def addOrBuildBOWToIndex(writer, guid, index_data, full_corpus=False):
"""
Loads JSON file with BOW data to doc in index, NOT filtering for anything
"""
bow_filename=cp.Corpus.cachedDataIDString("bow",guid,index_data)
try:
bows=cp.Corpus.loadCachedJson(bow_filename)
except:
bows=None
if bows is None:
print("BOW not found, rebuilding")
bows=prebuildMulti(index_data["method"],
index_data["parameters"],
index_data["function_name"],
None,
None,
guid,
False,
[]) #!TODO rhetorical_annotations here?
# Note: prebuildMulti will return a dict[param]=list of bows
bows=bows[index_data["parameter"]]
## if not isinstance(bows, list):
## print("BOWS IS NOT A LIST")
## print("guid:", guid)
## print("index_data:", index_data)
## print("Type:", type(bows))
assert isinstance(bows,list)
addLoadedBOWsToIndex(writer, guid, bows, index_data)
def addOrBuildBOWToIndexExcludingCurrent(writer, guid, exclude_list, max_year, index_data, full_corpus=False):
"""
Loads JSON file with BOW data to index, filtering for
inlink_context, excluding what bits
came from the current exclude_list, posterior year, same author, etc.
"""
bow_filename=cp.Corpus.cachedDataIDString("bow",guid,index_data)
try:
bows=cp.Corpus.loadCachedJson(bow_filename)
except:
bows=None
if not bows:
bows=prebuildMulti(index_data["method"],
index_data["parameters"],
index_data["function_name"],
None,
None,
guid,
False,
[]) #!TODO rhetorical_annotations here?
assert isinstance(bows,list)
# joinTogetherContext?
bows=doc_representation.filterInlinkContext(bows, exclude_list, max_year, full_corpus=full_corpus)
assert isinstance(bows,list)
addLoadedBOWsToIndex(writer, guid, bows, {"method":index_data["method"],
"parameter":index_data["parameter"]})
