annotate table-sort.py @ 0:402b58f45844 draft default tip

planemo upload commit 9cc4dc1db55299bf92ec6bd359161ece4592bd16-dirty
author jpayne
date Mon, 08 Dec 2025 15:03:06 +0000
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
402b58f45844 planemo upload commit 9cc4dc1db55299bf92ec6bd359161ece4592bd16-dirty
jpayne
parents:
diff changeset
1 #!/usr/bin/env python3
402b58f45844 planemo upload commit 9cc4dc1db55299bf92ec6bd359161ece4592bd16-dirty
jpayne
parents:
diff changeset
2
402b58f45844 planemo upload commit 9cc4dc1db55299bf92ec6bd359161ece4592bd16-dirty
jpayne
parents:
diff changeset
3 import csv
402b58f45844 planemo upload commit 9cc4dc1db55299bf92ec6bd359161ece4592bd16-dirty
jpayne
parents:
diff changeset
4 import sys
402b58f45844 planemo upload commit 9cc4dc1db55299bf92ec6bd359161ece4592bd16-dirty
jpayne
parents:
diff changeset
5
402b58f45844 planemo upload commit 9cc4dc1db55299bf92ec6bd359161ece4592bd16-dirty
jpayne
parents:
diff changeset
6
402b58f45844 planemo upload commit 9cc4dc1db55299bf92ec6bd359161ece4592bd16-dirty
jpayne
parents:
diff changeset
7 def main(headers):
402b58f45844 planemo upload commit 9cc4dc1db55299bf92ec6bd359161ece4592bd16-dirty
jpayne
parents:
diff changeset
8 rows = csv.DictReader(sys.stdin, delimiter="\t", dialect="excel-tab")
402b58f45844 planemo upload commit 9cc4dc1db55299bf92ec6bd359161ece4592bd16-dirty
jpayne
parents:
diff changeset
9
402b58f45844 planemo upload commit 9cc4dc1db55299bf92ec6bd359161ece4592bd16-dirty
jpayne
parents:
diff changeset
10 # More efficient header check using set intersection
402b58f45844 planemo upload commit 9cc4dc1db55299bf92ec6bd359161ece4592bd16-dirty
jpayne
parents:
diff changeset
11 if not set(headers).intersection(rows.fieldnames):
402b58f45844 planemo upload commit 9cc4dc1db55299bf92ec6bd359161ece4592bd16-dirty
jpayne
parents:
diff changeset
12 raise ValueError(
402b58f45844 planemo upload commit 9cc4dc1db55299bf92ec6bd359161ece4592bd16-dirty
jpayne
parents:
diff changeset
13 f"Couldn't find any of supplied headers ({', '.join(map(repr, headers))}) in the table."
402b58f45844 planemo upload commit 9cc4dc1db55299bf92ec6bd359161ece4592bd16-dirty
jpayne
parents:
diff changeset
14 )
402b58f45844 planemo upload commit 9cc4dc1db55299bf92ec6bd359161ece4592bd16-dirty
jpayne
parents:
diff changeset
15
402b58f45844 planemo upload commit 9cc4dc1db55299bf92ec6bd359161ece4592bd16-dirty
jpayne
parents:
diff changeset
16 # Optimized sorting using tuple comparison (generally faster than list comparison)
402b58f45844 planemo upload commit 9cc4dc1db55299bf92ec6bd359161ece4592bd16-dirty
jpayne
parents:
diff changeset
17 items = sorted(rows, key=lambda d: tuple(d.get(h, "") for h in headers))
402b58f45844 planemo upload commit 9cc4dc1db55299bf92ec6bd359161ece4592bd16-dirty
jpayne
parents:
diff changeset
18
402b58f45844 planemo upload commit 9cc4dc1db55299bf92ec6bd359161ece4592bd16-dirty
jpayne
parents:
diff changeset
19 wr = csv.DictWriter(sys.stdout, dialect="excel-tab", fieldnames=rows.fieldnames)
402b58f45844 planemo upload commit 9cc4dc1db55299bf92ec6bd359161ece4592bd16-dirty
jpayne
parents:
diff changeset
20 wr.writeheader()
402b58f45844 planemo upload commit 9cc4dc1db55299bf92ec6bd359161ece4592bd16-dirty
jpayne
parents:
diff changeset
21 wr.writerows(items)
402b58f45844 planemo upload commit 9cc4dc1db55299bf92ec6bd359161ece4592bd16-dirty
jpayne
parents:
diff changeset
22 # sys.stdout.flush()
402b58f45844 planemo upload commit 9cc4dc1db55299bf92ec6bd359161ece4592bd16-dirty
jpayne
parents:
diff changeset
23
402b58f45844 planemo upload commit 9cc4dc1db55299bf92ec6bd359161ece4592bd16-dirty
jpayne
parents:
diff changeset
24
402b58f45844 planemo upload commit 9cc4dc1db55299bf92ec6bd359161ece4592bd16-dirty
jpayne
parents:
diff changeset
25 if __name__ == "__main__":
402b58f45844 planemo upload commit 9cc4dc1db55299bf92ec6bd359161ece4592bd16-dirty
jpayne
parents:
diff changeset
26 main(sys.argv[1:])