Mercurial > repos > guerler > dbkit
annotate dbkit_package/DBKit.py @ 3:03e124ff7e26 draft
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
author | guerler |
---|---|
date | Wed, 16 Dec 2020 13:11:35 +0000 |
parents | 81c7d4668a7e |
children |
rev | line source |
---|---|
3
03e124ff7e26
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
2
diff
changeset
|
1 from os.path import isfile, getsize |
03e124ff7e26
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
2
diff
changeset
|
2 |
03e124ff7e26
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
2
diff
changeset
|
3 |
2
81c7d4668a7e
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff
changeset
|
4 class DBKit: |
81c7d4668a7e
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff
changeset
|
5 def __init__(self, indexFile, databaseFile): |
81c7d4668a7e
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff
changeset
|
6 self.databaseFile = databaseFile |
81c7d4668a7e
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff
changeset
|
7 self.index = dict() |
81c7d4668a7e
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff
changeset
|
8 with open(indexFile) as file: |
81c7d4668a7e
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff
changeset
|
9 for line in file: |
81c7d4668a7e
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff
changeset
|
10 cols = line.split() |
81c7d4668a7e
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff
changeset
|
11 try: |
81c7d4668a7e
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff
changeset
|
12 identifier = cols[0] |
81c7d4668a7e
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff
changeset
|
13 start = int(cols[1]) |
81c7d4668a7e
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff
changeset
|
14 size = int(cols[2]) |
81c7d4668a7e
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff
changeset
|
15 self.index[identifier] = [start, size] |
81c7d4668a7e
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff
changeset
|
16 except Exception: |
81c7d4668a7e
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff
changeset
|
17 raise Exception("Invalid DBKit Index file format: %s." % line) |
81c7d4668a7e
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff
changeset
|
18 |
81c7d4668a7e
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff
changeset
|
19 def createFile(self, identifier, outputName): |
81c7d4668a7e
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff
changeset
|
20 if identifier in self.index: |
81c7d4668a7e
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff
changeset
|
21 entry = self.index[identifier] |
81c7d4668a7e
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff
changeset
|
22 start = entry[0] |
81c7d4668a7e
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff
changeset
|
23 size = entry[1] |
81c7d4668a7e
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff
changeset
|
24 with open(self.databaseFile) as file: |
81c7d4668a7e
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff
changeset
|
25 file.seek(start) |
81c7d4668a7e
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff
changeset
|
26 content = file.read(size) |
81c7d4668a7e
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff
changeset
|
27 outputFile = open(outputName, "w") |
81c7d4668a7e
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff
changeset
|
28 outputFile.write(content) |
81c7d4668a7e
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff
changeset
|
29 outputFile.close() |
81c7d4668a7e
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff
changeset
|
30 return True |
81c7d4668a7e
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff
changeset
|
31 else: |
81c7d4668a7e
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff
changeset
|
32 return False |
81c7d4668a7e
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff
changeset
|
33 |
81c7d4668a7e
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff
changeset
|
34 def getIndex(self): |
81c7d4668a7e
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
diff
changeset
|
35 return self.index |
3
03e124ff7e26
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
2
diff
changeset
|
36 |
03e124ff7e26
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
2
diff
changeset
|
37 |
03e124ff7e26
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
2
diff
changeset
|
38 def writeEntry(identifier, fileName, outputIndex, outputDatabase): |
03e124ff7e26
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
2
diff
changeset
|
39 if isfile(outputDatabase): |
03e124ff7e26
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
2
diff
changeset
|
40 currentSize = getsize(outputDatabase) |
03e124ff7e26
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
2
diff
changeset
|
41 else: |
03e124ff7e26
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
2
diff
changeset
|
42 currentSize = 0 |
03e124ff7e26
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
2
diff
changeset
|
43 if isfile(fileName): |
03e124ff7e26
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
2
diff
changeset
|
44 entrySize = getsize(fileName) |
03e124ff7e26
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
2
diff
changeset
|
45 else: |
03e124ff7e26
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
2
diff
changeset
|
46 entrySize = 0 |
03e124ff7e26
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
2
diff
changeset
|
47 if entrySize > 0: |
03e124ff7e26
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
2
diff
changeset
|
48 outputIndexFile = open(outputIndex, "a+") |
03e124ff7e26
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
2
diff
changeset
|
49 outputIndexFile.write("%s\t%s\t%s\n" % (identifier, currentSize, entrySize)) |
03e124ff7e26
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
2
diff
changeset
|
50 tempFile = open(fileName, "r") |
03e124ff7e26
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
2
diff
changeset
|
51 databaseFile = open(outputDatabase, "a+") |
03e124ff7e26
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
2
diff
changeset
|
52 databaseFile.write(tempFile.read()) |
03e124ff7e26
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
2
diff
changeset
|
53 databaseFile.close() |
03e124ff7e26
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
2
diff
changeset
|
54 tempFile.close() |
03e124ff7e26
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
2
diff
changeset
|
55 outputIndexFile.close() |
03e124ff7e26
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
2
diff
changeset
|
56 return True |
03e124ff7e26
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
2
diff
changeset
|
57 else: |
03e124ff7e26
"planemo upload commit bd03b7888eab0b010acfc3affd38bf4d4e2bb1ef-dirty"
guerler
parents:
2
diff
changeset
|
58 return False |