Feature #369
restinout json/UBF remap
| Status: | Rejected | Start date: | 01/04/2019 | |
|---|---|---|---|---|
| Priority: | Normal (Code 4) | Due date: | ||
| Assignee: | - | % Done: | 0% | |
| Category: | - | |||
| Target version: | - |
Description
We shall provide buffer remap server that could be called by restin/out to convert format before server calls and when generating responses.
For UBF/JSON mappings we could create new C/PS binary which would perform two way mappings with help of Parson for following json descriptor:
#!/usr/bin/pscript
a <- {
//Single field
"a": {"_ubf":"T_A_FLD[0]", "_opt":"y", "_min":"-1", "_max":"100", "_lenmin":"1", "_lenmax":"5"},
//Single sub-object
"b":
{
"username" : {"_ubf":"T_FIRST_NAME[0]", "_opt":"y"},
"lastname" : {"_ubf":"T_LAST_NAME[0]", "_opt":"n"}
},
//Array with singles
"c": {
"_minocc":"5",
"_array": [{"_ubf":"T_C_FLD"}]
},
//Array with objects
"d": {
"_minocc":"5",
"_maxocc":"10",
"_array": [
{
"test": {"_ubf":"C_TEST"},
"field2": {"_ubf":"C_FIELD"},
//sub field
"sub_obj": {
"car" : {"_ubf":"T_CAR", "_opt":"y"},
"model" : {"_ubf":"T_MODEL", "_opt":"n"}
},
//sub array
"sub_array_of_obj": {
"_array": [
{
"car" : {"_ubf":"T_CAR", "_opt":"y"},
"model" : {"_ubf":"T_MODEL", "_opt":"n"}
}
]
},
}
]
}
};
/*
We could use PS as Abstract Sytax tree, before loading/compiling the number
like field IDs shall be resolved.
_minocc shall generate reject if missing
_maxocc shall tell until which step to process.
each UBF field occurrence shall be counted on further...
Valid json for above:
{
"a": "hello world",
"b":
{
"username" : "John",
"lastname" : "Doe"
},
"c": ["A", "B", "C", "D", "E"],
"d":
[
{
"test": "hello",
"field2": "world",
"sub_obj":
{
"car":"Volvo",
"mode":"S80"
},
"sub_array_of_obj":
[
{
"car":"Volvo",
"mode":"S80"
},
{
"car":"Volvo",
"mode":"S60"
}
]
},
{
"test": "hello2",
"field2": "world2",
"sub_obj":
{
"car":"Volvo2",
"mode":"S802"
},
"sub_array_of_obj":
[
{
"car":"Volvo2",
"mode":"S802"
},
{
"car":"Volvo2",
"mode":"S602"
}
]
}
]
}
*/
History
#1 Updated by Madars over 5 years ago
- Tracker changed from Bug to Feature
#2 Updated by Madars about 5 years ago
Other option would be to describe the mapping with Squirel script classes and then drive them recursively for build and parse:
mapping could look like:
class Transaction
{
</ field = "C_RRN", min=12, max=12 />
rrn;
</ field = "C_AMOUNT", min=12, max=12/>
amount;
</ field = "C_CURRCODE", min=3, max=3/>
currcode;
}
class User
{
</ field = "C_FIRSTNAME", min=1, max=32/>
firstname;
</ field = "C_USERNAME", min=1, max=32/>
username;
Transaction [] tran;
}
class Account
{
</ field = "C_ACCNUM", min=1, max=28>
accnum;
Transaction [] tran;
}
When parsed in, the buffer could contain:
# Top level class
EX_CLASS User
C_FIRSTNAME JIM
C_USERNAME MORRISON
C_RRN 123456789012
C_RRN 123456789013
C_RRN 123456789014
C_AMOUNT 1.111
C_AMOUNT 1.112
C_AMOUNT 1.113
C_CURRCODE 643
C_CURRCODE 643
C_CURRCODE 643
When responding, we may respond with other object projection, for example
EX_CLASS Transaction
# and send only response data (single tran):
C_RRN 123456789012
C_AMOUNT 1.111
C_CURRCODE 643
{
"ex_class":"User",
"firstname":"JIM",
"username":"MORRISON"
"tran": [
{
"rrn":"123456789012",
"amount":"1.111",
"currcode":"643"
}
,
{
"rrn":"123456789013",
"amount":"1.133",
"currcode":"643"
}
,
{
"rrn":"123456789014",
"amount":"1.144",
"currcode":"643"
}
]
}
#3 Updated by Madars about 5 years ago
sub-Object occurrence:
</ occ_field="C_ACCTYPE" occ_value="1" /> Transaction tran;