def _generate_diff(self, hunks):
"""Generate a diff text for the given hunks."""
def extend_context(start, end):
"""Add context lines."""
return ''.join(' {0}\n'.format(line.rstrip())
for line in self.a[start:end])
context_range = self._get_context_range(hunks)
output = color_format(
'{aqua}{0}{default}\n{1}', Hunk.get_header_text(*context_range),
extend_context(context_range[0][0], hunks[0].a_rng[0]))
previous_hunk = None
for hunk in hunks:
if previous_hunk:
output += extend_context(previous_hunk.a_rng[1], hunk.a_rng[0])
previous_hunk = hunk
output += hunk.diff_text
output += extend_context(hunks[-1].a_rng[1], context_range[0][1])
if self._replace_invisible:
output = chars.replace_invisible(output)
return output
def _generate_diff(self, hunks):
"""Generate a diff text for the given hunks."""
def extend_context(start, end):
"""Add context lines."""
return ''.join(' {0}\n'.format(line.rstrip())
for line in self.a[start:end])
context_range = self._get_context_range(hunks)
output = ('\03{aqua}' +
Hunk.get_header_text(*context_range) + '\03{default}\n' +
extend_context(context_range[0][0], hunks[0].a_rng[0]))
previous_hunk = None
for hunk in hunks:
if previous_hunk:
output += extend_context(previous_hunk.a_rng[1], hunk.a_rng[0])
previous_hunk = hunk
output += hunk.diff_text
output += extend_context(hunks[-1].a_rng[1], context_range[0][1])
if self._replace_invisible:
output = chars.replace_invisible(output)
return output
def main(*args):
"""
Process command line arguments and invoke bot.
If args is an empty list, sys.argv is used.
@param args: command line arguments
@type args: list of unicode
"""
add_cat = None
gen = None
# summary message
edit_summary = ""
# Array which will collect commandline parameters.
# First element is original text, second element is replacement text.
commandline_replacements = []
# A list of 2-tuples of original text and replacement text.
replacements = []
# Don't edit pages which contain certain texts.
exceptions = {
"title": [],
"text-contains": [],
"inside": [],
"inside-tags": [],
"require-title": [], # using a seperate requirements dict needs some
} # major refactoring of code.
# Should the elements of 'replacements' and 'exceptions' be interpreted
# as regular expressions?
regex = False
# Predefined fixes from dictionary 'fixes' (see above).
fixes_set = []
# the dump's path, either absolute or relative, which will be used
# if -xml flag is present
xmlFilename = None
useSql = False
# will become True when the user presses a ('yes to all') or uses the
# -always flag.
acceptall = False
# Will become True if the user inputs the commandline parameter -nocase
caseInsensitive = False
# Will become True if the user inputs the commandline parameter -dotall
dotall = False
# Will become True if the user inputs the commandline parameter -multiline
multiline = False
# Do all hits when they overlap
allowoverlap = False
# Do not recurse replacement
recursive = False
# Between a regex and another (using -fix) sleep some time (not to waste
# too much CPU
sleep = None
# Request manual replacements even if replacements are already defined
manual_input = False
# Replacements loaded from a file
replacement_file = None
replacement_file_arg_misplaced = False
# Read commandline parameters.
local_args = pywikibot.handle_args(args)
genFactory = pagegenerators.GeneratorFactory()
for arg in local_args:
if genFactory.handleArg(arg):
continue
if arg == "-regex":
regex = True
elif arg.startswith("-xmlstart"):
if len(arg) == 9:
xmlStart = pywikibot.input("Please enter the dumped article to start with:")
else:
xmlStart = arg[10:]
elif arg.startswith("-xml"):
if len(arg) == 4:
xmlFilename = i18n.input("pywikibot-enter-xml-filename")
else:
xmlFilename = arg[5:]
elif arg == "-sql":
useSql = True
elif arg.startswith("-excepttitle:"):
exceptions["title"].append(arg[13:])
elif arg.startswith("-requiretitle:"):
exceptions["require-title"].append(arg[14:])
elif arg.startswith("-excepttext:"):
exceptions["text-contains"].append(arg[12:])
elif arg.startswith("-exceptinside:"):
exceptions["inside"].append(arg[14:])
elif arg.startswith("-exceptinsidetag:"):
exceptions["inside-tags"].append(arg[17:])
elif arg.startswith("-fix:"):
fixes_set += [arg[5:]]
elif arg.startswith("-sleep:"):
sleep = float(arg[7:])
elif arg == "-always":
acceptall = True
elif arg == "-recursive":
recursive = True
elif arg == "-nocase":
caseInsensitive = True
elif arg == "-dotall":
dotall = True
elif arg == "-multiline":
multiline = True
elif arg.startswith("-addcat:"):
add_cat = arg[8:]
elif arg.startswith("-summary:"):
edit_summary = arg[9:]
elif arg.startswith("-automaticsummary"):
edit_summary = True
elif arg.startswith("-allowoverlap"):
allowoverlap = True
elif arg.startswith("-manualinput"):
manual_input = True
elif arg.startswith("-replacementfile"):
issue_deprecation_warning("-replacementfile", "-pairsfile", 2, ArgumentDeprecationWarning)
elif arg.startswith("-pairsfile"):
if len(commandline_replacements) % 2:
replacement_file_arg_misplaced = True
if arg == "-pairsfile":
replacement_file = pywikibot.input("Please enter the filename to read replacements from:")
else:
replacement_file = arg[len("-pairsfile:") :]
else:
commandline_replacements.append(arg)
site = pywikibot.Site()
if len(commandline_replacements) % 2:
pywikibot.error("Incomplete command line pattern replacement pair.")
return False
if replacement_file_arg_misplaced:
pywikibot.error("-pairsfile used between a pattern replacement pair.")
