Mangled text output

Hello,

I'm trying to extract text from a generated PDF document. It appears to be a well-formatted document, although it's v1.3. I've opened it in acrobat and copy/pasted into textedit to verify that it's not an embedded font issue. That went fine. I don't know if the information below helps shed any light, or if there's anything else I can do to provide more detail?

aaa@bbb:/path# head -n 15 /xxx/yyy.pdf 
%PDF-1.3
%????
7 0 obj
<<
/Predictor 15
/Colors 1
/BitsPerComponent 8
/Columns 320
>>
endobj
11 0 obj
<<
/Type /ExtGState
/CA 1
>>

This is the output from getDetails():

Producer:	 PDFKit
Creator:	 PDFKit
CreationDate:	 2021-11-06T20:36:50+00:00
Pages:	 9

This is a sample of the mangled text:

>K?)0'C%A0CD#K)SA#.)CB%)0E#Q%)-$%"DSD"0CD#K-)0KM)"#AA%-$#KMDKF)$AD"%)0M/V-C.%KC)cDS)K%"%--0A?d)&D'')E%).0M%)#K'?)0C)CB%)+V-C#.%Ae-)A%[V%-C)#A)0$$A#Q0',)
+#.$'%CDKF)CB%)&#A1)DK)CBD-)IA#$#-0')0C)CB%)CD.%)-"B%MV'%M)D-)"#KCDKF%KC)V$#K)0""DM%KC-)#A)M%'0?-)E%?#KM)#VA)"#KCA#',)7BD-)IA#$#-0')D-)E0-%M)$AD.0AD'?)#K)
CB%)+V-C#.%Ae-)M%-"AD$CD#K)#S)CB%)$A#E'%.,)7BD-)IA#$#-0').0?)E%)&DCBMA0&K)DS)K#C)0""%$C%M)E?)CB%)+V-C#.%A)&DCBDK)34;)M0?-,
>""%$C0K"%)#S)+#KCA0"Cf)X)0.T&%)0A%)0&0A%)#S)0KM)0FA%%)C#)CB%)"#KC%KC-)#S)CBD-)IA#$#-0'()CB%)0CC0"B%M)^#E)N%C0D')-B%%Cc-d()0KM)CB%)0CC0"B%M)*D.DC%M)
R0AA0KC?()cC#F%CB%A()CB%)g+#KCA0"Cgd,)h#V)0A%)0VCB#ADi%M)C#)M#)CB%)&#A1)0-)-$%"DSD%M)DK)CB%)+#KCA0"C,)XT&%)&D'').01%)CB%)$0?.%KC)-%C)S#ACB)DK)CBD-)+#KCA0"C)
0C)CB%)CD.%)DC)D-)MV%,)XT&%)&D'')$0?)?#VA)-%AQD"%)"B0AF%)#S)353THj)$%A).#KCB)c36j)$%A)0KKV.d)DS).?T#VA)0""#VKC)D-)H;)M0?-)#A).#A%)$0-C)MV%()$'V-)?#VA)

The report names getDetails() and includes PDF 1.3 metadata plus mangled extracted text, but no source file, test, or PDF attachment is provided. Start by obtaining the generated PDF and reproducing extraction, then trace the parser path used for its text streams. Done means this document's extracted text is readable and the failure is covered by a regression check.