I am having some issues with Oracle SQL as I am learning how to use it, but one thing seems not to be working and I cant see why:
I am trying to use the SDO_GEOMETRY objects to update a coordinate value in a table. I am trying to declare VARRAYs, filling these arrays with the coordinates and ELEM_INFO needed, and passing these arrays into the SDO_GEOMETRY obejct in the update part, like this:
CREATE OR REPLACE TYPE decimal_array is varray(728) OF NUMBER (10,9);
CREATE OR REPLACE TYPE int_array IS VARRAY(67) OF INTEGER;
> /
DECLARE
sdo_ordinate_arr decimal_array := decimal_array(2,4, 4,3, 10,3, 13,5, 13,9, 11,13, 5,13, 2,11, 2,4, 7,5, 7,10, 10,10, 10,5, 7,5);
info_arr int_array := int_array(1,1003,1, 19,2003,1);
begin
update schema.table set geo_coord = (SDO_GEOMETRY(2003, 25832, NULL,
SDO_ELEM_INFO_ARRAY(info_arr), SDO_ORDINATE_ARRAY(sdo_ordinate_arr))) where id = 123456;
END;
> /
I tried the SQL above, and expected it to work, but it did not. There are no errors being raised, but the value in the table is not being updated.
If i do it like this, without the arrays but using the same values, it works:
update schema.table set geo_coord_ = (SDO_GEOMETRY(2003, 25832, NULL,
SDO_ELEM_INFO_ARRAY(1,1003,1, 19,2003,1), SDO_ORDINATE_ARRAY(2,4, 4,3, 10,3, 13,5, 13,9, 11,13, 5,13, 2,11, 2,4, 7,5, 7,10, 10,10, 10,5, 7,5))) where id = 123456;
As I understand from the documentation, SDO_ELEM_INFO_ARRAY and SDO_ORDINATE_ARRAY are both declared as arrays. Why doesnt it work if passed an array as value?
Got it, you need to declare your arrays as SDO_ELEM_INFO_ARRAY and SDO_ORDINATE_ARRAY, not as VARRAYs:
sdo_ordinate_arr SDO_ORDINATE_ARRAY:= SDO_ORDINATE_ARRAY(2,4, 4,3, 10,3, 13,5, 13,9, 11,13, 5,13, 2,11, 2,4, 7,5, 7,10, 10,10, 10,5, 7,5)
info_arr SDO_ELEM_INFO_ARRAY := SDO_ELEM_INFO_ARRAY(1,1003,1, 19,2003,1)
Oracle is confusing.
Related
I have a query that validates an SDO_GEOMETRY in Oracle 18c:
select
sdo_geom.validate_geometry_with_context(
sdo_geometry ('polygon ((676832.320 4857578.086, 665287.423 4857578.086, 665277.423 4878109.585,
676832.320 4878119.585, 676842.320 4857588.086))', 26917)
, 0.005) as validation
from
dual
VALIDATION
-----------------------------
13348 [Element <1>] [Ring <1>]
(1 row selected.)
db<>fiddle
The query produces an error code in a text column, but it doesn't describe what the code means.
I am able look up the error manually in the docs: 82 ORA-12700 to ORA-19400
ORA-13348: polygon boundary is not closed
Cause: The boundary of a
polygon does not close.
Action: Alter the coordinate values or the
definition of the SDO_GTYPE or SDO_ETYPE attribute of the geometry.
But manually looking up those error codes is inconvenient.
Is there a way to enhance the query so that it returns the full error description? (get the description from the database)
Assuming you can parse the string to pull out the error message, you can pass it to sqlerrm to get the text of the error (note that you're apparently getting a positive value, you'd need to negate that value to pass it to sqlerrm). I would assume that you could just look for everything before the first space to get the error number but I don't have a huge sample set to work with.
declare
l_message varchar2(1000);
begin
l_message := sqlerrm( -13348 );
dbms_output.put_line( l_message );
end;
/
will print
ORA-13348: polygon boundary is not closed
Building on #JustinCave's answer, here's a custom function that gets the error description from the validation text:
with function error_description(validation in varchar2) return varchar2 is
begin
return sqlerrm(substr(validation, 1, instr(validation,' ') - 1) * -1); --Multiply by -1. Oracle error codes seem to be "negative".
end;
select
error_description(validation) as error_description
from
(select
sdo_geom.validate_geometry_with_context(
sdo_geometry ('polygon ((676832.320 4857578.086, 665287.423 4857578.086, 665277.423 4878109.585, 676832.320 4878119.585, 676842.320 4857588.086))', 26917), 0.005) as validation
from dual)
ERROR_DESCRIPTION
-------------------
ORA-13348: polygon boundary is not closed
Edit:
As pointed out by #SolomonYakobson in a related post, the SQLERRM() function can also be used in a SELECT query (without the need for a custom function).
