select_cohort.py

skills/partner-pipeline-push/scripts/select_cohort.py

1. #!/usr/bin/env python3
2. """select_cohort.py: pick a cohort from the vault's client stubs by tier fields
3. and emit the cohort mapping XML.
5. Usage:
6. python3 select_cohort.py \
7. --vault "/path/to/03-Clients/client-stubs" \
8. --user-group Retail \
9. --user-group-correct "Palladium Partner" \
10. --pipeline partner_outreach \
11. --out cohort-mapping.xml
13. Match rule: a stub is in the cohort if its frontmatter user_group == --user-group
14. (when given) AND user_group_correct == --user-group-correct (when given). Either
15. filter may be omitted. Prints the matching stubs and writes the XML.
17. Notes
18. -----
19. * Stubs live flat in `03-Clients/client-stubs/`, one .md per client.
20. * `user_group_correct` is the computed, true tier and is the one to filter on.
21. Raw `user_group` is blank on most stubs.
22. * Tier values are the full labels "Palladium Partner", "Platinum Partner", NOT the
23. bare tier ("Palladium"). The Bitrix User Group enum matches these.
24. * The output is a THROWAWAY working file for one push. It is NOT the vault census
25. (`03-Clients/partner-client-master-list.xml`). Never write one over the other.
26. * `bitrix_id` is copied through from the stub. Where populated it is a verified
27. Bitrix contact id, but coverage is partial: link contacts by EMAIL, not by this.
28. """
29. import argparse, os, re, html
31. STUB_DIR_RELATIVE = "/03-Clients/client-stubs"
34. def fm(text):
35. m = re.match(r'^---\n(.*?)\n---', text, re.S)
36. return m.group(1) if m else ''
39. def g(f, k):
40. m = re.search(r'(?m)^' + re.escape(k) + r':\s*(.*)$', f)
41. return (m.group(1).strip().strip('"').strip("'").strip() if m else '')
44. def main():
45. ap = argparse.ArgumentParser()
46. ap.add_argument("--vault", required=True,
47. help="path to 03-Clients/client-stubs")
48. ap.add_argument("--user-group", default="")
49. ap.add_argument("--user-group-correct", default="",
50. help='e.g. "Palladium Partner" (full label, not the bare tier)')
51. ap.add_argument("--pipeline", default="partner_outreach",
52. help="vault pipeline key: partner_outreach | quarterly_checkin | winback")
53. ap.add_argument("--out", default="cohort-mapping.xml")
54. a = ap.parse_args()
56. rows = []
57. for fn in sorted(os.listdir(a.vault)):
58. if not fn.endswith('.md') or fn.startswith('CLAUDE'):
59. continue
60. with open(os.path.join(a.vault, fn), encoding='utf-8') as fh:
61. F = fm(fh.read())
62. ug, ugc = g(F, 'user_group'), g(F, 'user_group_correct')
63. if a.user_group and ug.lower() != a.user_group.lower():
64. continue
65. if a.user_group_correct and ugc.lower() != a.user_group_correct.lower():
66. continue
67. rows.append((fn, g(F, 'bitrix_id'), g(F, 'first_name'),
68. g(F, 'last_name'), g(F, 'email')))
70. blocks = []
71. for fn, bid, first, last, email in rows:
72. blocks.append(
73. f' <client bitrix_id="{html.escape(bid)}">\n'
74. f' <first_name>{html.escape(first)}</first_name>\n'
75. f' <last_name>{html.escape(last)}</last_name>\n'
76. f' <outreach_pipeline>{html.escape(a.pipeline)}</outreach_pipeline>\n'
77. f' <stub_file_name>{html.escape(fn)}</stub_file_name>\n'
78. f' <relative_path>{STUB_DIR_RELATIVE}/{html.escape(fn)}</relative_path>\n'
79. f' </client>')
80. doc = ('<?xml version="1.0" encoding="UTF-8"?>\n<cohort_mapping>\n'
81. ' <client_list>\n' + "\n\n".join(blocks) +
82. '\n </client_list>\n</cohort_mapping>\n')
83. with open(a.out, "w", encoding='utf-8') as fh:
84. fh.write(doc)
86. print(f"cohort: {len(rows)} clients "
87. f"(user_group={a.user_group or '*'}, "
88. f"user_group_correct={a.user_group_correct or '*'})")
89. for fn, bid, first, last, email in rows:
90. print(f" - {first} {last} <{email}> [{fn}]")
91. print(f"\nwrote {a.out} (cohort working file, NOT the vault census)")
94. if __name__ == "__main__":
95. main()