def filter_junk_strings(substrings):
"""Filter out miscellaneous junk strings"""
filtered_substrings = {}
for k,v in substrings.items():
if not common.is_junk(k):
filtered_substrings[k] = v
return filtered_substrings
def get_substrings(string, k, allowed_substrings):
"""Return all substrings of length >= <k> in <string> as a dict of
substring:count where there are count occurrences of substring in
<string>
If <allowed_substrings> then only allow substrings from <allowed_substrings>.
Performance
-----------
The returned substring:count dict will no longer than
<allowed_substrings> or <parent_keys> so the best way to guarantee
performance is to find short key sets.
"""
common.report('get_substrings:k=%2d,allowed_substrings=%5d,size=%7d' %
(k, len(allowed_substrings) if allowed_substrings else -1, len(string)))
substrings = {}
n = len(string)
for i in range(n-k):
pattern = string[i:i+k]
if common.is_junk(pattern):
continue
if allowed_substrings:
if not pattern in allowed_substrings:
continue
if not pattern in substrings:
substrings[pattern] = 0
substrings[pattern] += 1
return substrings
def get_child_offsets(file_names, test_files, offsets_dict, k):
""" Given a set of substrings of length <k> defined by offsets in a set of
test_files, return a dict of substrings of length k+1
where
offsets_dict[<filename>][<substring>] is the set of offsets of <substring>
in test_files[<filename>]
<file_names> is keys of test_files in the desired sort order (shorter first)
Performance
-----------
This is the inner loop of the program.
The returned dict will no longer than offsets_dict and string searches are on existing
substrings + 1 character to left or right so there is not that much text to search.
"""
common.report('get_child_offsets(file_names=%d,test_files=%d,%d,substrings=%d,k=%d)' %
(len(file_names), len(test_files), len(offsets_dict), len(offsets_dict.values()[0]), k))
parent_substrings = offsets_dict[file_names[0]].keys()
child_offsets_dict = {}
allowed_substrings = None
for name in file_names:
x = test_files[name]
child_offsets_dict[name] = {}
for key, ofs_set in offsets_dict[name].items():
# Use a list which unlike a set can be indexed and sorted
ofs_list = sorted(ofs_set)
# Remove parent offsets that would truncate substrings of length k+1
if ofs_list[0] == 0:
del(ofs_list[0])
if ofs_list[-1]+k+1 == len(x['text']):
del(ofs_list[-1])
# Create the child length k+1 substrings and add them to the child offsets dict
# ofs1 is the offset of the k+1 substring key1
for ofs in ofs_list:
for ofs1 in [ofs-1, ofs]:
key1 = x['text'][ofs1:ofs1+k+1]
#if len(key1) != k+1:
# print 'key="%s", key1="%s"' % (key, key1)
assert(len(key1) == k+1)
if allowed_substrings:
if not key1 in allowed_substrings:
continue
# Only allow keys with valid parents
if not key1[1:] in parent_substrings or not key1[:-1] in parent_substrings:
continue
# Get rid of the junk too
if common.is_junk(key1):
continue
# Got through all the filters. Add the new offset to the child dict
if not key1 in child_offsets_dict[name].keys():
child_offsets_dict[name][key1] = set([])
child_offsets_dict[name][key1].add(ofs1)
# Prune the entries with insufficient repeats
unpruned_len = len(child_offsets_dict[name].keys())
for key, ofs_set in child_offsets_dict[name].items():
if len(ofs_set) < x['repeats']:
del(child_offsets_dict[name][key])
# allowed_substrings is used as a filter in all but first pass through this loop
allowed_substrings = child_offsets_dict[name].keys()
common.report(' allowed_substrings=%3d,%3d,size=%7d' %
(unpruned_len, len(allowed_substrings), len(x['text'])))
# Need to go back and trim the substrings lists to allowed_substrings
# If this results in a zero length list for any file then returns
for name in file_names:
for key in child_offsets_dict[name].keys():
if not key in allowed_substrings:
del(child_offsets_dict[name][key])
if len(child_offsets_dict[name]) == 0:
return None
common.dump_dict('dumpfile_%03d' % (k+1), file_names, test_files, child_offsets_dict)
if common.is_validate():
if not validate_child_offsets(file_names, offsets_dict, child_offsets_dict, k):
raise ValueError
for name in file_names:
common.report('before=%3d,after=%3d,file=%s' % (len(offsets_dict[name]),
len(child_offsets_dict[name]),name))
return child_offsets_dict
