The part number is always closest to the logo or company name and the CWN1 = factory/date/batch codes.
I recommend the PC817X3NSZ1B or PC817X4NSZ1B replacement.
Then do a diode test on all other parts or use a conservatively limited CC limited lab supply to bring up DC power externally to locate defective parts.
CWN1 C means CTR = 2 to 4 = Ic=/If= 10~20mA / 5 mA = 2 to 4 @ Vce= 5V but as a switch CTR, like hFE, always reduces to << 10% CTR @ Vce(sat)
Proof: from datasheet ... Ic/If= 20mA/1mA= 0.05 @ Vce(sat)= 0.1 to 0.2V so at Vce=Vce(sat) @ Ic=1mA CTR= 0.05/2~4= 2.5% to 1.25% Although the new datasheets do not show any distinction for CTR saturated vs different ranks, they do exist and will be around 10% of the max linear CTR for a fixed Vce(sat) max. They just overdrive the LEDs to ensure low Vce(Sat) but you app might not.
W is the factory now made by LITE-ON in China
N1 is the date code for 2001 January
still supplies them to Mouser, but they have new part numbers and they are most likely DIP for HV spacing of 5 kV could easily have been compromised by dust and humidity. over this period. No doubt this will cause a cascade of component secondary failures.
The new stock part number uses 3 for X as Rank C Ic= 10 to 20mA
PC817X3NSZ1B for DIP which are the only ones in stock at this time.
You could also use 4 as Rank D Ic= 15 to 30 mA both for If= 5mA which is slightly better and also cheaper a few cents perhaps due to high yield and more common, which is also stocked. PC817X4NSZ1B
other misc info
There are a few other Optos rated for 5kVrms but no others at Mouser are ranked as well for CTR, Onsemi= CTR= 1 to 6 , Everlight = 1 to 2, . Sharp makes some only in Japan also which are also good, maybe better quality PC123X5YFZ1B where 5 is Rank N 10 to 20 mA / 5mA similar to Rank C in the other p/n. but is non-stock. Nobody is better than Sharp in these IR optos. Sharp IR LEDs are now licensed to Vishay.