return False
if replacement_file:
try:
with codecs.open(replacement_file, "r", "utf-8") as f:
# strip newlines, but not other characters
file_replacements = f.read().splitlines()
except (IOError, OSError) as e:
pywikibot.error("Error loading {0}: {1}".format(replacement_file, e))
return False
if len(file_replacements) % 2:
pywikibot.error("{0} contains an incomplete pattern replacement pair.".format(replacement_file))
return False
# Strip BOM from first line
file_replacements[0].lstrip("\uFEFF")
commandline_replacements.extend(file_replacements)
if not (commandline_replacements or fixes_set) or manual_input:
old = pywikibot.input("Please enter the text that should be replaced:")
while old:
new = pywikibot.input("Please enter the new text:")
commandline_replacements += [old, new]
old = pywikibot.input("Please enter another text that should be replaced," "\nor press Enter to start:")
# The summary stored here won't be actually used but is only an example
single_summary = None
for i in range(0, len(commandline_replacements), 2):
replacement = Replacement(commandline_replacements[i], commandline_replacements[i + 1])
if not single_summary:
single_summary = i18n.twtranslate(
site, "replace-replacing", {"description": " (-%s +%s)" % (replacement.old, replacement.new)}
)
replacements.append(replacement)
# Perform one of the predefined actions.
missing_fixes_summaries = [] # which a fixes/replacements miss a summary
for fix_name in fixes_set:
try:
fix = fixes.fixes[fix_name]
except KeyError:
pywikibot.output("Available predefined fixes are: %s" % ", ".join(fixes.fixes.keys()))
if not fixes.user_fixes_loaded:
pywikibot.output("The user fixes file could not be found: " "{0}".format(fixes.filename))
return
if not fix["replacements"]:
pywikibot.warning("No replacements defined for fix " '"{0}"'.format(fix_name))
continue
if "msg" in fix:
if isinstance(fix["msg"], basestring):
set_summary = i18n.twtranslate(site, str(fix["msg"]))
else:
set_summary = i18n.translate(site, fix["msg"], fallback=True)
else:
set_summary = None
replacement_set = ReplacementList(
fix.get("regex"), fix.get("exceptions"), fix.get("nocase"), set_summary, name=fix_name
)
# Whether some replacements have a summary, if so only show which
# have none, otherwise just mention the complete fix
missing_fix_summaries = []
for index, replacement in enumerate(fix["replacements"], start=1):
summary = None if len(replacement) < 3 else replacement[2]
if not set_summary and not summary:
missing_fix_summaries.append('"{0}" (replacement #{1})'.format(fix_name, index))
if chars.contains_invisible(replacement[0]):
pywikibot.warning(
'The old string "{0}" contains formatting '
"characters like U+200E".format(chars.replace_invisible(replacement[0]))
)
if not callable(replacement[1]) and chars.contains_invisible(replacement[1]):
pywikibot.warning(
'The new string "{0}" contains formatting '
"characters like U+200E".format(chars.replace_invisible(replacement[1]))
)
replacement_set.append(
ReplacementListEntry(
old=replacement[0], new=replacement[1], fix_set=replacement_set, edit_summary=summary
)
)
if replacement_set:
replacements.extend(replacement_set)
if len(fix["replacements"]) == len(missing_fix_summaries):
missing_fixes_summaries.append('"{0}" (all replacements)'.format(fix_name))
else:
missing_fixes_summaries += missing_fix_summaries
if (not edit_summary or edit_summary is True) and (missing_fixes_summaries or single_summary):
if single_summary:
pywikibot.output(
"The summary message for the command line " "replacements will be something like: %s" % single_summary
)
if missing_fixes_summaries:
pywikibot.output(
"The summary will not be used when the fix has "
"one defined but the following fix(es) do(es) not "
"have a summary defined: "
"{0}".format(", ".join(missing_fixes_summaries))
)
if edit_summary is not True:
edit_summary = pywikibot.input(
"Press Enter to use this automatic message, or enter a "
"description of the\nchanges your bot will make:"
)
else:
edit_summary = ""
# Set the regular expression flags
flags = re.UNICODE
if caseInsensitive:
flags = flags | re.IGNORECASE
if dotall:
flags = flags | re.DOTALL
if multiline:
flags = flags | re.MULTILINE
# Pre-compile all regular expressions here to save time later
for replacement in replacements:
replacement.compile(regex, flags)
precompile_exceptions(exceptions, regex, flags)
if xmlFilename:
try:
xmlStart
except NameError:
xmlStart = None
gen = XmlDumpReplacePageGenerator(xmlFilename, xmlStart, replacements, exceptions, site)
elif useSql:
whereClause = "WHERE (%s)" % " OR ".join(
[
"old_text RLIKE '%s'" % prepareRegexForMySQL(old_regexp.pattern)
for (old_regexp, new_text) in replacements
]
)
if exceptions:
exceptClause = "AND NOT (%s)" % " OR ".join(
["old_text RLIKE '%s'" % prepareRegexForMySQL(exc.pattern) for exc in exceptions]
)
else:
exceptClause = ""
query = """
SELECT page_namespace, page_title
FROM page
JOIN text ON (page_id = old_id)
%s
%s
LIMIT 200""" % (
whereClause,
exceptClause,
)
gen = pagegenerators.MySQLPageGenerator(query)
gen = genFactory.getCombinedGenerator(gen)
if not gen:
pywikibot.bot.suggest_help(missing_generator=True)
return False
preloadingGen = pagegenerators.PreloadingGenerator(gen)
bot = ReplaceRobot(
preloadingGen,
replacements,
exceptions,
allowoverlap,
recursive,
add_cat,
sleep,
edit_summary,
always=acceptall,
site=site,
)
site.login()
bot.run()
# Explicitly call pywikibot.stopme().
# It will make sure the callback is triggered before replace.py is unloaded.
pywikibot.stopme()
pywikibot.output("\n%s pages changed." % bot.changed_pages)
def test_replace(self):
"""Test replace_invisible."""
self.assertEqual(chars.replace_invisible('Hello world!'),
'Hello world!')
self.assertEqual(chars.replace_invisible('\u200eRTL\u200f'),
'<200e>RTL<200f>')
def main(*args):
"""
Process command line arguments and invoke bot.
