def parse_rcs_tree(self):
while 1:
revision = self.ts.get()
# End of RCS tree description ?
if revision == 'desc':
self.ts.unget(revision)
return
# Parse date
semi, date, sym = self.ts.mget(3)
if sym != 'date':
raise RCSExpected(sym, 'date')
if semi != ';':
raise RCSExpected(semi, ';')
# Convert date into timestamp
date_fields = string.split(date, '.') + ['0', '0', '0']
date_fields = map(string.atoi, date_fields)
# need to make the date four digits for timegm
EPOCH = 1970
if date_fields[0] < EPOCH:
if date_fields[0] < 70:
date_fields[0] = date_fields[0] + 2000
else:
date_fields[0] = date_fields[0] + 1900
if date_fields[0] < EPOCH:
raise ValueError, 'invalid year'
timestamp = compat.timegm(tuple(date_fields))
# Parse author
### NOTE: authors containing whitespace are violations of the
### RCS specification. We are making an allowance here because
### CVSNT is known to produce these sorts of authors.
self.ts.match('author')
author = ''
while 1:
token = self.ts.get()
if token == ';':
break
author = author + token + ' '
author = author[:-1] # toss the trailing space
# Parse state
self.ts.match('state')
state = ''
while 1:
token = self.ts.get()
if token == ';':
break
state = state + token + ' '
state = state[:-1] # toss the trailing space
# Parse branches
self.ts.match('branches')
branches = [ ]
while 1:
token = self.ts.get()
if token == ';':
break
branches.append(token)
# Parse revision of next delta in chain
next, sym = self.ts.mget(2)
if sym != 'next':
raise RCSExpected(sym, 'next')
if next == ';':
next = None
else:
self.ts.match(';')
# there are some files with extra tags in them. for example:
# owner 640;
# group 15;
# permissions 644;
# hardlinks @configure.in@;
# this is "newphrase" in RCSFILE(5). we just want to skip over these.
while 1:
token = self.ts.get()
if token == 'desc' or token[0] in string.digits:
self.ts.unget(token)
break
# consume everything up to the semicolon
while self.ts.get() != ';':
pass
self.sink.define_revision(revision, timestamp, author, state, branches,
next)
def _parse_log_entry(fp):
"""Parse a single log entry.
On entry, fp should point to the first line of the entry (the "revision"
line).
On exit, fp will have consumed the log separator line (dashes) or the
end-of-file marker (equals).
Returns: Revision object and eof flag (see _EOF_*)
"""
rev = None
line = fp.readline()
if not line:
return None, _EOF_LOG
if line == LOG_END_MARKER:
# Needed because some versions of RCS precede LOG_END_MARKER
# with ENTRY_END_MARKER
return None, _EOF_FILE
if line[:8] == 'revision':
match = _re_rev.match(line)
if not match:
return None, _EOF_LOG
rev = match.group(1)
line = fp.readline()
if not line:
return None, _EOF_LOG
match = _re_log_info.match(line)
eof = None
log = ''
while 1:
line = fp.readline()
if not line:
# true end-of-file
eof = _EOF_LOG
break
if line[:9] == 'branches:':
continue
if line == ENTRY_END_MARKER:
break
if line == LOG_END_MARKER:
# end of this file's log information
eof = _EOF_FILE
break
log = log + line
if not rev or not match:
# there was a parsing error
return None, eof
# parse out a time tuple for the local time
tm = compat.cvs_strptime(match.group(1))
# rlog seems to assume that two-digit years are 1900-based (so, "04"
# comes out as "1904", not "2004").
EPOCH = 1970
if tm[0] < EPOCH:
tm = list(tm)
if (tm[0] - 1900) < 70:
tm[0] = tm[0] + 100
if tm[0] < EPOCH:
raise ValueError, 'invalid year'
date = compat.timegm(tm)
return Revision(rev, date,
# author, state, lines changed
match.group(2), match.group(3) == "dead", match.group(5),
log), eof