Are certain kinds of Oracle functions only available in PL/SQL, not
SQL?
Many of the functions are defined in Oracle supplied package
SYS.STANDARD.
Example: SELECT SYS.STANDARD.SQLERRM(-1422) FROM DUAL;
So we just need to fully qualify the function: SYS.STANDARD.SQLERRM()
select
sys.standard.sqlerrm(substr(validation, 1, instr(validation,' ') - 1) * -1) error_description
from
(select
sdo_geom.validate_geometry_with_context(
sdo_geometry ('polygon ((676832.320 4857578.086, 665287.423 4857578.086, 665277.423 4878109.585, 676832.320 4878119.585, 676842.320 4857588.086))', 26917), 0.005) as validation
from dual)
ERROR_DESCRIPTION
---------------------
ORA-13348: polygon boundary is not closed
So this is the documentation of SDO_INTERSECTION : https://docs.oracle.com/cd/B28359_01/appdev.111/b28400/sdo_objgeom.htm#SPATL1118
And this is my code:
PROCEDURE ciudadInterseccionCarretera (carretera1 IN VARCHAR2,carretera2 IN
VARCHAR2) IS
tupla Extremadura%ROWTYPE;
geomResultado SDO_GEOMETRY;
BEGIN
SELECT SDO_GEOM.SDO_INTERSECTION(ex1.Geom,ex2.Geom,0.05) INTO geomResultado
FROM Extremadura ex1, Extremadura ex2
WHERE ex1.Nombre = carretera1 AND ex2.Nombre = carretera2;
DBMS_OUTPUT.PUT_LINE('ciudad Intersección : '||geomResultado.SDO_GTYPE||' '||geomResultado.SDO_POINT.X);
END ciudadInterseccionCarretera;
ciudad Intersección : 2001
geomResultado is returning SDO_GTYPE properly when I execute the code, it returns 2001 as the INTERSECTION between carretera1 and carretera2 is a POINT. However I'm not able to return, for example the X coordinate of such POINT using geomResultado.SDO_POINT.X . Any ideas? I think my code is ok..
EDIT: Furthermore, when I try to use this code trying to find the intersection of two lines that doesn't really intersect, the result of the query is empty, so the problem is in SDO_POINT 100%.
Oracle has two ways of storing a point geometry
The first is in the sdo_point part of the sdo_geometry.
sdo_geometry(2001,null,sdo_point(x,y,null),null,null)
The other is as ordinates in the ordinate array:
sdo_geometry(2001,null,null,sdo_elem_info_array(1,1,1),sdo_ordinate_array(x,y))
In this case sdo_intersection returns a geometry with coordinates in the ordinate array. You have to get the values from there.
Example:
declare
l_geo1 sdo_geometry := sdo_geometry(2001
,null
,null
,sdo_elem_info_array(1, 2, 1)
,sdo_ordinate_array(0, 0, 10, 10));
l_geo2 sdo_geometry := sdo_geometry(2001
,null
,null
,sdo_elem_info_array(1, 2, 1)
,sdo_ordinate_array(0, 10, 10, 0));
l_geo3 sdo_geometry;
l_x number;
l_y number;
begin
l_geo3 := sdo_geom.sdo_intersection(l_geo1, l_geo2, 0.05);
l_x := l_geo3.sdo_ordinates(1);
l_y := l_geo3.sdo_ordinates(2);
dbms_output.put_line(l_x || ' ' || l_y);
end;
As Rene suggested, the solution of this problem was just to ask for the coordinates using:
geomResultado.SDO_ORDINATES(1);
geomResultado.SDO_ORDINATES(2);
For X and Y coordinates.
I'm able to insert values into table 2 from table 1 and execute the PL/SQL procedure successfully but somehow the output is clunky. I don't know why?
Below is the code :
create table airports_2_xml
(
airport xmltype
);
declare
cursor insert_xml_cr is select * from airports_1_orcl;
begin
for i in insert_xml_cr
loop
insert into airports_2_xml values
(
xmlelement("OneAirport",
xmlelement("Rank", i.Rank) ||
xmlelement("airport",i.airport) ||
xmlelement("Location",i.Location) ||
xmlelement("Country", i.Country) ||
xmlelement("Code_iata",i.code_iata) ||
xmlelement("Code_icao", i.code_icao) ||
xmlelement("Total_Passenger",i.Total_Passenger) ||
xmlelement("Rank_change", i.Rank_change) ||
xmlelement("Percent_Change", i.Percent_change)
));
end loop;
end;
/
select * from airports_2_xml;
Output:
Why it is showing < ,> in the output ? And why am I unable to see the output fully?