If args is an empty list, sys.argv is used.
@param args: command line arguments
@type args: unicode
"""
add_cat = None
gen = None
# summary message
edit_summary = ''
# Array which will collect commandline parameters.
# First element is original text, second element is replacement text.
commandline_replacements = []
# A list of 2-tuples of original text and replacement text.
replacements = []
# Don't edit pages which contain certain texts.
exceptions = {
'title': [],
'text-contains': [],
'inside': [],
'inside-tags': [],
'require-title': [], # using a separate requirements dict needs some
} # major refactoring of code.
# Should the elements of 'replacements' and 'exceptions' be interpreted
# as regular expressions?
regex = False
# Predefined fixes from dictionary 'fixes' (see above).
fixes_set = []
# the dump's path, either absolute or relative, which will be used
# if -xml flag is present
xmlFilename = None
useSql = False
sql_query = None
# will become True when the user presses a ('yes to all') or uses the
# -always flag.
acceptall = False
# Will become True if the user inputs the commandline parameter -nocase
caseInsensitive = False
# Will become True if the user inputs the commandline parameter -dotall
dotall = False
# Will become True if the user inputs the commandline parameter -multiline
multiline = False
# Do all hits when they overlap
allowoverlap = False
# Do not recurse replacement
recursive = False
# Between a regex and another (using -fix) sleep some time (not to waste
# too much CPU
sleep = None
# Request manual replacements even if replacements are already defined
manual_input = False
# Replacements loaded from a file
replacement_file = None
replacement_file_arg_misplaced = False
# Read commandline parameters.
local_args = pywikibot.handle_args(args)
genFactory = pagegenerators.GeneratorFactory()
for arg in local_args:
if genFactory.handleArg(arg):
continue
if arg == '-regex':
regex = True
elif arg.startswith('-xmlstart'):
if len(arg) == 9:
xmlStart = pywikibot.input(
'Please enter the dumped article to start with:')
else:
xmlStart = arg[10:]
elif arg.startswith('-xml'):
if len(arg) == 4:
xmlFilename = i18n.input('pywikibot-enter-xml-filename')
else:
xmlFilename = arg[5:]
elif arg.startswith(('-sql', '-mysqlquery')):
if arg.startswith('-sql'):
issue_deprecation_warning('The usage of "-sql"',
'-mysqlquery',
1,
ArgumentDeprecationWarning,
since='20180617')
useSql = True
sql_query = arg.partition(':')[2]
elif arg.startswith('-excepttitle:'):
exceptions['title'].append(arg[13:])
elif arg.startswith('-requiretitle:'):
exceptions['require-title'].append(arg[14:])
elif arg.startswith('-excepttext:'):
exceptions['text-contains'].append(arg[12:])
elif arg.startswith('-exceptinside:'):
exceptions['inside'].append(arg[14:])
elif arg.startswith('-exceptinsidetag:'):
exceptions['inside-tags'].append(arg[17:])
elif arg.startswith('-fix:'):
fixes_set += [arg[5:]]
elif arg.startswith('-sleep:'):
sleep = float(arg[7:])
elif arg == '-always':
acceptall = True
elif arg == '-recursive':
recursive = True
elif arg == '-nocase':
caseInsensitive = True
elif arg == '-dotall':
dotall = True
elif arg == '-multiline':
multiline = True
elif arg.startswith('-addcat:'):
add_cat = arg[8:]
elif arg.startswith('-summary:'):
edit_summary = arg[9:]
elif arg.startswith('-automaticsummary'):
edit_summary = True
elif arg.startswith('-allowoverlap'):
allowoverlap = True
elif arg.startswith('-manualinput'):
manual_input = True
elif arg.startswith('-replacementfile'):
issue_deprecation_warning('-replacementfile',
'-pairsfile',
2,
ArgumentDeprecationWarning,
since='20160304')
elif arg.startswith('-pairsfile'):
if len(commandline_replacements) % 2:
replacement_file_arg_misplaced = True
if arg == '-pairsfile':
replacement_file = pywikibot.input(
'Please enter the filename to read replacements from:')
else:
replacement_file = arg[len('-pairsfile:'):]
else:
commandline_replacements.append(arg)
site = pywikibot.Site()
if len(commandline_replacements) % 2:
pywikibot.error('Incomplete command line pattern replacement pair.')
return False
if replacement_file_arg_misplaced:
pywikibot.error('-pairsfile used between a pattern replacement pair.')
return False
if replacement_file:
try:
with codecs.open(replacement_file, 'r', 'utf-8') as f:
# strip newlines, but not other characters
file_replacements = f.read().splitlines()
except (IOError, OSError) as e:
pywikibot.error('Error loading {0}: {1}'.format(
replacement_file, e))
return False
if len(file_replacements) % 2:
pywikibot.error(
'{0} contains an incomplete pattern replacement pair.'.format(
replacement_file))
return False
# Strip BOM from first line
file_replacements[0].lstrip('\uFEFF')
commandline_replacements.extend(file_replacements)
if not (commandline_replacements or fixes_set) or manual_input:
old = pywikibot.input('Please enter the text that should be replaced:')
while old:
new = pywikibot.input('Please enter the new text:')
commandline_replacements += [old, new]
old = pywikibot.input(
'Please enter another text that should be replaced,'
'\nor press Enter to start:')
# The summary stored here won't be actually used but is only an example
single_summary = None
for i in range(0, len(commandline_replacements), 2):
replacement = Replacement(commandline_replacements[i],
commandline_replacements[i + 1])
if not single_summary:
single_summary = i18n.twtranslate(
site, 'replace-replacing', {
'description':
' (-{0} +{1})'.format(replacement.old, replacement.new)
})
replacements.append(replacement)
