def __find_best_series(book, config):
'''
Queries the databse to find a best guess for a series matching the given
ComicBook, based on its name, year, issue number, and other text attributes.
Returns SeriesRef if a reasonable guess was found, or None if one wasn't.
'''
# 1. obtain SeriesRefs for this book, removing some as dictated by prefs
series_refs = db.query_series_refs(book.series_s,
config.ignored_searchterms_sl)
series_refs = dbutils.filter_series_refs(series_refs,
config.ignored_publishers_sl,
config.ignored_before_year_n,
config.ignored_after_year_n,
config.never_ignore_threshold_n)
# 2. obtain the first, second, and third best matching SeriesRefs for the
# given book, if there are any.
primary = None
secondary = None
tertiary = None
if len(series_refs) > 0:
mscore = MatchScore()
def find_best_score(refs):
return reduce(
lambda x, y: x if mscore.compute_n(book, x) >= mscore.
compute_n(book, y) else y, refs) if refs else None
primary = find_best_score(series_refs)
if primary:
series_refs.remove(primary)
secondary = find_best_score(series_refs)
if secondary:
series_refs.remove(secondary)
tertiary = find_best_score(series_refs)
# 3. if our book is the first (or unknown) issue, figure out if the best
# matching series has a similar cover to the second or third best.
# if it does, we're probably dealing with a trade paperback and a
# regular issue, and we can't find the best series reliably, so we bail
is_first_issue = (lambda i : not i or \
(utils.is_number(i) and float(i)==1.0))(book.issue_num_s)
if is_first_issue and primary and secondary:
too_similar = False
SIMILARITY_THRESHOLD = __MATCH_THRESHOLD - 0.10
hash1 = __get_remote_hash(primary)
hash2 = __get_remote_hash(secondary)
if imagehash.similarity(hash1, hash2) > SIMILARITY_THRESHOLD:
too_similar = True
elif tertiary:
hash3 = __get_remote_hash(tertiary)
if imagehash.similarity(hash1, hash3) > SIMILARITY_THRESHOLD:
too_similar = True
primary = None if too_similar else primary
return primary
def __find_best_series(book, config):
'''
Queries the databse to find a best guess for a series matching the given
ComicBook, based on its name, year, issue number, and other text attributes.
Returns SeriesRef if a reasonable guess was found, or None if one wasn't.
'''
# 1. obtain SeriesRefs for this book, removing some as dictated by prefs
series_refs = db.query_series_refs( book.series_s,
config.ignored_searchterms_sl )
series_refs = dbutils.filter_series_refs(
series_refs,
config.ignored_publishers_sl,
config.ignored_before_year_n,
config.ignored_after_year_n,
config.never_ignore_threshold_n)
# 2. obtain the first, second, and third best matching SeriesRefs for the
# given book, if there are any.
primary = None
secondary = None
tertiary = None
if len(series_refs) > 0:
mscore = MatchScore()
def find_best_score( refs ):
return reduce( lambda x,y: x if mscore.compute_n(book, x)
>= mscore.compute_n(book,y) else y, refs) if refs else None
primary = find_best_score(series_refs)
if primary:
series_refs.remove(primary)
secondary = find_best_score(series_refs)
if secondary:
series_refs.remove(secondary)
tertiary = find_best_score(series_refs)
# 3. if our book is the first (or unknown) issue, figure out if the best
# matching series has a similar cover to the second or third best.
# if it does, we're probably dealing with a trade paperback and a
# regular issue, and we can't find the best series reliably, so we bail
is_first_issue = (lambda i : not i or \
(utils.is_number(i) and float(i)==1.0))(book.issue_num_s)
if is_first_issue and primary and secondary:
too_similar = False
SIMILARITY_THRESHOLD = __MATCH_THRESHOLD - 0.10
hash1 = __get_remote_hash(primary)
hash2 = __get_remote_hash(secondary)
if imagehash.similarity(hash1, hash2) > SIMILARITY_THRESHOLD:
too_similar = True
elif tertiary:
hash3 = __get_remote_hash(tertiary)
if imagehash.similarity(hash1, hash3) > SIMILARITY_THRESHOLD:
too_similar = True
primary = None if too_similar else primary
return primary
def are_the_same(hash1, hash2):
x = imagehash.similarity(hash1, hash2)
return x > __MATCH_THRESHOLD
def are_the_same(hash1, hash2):
x = imagehash.similarity(hash1, hash2)
return x > __MATCH_THRESHOLD