Expected output:
<OneAirport>
<Rank>3</Rank>
<Airport>Dubai International</Airport>
<Location>Garhoud</Location>
<Country>United Arab Emirates</Country>
<Code_IATA>DXB</Code_IATA>
<Code_ICAO>OMDB</Code_ICAO>
<Total_passenger>88242099</Total_passenger>
<Rank_change>0</Rank_change>
<Percent_Change>5.5</Percent_Change>
</OneAirport>
The main issue is how you are constructnig the XML. You have an outer XMLElement for OneAirport, and the content of that element is a single string.
You are generating individual XMLElements from the cursor fields, but then you are concenating those together, which gives you a single string which still has the angle brackets you're expecting. So you're trying to do something like, simplified a bit:
select
xmlelement("OneAirport", '<Rank>1</Rank><airport>Hartsfield-Jackson</airport>')
from dual;
XMLELEMENT("ONEAIRPORT",'<RANK>1</RANK><AIRPORT>HARTSFIELD-JACKSON</AIRPORT>')
--------------------------------------------------------------------------------
<OneAirport><Rank>1</Rank><airport>Hartsfield-Jackson</airp
and by default XMLElement() escapes entities in the passed-in values, so the angle-brackets are being converted to 'safe' equivalents like <. If it didn't do that, or you told it not to with noentityescaping:
select xmlelement(noentityescaping "OneAirport", '<Rank>1</Rank><airport>Hartsfield-Jackson</airport>')
from dual;
XMLELEMENT(NOENTITYESCAPING"ONEAIRPORT",'<RANK>1</RANK><AIRPORT>HARTSFIELD-JACKS
--------------------------------------------------------------------------------
<OneAirport><Rank>1</Rank><airport>Hartsfield-Jackson</airport></OneAirport>
then that would appear to be better, but you still actually have a single element with a single string (with characters that are likely to cause problems down the line), rather than the XML structure you almost certainly intended.
A simple way to get an zctual structure is with XMLForest():
xmlelement("OneAirport",
xmlforest(i.Rank, i.airport, i.Location, i.Country, i.code_iata,
i.code_icao, i.Total_Passenger, i.Rank_change, i.Percent_change)
)
You don't need the cursor loop, or any PL/SQL; you can just do:
insert into airports_2_xml (airport)
select xmlelement("OneAirport",
xmlforest(i.Rank, i.airport, i.Location, i.Country, i.code_iata,
i.code_icao, i.Total_Passenger, i.Rank_change, i.Percent_change)
)
from airports_1_orcl i;
The secondary issue is the display. You'll see more data if you issue some formatting commands, such as:
set lines 120
set long 32767
set longchunk 32767
Those will tell your client to retrieve and show more of the long (XMLType here) data, rather the default 80 characters it's giving you now.
Once you are generating a nested XML structure you can use XMLSerialize() to display that more readable when you query your second table.
Try this below block :
declare
cursor insert_xml_cr is select * from airports_1_orcl;
v_airport_xml SYS.XMLTYPE;
begin
for i in insert_xml_cr
loop
SELECT XMLELEMENT ( "OneAirport",
XMLFOREST(i.Rank as "Rank"
,i.airport as "Airport"
,i.Location as "Location"
,i.Country as "Country"
,i.code_iata as "Code_iata"
,i.code_icao as "code_icao"
,i.Total_Passenger as "Total_Passenger"
, i.Rank_change as "Rank_change"
,i.Percent_change as "Percent_Change"
))
into v_airport_xml
FROM DUAL;
insert into airports_2_xml values (v_airport_xml);
end loop;
end;
I have some SQL Command that contains a parameter such:(Note that myID has "int" type in SQL)
vSqlString :='Select * From myTable Where myID= :paramID';
and Use ParseSQL Command to execute this command:
myADOQuery.Parameters.ParseSQL(vSqlString , True);
Now myADOQuery.Parameters.ParamByName('paramID').DataType is smallint Type and it can't accept negative integer values.
I can exactly show to compiler that my Parameter[0].DataType is ftIneteger and it works properly, but what is a good solution for this problem?
My question is asked by Mr. imanShadabi Here: Using TAdoQuery.ParseSql and has resolved by user1008646
Hope this will help.
If you want to create in runtime parameters, you can use something like this:
ADOQuery1.Close;
ADOQuery1.SQL.Text := vSqlString;
ADOQuery1.Parameters.Clear;
ADOQuery1.Parameters.CreateParameter('paramID', ftInteger, pdInput, 10, vIntegerValue);
ADOQuery1.Open;
Or you can concatenate values to the query.
