def check_isbn(item, errors):
"""
Checks ISBN for validity.
Will accept both ISBN-10 and ISBN-13 formats
"""
isbn_list = item.get("isbn")
if isbn_list is None:
return
for idx, single_isbn in enumerate(isbn_list):
if not isbn.is_valid(single_isbn):
errors.add("ISBN #{idx} isn't valid".format(
idx=idx
))
continue
formatted = isbn.format(single_isbn)
if (formatted != single_isbn):
errors.add("ISBN #{idx} ({single_isbn}) should be reformatted to {formatted}".format(
idx=idx,
single_isbn=single_isbn,
formatted=formatted
))
if (isbn.isbn_type(single_isbn) != 'ISBN13'):
errors.add("ISBN-10 #{idx} ({single_isbn}) should be reformatted to ISBN-13 {formatted}".format(
idx=idx,
single_isbn=single_isbn,
formatted=isbn.to_isbn13(single_isbn)
))
def create_isbn(self, key, var_list=None):
tmp_list = []
for var in var_list:
# print(self.csvRow[var].strip())
tmp_elem = re.split('; |, ', self.csvRow[var].strip())
for i in range(len(tmp_elem)):
"""
Filter ISSN and ISBN only
"""
tmp_var_isbn = self.remove_non_isbn_chars(tmp_elem[i].upper())
if isbn.is_valid(tmp_var_isbn):
# print("Valid ISBN: " + tmp_var_isbn)
tmp_elem[i] = isbn.format(tmp_var_isbn)
else:
tmp_elem[i] = None
# Filter all elements, remove text chars, parenthesis etc
# print(tmp_elem)
tmp_list += tmp_elem
# print(tmp_list)
tmp_list = list(filter(None, tmp_list)) # remove empty list elements
# print(tmp_list)
# No semi-colon, so do a join here. Cannot use the function to split semi-colons
return '||'.join(filter(None, tmp_list)).strip()
def _format_isbn_match(match, strict=True):
"""Helper function to validate and format a single matched ISBN."""
isbn = match.group('code')
if stdnum_isbn:
try:
stdnum_isbn.validate(isbn)
except stdnum_isbn.ValidationError as e:
if strict:
raise
pywikibot.log('ISBN "%s" validation error: %s' % (isbn, e))
return isbn
return stdnum_isbn.format(isbn)
else:
try:
scripts_isbn.is_valid(isbn)
except scripts_isbn.InvalidIsbnException as e:
if strict:
raise
pywikibot.log('ISBN "%s" validation error: %s' % (isbn, e))
return isbn
isbn = scripts_isbn.getIsbn(isbn)
try:
isbn.format()
except scripts_isbn.InvalidIsbnException as e:
if strict:
raise
pywikibot.log('ISBN "%s" validation error: %s' % (isbn, e))
return isbn.code
def validate_isbn(item, errors):
"""
Checks ISBN for validity.
Will accept both ISBN-10 and ISBN-13 formats
"""
isbn_list = item.get("isbn")
if isbn_list is None:
return
for idx, single_isbn in enumerate(isbn_list):
if not isbn.is_valid(single_isbn):
errors.add("ISBN #{idx} isn't valid".format(
idx=idx
))
continue
formatted = isbn.format(single_isbn)
if (formatted != single_isbn):
errors.add("ISBN #{idx} ({single_isbn}) should be reformatted to {formatted}".format(
idx=idx,
single_isbn=single_isbn,
formatted=formatted
))
if (isbn.isbn_type(single_isbn) != 'ISBN13'):
errors.add("ISBN-10 #{idx} ({single_isbn}) should be reformatted to ISBN-13 {formatted}".format(
idx=idx,
single_isbn=single_isbn,
formatted=isbn.to_isbn13(single_isbn)
))
def _format_isbn_match(match, strict=True):
"""Helper function to validate and format a single matched ISBN."""
