def process_one(recid, tags, citations_fun):
"""Self-cites core func, executed on each recid"""
# First update this record then all its references
compute_and_store_self_citations(recid, tags, citations_fun)
references = get_refers_to(recid)
for recordid in references:
compute_and_store_self_citations(recordid, tags, citations_fun)
def process_one(recid, tags, citations_fun):
"""Self-cites core func, executed on each recid"""
# First update this record then all its references
compute_and_store_self_citations(recid, tags, citations_fun)
references = get_refers_to(recid)
for recordid in references:
compute_and_store_self_citations(recordid, tags, citations_fun)
def compute_and_store_self_citations(recid,
tags,
citations_fun,
verbose=False):
"""Compute and store self-cites in a table
Args:
- recid
- tags: used when bibauthorid is desactivated see get_author_tags()
in bibrank_selfcites_indexer
"""
assert recid
if verbose:
write_message("* processing %s" % recid)
references = get_refers_to(recid)
recids_to_check = set([recid]) | set(references)
placeholders = ','.join('%s' for r in recids_to_check)
rec_row = run_sql(
"SELECT MAX(`modification_date`) FROM `bibrec`"
" WHERE `id` IN (%s)" % placeholders, recids_to_check)
try:
rec_timestamp = rec_row[0]
except IndexError:
write_message("record not found")
return
cached_citations_row = run_sql("SELECT `count` FROM `rnkSELFCITES`"
" WHERE `last_updated` >= %s" \
" AND `id_bibrec` = %s", (rec_timestamp[0], recid))
if cached_citations_row and cached_citations_row[0][0]:
if verbose:
write_message("%s found (cached)" % cached_citations_row[0])
else:
cites = citations_fun(recid, tags)
sql = """REPLACE INTO rnkSELFCITES (`id_bibrec`, `count`, `references`,
`last_updated`) VALUES (%s, %s, %s, NOW())"""
references_string = ','.join(str(r) for r in references)
run_sql(sql, (recid, len(cites), references_string))
if verbose:
write_message("%s found" % len(cites))
def compute_and_store_self_citations(recid, tags, citations_fun, selfcites_dic,
verbose=False):
"""Compute and store self-cites in a table
Args:
- recid
- tags: used when bibauthorid is desactivated see get_author_tags()
in bibrank_selfcites_indexer
"""
assert recid
if verbose:
write_message("* processing %s" % recid)
references = get_refers_to(recid)
recids_to_check = set([recid]) | set(references)
placeholders = ','.join('%s' for r in recids_to_check)
rec_row = run_sql("SELECT MAX(`modification_date`) FROM `bibrec`"
" WHERE `id` IN (%s)" % placeholders, recids_to_check)
try:
rec_timestamp = rec_row[0]
except IndexError:
write_message("record not found")
return
cached_citations_row = run_sql("SELECT `count` FROM `rnkSELFCITES`"
" WHERE `last_updated` >= %s"
" AND `id_bibrec` = %s", (rec_timestamp[0], recid))
if cached_citations_row and cached_citations_row[0][0]:
if verbose:
write_message("%s found (cached)" % cached_citations_row[0])
else:
cites = citations_fun(recid, tags)
selfcites_dic[recid] = len(cites)
replace_cites(recid, cites)
sql = """REPLACE INTO rnkSELFCITES (`id_bibrec`, `count`, `references`,
`last_updated`) VALUES (%s, %s, %s, NOW())"""
references_string = ','.join(str(r) for r in references)
run_sql(sql, (recid, len(cites), references_string))
if verbose:
write_message("%s found" % len(cites))
def _find_citations_by(bib):
return get_refers_to(bib[2])
def _find_citations_by(bib):
return get_refers_to(bib[2])