# Perform one of the predefined actions.
missing_fixes_summaries = [] # which a fixes/replacements miss a summary
generators_given = bool(genFactory.gens)
for fix_name in fixes_set:
try:
fix = fixes.fixes[fix_name]
except KeyError:
pywikibot.output('Available predefined fixes are: {0}'.format(
', '.join(fixes.fixes.keys())))
if not fixes.user_fixes_loaded:
pywikibot.output('The user fixes file could not be found: '
'{0}'.format(fixes.filename))
return
if not fix['replacements']:
pywikibot.warning('No replacements defined for fix '
'"{0}"'.format(fix_name))
continue
if 'msg' in fix:
if isinstance(fix['msg'], basestring):
set_summary = i18n.twtranslate(site, str(fix['msg']))
else:
set_summary = i18n.translate(site, fix['msg'], fallback=True)
else:
set_summary = None
if not generators_given and 'generator' in fix:
gen_args = fix['generator']
if isinstance(gen_args, basestring):
gen_args = [gen_args]
for gen_arg in gen_args:
genFactory.handleArg(gen_arg)
replacement_set = ReplacementList(fix.get('regex'),
fix.get('exceptions'),
fix.get('nocase'),
set_summary,
name=fix_name)
# Whether some replacements have a summary, if so only show which
# have none, otherwise just mention the complete fix
missing_fix_summaries = []
for index, replacement in enumerate(fix['replacements'], start=1):
summary = None if len(replacement) < 3 else replacement[2]
if not set_summary and not summary:
missing_fix_summaries.append('"{0}" (replacement #{1})'.format(
fix_name, index))
if chars.contains_invisible(replacement[0]):
pywikibot.warning('The old string "{0}" contains formatting '
'characters like U+200E'.format(
chars.replace_invisible(replacement[0])))
if (not callable(replacement[1])
and chars.contains_invisible(replacement[1])):
pywikibot.warning('The new string "{0}" contains formatting '
'characters like U+200E'.format(
chars.replace_invisible(replacement[1])))
replacement_set.append(
ReplacementListEntry(
old=replacement[0],
new=replacement[1],
fix_set=replacement_set,
edit_summary=summary,
))
# Exceptions specified via 'fix' shall be merged to those via CLI.
if replacement_set:
replacements.extend(replacement_set)
if replacement_set._exceptions is not None:
for k, v in replacement_set._exceptions.items():
if k in exceptions:
exceptions[k] = list(set(exceptions[k]) | set(v))
else:
exceptions[k] = v
if len(fix['replacements']) == len(missing_fix_summaries):
missing_fixes_summaries.append(
'"{0}" (all replacements)'.format(fix_name))
else:
missing_fixes_summaries += missing_fix_summaries
if ((not edit_summary or edit_summary is True)
and (missing_fixes_summaries or single_summary)):
if single_summary:
pywikibot.output('The summary message for the command line '
'replacements will be something like: ' +
single_summary)
if missing_fixes_summaries:
pywikibot.output('The summary will not be used when the fix has '
'one defined but the following fix(es) do(es) '
'not have a summary defined: '
'{0}'.format(', '.join(missing_fixes_summaries)))
if edit_summary is not True:
edit_summary = pywikibot.input(
'Press Enter to use this automatic message, or enter a '
'description of the\nchanges your bot will make:')
else:
edit_summary = ''
# Set the regular expression flags
flags = re.UNICODE
if caseInsensitive:
flags = flags | re.IGNORECASE
if dotall:
flags = flags | re.DOTALL
if multiline:
flags = flags | re.MULTILINE
# Pre-compile all regular expressions here to save time later
for replacement in replacements:
replacement.compile(regex, flags)
precompile_exceptions(exceptions, regex, flags)
if xmlFilename:
try:
xmlStart
except NameError:
xmlStart = None
gen = XmlDumpReplacePageGenerator(xmlFilename, xmlStart, replacements,
exceptions, site)
elif useSql:
if not sql_query:
whereClause = 'WHERE (%s)' % ' OR '.join([
"old_text RLIKE '%s'" %
prepareRegexForMySQL(old_regexp.pattern)
for (old_regexp, new_text) in replacements
])
if exceptions:
exceptClause = 'AND NOT (%s)' % ' OR '.join([
"old_text RLIKE '%s'" % prepareRegexForMySQL(exc.pattern)
for exc in exceptions
])
else:
exceptClause = ''
query = sql_query or """
SELECT page_namespace, page_title
FROM page
JOIN text ON (page_id = old_id)
%s
%s
LIMIT 200""" % (whereClause, exceptClause)
gen = pagegenerators.MySQLPageGenerator(query)
gen = genFactory.getCombinedGenerator(gen, preload=True)
if not gen:
pywikibot.bot.suggest_help(missing_generator=True)
return False
bot = ReplaceRobot(gen,
replacements,
exceptions,
allowoverlap,
recursive,
add_cat,
sleep,
edit_summary,
always=acceptall,
site=site)
site.login()
bot.run()
# Explicitly call pywikibot.stopme(). It will make sure the callback is
# triggered before replace.py is unloaded.
pywikibot.stopme()
pywikibot.output('\n{0} pages changed.'.format(bot.changed_pages))
def main(*args):
"""
Process command line arguments and invoke bot.
If args is an empty list, sys.argv is used.
@param args: command line arguments
@type args: list of unicode
"""
add_cat = None
gen = None
# summary message
edit_summary = u""
# Array which will collect commandline parameters.
# First element is original text, second element is replacement text.
commandline_replacements = []
# A list of 2-tuples of original text and replacement text.
replacements = []
# Don't edit pages which contain certain texts.
exceptions = {
'title': [],
'text-contains': [],
'inside': [],
'inside-tags': [],
'require-title': [], # using a seperate requirements dict needs some
} # major refactoring of code.