isbn = match.group('code')
if stdnum_isbn:
try:
stdnum_isbn.validate(isbn)
except stdnum_isbn.ValidationError as e:
if strict:
raise
pywikibot.log('ISBN "%s" validation error: %s' % (isbn, e))
return isbn
return stdnum_isbn.format(isbn)
else:
try:
scripts_isbn.is_valid(isbn)
except scripts_isbn.InvalidIsbnException as e:
if strict:
raise
pywikibot.log('ISBN "%s" validation error: %s' % (isbn, e))
return isbn
isbn = scripts_isbn.getIsbn(isbn)
try:
isbn.format()
except scripts_isbn.InvalidIsbnException as e:
if strict:
raise
pywikibot.log('ISBN "%s" validation error: %s' % (isbn, e))
return isbn.code
def show_isbn(_isbn):
isbn = ''
for code in _isbn.split(';'):
if stdisbn.is_valid(code):
isbn += stdisbn.format(code) + '; '
else:
isbn += code + '; '
if isbn:
# chop trailing '; '
return isbn[:-2]
def show_isbn(_isbn):
isbn = ''
for code in _isbn.split(';'):
if stdisbn.is_valid(code):
isbn += stdisbn.format(code) + '; '
else:
isbn += code + '; '
if isbn:
# chop trailing '; '
return isbn[:-2]
def _format_isbn_match(match, strict=True):
"""Helper function to validate and format a single matched ISBN."""
if not stdnum_isbn:
raise NotImplementedError(
'ISBN functionality not available. Install stdnum package.')
isbn = match.group('code')
try:
stdnum_isbn.validate(isbn)
except stdnum_isbn.ValidationError as e:
if strict:
raise
pywikibot.log('ISBN "%s" validation error: %s' % (isbn, e))
return isbn
return stdnum_isbn.format(isbn)
def main(raw_url, convert=True):
url = get_page_url(raw_url)
wikitext, times = get_wikitext(url)
code = mwparserfromhell.parse(wikitext)
count = 0
for template, raw_isbn, para in find_isbns(code):
if not check_isbn(raw_isbn):
continue
new_isbn = isbn.format(raw_isbn, convert=convert)
if raw_isbn != new_isbn:
count += 1
template.add(para, new_isbn)
return code, times, count, url
def _format_isbn_match(match, strict=True):
"""Helper function to validate and format a single matched ISBN."""
scripts_isbn = None
if not stdnum_isbn:
# For backwards compatibility, if stdnum.isbn is not available
# attempt loading scripts.isbn as an alternative implementation.
try:
import scripts.isbn as scripts_isbn
except ImportError:
raise NotImplementedError(
'ISBN functionality not available. Install stdnum package.')
warn('package stdnum.isbn not found; using scripts.isbn',
ImportWarning)
isbn = match.group('code')
if stdnum_isbn:
try:
stdnum_isbn.validate(isbn)
except stdnum_isbn.ValidationError as e:
if strict:
raise
pywikibot.log('ISBN "%s" validation error: %s' % (isbn, e))
return isbn
return stdnum_isbn.format(isbn)
else:
try:
scripts_isbn.is_valid(isbn)
except scripts_isbn.InvalidIsbnException as e:
if strict:
raise
pywikibot.log('ISBN "%s" validation error: %s' % (isbn, e))
return isbn
isbn = scripts_isbn.getIsbn(isbn)
try:
isbn.format()
except scripts_isbn.InvalidIsbnException as e:
if strict:
raise
pywikibot.log('ISBN "%s" validation error: %s' % (isbn, e))
return isbn.code
def _format_isbn_match(match, strict=True):
"""Helper function to validate and format a single matched ISBN."""
scripts_isbn = None
if not stdnum_isbn:
# For backwards compatibility, if stdnum.isbn is not available
# attempt loading scripts.isbn as an alternative implementation.
try:
import scripts.isbn as scripts_isbn
except ImportError:
raise NotImplementedError(
'ISBN functionality not available. Install stdnum package.')
