code/coalesce_new_source_fields.py

"""Fill selected blank legacy fields from new PATSTAT source while preserving nonblank legacy values."""
import csv,gzip,json,os
from collections import defaultdict,Counter
from pathlib import Path
csv.field_size_limit(100_000_000)
BASE=Path('/Users/deep1003/data3/webofscience_ai_global_export/bibtex/ai_policy_organized_20260619/patstat')
OUT=BASE/'final_master_20260712'; RAW=BASE/'online_sql_exports/patstat_2026_spring_ai_expanded_v2_full_fields'
FINAL=OUT/'patstat_ai_complete_master_legacy_priority_20260712.csv.gz';TMP=OUT/'_coalesce.tmp.csv.gz'
NEW=OUT/'patstat_ai_new_collection_2023_2026_integrated_country_filled.csv.gz'
YEARS=(2023,2024,2025,2026)
def add(d,k,v):
 v=(v or '').strip()
 if v and v not in d[k]:d[k].append(v)
def code_count(v):
 return str(len({x.strip() for x in (v or '').split(';') if x.strip()})) if (v or '').strip() else ''
new={}
with gzip.open(NEW,'rt',encoding='utf-8',newline='') as f:
 for r in csv.DictReader(f):new[r['app_id']]={'all_ipc_codes':r.get('all_ipc_codes',''),'all_cpc_codes':r.get('all_cpc_codes','')}
people=defaultdict(lambda:defaultdict(list))
for y in YEARS:
 for p in sorted(RAW.glob(f'patstat_ai_person_{y}_{y}_rows_*.csv')):
  with p.open(encoding='utf-8-sig',newline='') as f:
   for r in csv.DictReader(f,delimiter=';'):
    a=r.get('app_id')
    if a not in new:continue
    for pref,seq in [('applicant','applt_seq_nr'),('inventor','invt_seq_nr')]:
     try:on=int(r.get(seq) or 0)>0
     except:on=False
     if not on:continue
     add(people[a],f'{pref}_std_ids',r.get('doc_std_name_id'))
     add(people[a],f'{pref}_countries',r.get('person_ctry_code'))
     add(people[a],f'{pref}_addresses',r.get('person_address'))
     name=r.get('doc_std_name') or r.get('psn_name') or r.get('person_name') or ''
     info='|'.join([r.get('person_id','').strip(),name.strip(),r.get('person_ctry_code','').strip(),r.get('person_address','').strip()]).rstrip('|')
     add(people[a],f'{pref}_info',info)
for a,x in new.items():
 p=people.get(a,{})
 x.update({'applicant_info':'; '.join(p.get('applicant_info',[])),'applicant_std_name_ids':'; '.join(p.get('applicant_std_ids',[])),'applicant_ctry_codes':'; '.join(p.get('applicant_countries',[])),'inventor_info':'; '.join(p.get('inventor_info',[])),'inventor_ctry_codes':'; '.join(p.get('inventor_countries',[])),'inventor_addresses':'; '.join(p.get('inventor_addresses',[]))})
targets=['all_ipc_codes','ipc_count','all_cpc_codes','cpc_count','applicant_info','applicant_std_name_ids','applicant_ctry_codes','inventor_info','inventor_ctry_codes','inventor_addresses']
filled=Counter();before=Counter();after=Counter();rows=0
with gzip.open(FINAL,'rt',encoding='utf-8',newline='') as src,gzip.open(TMP,'wt',encoding='utf-8',newline='') as dst:
 rd=csv.DictReader(src);w=csv.DictWriter(dst,fieldnames=rd.fieldnames);w.writeheader()
 for r in rd:
  rows+=1;n=new.get(r['app_id'],{})
  for c in targets:before[c]+=bool((r.get(c) or '').strip())
  for c in ['all_ipc_codes','all_cpc_codes','applicant_info','applicant_std_name_ids','applicant_ctry_codes','inventor_info','inventor_ctry_codes','inventor_addresses']:
   if not (r.get(c) or '').strip() and (n.get(c) or '').strip():r[c]=n[c];filled[c]+=1
  if not (r.get('ipc_count') or '').strip() and (r.get('all_ipc_codes') or '').strip():r['ipc_count']=code_count(r['all_ipc_codes']);filled['ipc_count']+=1
  if not (r.get('cpc_count') or '').strip() and (r.get('all_cpc_codes') or '').strip():r['cpc_count']=code_count(r['all_cpc_codes']);filled['cpc_count']+=1
  for c in targets:after[c]+=bool((r.get(c) or '').strip())
  w.writerow(r)
os.replace(TMP,FINAL)
report={'rows':rows,'rule':'preserve existing nonblank; fill blank from new PATSTAT source; derive counts from final code lists','targets':targets,'before_nonempty':dict(before),'filled_rows':dict(filled),'after_nonempty':dict(after),'remaining_missing':{c:rows-after[c] for c in targets}}
(OUT/'coalesced_fields_report.json').write_text(json.dumps(report,indent=2),encoding='utf-8')
print(json.dumps(report,indent=2))