JSON Parse() raises error after CODEPAGE-CONVERT of LONGCHAR to UTF

When using CODEPAGE-CONVERT with LONGCHAR variables: 1. The codepage of the LONGCHAR variable is not fixed to the same codepage to which the contents are converted. 2. When the client is running with a non-Unicode code page, the LONGCHAR variable is initialized to a non-empty string, and the LONGCHAR is used as input to Progress.Json.ObjectModel.ObjectModelParser:Parse(), no codepage errors are raised by Parse() in spite of observation 1. However, if the LONGCHAR is uninitialized or is initialized to an empty string, Parse() raises error (16063).

Dec 8, 2016
Knowledge

Title

URL Name

JSON-Parse-raises-error-after-CODEPAGE-CONVERT-of-LONGCHAR-to-UTF

Article Number

000176074

Environment

Product: OpenEdge Version: 11.x OS: All supported platforms

Question/Problem Description

When using CODEPAGE-CONVERT with LONGCHAR variables:

1. The codepage of the LONGCHAR variable is not fixed to the same codepage to which the contents are converted.

2. When the client is running with -cpinternal iso8859-1, the LONGCHAR variable is initialized to a non-empty string, and the LONGCHAR is used as input to Progress.Json.ObjectModel.ObjectModelParser:Parse(), no codepage errors are raised by Parse() in spite of observation 1. However, if the LONGCHAR is uninitialized or is initialized to an empty string, Parse() raises this error:

Invalid encoding parameter to
Progress.Json.ObjectModel.ObjecteModelParser:Parse().'ISO8859-1' is
not a valid UTF encoding. (16063)

Steps to Reproduce

1. Unpack the attached file 000073242.zip into the OpenEdge work directory. 2. proenv> prowin -cpinternal iso8859-1 3. Open repro1.p and examine the code. 4. Run the code. Note that GET-CODEPAGE returns "?" for all three LONGCHAR variables. 5. Open repro2.p and examine the code. Note the 3 alternative invocations of Parse() labeled "#1", "#2" and "#3". 6. Uncomment each of the Parse() invocations in turn, comment out the other two, and run the code. Note that #1 does not raise any error, but #2 and #3 raise the error noted in the problem description.

Clarifying Information

When the parameter to the Parse() method is a LONGCHAR, it must be encoded in a Unicode code page.

Error Message

Invalid encoding parameter to Progress.Json.ObjectModel.ObjecteModelParser:Parse().'<non-Unicode codepage>' is not a valid UTF encoding. (16063)

Defect Number

Defect PSC00350164

Enhancement Number

Cause

The Parse() method was supposed to check for an empty string and raise an error if encountered. Instead, the AVM was continuing on to the code that checks the code-page where it raises an error. It is the code-page-related error that gets thrown instead of the error related to the empty string.

Resolution

Upgrade to OpenEdge 11.7.0 or later.

Workaround

Depending on application needs, use one of the following workarounds:

1. Start the client in a Unicode code page such as UTF-8.
2. Use the FIX-CODEPAGE statement to set the code page of the LONGCHAR variable to a Unicode code page before assigning it a value. Note that FIX-CODEPAGE works only on an empty LONGCHAR variable, while CODEPAGE-CONVERT works on the string value contained in a LONGCHAR (or CHARACTER) variable.

Notes

References to Other Documentation:

OpenEdge Development: ABL Reference
OpenEdge Development: Working with JSON

Progress article(s):
Error (16063) raised when parsing JSON input.

Keyword Phrase

Last Modified Date

11/20/2020 7:19 AM

JSON Parse() raises error after CODEPAGE-CONVERT of LONGCHAR to UTF

Files(1)

Was this article helpful?

Was this article helpful?

JSON Parse() raises error after CODEPAGE-CONVERT of LONGCHAR to UTF

Information

Related Articles