def process_pagedirs(dir_from, dir_to):
pagelist = listdir(dir_from)
for pagename in pagelist:
pagedir_from = opj(dir_from, pagename)
pagedir_to = opj(dir_to, pagename)
# first we copy all, even the stuff we convert later:
copy_dir(pagedir_from, pagedir_to)
rev_from = opj(pagedir_from, 'revisions')
rev_to = opj(pagedir_to, 'revisions')
if os.path.exists(rev_from):
revlist = listdir(rev_from)
for rfile in revlist:
rev = int(rfile)
r_from = opj(rev_from, rfile)
r_to = opj(rev_to, rfile)
tocrlf(r_from, r_to)
def generate_pages(dir_from, dir_to):
for pagename in info2:
entry = info2.get(pagename, {})
tslist = entry.keys()
if tslist:
pagedir = opj(dir_to, 'pages', pagename)
os.makedirs(opj(pagedir, 'revisions'))
editlog_file = opj(pagedir, 'edit-log')
f = open(editlog_file, 'w')
rev = 0
tslist.sort()
for ts in tslist:
rev += 1
revstr = '%08d' % rev
file_from, data = entry[ts]
data[0] = str(ts)
data[1] = revstr
if data[2].endswith('/REVERT'):
# replace the timestamp with the revision number
revertts = long(data[7]) # must be long for py 2.2.x
try:
revertrev = int(entry[revertts][1][1])
except KeyError:
# never should trigger...
print "********* KeyError %s entry[%d][1][1] **********" % (pagename, revertts)
revertrev = 0
data[7] = '%08d' % revertrev
f.write('\t'.join(data)+'\n')
if file_from is not None:
file_to = opj(pagedir, 'revisions', revstr)
copy_file(file_from, file_to)
f.close()
curr_file = opj(pagedir, 'current')
f = open(curr_file, 'w')
f.write(revstr)
f.close()
att_from = opj(dir_from, 'pages', pagename, 'attachments')
if os.path.exists(att_from):
att_to = opj(pagedir, 'attachments')
copy_dir(att_from, att_to)
def convert_pagedir(dir_from, dir_to, is_backupdir=0):
os.mkdir(dir_to)
for pagedir in listdir(dir_from):
text_from = opj(dir_from, pagedir, 'text')
text_to = opj(dir_to, pagedir, 'text')
os.mkdir(opj(dir_to, pagedir))
copy_file(text_from, text_to)
backupdir_from = opj(dir_from, pagedir, 'backup')
backupdir_to = opj(dir_to, pagedir, 'backup')
if os.path.exists(backupdir_from):
os.mkdir(backupdir_to)
for ts in listdir(backupdir_from):
ts_usec = str(convert_ts(float(ts)))
backup_from = opj(backupdir_from, ts)
backup_to = opj(backupdir_to, ts_usec)
copy_file(backup_from, backup_to)
editlog_from = opj(dir_from, pagedir, 'edit-log')
editlog_to = opj(dir_to, pagedir, 'edit-log')
convert_editlog(editlog_from, editlog_to)
#cachedir_from = opj(dir_from, pagedir, 'cache')
#cachedir_to = opj(dir_to, pagedir, 'cache')
#if os.path.exists(cachedir_from):
# os.mkdir(cachedir_to)
# try:
# copy_file(
# opj(cachedir_from, 'hitcounts'),
# opj(cachedir_to, 'hitcounts'))
# except: pass
attachdir_from = opj(dir_from, pagedir, 'attachments')
attachdir_to = opj(dir_to, pagedir, 'attachments')
if os.path.exists(attachdir_from):
try:
copy_dir(attachdir_from, attachdir_to)
except: pass
revlist = listdir(rev_from)
for rfile in revlist:
rev = int(rfile)
r_from = opj(rev_from, rfile)
r_to = opj(rev_to, rfile)
tocrlf(r_from, r_to)
origdir = 'data.pre-mig7'
try:
os.rename('data', origdir)
except OSError:
print "You need to be in the directory where your copy of the 'data' directory is located."
sys.exit(1)
os.makedirs(opj('data','pages'))
process_pagedirs(opj(origdir, 'pages'), opj('data', 'pages'))
copy_dir(opj(origdir, 'plugin'), opj('data', 'plugin'))
copy_dir(opj(origdir, 'user'), opj('data', 'user'))
copy_file(opj(origdir, 'edit-log'), opj('data', 'edit-log'))
copy_file(opj(origdir, 'event-log'), opj('data', 'event-log'))
copy_file(opj(origdir, 'intermap.txt'), opj('data', 'intermap.txt'))
try:
os.remove(editlock)
except:
pass
attachdir = os.path.join(pagedir, 'attachments')
for root, dirs, files in walk(attachdir):
for f in files:
try:
f.decode(to_encoding)
except UnicodeDecodeError:
fnew = f.decode(from_encoding).encode(to_encoding)
os.rename(os.path.join(root,f), os.path.join(root, fnew))
print 'renamed', f, '\n ->', fnew, ' in dir:', root
origdir = 'data.pre-mig10'
destdir = 'data'
# Backup original dir and create new empty dir
try:
os.rename(destdir, origdir)
except OSError:
print "You need to be in the directory where your copy of the 'data' directory is located."
