def bracketpaths(paths):
"add brackets to disambiguate paths (and remove Eq wrapper)"
spans = dct.count(concat(paths))
hapax = set(node for (node,n) in spans.items() if n==1)
firsts = dict((node,findif(elem(node),paths)[-1]) for node in spans)
lasts = dict((node,findif(elem(node),reversed(paths))[-1]) for node in spans)
@typecheck([Eq], [str])
def bracket(path):
first = edge(path, firsts, hapax)
last = edge(path, lasts, hapax)
if first != -1:
return map(Eq.get,path[:first+1])+["["]+map(Eq.get,path[first+1:])
elif last != -1:
return map(Eq.get,path[:last])+["]"]+map(Eq.get,path[last:])
else:
return map(Eq.get, path)
return map(bracket, paths)
def groupedSites(path, sites):
"path*[site] -> {site:[filename]}"
corpora = dct.collapse(filter(visible, os.listdir(path)),
keymap=lambda f: findif(f.startswith, sites))
if None in corpora:
# print ("Missing:", corpora[None])
del corpora[None]
return corpora