def _build_collections(self, branch_forms):
# parse into high-level records (collections, list collections, and singletons)
collections = set(k.split("_")[0] for k in branch_forms)
collections -= set(k for k in collections
if k.startswith("n") and k[1:] in collections)
# Create offsets virtual arrays
for name in collections:
if "n" + name in branch_forms:
branch_forms["o" + name] = transforms.counts2offsets_form(
branch_forms["n" + name])
# Create global index virtual arrays for indirection
idxbranches = [k for k in branch_forms if "Idx" in k]
for name in collections:
indexers = [k for k in idxbranches if k.startswith(name + "_")]
for k in indexers:
target = k[len(name) + 1:k.find("Idx")]
target = target[0].upper() + target[1:]
if target not in collections:
raise RuntimeError(
"Parsing indexer %s, expected to find collection %s but did not"
% (k, target))
branch_forms[k + "G"] = transforms.local2global_form(
branch_forms[k], branch_forms["o" + target])
# Create nested indexer from Idx1, Idx2, ... arrays
for name, indexers in self.nested_items.items():
if all(idx in branch_forms for idx in indexers):
branch_forms[name] = transforms.nestedindex_form(
[branch_forms[idx] for idx in indexers])
# Create any special arrays
for name, (fcn, args) in self.special_items.items():
if all(k in branch_forms for k in args):
branch_forms[name] = fcn(*(branch_forms[k] for k in args))
output = {}
for name in collections:
mixin = self.mixins.get(name, "NanoCollection")
if "o" + name in branch_forms and name not in branch_forms:
# list collection
offsets = branch_forms["o" + name]
content = {
k[len(name) + 1:]: branch_forms[k]
for k in branch_forms if k.startswith(name + "_")
}
output[name] = zip_forms(content,
name,
record_name=mixin,
offsets=offsets)
output[name]["content"]["parameters"].update({
"__doc__":
offsets["parameters"]["__doc__"],
"collection_name":
name,
})
elif "o" + name in branch_forms:
# list singleton, can use branch's own offsets
output[name] = branch_forms[name]
output[name]["parameters"].update({
"__array__": mixin,
"collection_name": name
})
elif name in branch_forms:
# singleton
output[name] = branch_forms[name]
else:
# simple collection
output[name] = zip_forms(
{
k[len(name) + 1:]: branch_forms[k]
for k in branch_forms if k.startswith(name + "_")
},
name,
record_name=mixin,
)
output[name]["parameters"].update({"collection_name": name})
return output
def _build_collections(self, branch_forms):
# parse into high-level records (collections, list collections, and singletons)
collections = set(k.split("_")[0] for k in branch_forms)
collections -= set(
k for k in collections if k.startswith("n") and k[1:] in collections
)
isData = "GenPart" not in collections
# Create offsets virtual arrays
for name in collections:
if "n" + name in branch_forms:
branch_forms["o" + name] = transforms.counts2offsets_form(
branch_forms["n" + name]
)
# Create global index virtual arrays for indirection
for indexer, target in self.cross_references.items():
if target.startswith("Gen") and isData:
continue
if indexer not in branch_forms:
if self.warn_missing_crossrefs:
warnings.warn(
f"Missing cross-reference index for {indexer} => {target}",
RuntimeWarning,
)
continue
if "o" + target not in branch_forms:
if self.warn_missing_crossrefs:
warnings.warn(
f"Missing cross-reference target for {indexer} => {target}",
RuntimeWarning,
)
continue
branch_forms[indexer + "G"] = transforms.local2global_form(
branch_forms[indexer], branch_forms["o" + target]
)
# Create nested indexer from Idx1, Idx2, ... arrays
for name, indexers in self.nested_items.items():
if all(idx in branch_forms for idx in indexers):
branch_forms[name] = transforms.nestedindex_form(
[branch_forms[idx] for idx in indexers]
)
# Create nested indexer from n* counts arrays
for name, (local_counts, target) in self.nested_index_items.items():
if local_counts in branch_forms and "o" + target in branch_forms:
branch_forms[name] = transforms.counts2nestedindex_form(
branch_forms[local_counts], branch_forms["o" + target]
)
# Create any special arrays
for name, (fcn, args) in self.special_items.items():
if all(k in branch_forms for k in args):
branch_forms[name] = fcn(*(branch_forms[k] for k in args))
output = {}
for name in collections:
mixin = self.mixins.get(name, "NanoCollection")
if "o" + name in branch_forms and name not in branch_forms:
# list collection
offsets = branch_forms["o" + name]
content = {
k[len(name) + 1 :]: branch_forms[k]
for k in branch_forms
if k.startswith(name + "_")
}
output[name] = zip_forms(
content, name, record_name=mixin, offsets=offsets
)
output[name]["content"]["parameters"].update(
{
"__doc__": offsets["parameters"]["__doc__"],
"collection_name": name,
}
)
elif "o" + name in branch_forms:
# list singleton, can use branch's own offsets
output[name] = branch_forms[name]
output[name]["parameters"].update(
{"__array__": mixin, "collection_name": name}
)
elif name in branch_forms:
# singleton
output[name] = branch_forms[name]
else:
# simple collection
output[name] = zip_forms(
{
k[len(name) + 1 :]: branch_forms[k]
for k in branch_forms
if k.startswith(name + "_")
},
name,
record_name=mixin,
)
output[name]["parameters"].update({"collection_name": name})
return output