sys.exit(1)
copy_dir(origdir, destdir)
migrate(destdir)
key, val = kvpair.split('=')
key = urllib.unquote(key)
val = urllib.unquote(val)
key = convert_string(key, enc_from, enc_to)
val = convert_string(val, enc_from, enc_to)
key = urllib.quote(key)
val = urllib.quote(val)
kvlist.append("%s=%s" % (key,val))
fields[2] = '&'.join(kvlist)
line = '\t'.join(fields) + '\n'
file_to.write(line)
file_to.close()
file_from.close()
st=os.stat(fname_from)
os.utime(fname_to, (st.st_atime,st.st_mtime))
origdir = 'data.pre-mig6'
try:
os.rename('data', origdir)
except OSError:
print "You need to be in the directory where your copy of the 'data' directory is located."
sys.exit(1)
copy_dir(origdir, 'data')
os.remove(opj('data','event-log')) # old format
convert_eventlog(opj(origdir, 'event-log'), opj('data', 'event-log'), from_encoding, to_encoding)
f = open(file_to2, "a")
f.write(data)
f.close()
except:
pass
# Backup original dir and create new empty dir
try:
os.rename("data", origdir)
os.mkdir("data")
except OSError:
print "You need to be in the directory where your copy of the 'data' directory is located."
sys.exit(1)
convert_pagedir(opj(origdir, "pages"), opj("data", "pages"))
convert_textdir(opj(origdir, "text"), "data")
convert_textdir(opj(origdir, "backup"), "data", 1)
convert_editlog(opj(origdir, "editlog"), opj("data", "edit-log"), opj("data", "pages"))
copy_file(opj(origdir, "event.log"), opj("data", "event.log"))
copy_dir(opj(origdir, "plugin"), opj("data", "plugin"))
copy_dir(opj(origdir, "user"), opj("data", "user"))
copy_file(opj(origdir, "intermap.txt"), opj("data", "intermap.txt"))
def generate_pages(dir_from, dir_to):
revactions = ['SAVE','SAVENEW','SAVE/REVERT',] # these actions create revisions
for pn in info:
entry = info.get(pn, {})
tslist = entry.keys()
if tslist:
pagedir = opj(dir_to, 'pages', pn)
revdir = opj(pagedir, 'revisions')
os.makedirs(revdir)
editlog_file = opj(pagedir, 'edit-log')
f = open(editlog_file, 'w')
revnew = 0
tslist.sort()
for ts in tslist:
data = entry[ts][1]
datanew = data[:]
(timestamp,rev,action,pagename,ip,host,id,extra,comment) = data
revstr = '%08d' % rev
if action in revactions:
revnew += 1
revnewstr = '%08d' % revnew
entry[ts][0] = revnew # remember what new revno we chose
else: # ATTNEW,ATTDRW,ATTDEL
revnewstr = '99999999'
if action.endswith('/REVERT'):
# replace the old revno with the correct new revno
revertrevold = int(extra)
revertrevnew = 0
for ts2 in tslist:
data2 = entry[ts2][1]
(timestamp2,rev2,action2,pagename2,ip2,host2,id2,extra2,comment2) = data2
if rev2 == revertrevold:
revertrevnew = entry[ts2][0]
datanew[7] = '%08d' % revertrevnew
datanew[1] = revnewstr
f.write('\t'.join(datanew)+'\n') # does make a CRLF on win32 in the file
if action in revactions: # we DO have a page rev for this one
file_from = opj(dir_from, 'pages', pn, 'revisions', revstr)
file_to = opj(revdir, revnewstr)
copy_file(file_from, file_to)
f.close()
# check if page exists or is deleted in orig dir
pagedir_from = opj(dir_from, 'pages', pn)
revdir_from = opj(pagedir_from, 'revisions')
try:
curr_file_from = opj(pagedir_from, 'current')
currentfrom = open(curr_file_from).read().strip() # try to access it
page_exists = 1
except:
page_exists = 0
# re-make correct DELETED status!
if page_exists:
curr_file = opj(pagedir, 'current')
f = open(curr_file, 'w')
f.write("%08d\n" % revnew) # we add a \n, so it is easier to hack in there manually
f.close()
att_from = opj(dir_from, 'pages', pn, 'attachments')
if os.path.exists(att_from):
att_to = opj(pagedir, 'attachments')
copy_dir(att_from, att_to)