[ibmi] Fix error binding Unicode data causing truncation

kadler · kadler · commit 8e41506a53a6 · 2019-04-25T18:36:29.000-05:00
The PASE binding code converts the Python Unicode object to UTF-8,
but passes the size of the PyUnicode object, which is in UCS-2
code units. Thus, if a Python Unicode object has any code points
above U+007F, the string lengths will not match and various errors
can occur, ranging from data trunctation to UTF-8 not properly
formed errors.
diff --git a/IBM_DB/ibm_db/ibm_db.c b/IBM_DB/ibm_db/ibm_db.c
@@ -5669,8 +5669,11 @@ static int _python_ibm_db_bind_data( stmt_handle *stmt_res, param_node *curr, Py
 						curr->uvalue = NULL;
 					}
 					curr->uvalue = getUnicodeDataAsSQLTCHAR(bind_data, &isNewBuffer);
-					curr->ivalue = PyUnicode_GetSize(bind_data);
-					curr->ivalue = curr->ivalue * sizeof(SQLTCHAR);
+#ifdef __PASE__
+					curr->ivalue = strlen(curr->uvalue);
+#else
+					curr->ivalue = PyUnicode_GetSize(bind_data) * sizeof(SQLTCHAR);
+#endif
 				}
 				param_length = curr->ivalue;
 				if (curr->size != 0) {

Original file line number	Diff line number	Diff line change
`@@ -5669,8 +5669,11 @@ static int _python_ibm_db_bind_data( stmt_handle stmt_res, param_node curr, Py`
`5669`	`5669`	`curr->uvalue = NULL;`
`5670`	`5670`	`}`
`5671`	`5671`	`curr->uvalue = getUnicodeDataAsSQLTCHAR(bind_data, &isNewBuffer);`
`5672`		`- curr->ivalue = PyUnicode_GetSize(bind_data);`
`5673`		`- curr->ivalue = curr->ivalue * sizeof(SQLTCHAR);`
	`5672`	`+#ifdef __PASE__`
	`5673`	`+ curr->ivalue = strlen(curr->uvalue);`
	`5674`	`+#else`
	`5675`	`+ curr->ivalue = PyUnicode_GetSize(bind_data) * sizeof(SQLTCHAR);`
	`5676`	`+#endif`
`5674`	`5677`	`}`
`5675`	`5678`	`param_length = curr->ivalue;`
`5676`	`5679`	`if (curr->size != 0) {`