def _format_download_uri(etextno, mirror=None, prefer_ascii=False):
"""Returns the download location on the Project Gutenberg servers for a
given text.
Use prefer_ascii to control whether you want to fetch plaintext us-ascii
file first (default old behavior) or if you prefer UTF-8 then 8-bits then
plaintext.
Raises:
UnknownDownloadUri: If no download location can be found for the text.
"""
mirror = mirror or _GUTENBERG_MIRROR
if not _does_mirror_exist(mirror):
raise UnknownDownloadUriException(
'Could not reach Gutenberg mirror "{:s}". Try setting a '
'different mirror (https://www.gutenberg.org/MIRRORS.ALL) for '
'--mirror flag or GUTENBERG_MIRROR environment variable.'.format(
mirror))
# Check https://www.gutenberg.org/files/ for details about available
# extensions ;
# - .txt is plaintext us-ascii
# - -8.txt is 8-bit plaintext, multiple encodings
# - -0.txt is UTF-8
ascii_first = ('.txt', '-0.txt', '-8.txt')
utf8_first = ('-0.txt', '-8.txt', '.txt')
extensions = ascii_first if prefer_ascii else utf8_first
for extension in extensions:
uri = _format_download_uri_for_extension(etextno, extension, mirror)
if _does_uri_exist(uri):
return uri
raise UnknownDownloadUriException(
'Failed to find a textual download candidate for {} on {}. '
'Either the book does not exist or it is only available in '
'non-textual formats.'.format(etextno, mirror))
def _format_download_uri(etextno, mirror=None):
"""Returns the download location on the Project Gutenberg servers for a
given text.
Raises:
UnknownDownloadUri: If no download location can be found for the text.
"""
uri_root = mirror or _GUTENBERG_MIRROR
uri_root = uri_root.strip().rstrip('/')
_check_mirror_exists(uri_root)
extensions = ('.txt', '-8.txt', '-0.txt')
for extension in extensions:
path = _etextno_to_uri_subdirectory(etextno)
uri = '{root}/{path}/{etextno}{extension}'.format(
root=uri_root,
path=path,
etextno=etextno,
extension=extension)
response = requests.head(uri)
if response.ok:
return uri
raise UnknownDownloadUriException('Failed to find {0} on {1}.'
.format(etextno, uri_root))
def _check_mirror_exists(mirror):
response = requests.head(mirror)
if not response.ok:
raise UnknownDownloadUriException(
'Could not reach Gutenberg mirror "{0:s}". Try setting a '
'different mirror (https://www.gutenberg.org/MIRRORS.ALL) for '
'--mirror flag or GUTENBERG_MIRROR environment variable.'
.format(mirror))