xD800 <= ord(ch) <= 0xDFFF))
  66. def _format_date_only(value: str) -> str:
  67. """
  68. 将各种时间字符串格式化为 YYYY-MM-DD,仅保留日期部分。
  69. 如果无法解析,则尽量取前 10 位作为日期返回。
  70. """
  71. if not value:
  72. return ""
  73. s = str(value).strip()
  74. try:
  75. if s.endswith("Z"):
  76. s_clean = s[:-1]
  77. else:
  78. s_clean = s
  79. s_clean = s_clean.replace(" ", "T")
  80. dt = datetime.fromisoformat(s_clean)
  81. return dt.strftime("%Y-%m-%d")
  82. except Exception:
  83. pass
  84. if len(s) >= 10:
  85. return s[:10]
  86. return s
  87. def build_xlsx(platforms):
  88. platform_name_map = {
  89. "douyin": "抖音",
  90. "baijiahao": "百家号",
  91. "weixin_video": "视频号",
  92. "xiaohongshu": "小红书",
  93. "bilibili": "哔哩哔哩",
  94. "kuaishou": "快手",
  95. }
  96. wb = Workbook()
  97. ws = wb.active
  98. ws.title = "平台数据"
  99. ws.append(HEADERS)
  100. header_font = Font(bold=True)
  101. header_fill = PatternFill("solid", fgColor="F2F2F2")
  102. center = Alignment(horizontal="center", vertical="center", wrap_text=False)
  103. left = Alignment(horizontal="left", vertical="center", wrap_text=False)
  104. thin = Side(style="thin", color="D9D9D9")
  105. border = Border(left=thin, right=thin, top=thin, bottom=thin)
  106. for col_idx in range(1, len(HEADERS) + 1):
  107. cell = ws.cell(row=1, column=col_idx)
  108. cell.font = header_font
  109. cell.fill = header_fill
  110. cell.alignment = center
  111. cell.border = border
  112. for i, w in enumerate(COL_WIDTHS, start=1):
  113. col_letter = get_column_letter(i)
  114. ws.column_dimensions[col_letter].width = w
  115. for p in platforms:
  116. platform_raw = (p.get("platform") or "").strip()
  117. platform_cn = platform_name_map.get(platform_raw, platform_raw)
  118. ws.append([
  119. _safe_str(platform_cn),
  120. _safe_int(p.get("viewsCount")) or 0,
  121. _safe_int(p.get("commentsCount")) or 0,
  122. _safe_int(p.get("likesCount")) or 0,
  123. _safe_int(p.get("fansIncrease")) or 0,
  124. _safe_str(_format_date_only(p.get("updateTime"))),
  125. ])
  126. int_cols = {"B", "C", "D", "E"}
  127. for row in range(2, ws.max_row + 1):
  128. for col in range(1, len(HEADERS) + 1):
  129. c = ws.cell(row=row, column=col)
  130. c.border = border
  131. if col == 1:
  132. c.alignment = left
  133. else:
  134. c.alignment = center
  135. for c_letter in int_cols:
  136. c = ws[f"{c_letter}{row}"]
  137. if c.value is not None:
  138. c.number_format = "0"
  139. ws.freeze_panes = "A2"
  140. bio = BytesIO()
  141. wb.save(bio)
  142. return bio.getvalue()
  143. def main():
  144. try:
  145. raw = sys.stdin.read()
  146. payload = json.loads(raw) if raw.strip() else {}
  147. except Exception as e:
  148. sys.stderr.write(f"Invalid JSON input: {e}\n")
  149. sys.exit(2)
  150. platforms = payload.get("platforms") or []
  151. xlsx_bytes = build_xlsx(platforms)
  152. sys.stdout.buffer.write(xlsx_bytes)
  153. if __name__ == "__main__":
  154. main()