def build_journals_re_kb(fpath):
"""Load journals regexps knowledge base
@see build_journals_kb
"""
def make_tuple(match):
regexp = match.group('seek')
repl = match.group('repl')
return regexp, repl
kb = []
if isinstance(fpath, basestring):
fpath_needs_closing = True
try:
fh = open(fpath, "r")
except IOError:
raise IOError("Error: Unable to open journal kb '%s'" % fpath)
else:
fpath_needs_closing = False
fh = fpath
try:
for rawline in fh:
if rawline.startswith('#'):
continue
# Extract the seek->replace terms from this KB line:
m_kb_line = re_kb_line.search(rawline.decode('utf-8'))
kb.append(make_tuple(m_kb_line))
finally:
if fpath_needs_closing:
fh.close()
return kb
def build_authors_kb(fpath):
replacements = []
if isinstance(fpath, basestring):
fpath_needs_closing = True
try:
fh = open(fpath, "r")
except IOError:
# problem opening KB for reading, or problem while reading from it:
emsg = "Error: Could not build list of authors - failed " \
"to read from KB %(kb)s." % {'kb' : fpath}
write_message(emsg, sys.stderr, verbose=0)
raise IOError("Error: Unable to open authors kb '%s'" % fpath)
else:
fpath_needs_closing = False
fh = fpath
try:
for rawline in fh:
if rawline.startswith('#'):
continue
# Extract the seek->replace terms from this KB line:
m_kb_line = re_kb_line.search(rawline.decode('utf-8'))
if m_kb_line:
seek = m_kb_line.group('seek')
repl = m_kb_line.group('repl')
replacements.append((seek, repl))
finally:
if fpath_needs_closing:
fh.close()
return replacements
def build_journals_re_kb(fpath):
"""Load journals regexps knowledge base
@see build_journals_kb
"""
def make_tuple(match):
regexp = match.group('seek')
repl = match.group('repl')
return regexp, repl
kb = []
if isinstance(fpath, basestring):
fpath_needs_closing = True
try:
fh = open(fpath, "r")
except IOError:
raise IOError("Error: Unable to open journal kb '%s'" % fpath)
else:
fpath_needs_closing = False
fh = fpath
try:
for rawline in fh:
if rawline.startswith('#'):
continue
# Extract the seek->replace terms from this KB line:
m_kb_line = re_kb_line.search(rawline.decode('utf-8'))
kb.append(make_tuple(m_kb_line))
finally:
if fpath_needs_closing:
fh.close()
return kb
def build_authors_kb(fpath):
replacements = []
if isinstance(fpath, basestring):
fpath_needs_closing = True
try:
fh = open(fpath, "r")
except IOError:
# problem opening KB for reading, or problem while reading from it:
emsg = "Error: Could not build list of authors - failed " \
"to read from KB %(kb)s." % {'kb' : fpath}
write_message(emsg, sys.stderr, verbose=0)
raise IOError("Error: Unable to open authors kb '%s'" % fpath)
else:
fpath_needs_closing = False
fh = fpath
try:
for rawline in fh:
if rawline.startswith('#'):
continue
# Extract the seek->replace terms from this KB line:
m_kb_line = re_kb_line.search(rawline.decode('utf-8'))
if m_kb_line:
seek = m_kb_line.group('seek')
repl = m_kb_line.group('repl')
replacements.append((seek, repl))
finally:
if fpath_needs_closing:
fh.close()
return replacements
def lazy_parser(fh):
for rawline in fh:
if rawline.startswith('#'):
continue
try:
rawline = rawline.decode("utf-8").rstrip("\n")
except UnicodeError:
raise StandardError("Unicode problems in kb %s at line %s" %
(path, rawline))
# Test line to ensure that it is a correctly formatted
# knowledge base line:
# Extract the seek->replace terms from this KB line
m_kb_line = re_kb_line.search(rawline)
if m_kb_line: # good KB line
yield m_kb_line.group('seek'), m_kb_line.group('repl')
else:
raise StandardError("Badly formatted kb '%s' at line %s" %
(path, rawline))
def lazy_parser(fh):
for rawline in fh:
if rawline.startswith('#'):
continue
try:
rawline = rawline.decode("utf-8").rstrip("\n")
except UnicodeError:
raise StandardError("Unicode problems in kb %s at line %s"
% (path, rawline))
# Test line to ensure that it is a correctly formatted
# knowledge base line:
# Extract the seek->replace terms from this KB line
m_kb_line = re_kb_line.search(rawline)
if m_kb_line: # good KB line
yield m_kb_line.group('seek'), m_kb_line.group('repl')
else:
raise StandardError("Badly formatted kb '%s' at line %s"
% (path, rawline))
