next rev to parse build log

parse_log24.py

import sys,re

# more hacks
#  remove the last two lines from the saved output
#  psql
#  > create database pkgs1;
#  > \c pkgs1
#  pkgs1=# create table raw_parse (script_name text, action text, pkg_name text);
#  pkgs1=# copy raw_parse from '/home/user/out1.csv' with CSV delimiter E'\t';
#  pkgs1=# select pkg_name from raw_parse group by pkg_name having every(action <> 'REMOVE') order by pkg_name;
#
# --

if len(sys.argv) > 1 and sys.argv[1] is not None:
  tF = sys.argv[1]
else:
  tF = 'chroot-build.log'

try:
  tLog = open( tF, 'r+')
except Exception, E:
  print str(E)
  sys.exit(1)
  

tScriptBlocksA = []
tPkgs_New = []
tPkgs_Extra = []
tPkgs_Rem = []

tLineBuf = []
tCurBlockName = None

tCnt = 0
bStartBlock = False
bInHdrBlock = False
bInFinishBlk = False
bInPkgsNewBlk  = False
bInPkgsExBlk  = False

for tLine in tLog.readlines():
  tCnt += 1
  resM_Hdr = re.search( '^[=]{64}$', tLine )
  resM_BlkTitle = re.search( '^Starting \"([\-a-zA-Z0-9_]+)\.sh', tLine )
  resM_PkgNew = re.search( '^The following NEW packages will be installed', tLine)
  resM_PkgEx = re.search( '^The following extra packages will be installed', tLine)
  resM_PkgRemove = re.search( '^Removing ([\-a-zA-Z0-9_]+) ...', tLine)

  if resM_PkgRemove is not None  and  tCurBlockName is not None:
    resM_hack = re.search( ' ([\-a-zA-Z0-9_]+) ', tLine)
    tRemElem = resM_hack.group(0)
    tRemElem = tRemElem[1:-1]
    tPkgs_Rem.append( tRemElem )
    print '\t'.join( [tCurBlockName,'REMOVE', tRemElem] )
    continue

  if resM_BlkTitle is not None:
    tCurBlockName = tLine[10:-9]
    tScriptBlocksA.append( tCurBlockName )
    continue

  if resM_PkgEx is not None:
    ##-- chroot starter breaks the pattern.. so check for that too
    if tCurBlockName is None:
      continue
    bInPkgsExBlk = True
    #print '  EXTRA'
    continue
  else:
    if tLine[0:2] != '  ':
      bInPkgsExBlk = False
      #print '  --'
    elif bInPkgsExBlk == True  and  tLine[0:2] == '  ':
      tSomePkgs = tLine[2:-1].split(' ')
      for n in tSomePkgs:
        tPkgs_Extra.append(n)
        print '\t'.join( [tCurBlockName,'EXTRA',n] )
    
  if resM_PkgNew is not None:
    ##-- chroot starter breaks the pattern.. so check for that too
    if tCurBlockName is None:
      continue
    bInPkgsNewBlk = True
    #print '  NEW'
    continue
  else:
    if tLine[0:2] != '  ':
      bInPkgsNewBlk = False
      #print '  --'
    elif bInPkgsNewBlk == True  and  tLine[0:2] == '  ':
      tSomePkgs = tLine[2:-1].split(' ')
      for n in tSomePkgs:
        tPkgs_New.append(n)
        print '\t'.join( [tCurBlockName,'NEW',n] )
    
  if resM_Hdr is not None:
    if bStartBlock == False and bInFinishBlk == False:
      bStartBlock = True
      bInHdrBlock = True
      continue
    elif bStartBlock == True:
      if bInHdrBlock == True:
        bInHdrBlock = False
        continue
      elif bInFinishBlk == False:
        bInFinishBlk = True
        continue
      elif bInFinishBlk == True:
        bInFinishBlk = False

        ##-- clean up a block here
        tCurBlockName = None
        tPkgs_New = []
        tPkgs_Extra = []
        tPkgs_Rem = []
        
        continue
      else:
        print tLine
        print 'PROBLEM'


  if bInFinishBlk == True:
    continue


print 'DONE'
print tScriptBlocksA