ferrero-opentext/Python-Version/config/country_code_mappings.yaml
DJP d5c71b4c45 Add country code mapping system (ISO -> DAM codes)
- Created config/country_code_mappings.yaml with 165 country mappings
- ISO codes (used in filenames) now map to DAM-specific codes
- Codes under review set to XX as placeholder
- Added load_country_code_mappings() to config_loader.py
- Updated MetadataExtractorMVP to load and apply country mappings
- Added _map_country_code() and _get_field_value() helper methods
- Country mapping applies in both full and folder-only modes

Key mappings:
- BD (Bangladesh) -> BG (DAM code, though appears incorrect)
- DE (Germany) -> DE (same)
- IT (Italy) -> IT (same)
- Most codes under review -> XX (placeholder)

Mapping file can be edited without code changes - updates apply automatically

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>
2025-11-05 19:15:26 -05:00

256 lines
7.6 KiB
YAML

# Country Code Mappings: ISO 3166-1 Alpha-2 -> DAM Codes
#
# Purpose: Maps standard ISO country codes (used in filenames) to DAM-specific codes
# Last Updated: November 5, 2025
#
# Special Values:
# - XX: Code under review/question (use XX as placeholder)
# - Same code: When ISO and DAM codes match, list explicitly for clarity
#
# Format:
# iso_code: dam_code
#
# IMPORTANT: Some DAM codes appear incorrect (e.g., BG for Bangladesh when BG should be Bulgaria)
# These mappings reflect current DAM system requirements, not correct ISO standards
# ============================================================================
# Special Cases & Global Entries
# ============================================================================
GL: XX # GLOBAL - Under review
# ANDHRA PRADESH - Not a country, state in India
# GOOBAL - Appears to be typo for GLOBAL
# ============================================================================
# Countries - ISO to DAM Code Mapping
# ============================================================================
# A
AF: XX # AFGHANISTAN - Under review
AL: XX # ALBANIA - Under review
DZ: XX # ALGERIA - Under review
AO: XX # ANGOLA - Under review
AR: XX # ARGENTINA - Under review
AM: XX # ARMENIA - Under review
AW: XX # ARUBA - Under review
AU: XX # AUSTRALIA - Under review
AT: XX # AUSTRIA - Under review
AZ: XX # AZERBAIJAN - Under review
# B
BS: XX # BAHAMAS - Under review
BH: XX # BAHRAIN - Under review
BD: BG # BANGLADESH - Maps to BG (NOTE: This appears incorrect as BG is Bulgaria ISO code)
BY: XX # BELARUS - Under review
BE: XX # BELGIUM - Under review
BZ: XX # BELIZE - Under review
BJ: XX # BENIN - Under review
BM: XX # BERMUDA - Under review
BO: XX # BOLIVIA - Under review
BA: XX # BOSNIA AND HERZEGOVINA - Under review
BW: XX # BOTSWANA - Under review
BR: XX # BRAZIL - Under review
BN: XX # BRUNEI - Under review
BG: XX # BULGARIA - Under review
# C
KH: XX # CAMBODIA - Under review
CM: XX # CAMEROON - Under review
CA: XX # CANADA - Under review
CV: XX # CAPE VERDE - Under review
CF: XX # CENTRAL AFRICAN REP - Under review
CO: XX # CONGO DEM REP - Under review (Note: Should be CD for Dem Rep, CG for Republic)
CR: XX # COSTA RICA - Under review
HR: XX # CROATIA - Under review
CW: XX # CURACAO - Under review
CY: XX # CYPRUS - Under review
CZ: XX # CZECH REPUBLIC - Under review
# D
DK: XX # DENMARK - Under review
DJ: XX # DJIBOUTI - Under review
DO: XX # DOMINICAN REPUBLIC - Under review
# E
EC: XX # ECUADOR - Under review
EG: XX # EGYPT - # Under review
SV: XX # EL SALVADOR - Under review
ER: XX # ERITREA - Under review
EE: XX # ESTONIA - Under review
ET: XX # ETHIOPIA - Under review
# F
FI: XX # FINLAND - Under review
FR: XX # FRANCE - Under review
GF: XX # FRENCH GUIANA - Under review
# G
GE: XX # GEORGIA - Under review
DE: DE # GERMANY - Matches
GH: XX # GHANA - Under review
GR: XX # GREECE - Under review
GL: XX # GREENLAND - Under review (conflicts with GLOBAL above)
GD: XX # GRENADA - Under review
GP: XX # GUADELOUPE - Under review
GT: XX # GUATEMALA - Under review
GN: XX # GUINEA - Under review
GQ: XX # GUINEA BISSAU - Under review (Note: Should be GW for Guinea-Bissau)
GY: XX # GUYANA - Under review
# H
HT: XX # HAITI - Under review
HN: XX # HONDURAS - Under review
HK: XX # HONG KONG - Under review
HU: XX # HUNGARY - Under review
# I
IS: XX # ICELAND - Under review