warn('package stdnum.isbn not found; using scripts.isbn',
ImportWarning)
isbn = match.group('code')
if stdnum_isbn:
try:
stdnum_isbn.validate(isbn)
except stdnum_isbn.ValidationError as e:
if strict:
raise
pywikibot.log('ISBN "%s" validation error: %s' % (isbn, e))
return isbn
return stdnum_isbn.format(isbn)
else:
try:
scripts_isbn.is_valid(isbn)
except scripts_isbn.InvalidIsbnException as e:
if strict:
raise
pywikibot.log('ISBN "%s" validation error: %s' % (isbn, e))
return isbn
isbn = scripts_isbn.getIsbn(isbn)
try:
isbn.format()
except scripts_isbn.InvalidIsbnException as e:
if strict:
raise
pywikibot.log('ISBN "%s" validation error: %s' % (isbn, e))
return isbn.code
def validate_isbn(item, errors):
"""
Checks ISBN for validity.
Will accept both ISBN-10 and ISBN-13 formats
"""
isbn_list = item.get("isbn")
if isbn_list is None:
return
for idx, single_isbn in enumerate(isbn_list):
try:
isbn.validate(single_isbn)
except isbn.ValidationError as ex:
errors.add(f"ISBN #{idx} ({single_isbn}) is not valid: {ex}")
continue
formatted = isbn.format(single_isbn)
if (formatted != single_isbn):
errors.add(f"ISBN #{idx} ({single_isbn}) should be reformatted to {formatted}")
if (isbn.isbn_type(single_isbn) != 'ISBN13'):
errors.add(f"ISBN-10 #{idx} ({single_isbn}) should be reformatted to ISBN-13 {formatted}")
def formatIdentifier(stringToFormat, type):
"""
This is used to format/check ISBN and ISSN numbers
"""
if type.lower() == "isbn":
if isbn.is_valid(stringToFormat):
log.debug("OK ISBN: {0}".format(stringToFormat))
return isbn.format(stringToFormat)
else:
log.warning("Malformed ISBN: {0}".format(stringToFormat))
return stringToFormat
elif type.lower() == "issn":
if issn.is_valid(stringToFormat):
log.debug("OK ISSN: {0}".format(stringToFormat))
return issn.format(stringToFormat)
else:
log.warning("Malformed ISSN: {0}".format(stringToFormat))
return stringToFormat
def set_isbn(self):
"""Add ISBN's, both 10 and 13 char long."""
self.isbn_13 = None
self.isbn_10 = None
raw_ids = self.raw_data[1].get("identifiedBy")
if not raw_ids:
return
for r_id in raw_ids:
if r_id.get("@type").lower() == "isbn":
raw_isbn = r_id.get("value")
isbn_type = isbn_tool.isbn_type(raw_isbn)
if isbn_type:
formatted = isbn_tool.format(raw_isbn)
if isbn_type == "ISBN13":
prop = "isbn_13"
self.isbn_13 = isbn_tool.compact(formatted)
elif isbn_type == "ISBN10":
prop = "isbn_10"
self.isbn_10 = isbn_tool.compact(formatted)
self.add_statement(prop, formatted, ref=self.source)
def validate_isbn(item, errors):
"""
Checks ISBN for validity.
Will accept both ISBN-10 and ISBN-13 formats
"""
isbn_list = item.get("isbn")
if isbn_list is None:
return
for idx, single_isbn in enumerate(isbn_list):
try:
isbn.validate(single_isbn)
except isbn.ValidationError as ex:
errors.add(f"ISBN #{idx} ({single_isbn}) is not valid: {ex}")
continue
formatted = isbn.format(single_isbn)
if (formatted != single_isbn):
errors.add(
f"ISBN #{idx} ({single_isbn}) should be reformatted to {formatted}"
)
if (isbn.isbn_type(single_isbn) != 'ISBN13'):
errors.add(
f"ISBN-10 #{idx} ({single_isbn}) should be reformatted to ISBN-13 {formatted}"
)
def formatted_isbn(self):
"""
Returns ISBN in proper format woth dashes.
"""
return isbn.format(self.isbn)
