def calculate_pagerank_scores(ref_topic_map):
from sefaria.pagesheetrank import pagerank_rank_ref_list
pr_map = {}
pr_seg_map = {} # keys are (topic, seg_tref). used for sheet relevance
for topic, ref_list in tqdm(ref_topic_map.items(), desc='calculate pr'):
oref_list = []
for tref in ref_list:
try:
oref_list += [Ref(tref)]
except InputError:
continue
oref_pr_list = pagerank_rank_ref_list(oref_list, normalize=True)
for oref, pr in oref_pr_list:
pr_map[(topic, oref.normal())] = pr
for seg_oref in oref.all_segment_refs():
pr_seg_map[(topic, seg_oref.normal())] = pr
return pr_map, pr_seg_map
def calculate_pagerank_scores(ref_topic_map):
from sefaria.pagesheetrank import pagerank_rank_ref_list
from statistics import mean
pr_map = {}
pr_seg_map = {} # keys are (topic, seg_tref). used for sheet relevance
for topic, ref_list in tqdm(ref_topic_map.items(), desc='calculate pr'):
oref_list = []
for tref in ref_list:
oref = get_ref_safely(tref)
if oref is None:
continue
oref_list += [oref]
seg_ref_map = {
r.normal(): [rr.normal() for rr in r.all_segment_refs()]
for r in oref_list
}
oref_pr_list = pagerank_rank_ref_list(oref_list,
normalize=True,
seg_ref_map=seg_ref_map)
for oref, pr in oref_pr_list:
pr_map[(topic, oref.normal())] = pr
for seg_tref in seg_ref_map[oref.normal()]:
pr_seg_map[(topic, seg_tref)] = pr
return pr_map, pr_seg_map