IN: XX # INDIA - Under review
ID: XX # INDONESIA - Under review
IQ: XX # IRAQ - Under review
IE: XX # IRELAND - Under review
IL: XX # ISRAEL - Under review
IT: IT # ITALY - Matches
# J
JM: XX # JAMAICA - Under review
JP: XX # JAPAN - Under review
JO: XX # JORDAN - Under review
# K
KZ: XX # KAZAKHSTAN - Under review
KE: XX # KENYA - Under review
KR: XX # KOREA - Under review (Note: Should be KR for South Korea, KP for North Korea)
KW: XX # KUWAIT - Under review
KG: XX # KYRGYZSTAN - Under review
# L
LA: XX # LAO - Under review
LV: XX # LATVIA - Under review
LB: XX # LEBANON - Under review
LR: XX # LIBERIA - Under review
LT: XX # LITHUANIA - Under review
LU: XX # LUXEMBOURG - Under review
# M
MK: XX # MACEDONIA - Under review
MG: XX # MADAGASCAR - Under review
MW: XX # MALAWI - Under review
MY: XX # MALAYSIA - Under review
MV: XX # MALDIVES - Under review
ML: XX # MALI - Under review
MT: XX # MALTA - Under review
MQ: XX # MARTINIQUE - Under review
MR: XX # MAURITANIA - Under review
MU: XX # MAURITIUS - Under review
YT: XX # MAYOTTE - Under review
MX: XX # MEXICO - Under review
MD: XX # MOLDOVA - Under review
MN: XX # MONGOLIA - Under review
ME: XX # MONTENEGRO - Under review
MA: XX # MOROCCO - Under review
MZ: XX # MOZAMBIQUE - Under review
# N
NA: XX # NAMIBIA - Under review
NP: XX # NEPAL - Under review
NL: XX # NETHERLANDS - Under review
NC: XX # NEW CALEDONIA - Under review
NZ: XX # NEW ZEALAND - Under review
NI: XX # NICARAGUA - Under review
NE: XX # NIGER - Under review
NG: XX # NIGERIA - Under review
NO: XX # NORWAY - Under review
# P
PK: XX # PAKISTAN - Under review
PA: XX # PANAMA - Under review
PG: XX # PAPUA NEW GUINEA - Under review
PY: XX # PARAGUAY - Under review
PE: XX # PERU - Under review
PH: XX # PHILIPPINES - Under review
PL: XX # POLAND - Under review
PT: XX # PORTUGAL - Under review
PR: XX # PUERTO RICO - Under review
# Q
QA: XX # QATAR - Under review
# R
RE: XX # REUNION - Under review
RO: XX # ROMANIA - Under review
RU: XX # RUSSIA - Under review
RW: XX # RWANDA - Under review
# S
LC: XX # SAINT LUCIA - Under review
WS: XX # SAMOA - Under review
SA: XX # SAUDI ARABIA - Under review
SN: XX # SENEGAL - Under review
RS: XX # SERBIA - Under review
SC: XX # SEYCHELLES - Under review
SL: XX # SIERRA LEONE - Under review
SG: XX # SINGAPORE - Under review
SK: XX # SLOVAKIA - Under review
SI: XX # SLOVENIA - Under review
SO: XX # SOMALIA - Under review
ZA: XX # SOUTH AFRICA - Under review
SS: XX # SOUTH SUDAN - Under review
ES: XX # SPAIN - Under review
LK: XX # SRI LANKA - Under review
SD: XX # SUDAN - Under review
SR: XX # SURINAME - Under review
SZ: XX # SWAZILAND - Under review
SE: XX # SWEDEN - Under review
CH: XX # SWITZERLAND - Under review
SY: XX # SYRIA - Under review
# T
TW: XX # TAIWAN - Under review
TJ: XX # TAJIKISTAN - Under review
TZ: XX # TANZANIA - Under review
TH: XX # THAILAND - Under review
TG: XX # TOGO - Under review
TT: XX # TRINIDAD AND TOBAGO - Under review
TN: XX # TUNISIA - Under review
TR: XX # TURKEY - Under review
TM: XX # TURKMENISTAN - Under review
# U
UG: XX # UGANDA - Under review
UA: XX # UKRAINE - Under review
AE: XX # UNITED ARAB EMIRATES - Under review
GB: XX # UNITED KINGDOM - Under review
US: XX # UNITED STATES - Under review
UY: XX # URUGUAY - Under review
UZ: XX # UZBEKISTAN - Under review
# V
VE: XX # VENEZUELA - Under review
VN: XX # VIETNAM - Under review
# Y
YE: XX # YEMEN - Under review
# Z
ZM: XX # ZAMBIA - Under review
ZW: XX # ZIMBABWE - Under review
# ============================================================================
# Notes on Problematic Mappings
# ============================================================================
#
# Issues identified in source data:
# 1. BANGLADESH (ISO: BD) -> DAM: BG
# Problem: BG is the ISO code for Bulgaria, not Bangladesh
# This suggests a data quality issue in the DAM system
#
# 2. Many entries marked for review (yellow highlight in source)
# These are set to XX until confirmed
#
# 3. Some entries like "ANDHRA PRADESH" are not countries
# These are Indian states and should not have country codes
#
# 4. "GOOBAL" appears to be a typo for "GLOBAL"
#
# When codes are confirmed, update this file and the changes will
# automatically apply to all uploads without code modifications.