annotate dbkit_extract.py @ 3:03e124ff7e26 draft

"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
author guerler
date Wed, 16 Dec 2020 13:11:35 +0000
parents 81c7d4668a7e
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
2
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
1 #! /usr/bin/env python3
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
2 import argparse
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
3
3
03e124ff7e26 "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents: 2
diff changeset
4 from dbkit_package.DBKit import DBKit, writeEntry
2
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
5
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
6
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
7 def main(args):
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
8 logFile = open(args.log, "w")
3
03e124ff7e26 "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents: 2
diff changeset
9 outputIndex = args.outputindex
03e124ff7e26 "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents: 2
diff changeset
10 outputDatabase = args.outputdatabase
2
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
11 entries = list()
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
12 with open(args.list, "r") as f:
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
13 for line in f:
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
14 name = line.split()[0]
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
15 entries.append(name)
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
16 logFile.write("Detected %s entries.\n" % len(entries))
3
03e124ff7e26 "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents: 2
diff changeset
17 fileName = "temp.dat"
2
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
18 count = 0
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
19 dbkit = DBKit(args.index, args.database)
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
20 for entry in sorted(entries):
3
03e124ff7e26 "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents: 2
diff changeset
21 success = dbkit.createFile(entry, fileName)
2
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
22 if success:
3
03e124ff7e26 "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents: 2
diff changeset
23 writeEntry(entry, fileName, outputIndex, outputDatabase)
2
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
24 count = count + 1
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
25 else:
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
26 logFile.write("Entry %s not found.\n" % entry)
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
27 logFile.write("Extracted %s entries.\n" % count)
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
28 logFile.close()
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
29
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
30
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
31 if __name__ == "__main__":
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
32 parser = argparse.ArgumentParser(description='DBKit - Merge database pair.')
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
33 parser.add_argument('-l', '--list', help='List of entries to be extracted', required=True)
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
34 parser.add_argument('-i', '--index', help='Database Index file (ffindex)', required=True)
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
35 parser.add_argument('-d', '--database', help='Database Data file (ffdata)', required=True)
3
03e124ff7e26 "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents: 2
diff changeset
36 parser.add_argument('-oi', '--outputindex', help='Output Index file', required=True)
03e124ff7e26 "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents: 2
diff changeset
37 parser.add_argument('-od', '--outputdatabase', help='Output Data file', required=True)
2
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
38 parser.add_argument('-g', '--log', help='Log file', required=True)
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
39 args = parser.parse_args()
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
40 main(args)