Fix bug #54379 PDO_OCI: UTF-8 output gets truncated

[gureedo]

Replacement for issue #542

[KalleZ]

Tagging Christopher as he is the Oracle guy; @cjbj

[nikic]

Ping @cjbj

[KalleZ]

@nikic@cjbj is currently on holiday, he said he would have a look once he returns

[cjbj]

I'm back; I've asked a developer to look at this. It's in a big queue of tasks TBD.

[cjbj]

Lines 602-606 need to be:

 } else if (dtype == SQLT_BIN){
S->cols[colno].datalen = (ub4) col->maxlen * 2; // raw characters to hex digits
} else {
S->cols[colno].datalen = (ub4) (col->maxlen * S->H->max_char_width);
}

[cjbj]

Rationale from the developer who looked at it is:
"Given raw data(SQLT_BIN) is defined as SQLT_CHR (See line #609), raw is returned as strings. So 1 byte raw is represented as two bytes of hex digits.
For example, a 1-byte raw data 'FF' is converted to two 'F' characters. So we need to multiply by 2 (not 3) to accommodate the translation of raw to hex digits. max_char_width should be used for character strings.

Apart from that, other changes all look good including the testcase. The test would normally be skipped in the test suite as a UTF-8 DSN is required.

But we can see the skip reason in the test result and run it manually if required."

[cjbj]

No update from filer on this. But no fear! We've got a corrected patch ready for testing and merge. I'll close this PR

[php-pulls]

Comment on behalf of sixd at php.net:

We will commit a corrected patch