# Should the elements of 'replacements' and 'exceptions' be interpreted
# as regular expressions?
regex = False
# Predefined fixes from dictionary 'fixes' (see above).
fixes_set = []
# the dump's path, either absolute or relative, which will be used
# if -xml flag is present
xmlFilename = None
useSql = False
# will become True when the user presses a ('yes to all') or uses the
# -always flag.
acceptall = False
# Will become True if the user inputs the commandline parameter -nocase
caseInsensitive = False
# Will become True if the user inputs the commandline parameter -dotall
dotall = False
# Will become True if the user inputs the commandline parameter -multiline
multiline = False
# Do all hits when they overlap
allowoverlap = False
# Do not recurse replacement
recursive = False
# Between a regex and another (using -fix) sleep some time (not to waste
# too much CPU
sleep = None
# Request manual replacements even if replacements are already defined
manual_input = False
# Replacements loaded from a file
replacement_file = None
replacement_file_arg_misplaced = False
# Read commandline parameters.
local_args = pywikibot.handle_args(args)
genFactory = pagegenerators.GeneratorFactory()
for arg in local_args:
if genFactory.handleArg(arg):
continue
if arg == '-regex':
regex = True
elif arg.startswith('-xmlstart'):
if len(arg) == 9:
xmlStart = pywikibot.input(
u'Please enter the dumped article to start with:')
else:
xmlStart = arg[10:]
elif arg.startswith('-xml'):
if len(arg) == 4:
xmlFilename = i18n.input('pywikibot-enter-xml-filename')
else:
xmlFilename = arg[5:]
elif arg == '-sql':
useSql = True
elif arg.startswith('-excepttitle:'):
exceptions['title'].append(arg[13:])
elif arg.startswith('-requiretitle:'):
exceptions['require-title'].append(arg[14:])
elif arg.startswith('-excepttext:'):
exceptions['text-contains'].append(arg[12:])
elif arg.startswith('-exceptinside:'):
exceptions['inside'].append(arg[14:])
elif arg.startswith('-exceptinsidetag:'):
exceptions['inside-tags'].append(arg[17:])
elif arg.startswith('-fix:'):
fixes_set += [arg[5:]]
elif arg.startswith('-sleep:'):
sleep = float(arg[7:])
elif arg == '-always':
acceptall = True
elif arg == '-recursive':
recursive = True
elif arg == '-nocase':
caseInsensitive = True
elif arg == '-dotall':
dotall = True
elif arg == '-multiline':
multiline = True
elif arg.startswith('-addcat:'):
add_cat = arg[8:]
elif arg.startswith('-summary:'):
edit_summary = arg[9:]
elif arg.startswith('-allowoverlap'):
allowoverlap = True
elif arg.startswith('-manualinput'):
manual_input = True
elif arg.startswith('-replacementfile'):
if len(commandline_replacements) % 2:
replacement_file_arg_misplaced = True
if arg == '-replacementfile':
replacement_file = pywikibot.input(
u'Please enter the filename to read replacements from:')
else:
replacement_file = arg[len('-replacementfile:'):]
else:
commandline_replacements.append(arg)
site = pywikibot.Site()
if len(commandline_replacements) % 2:
pywikibot.error('Incomplete command line pattern replacement pair.')
return False
if replacement_file_arg_misplaced:
pywikibot.error(
'-replacementfile used between a pattern replacement pair.')
return False
if replacement_file:
try:
with codecs.open(replacement_file, 'r', 'utf-8') as f:
# strip newlines, but not other characters
file_replacements = f.read().splitlines()
except (IOError, OSError) as e:
pywikibot.error(u'Error loading {0}: {1}'.format(
replacement_file, e))
return False
if len(file_replacements) % 2:
pywikibot.error(
'{0} contains an incomplete pattern replacement pair.'.format(
replacement_file))
return False
# Strip BOM from first line
file_replacements[0].lstrip(u'\uFEFF')
commandline_replacements.extend(file_replacements)
if not(commandline_replacements or fixes_set) or manual_input:
old = pywikibot.input(u'Please enter the text that should be replaced:')
while old:
new = pywikibot.input(u'Please enter the new text:')
commandline_replacements += [old, new]
old = pywikibot.input(
'Please enter another text that should be replaced,'
'\nor press Enter to start:')
single_summary = None
for i in range(0, len(commandline_replacements), 2):
replacement = Replacement(commandline_replacements[i],
commandline_replacements[i + 1])
if not single_summary:
single_summary = i18n.twtranslate(
site, 'replace-replacing',
{'description':
' (-%s +%s)' % (replacement.old, replacement.new)}
)
replacements.append(replacement)
if not edit_summary:
if single_summary:
pywikibot.output(u'The summary message for the command line '
'replacements will be something like: %s'
% single_summary)
if fixes_set:
pywikibot.output('If a summary is defined for the fix, this '
'default summary won\'t be applied.')
edit_summary = pywikibot.input(
'Press Enter to use this automatic message, or enter a '
'description of the\nchanges your bot will make:')
