def make_alignments(verbose=False):
wl = load_burmish(sqlite=False, remote=False)
blacklist = []
renumber = {0: 0}
for k in wl:
cogids = wl[k, 'cogids'].strip()
concept = wl[k, 'concept']
wl[k][wl.header['tokens']] = clean_tokens(wl[k, 'tokens'])
if not cogids or cogids == 0:
blacklist += [(k, '?')]
else:
tokens = clean_tokens(wl[k, 'tokens'])
morphemes = split_tokens(tokens)
if '0' in tokens2class(tokens, 'sca'):
blacklist += [(k, '0')]
elif len(morphemes) != len(cogids.split(' ')):
blacklist += [(k, 'M')]
else:
for c in cogids.split(' '):
cogid = c+':'+concept
if cogid not in renumber:
new_val = max(renumber.values())+1
renumber[cogid] = new_val
else:
pass
C = {}
blist = [k[0] for k in blacklist]
for k in wl:
if k not in blist:
C[k] = [renumber[c+':'+wl[k, 'concept']] for c in wl[k,
'cogids'].split()]
else:
C[k] = [0]
wl.add_entries('pcogids', C, lambda x: x)
D = {}
D[0] = [h for h in sorted(wl.header, key=lambda x: wl.header[x]) if h not
in ['alignment']]
for k in wl:
if k not in blacklist:
D[k] = [wl[k, h] for h in D[0]]
if verbose: print(D[0])
alm = Alignments(D, ref='pcogids', conf=burmish_path('conf',
'wordlist.rc'))
if verbose: print(alm._mode)
if verbose:
for cogid, msa in alm.msa['pcogids'].items():
sca = SCA(msa)
sca.prog_align()
alm.align(method='library', iterate=True)
alm.output('tsv', filename=burmish_path('dumps', 'alignments'),
ignore='all', prettify=False)
for i, (k, r) in enumerate(blacklist):
if wl[k, 'cogids']:
print(i+1, r, k, wl[k, 'concept'], wl[k, 'doculect'], wl[k,
'tokens'], repr(wl[k, 'cogids']))
def compare_partial_colexifications():
alms = get_alignments()
G = nx.Graph()
for k in alms:
concept = alms[k, "concept"]
morphemes = split_tokens(alms[k, "tokens"])
doculect = alms[k, "doculect"]
if concept not in G.node:
G.add_node(concept, bipartite=0)
for m in morphemes:
idf = "".join(m) + "/" + doculect
if idf not in G.node:
G.add_node(idf, bipartite=1)
G.add_edge(concept, idf)
print("printing now")
save_network("bipartite.gml", G)
nodes = {n for n, d in G.nodes(data=True) if d["bipartite"] == 0}
G2 = nx.bipartite.collaboration_weighted_projected_graph(G, nodes)
save_network("projected.gml", G2)
return G
def make_alignments(verbose=False):
wl = load_burmish(sqlite=False, remote=False)
blacklist = []
renumber = {0: 0}
for k in wl:
cogids = wl[k, 'cogids'].strip()
concept = wl[k, 'concept']
wl[k][wl.header['tokens']] = clean_tokens(wl[k, 'tokens'])
if not cogids or cogids == 0:
blacklist += [(k, '?')]
else:
tokens = clean_tokens(wl[k, 'tokens'])
morphemes = split_tokens(tokens)
if '0' in tokens2class(tokens, 'sca'):
blacklist += [(k, '0')]
elif len(morphemes) != len(cogids.split(' ')):
blacklist += [(k, 'M')]
else:
for c in cogids.split(' '):
cogid = c + ':' + concept
if cogid not in renumber:
new_val = max(renumber.values()) + 1
renumber[cogid] = new_val
else:
pass
C = {}
blist = [k[0] for k in blacklist]
for k in wl:
if k not in blist:
C[k] = [
renumber[c + ':' + wl[k, 'concept']]
for c in wl[k, 'cogids'].split()
]
else:
C[k] = [0]
wl.add_entries('pcogids', C, lambda x: x)
D = {}
D[0] = [
h for h in sorted(wl.header, key=lambda x: wl.header[x])
if h not in ['alignment']
]
for k in wl:
if k not in blacklist:
D[k] = [wl[k, h] for h in D[0]]
if verbose: print(D[0])
alm = Alignments(D,
ref='pcogids',
conf=burmish_path('conf', 'wordlist.rc'))
if verbose: print(alm._mode)
if verbose:
for cogid, msa in alm.msa['pcogids'].items():
sca = SCA(msa)
sca.prog_align()
alm.align(method='library', iterate=True)
alm.output('tsv',
filename=burmish_path('dumps', 'alignments'),
ignore='all',
prettify=False)
for i, (k, r) in enumerate(blacklist):
if wl[k, 'cogids']:
print(i + 1, r, k, wl[k, 'concept'], wl[k, 'doculect'],
wl[k, 'tokens'], repr(wl[k, 'cogids']))