spacepaste

Paste details

reply | raw

  1.  
  2. #
  3. #
  4. # Mark Grandi - Oct 14, 2014
  5. #
  8. import argparse, sys, zlib, pprint
  10. HEADER="B+Tree Graph Index 2"
  11. NODEREF = "node_ref_lists="
  12. KEYELEMENTS = "key_elements="
  13. LEN = "len="
  14. ROWLENGTH = "row_lengths="
  16. def printOutIndexKeys(args):
  17. '''print out bzr index keys
  18. @param args - the namespace object we get from argparse.parse_args()
  19. '''
  21. def readUntilNewline(fileObj):
  23. ba = bytearray()
  25. while True:
  26. tmp = fileObj.read(1)
  27. if tmp == "\n".encode("utf-8"):
  28. return ba.decode("utf-8")
  29. else:
  31. ba.append(ord(tmp))
  33. def testString(fileObj, expectedString):
  34. test = fileObj.read(len(expectedString))
  35. if not test.decode("utf-8") == expectedString:
  36. sys.exit("String incorrect, expected {} but got {}".format(expectedString, test.decode("utf-8")))
  39. for iterFile in args.infileList:
  40. testString(iterFile, HEADER)
  42. testString(iterFile, "\n")
  44. testString(iterFile, NODEREF)
  46. noderef_val = int(readUntilNewline(iterFile))
  48. testString(iterFile, KEYELEMENTS)
  49. keyelements_val = int(readUntilNewline(iterFile))
  51. testString(iterFile, LEN)
  52. len_val = int(readUntilNewline(iterFile))
  54. testString(iterFile, ROWLENGTH)
  55. rowlength_val = readUntilNewline(iterFile)
  57. print("noderef: {}, keyelements: {}, len: {}, rowlength: {}".format(noderef_val, keyelements_val, len_val, rowlength_val))
  60. start = iterFile.tell()
  61. finish = 4096
  62. numLeaf = 0
  63. numInternal = 0
  64. totalKeysInInternal = 0
  66. while True:
  68. print("######################################")
  69. uncomp = iterFile.read(finish-start)
  71. if not uncomp:
  72. print("eof")
  73. break
  75. print("read from {} to {}".format(start, iterFile.tell()))
  77. decomp = zlib.decompress(uncomp)
  79. splitList = decomp.split(b'\n')
  80. if splitList[0].decode("utf-8") == "type=internal":
  81. numInternal += 1
  83. totalKeysInInternal += len(splitList) - 3 # remove type=, offset= and empty string at end
  84. if splitList[0].decode("utf-8") == "type=leaf":
  85. numLeaf += 1
  87. pprint.pprint(splitList)
  89. start = iterFile.tell()
  90. finish += 4096
  92. print("done, numLeaf: {}, numInternal: {}, totalKeysInInternal: {}".format(numLeaf, numInternal, totalKeysInInternal))
  95. if __name__ == "__main__":
  96. # if we are being run as a real program
  98. parser = argparse.ArgumentParser(description="print out bzr index keys",
  99. epilog="Copyright Oct 14, 2014 - Mark Grandi")
  101. parser.add_argument('infileList', type=argparse.FileType('rb'), nargs="+", help="bzr index to read the keys from")
  104. printOutIndexKeys(parser.parse_args())
  105.  
Powered by spacepaste. Paste removal.