# Perform one of the predefined actions.
for fix in fixes_set:
try:
fix = fixes.fixes[fix]
except KeyError:
pywikibot.output(u'Available predefined fixes are: %s'
% ', '.join(fixes.fixes.keys()))
if not fixes.user_fixes_loaded:
pywikibot.output('The user fixes file could not be found: '
'{0}'.format(fixes.filename))
return
if "msg" in fix:
if isinstance(fix['msg'], basestring):
set_summary = i18n.twtranslate(site, str(fix['msg']))
else:
set_summary = i18n.translate(site, fix['msg'], fallback=True)
else:
set_summary = None
replacement_set = ReplacementList(fix.get('regex'),
fix.get('exceptions'),
fix.get('nocase'),
set_summary)
for replacement in fix['replacements']:
summary = None if len(replacement) < 3 else replacement[2]
if chars.contains_invisible(replacement[0]):
pywikibot.warning('The old string "{0}" contains formatting '
'characters like U+200E'.format(
chars.replace_invisible(replacement[0])))
if chars.contains_invisible(replacement[1]):
pywikibot.warning('The new string "{0}" contains formatting '
'characters like U+200E'.format(
chars.replace_invisible(replacement[1])))
replacements.append(ReplacementListEntry(
old=replacement[0],
new=replacement[1],
fix_set=replacement_set,
edit_summary=summary,
))
# Set the regular expression flags
flags = re.UNICODE
if caseInsensitive:
flags = flags | re.IGNORECASE
if dotall:
flags = flags | re.DOTALL
if multiline:
flags = flags | re.MULTILINE
# Pre-compile all regular expressions here to save time later
for replacement in replacements:
replacement.compile(regex, flags)
precompile_exceptions(exceptions, regex, flags)
if xmlFilename:
try:
xmlStart
except NameError:
xmlStart = None
gen = XmlDumpReplacePageGenerator(xmlFilename, xmlStart,
replacements, exceptions, site)
elif useSql:
whereClause = 'WHERE (%s)' % ' OR '.join(
["old_text RLIKE '%s'" % prepareRegexForMySQL(old_regexp.pattern)
for (old_regexp, new_text) in replacements])
if exceptions:
exceptClause = 'AND NOT (%s)' % ' OR '.join(
["old_text RLIKE '%s'" % prepareRegexForMySQL(exc.pattern)
for exc in exceptions])
else:
exceptClause = ''
query = u"""
SELECT page_namespace, page_title
FROM page
JOIN text ON (page_id = old_id)
%s
%s
LIMIT 200""" % (whereClause, exceptClause)
gen = pagegenerators.MySQLPageGenerator(query)
gen = genFactory.getCombinedGenerator(gen)
if not gen:
# syntax error, show help text from the top of this file
pywikibot.showHelp('replace')
return
preloadingGen = pagegenerators.PreloadingGenerator(gen)
bot = ReplaceRobot(preloadingGen, replacements, exceptions, acceptall,
allowoverlap, recursive, add_cat, sleep, edit_summary,
site)
site.login()
bot.run()
# Explicitly call pywikibot.stopme().
# It will make sure the callback is triggered before replace.py is unloaded.
pywikibot.stopme()
pywikibot.output(u'\n%s pages changed.' % bot.changed_pages)
def main(*args):
"""
Process command line arguments and invoke bot.
If args is an empty list, sys.argv is used.
@param args: command line arguments
@type args: list of unicode
"""
add_cat = None
gen = None
# summary message
edit_summary = u""
# Array which will collect commandline parameters.
# First element is original text, second element is replacement text.
commandline_replacements = []
# A list of 2-tuples of original text and replacement text.
replacements = []
# Don't edit pages which contain certain texts.
exceptions = {
'title': [],
'text-contains': [],
'inside': [],
'inside-tags': [],
'require-title': [], # using a seperate requirements dict needs some
} # major refactoring of code.
# Should the elements of 'replacements' and 'exceptions' be interpreted
# as regular expressions?
regex = False
# Predefined fixes from dictionary 'fixes' (see above).
fixes_set = []
# the dump's path, either absolute or relative, which will be used
# if -xml flag is present
xmlFilename = None
useSql = False
# will become True when the user presses a ('yes to all') or uses the
# -always flag.
acceptall = False
# Will become True if the user inputs the commandline parameter -nocase
caseInsensitive = False
# Will become True if the user inputs the commandline parameter -dotall
dotall = False
# Will become True if the user inputs the commandline parameter -multiline
multiline = False
# Do all hits when they overlap
allowoverlap = False
# Do not recurse replacement
recursive = False
# Between a regex and another (using -fix) sleep some time (not to waste
# too much CPU
sleep = None
# Request manual replacements even if replacements are already defined
manual_input = False
# Replacements loaded from a file
replacement_file = None
replacement_file_arg_misplaced = False
# Read commandline parameters.
local_args = pywikibot.handle_args(args)
genFactory = pagegenerators.GeneratorFactory()
for arg in local_args:
if genFactory.handleArg(arg):
continue
if arg == '-regex':
regex = True
elif arg.startswith('-xmlstart'):
if len(arg) == 9:
xmlStart = pywikibot.input(
u'Please enter the dumped article to start with:')
else:
xmlStart = arg[10:]
elif arg.startswith('-xml'):
if len(arg) == 4:
xmlFilename = i18n.input('pywikibot-enter-xml-filename')
else:
xmlFilename = arg[5:]
elif arg == '-sql':
useSql = True
elif arg.startswith('-excepttitle:'):
exceptions['title'].append(arg[13:])
elif arg.startswith('-requiretitle:'):
exceptions['require-title'].append(arg[14:])
elif arg.startswith('-excepttext:'):
exceptions['text-contains'].append(arg[12:])
elif arg.startswith('-exceptinside:'):
exceptions['inside'].append(arg[14:])
elif arg.startswith('-exceptinsidetag:'):
exceptions['inside-tags'].append(arg[17:])
elif arg.startswith('-fix:'):
fixes_set += [arg[5:]]
elif arg.startswith('-sleep:'):
sleep = float(arg[7:])
elif arg == '-always':
acceptall = True
elif arg == '-recursive':
recursive = True
elif arg == '-nocase':
caseInsensitive = True
elif arg == '-dotall':
dotall = True
elif arg == '-multiline':
multiline = True
elif arg.startswith('-addcat:'):
add_cat = arg[8:]
elif arg.startswith('-summary:'):
edit_summary = arg[9:]
elif arg.startswith('-allowoverlap'):
allowoverlap = True
elif arg.startswith('-manualinput'):
manual_input = True
elif arg.startswith('-replacementfile'):
if len(commandline_replacements) % 2:
replacement_file_arg_misplaced = True
if arg == '-replacementfile':
replacement_file = pywikibot.input(
u'Please enter the filename to read replacements from:')
else:
replacement_file = arg[len('-replacementfile:'):]
else:
commandline_replacements.append(arg)
site = pywikibot.Site()
if len(commandline_replacements) % 2:
pywikibot.error('Incomplete command line pattern replacement pair.')
return False
if replacement_file_arg_misplaced:
pywikibot.error(
'-replacementfile used between a pattern replacement pair.')