For example:
//For Integer values:
vSqlString: = 'Select * From myTable Where myID =' + IntToStr (vIntegerValue);
//For String values:
vSqlString: = 'Select * From myTable Where myID =' + QuotedStr (vStringValue);
//For float values:
//Be careful with this, usually in a query, the comma is separator values,
//so make sure that the decimal separator is '.'
vDS := DecimalSeparator; //I keep the value it had
DecimalSeparator := '.';
try
ADOQuery1.close;
ADOQuery1.SQL.Text := 'Select * From myTable Where myID='+FloatToStr(vFloatValue);
ADOQuery1.Open;
finally
DecimalSeparator := vDS; //Restore the value that had
end;
The third option is to set the parameters at design time.
But I think this is not what you want.
A lot of times when we query the database, we just need one column with varchar.
So I've made a nice function for querying the database and putting the results in a stringlist:
function Getdatatostringlist(sqlcomponent, sqlquery: string): TStringlist;
What I'm looking for now is basically the same function but for results with multiple columns where you don't know in advance what type the data is, be it varchar, int, datetime.
What kind of datastructure would be good to use here.
The reason I want this is that I try not to work on open datasets. I like much more to fetch all results into a temporary structure, close the dataset and work on the results.
After Kobiks reply about using in Memory datasets I came up with the following, it's fast put together to test the concept:
procedure TForm1.Button2Click(Sender: TObject);
var
MyDataSet : TAdoDataSet;
begin
MyDataSet := GetDataToDataSet('SELECT naam FROM user WHERE userid = 1', ADOConnection1);
try
Form1.Caption := MyDataSet.FieldByName('naam').AsString;
finally
MyDataSet.free;
end;
end;
function TForm1.GetDataToDataSet(sSql: string; AdoConnection: TADOConnection): TAdoDataSet;
begin
Result := TAdoDataSet.Create(nil);
Result.LockType := ltBatchOptimistic;
Result.Connection := AdoConnection;
Result.CommandText := sSql;
Result.Open;
Result.Connection := nil;
end;
I think this is something to build on.
You should use any disconnected in-memory TDataSet descendant, such as TClientDataSet.
Do not attempt to re-invent the wheel by storing a record-set in some new "Variant" structure. A TClientDataSet already contains all features you need to manipulate a "temporary" data structure.
Here is how you create a TClientDataSet structure:
cds.FieldDefs.Add('id', ftInteger);
cds.FieldDefs.Add('name', ftString, 100);
// ...
// create it
cds.CreateDataSet;
// add some data records
cds.AppendRecord([1, 'Foo']);
cds.AppendRecord([2, 'Bar']);
Many TDataSets has an ability to be used as an in-memory (client) datasets depending on the provider and LockType, for example a TADODataSet with LockType=ltBatchOptimistic could fetch results-set from the server, and then remain disconnected.
For exchanging Data with Excel this structure is usefull, might be useful for other purposes.
Function GetDatasetasDynArray(Ads: TDataset; WithHeader: Boolean = true): Variant;
// 20130118 by Thomas Wassermann
var
i, x, y: Integer;
Fields: Array of Integer;
begin
x := 0;
y := Ads.RecordCount;
if WithHeader then
inc(y);
SetLength(Fields, Ads.FieldCount);
for i := 0 to Ads.FieldCount - 1 do
if Ads.Fields[i].Visible then
begin
Fields[x] := i;
inc(x);
end;
SetLength(Fields, x);
Result := VarArrayCreate([0, y - 1 , 0, length(Fields) - 1], VarVariant);
y := 0;
if WithHeader then
begin
for i := Low(Fields) to High(Fields) do
begin
Result[y, i] := Ads.Fields[Fields[i]].DisplayLabel;
end;
inc(y);
end;
try
Ads.DisableControls;
Ads.First;
while not Ads.EOF do
begin
for i := Low(Fields) to High(Fields) do
begin
Result[y, i] := Ads.Fields[Fields[i]].Value;
end;
Ads.Next;
inc(y);
end;
finally
Ads.EnableControls;
end;
end;
procedure TForm1.Button1Click(Sender: TObject);
var
DynArray:Variant;
begin
DynArray := GetDatasetasDynArray(Adodataset1,true);
//DynArray[0,x] Header or First row
//DynArray[1,x] First row or SecondRow
Excel.Range.Value := DynArray;
end;
Why don't you like working with open datasets? They usually do not block the server. Copying the data from the dataset to whatever you want is extra overhead which is most likely not necessary.
A dataset provides exactly the functionality you want: A matrix with variable columns and rows.
EDIT: However, if you have iterate through the dataset often, you should consider creating a class holding the relevant information and then copy the data into a generic list, dictionary, tree or whatever you need as fast lookup structure.
Of course you could think of building something smart which can be as flexible as a dataset but: The more general things get, the poorer the performance (usually).