Mercurial > repos > guerler > dbkit

annotate dbkit_merge.py @ 2:81c7d4668a7e draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
author guerler
date Wed, 16 Dec 2020 12:02:48 +0000
parents
children 03e124ff7e26
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 from os import system
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
4 from os.path import getsize
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
5
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
6 from dbkit_package.DBKit import DBKit
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
7
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
8
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
9 def main(args):
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
10 logFile = open(args.log, "w")
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
11 outIndex = args.outindex
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
12 outData = args.outdata
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
13 if getsize(args.firstindex) > getsize(args.secondindex):
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
14 firstIndex = args.firstindex
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
15 firstData = args.firstdata
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
16 secondIndex = args.secondindex
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
17 secondData = args.seconddata
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
18 else:
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
19 firstIndex = args.secondindex
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
20 firstData = args.seconddata
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
21 secondIndex = args.firstindex
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
22 secondData = args.firstdata
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
23 system("cp %s %s" % (firstIndex, outIndex))
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
24 system("cp %s %s" % (firstData, outData))
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
25 firstEntries = set()
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
26 with open(firstIndex, "r") as f:
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
27 for line in f:
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
28 name = line.split()[0]
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
29 firstEntries.add(name)
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
30 logFile.write("Detected %s entries.\n" % len(firstEntries))
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
31 secondEntries = list()
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
32 with open(secondIndex, "r") as f:
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
33 for line in f:
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
34 name = line.split()[0]
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
35 secondEntries.append(name)
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
36 tempFile = "temp.dat"
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
37 count = 0
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
38 dbkit = DBKit(secondIndex, secondData)
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
39 for secondKey in secondEntries:
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
40 if secondKey not in firstEntries:
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
41 dbkit.createFile(secondKey, tempFile)
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
42 entrySize = getsize(tempFile)
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
43 currentSize = getsize(outData)
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
44 system("cat %s >> %s" % (tempFile, outData))
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
45 system("echo '%s\t%s\t%s' >> %s" % (secondKey, currentSize, entrySize, outIndex))
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
46 count = count + 1
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
47 else:
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
48 logFile.write("Skipping existing entry %s.\n" % secondKey)
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
49 logFile.write("Added %s entries.\n" % count)
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
50 logFile.close()
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
51
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
52
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
53 if __name__ == "__main__":
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
54 parser = argparse.ArgumentParser(description='DBKit - Merge database pair.')
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
55 parser.add_argument('-fi', '--firstindex', help='First Index file', required=True)
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
56 parser.add_argument('-fd', '--firstdata', help='First Data file', required=True)
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
57 parser.add_argument('-si', '--secondindex', help='Second Index file', required=True)
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
58 parser.add_argument('-sd', '--seconddata', help='Second Data file', required=True)
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
59 parser.add_argument('-oi', '--outindex', help='Output Index file', required=True)
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
60 parser.add_argument('-od', '--outdata', help='Output Data file', required=True)
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
61 parser.add_argument('-log', '--log', help='Log file', required=True)
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
62 args = parser.parse_args()
81c7d4668a7e "planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff changeset
63 main(args)