return False
if replacement_file:
try:
with codecs.open(replacement_file, 'r', 'utf-8') as f:
# strip newlines, but not other characters
file_replacements = f.read().splitlines()
except (IOError, OSError) as e:
pywikibot.error(u'Error loading {0}: {1}'.format(
replacement_file, e))
return False
if len(file_replacements) % 2:
pywikibot.error(
'{0} contains an incomplete pattern replacement pair.'.format(
replacement_file))
return False
# Strip BOM from first line
file_replacements[0].lstrip(u'\uFEFF')
commandline_replacements.extend(file_replacements)
if not (commandline_replacements or fixes_set) or manual_input:
old = pywikibot.input(
u'Please enter the text that should be replaced:')
while old:
new = pywikibot.input(u'Please enter the new text:')
commandline_replacements += [old, new]
old = pywikibot.input(
'Please enter another text that should be replaced,'
'\nor press Enter to start:')
single_summary = None
for i in range(0, len(commandline_replacements), 2):
replacement = Replacement(commandline_replacements[i],
commandline_replacements[i + 1])
if not single_summary:
single_summary = i18n.twtranslate(site, 'replace-replacing', {
'description':
' (-%s +%s)' % (replacement.old, replacement.new)
})
replacements.append(replacement)
if not edit_summary:
if single_summary:
pywikibot.output(u'The summary message for the command line '
'replacements will be something like: %s' %
single_summary)
if fixes_set:
pywikibot.output('If a summary is defined for the fix, this '
'default summary won\'t be applied.')
edit_summary = pywikibot.input(
'Press Enter to use this automatic message, or enter a '
'description of the\nchanges your bot will make:')
# Perform one of the predefined actions.
for fix in fixes_set:
try:
fix = fixes.fixes[fix]
except KeyError:
pywikibot.output(u'Available predefined fixes are: %s' %
', '.join(fixes.fixes.keys()))
if not fixes.user_fixes_loaded:
pywikibot.output('The user fixes file could not be found: '
'{0}'.format(fixes.filename))
return
if "msg" in fix:
if isinstance(fix['msg'], basestring):
set_summary = i18n.twtranslate(site, str(fix['msg']))
else:
set_summary = i18n.translate(site, fix['msg'], fallback=True)
else:
set_summary = None
replacement_set = ReplacementList(fix.get('regex'),
fix.get('exceptions'),
fix.get('nocase'), set_summary)
for replacement in fix['replacements']:
summary = None if len(replacement) < 3 else replacement[2]
if chars.contains_invisible(replacement[0]):
pywikibot.warning('The old string "{0}" contains formatting '
'characters like U+200E'.format(
chars.replace_invisible(replacement[0])))
if chars.contains_invisible(replacement[1]):
pywikibot.warning('The new string "{0}" contains formatting '
'characters like U+200E'.format(
chars.replace_invisible(replacement[1])))
replacements.append(
ReplacementListEntry(
old=replacement[0],
new=replacement[1],
fix_set=replacement_set,
edit_summary=summary,
))
# Set the regular expression flags
flags = re.UNICODE
if caseInsensitive:
flags = flags | re.IGNORECASE
if dotall:
flags = flags | re.DOTALL
if multiline:
flags = flags | re.MULTILINE
# Pre-compile all regular expressions here to save time later
for replacement in replacements:
replacement.compile(regex, flags)
precompile_exceptions(exceptions, regex, flags)
if xmlFilename:
try:
xmlStart
except NameError:
xmlStart = None
gen = XmlDumpReplacePageGenerator(xmlFilename, xmlStart, replacements,
exceptions, site)
elif useSql:
whereClause = 'WHERE (%s)' % ' OR '.join([
"old_text RLIKE '%s'" % prepareRegexForMySQL(old_regexp.pattern)
for (old_regexp, new_text) in replacements
])
if exceptions:
exceptClause = 'AND NOT (%s)' % ' OR '.join([
"old_text RLIKE '%s'" % prepareRegexForMySQL(exc.pattern)
for exc in exceptions
])
else:
exceptClause = ''
query = u"""
SELECT page_namespace, page_title
FROM page
JOIN text ON (page_id = old_id)
%s
%s
LIMIT 200""" % (whereClause, exceptClause)
gen = pagegenerators.MySQLPageGenerator(query)
gen = genFactory.getCombinedGenerator(gen)
if not gen:
# syntax error, show help text from the top of this file
pywikibot.showHelp('replace')
return
preloadingGen = pagegenerators.PreloadingGenerator(gen)
bot = ReplaceRobot(preloadingGen, replacements, exceptions, acceptall,
allowoverlap, recursive, add_cat, sleep, edit_summary,
site)
site.login()
bot.run()
# Explicitly call pywikibot.stopme().
# It will make sure the callback is triggered before replace.py is unloaded.
pywikibot.stopme()
pywikibot.output(u'\n%s pages changed.' % bot.changed_pages)
def test_replace(self):
"""Test replace_invisible."""
self.assertEqual(chars.replace_invisible('Hello world!'), 'Hello world!')
self.assertEqual(chars.replace_invisible('\u200eRTL\u200f'), '<200e>RTL<200f>')
def main(*args: str) -> None:
"""
Process command line arguments and invoke bot.
If args is an empty list, sys.argv is used.
