gigafibre-fsm/scripts/migration/update_item_descriptions.py

"""
Update ERPNext Item descriptions from legacy product data.
Uses hijack_desc from services and product_translate for French names.

Run inside erpnext-backend-1:
  /home/frappe/frappe-bench/env/bin/python /home/frappe/frappe-bench/update_item_descriptions.py
"""
import frappe
import pymysql
import os
import html

os.chdir("/home/frappe/frappe-bench/sites")
frappe.init(site="erp.gigafibre.ca", sites_path=".")
frappe.connect()
frappe.local.flags.ignore_permissions = True
print("Connected:", frappe.local.site)

conn = pymysql.connect(
    host="legacy-db",
    user="facturation",
    password="*******",
    database="gestionclient",
    cursorclass=pymysql.cursors.DictCursor
)

# ═══════════════════════════════════════════════════════════════
# STEP 1: Build a description map from legacy data
# ═══════════════════════════════════════════════════════════════
print("\n" + "=" * 60)
print("STEP 1: BUILD DESCRIPTION MAP")
print("=" * 60)

with conn.cursor() as cur:
    # Get all products with French translations
    cur.execute("""
        SELECT p.id, p.sku, p.price, p.category,
               pt.name as fr_name, pt.description_short as fr_desc
        FROM product p
        LEFT JOIN product_translate pt ON pt.product_id = p.id AND pt.language_id = 'fr'
        WHERE p.active = 1
    """)
    products = cur.fetchall()

    # Get category names
    cur.execute("SELECT id, name FROM product_cat")
    cats = {r["id"]: html.unescape(r["name"]) for r in cur.fetchall()}

    # Get most common hijack_desc per product (the custom description used on services)
    cur.execute("""
        SELECT product_id, hijack_desc, COUNT(*) as cnt
        FROM service
        WHERE hijack = 1 AND hijack_desc IS NOT NULL AND hijack_desc != ''
        GROUP BY product_id, hijack_desc
        ORDER BY product_id, cnt DESC
    """)
    hijack_descs = {}
    for r in cur.fetchall():
        pid = r["product_id"]
        if pid not in hijack_descs:
            hijack_descs[pid] = r["hijack_desc"]

conn.close()

# Build the best description for each SKU
desc_map = {}
for p in products:
    sku = p["sku"]
    if not sku:
        continue

    # Priority: French translation > hijack_desc > category name
    desc = p["fr_name"] or hijack_descs.get(p["id"]) or ""
    cat = cats.get(p["category"], "")

    desc_map[sku] = {
        "description": desc.strip() if desc else "",
        "item_group": cat,
        "price": float(p["price"] or 0),
    }

print("Products mapped: {}".format(len(desc_map)))

# Show samples
for sku in ["FTTB1000I", "TELEPMENS", "RAB24M", "HVIPFIXE", "FTT_HFAR", "CSERV", "RAB2X", "FTTH_LOCMOD"]:
    d = desc_map.get(sku, {})
    print("  {} -> desc='{}' group='{}' price={}".format(sku, d.get("description", ""), d.get("item_group", ""), d.get("price", "")))

# ═══════════════════════════════════════════════════════════════
# STEP 2: Update ERPNext Items
# ═══════════════════════════════════════════════════════════════
print("\n" + "=" * 60)
print("STEP 2: UPDATE ERPNEXT ITEMS")
print("=" * 60)

# Get all items that currently have "Legacy product ID" as description
items = frappe.db.sql("""
    SELECT name, item_name, description, item_group
    FROM "tabItem"
    WHERE description LIKE 'Legacy product ID%%'
""", as_dict=True)

print("Items with legacy placeholder description: {}".format(len(items)))

updated = 0
for item in items:
    sku = item["name"]
    legacy = desc_map.get(sku)
    if not legacy:
        continue

    new_desc = legacy["description"]
    new_name = new_desc if new_desc else sku

    if new_desc:
        frappe.db.sql("""
            UPDATE "tabItem"
            SET description = %s, item_name = %s
            WHERE name = %s
        """, (new_desc, new_name, sku))
        updated += 1

frappe.db.commit()
print("Updated: {} items".format(updated))

# Also update Subscription Plan descriptions (plan_name)
plans = frappe.db.sql("""
    SELECT name, plan_name, item, cost FROM "tabSubscription Plan"
""", as_dict=True)

plan_updated = 0
for plan in plans:
    item_sku = plan["item"]
    legacy = desc_map.get(item_sku)
    if not legacy or not legacy["description"]:
        continue

    new_plan_name = "PLAN-{}".format(item_sku)
    frappe.db.sql("""
        UPDATE "tabSubscription Plan"
        SET plan_name = %s
        WHERE name = %s
    """, (new_plan_name, plan["name"]))
    plan_updated += 1

frappe.db.commit()
print("Updated: {} subscription plans".format(plan_updated))

# ═══════════════════════════════════════════════════════════════
# STEP 3: VERIFY
# ═══════════════════════════════════════════════════════════════
print("\n" + "=" * 60)
print("STEP 3: VERIFY")
print("=" * 60)

# Sample items
sample = frappe.db.sql("""
    SELECT name, item_name, description, item_group
    FROM "tabItem"
    WHERE name IN ('FTTB1000I', 'TELEPMENS', 'RAB24M', 'HVIPFIXE', 'FTT_HFAR', 'CSERV', 'RAB2X')
    ORDER BY name
""", as_dict=True)
for s in sample:
    print("  {} | name={} | group={} | desc={}".format(
        s["name"], s["item_name"], s["item_group"], (s["description"] or "")[:80]))

# Count remaining legacy descriptions
remaining = frappe.db.sql("""
    SELECT COUNT(*) FROM "tabItem" WHERE description LIKE 'Legacy product ID%%'
""")[0][0]
print("\nRemaining with legacy placeholder: {}".format(remaining))

frappe.clear_cache()
print("Done — cache cleared")