:param args: command line arguments
"""
options = {}
gen = None
# summary message
edit_summary = ''
# Array which will collect commandline parameters.
# First element is original text, second element is replacement text.
commandline_replacements = []
file_replacements = []
# A list of 2-tuples of original text and replacement text.
replacements = []
# Should the elements of 'replacements' and 'exceptions' be interpreted
# as regular expressions?
regex = False
# Predefined fixes from dictionary 'fixes' (see above).
fixes_set = []
# the dump's path, either absolute or relative, which will be used
# if -xml flag is present
xmlFilename = None
xmlStart = None
sql_query = None # type: Optional[str]
# Set the default regular expression flags
flags = 0
# Request manual replacements even if replacements are already defined
manual_input = False
# Read commandline parameters.
genFactory = pagegenerators.GeneratorFactory(
disabled_options=['mysqlquery'])
local_args = pywikibot.handle_args(args)
local_args = genFactory.handle_args(local_args)
local_args, exceptions = handle_exceptions(*local_args)
for arg in local_args:
opt, _, value = arg.partition(':')
if opt == '-regex':
regex = True
elif opt == '-xmlstart':
xmlStart = value or pywikibot.input(
'Please enter the dumped article to start with:')
elif opt == '-xml':
xmlFilename = value or i18n.input('pywikibot-enter-xml-filename')
elif opt == '-mysqlquery':
sql_query = value
elif opt == '-fix':
fixes_set.append(value)
elif opt == '-sleep':
options['sleep'] = float(value)
elif opt in ('-allowoverlap', '-always', '-recursive'):
options[opt[1:]] = True
elif opt == '-nocase':
flags |= re.IGNORECASE
elif opt == '-dotall':
flags |= re.DOTALL
elif opt == '-multiline':
flags |= re.MULTILINE
elif opt == '-addcat':
options['addcat'] = value
elif opt == '-summary':
edit_summary = value
elif opt == '-automaticsummary':
edit_summary = True
elif opt == '-manualinput':
manual_input = True
elif opt == '-pairsfile':
file_replacements = handle_pairsfile(value)
else:
commandline_replacements.append(arg)
if file_replacements is None:
return
if len(commandline_replacements) % 2:
pywikibot.error('Incomplete command line pattern replacement pair.')
return
commandline_replacements += file_replacements
if not(commandline_replacements or fixes_set) or manual_input:
commandline_replacements += handle_manual()
# The summary stored here won't be actually used but is only an example
site = pywikibot.Site()
single_summary = None
for i in range(0, len(commandline_replacements), 2):
replacement = Replacement(commandline_replacements[i],
commandline_replacements[i + 1])
if not single_summary:
single_summary = i18n.twtranslate(
site, 'replace-replacing',
{'description':
' (-{} +{})'.format(replacement.old, replacement.new)}
)
replacements.append(replacement)
# Perform one of the predefined actions.
missing_fixes_summaries = [] # which a fixes/replacements miss a summary
generators_given = bool(genFactory.gens)
for fix_name in fixes_set:
try:
fix = fixes.fixes[fix_name]
except KeyError:
pywikibot.output('Available predefined fixes are: {}'
.format(', '.join(fixes.fixes.keys())))
if not fixes.user_fixes_loaded:
pywikibot.output('The user fixes file could not be found: {}'
.format(fixes.filename))
return
if not fix['replacements']:
pywikibot.warning('No replacements defined for fix "{}"'
.format(fix_name))
continue
if 'msg' in fix:
if isinstance(fix['msg'], str):
set_summary = i18n.twtranslate(site, str(fix['msg']))
else:
set_summary = i18n.translate(site, fix['msg'], fallback=True)
else:
set_summary = None
if not generators_given and 'generator' in fix:
gen_args = fix['generator']
if isinstance(gen_args, str):
gen_args = [gen_args]
for gen_arg in gen_args:
genFactory.handle_arg(gen_arg)
replacement_set = ReplacementList(fix.get('regex'),
fix.get('exceptions'),
fix.get('nocase'),
set_summary,
name=fix_name)
# Whether some replacements have a summary, if so only show which
# have none, otherwise just mention the complete fix
missing_fix_summaries = []
for index, replacement in enumerate(fix['replacements'], start=1):
summary = None if len(replacement) < 3 else replacement[2]
if not set_summary and not summary:
missing_fix_summaries.append(
'"{}" (replacement #{})'.format(fix_name, index))
if chars.contains_invisible(replacement[0]):
pywikibot.warning('The old string "{}" contains formatting '
'characters like U+200E'.format(
chars.replace_invisible(replacement[0])))
if (not callable(replacement[1])
and chars.contains_invisible(replacement[1])):
pywikibot.warning('The new string "{}" contains formatting '
'characters like U+200E'.format(
chars.replace_invisible(replacement[1])))
replacement_set.append(ReplacementListEntry(
old=replacement[0],
new=replacement[1],
fix_set=replacement_set,
edit_summary=summary,
))
# Exceptions specified via 'fix' shall be merged to those via CLI.
if replacement_set:
replacements.extend(replacement_set)
if replacement_set._exceptions is not None:
for k, v in replacement_set._exceptions.items():
if k in exceptions:
exceptions[k] = list(set(exceptions[k]) | set(v))
else:
exceptions[k] = v
if len(fix['replacements']) == len(missing_fix_summaries):
missing_fixes_summaries.append(
'"{}" (all replacements)'.format(fix_name))
else:
missing_fixes_summaries += missing_fix_summaries
if ((not edit_summary or edit_summary is True)
and (missing_fixes_summaries or single_summary)):
if single_summary:
pywikibot.output('The summary message for the command line '
'replacements will be something like: '
+ single_summary)
if missing_fixes_summaries:
pywikibot.output('The summary will not be used when the fix has '
'one defined but the following fix(es) do(es) '
'not have a summary defined: {}'
.format(', '.join(missing_fixes_summaries)))
if edit_summary is not True:
edit_summary = pywikibot.input(
'Press Enter to use this automatic message, or enter a '
'description of the\nchanges your bot will make:')
else:
edit_summary = ''
# Pre-compile all regular expressions here to save time later
for replacement in replacements:
replacement.compile(regex, flags)
precompile_exceptions(exceptions, regex, flags)
if xmlFilename:
gen = XmlDumpReplacePageGenerator(xmlFilename, xmlStart,
replacements, exceptions, site)
elif sql_query is not None:
# Only -excepttext option is considered by the query. Other
# exceptions are taken into account by the ReplaceRobot
gen = handle_sql(sql_query, replacements, exceptions['text-contains'])
gen = genFactory.getCombinedGenerator(gen, preload=True)
if pywikibot.bot.suggest_help(missing_generator=not gen):
return
bot = ReplaceRobot(gen, replacements, exceptions, site=site,
summary=edit_summary, **options)
site.login()
bot.run()
