PowerBuilder

curl How Known Variables Reduce Execution Steps

See more CURL Examples

This example demonstrates how previously resolved values can simplify future executions.

When DoYourThing is called for a curl command, it builds an execution plan by resolving any missing inputs. This may require running one or more helper curl commands to produce the needed values.

As each command runs, any defined output variables are stored in the HttpCurl object and remain available for subsequent calls.

If another curl command is executed afterward—and it requires variables that are already known from earlier executions—those inputs do not need to be resolved again. As a result, the dependency chain is shorter, and the execution plan contains fewer steps.

In other words, each call to DoYourThing benefits from the current set of known variables. The more values that have already been resolved, the simpler and more direct the execution plan becomes.

This demonstrates that dependency resolution is dynamic and incremental: only missing inputs trigger additional steps, while previously computed values are reused automatically.

PowerBuilder

integer li_rc
integer li_Success
oleobject loo_HttpCurl
string ls_TargetCurl
string ls_FnName
string ls_JsonPath
oleobject loo_JsonOAuth2
oleobject loo_PlanJson
string ls_TargetCurl2

li_Success = 0

li_Success = 0

// -----------------------------------------------------------------------------
// NOTE:
// This example builds on the earlier multi-step dependency example, where the
// execution plan required multiple curl commands (getSite → getDrives → target).
// 
// If you have not seen that example, refer to it here: curl Multi-Step Dependency
// 
// 
// In this example, we demonstrate how previously resolved variables (site_id,
// drive_id, etc.) are reused across multiple calls to DoYourThing, resulting
// in simpler execution plans for subsequent requests.
// -----------------------------------------------------------------------------

loo_HttpCurl = create oleobject
li_rc = loo_HttpCurl.ConnectToNewObject("Chilkat.HttpCurl")
if li_rc < 0 then
    destroy loo_HttpCurl
    MessageBox("Error","Connecting to COM object failed")
    return
end if

// The final curl command we want to execute.
// It depends on {{drive_id}}, which may or may not already be known.
ls_TargetCurl = "curl -X GET https://graph.microsoft.com/v1.0/drives/{{drive_id}}/root/children"

// Define helper function to get drives (produces drive_id, requires site_id)
ls_FnName = "getDrives"
loo_HttpCurl.AddFunction(ls_FnName,"curl -X GET https://graph.microsoft.com/v1.0/sites/{{site_id}}/drives")
ls_JsonPath = "value[0].id"
loo_HttpCurl.AddOutput(ls_FnName,ls_JsonPath,"drive_id")

// Define helper function to get site (produces site_id, requires site_name)
ls_FnName = "getSite"
loo_HttpCurl.AddFunction(ls_FnName,"GET https://graph.microsoft.com/v1.0/sites/root:/sites/{{site_name}}")
ls_JsonPath = "id"
loo_HttpCurl.AddOutput(ls_FnName,ls_JsonPath,"site_id")

// Provide the initial known input.
loo_HttpCurl.SetVar("site_name","test")

// Configure OAuth2 authentication (client credentials with secrets)
loo_JsonOAuth2 = create oleobject
li_rc = loo_JsonOAuth2.ConnectToNewObject("Chilkat.JsonObject")

loo_JsonOAuth2.EnableSecrets = 1
loo_JsonOAuth2.UpdateString("oauth2.client_id","!!sharepoint|oauth2|client_id")
loo_JsonOAuth2.UpdateString("oauth2.client_secret","!!sharepoint|oauth2|client_secret")
loo_JsonOAuth2.UpdateString("oauth2.scope","https://graph.microsoft.com/.default")
loo_JsonOAuth2.UpdateString("oauth2.token_endpoint","!!sharepoint|oauth2|token_endpoint")
loo_HttpCurl.SetAuth(loo_JsonOAuth2)

// -----------------------------------------------------------------------------
// EXAMINE THE EXECUTION PLAN (BEFORE RUNNING):
// At this point, drive_id and site_id are NOT known.
// We can inspect the plan that will be executed without sending any requests.
// -----------------------------------------------------------------------------
loo_PlanJson = create oleobject
li_rc = loo_PlanJson.ConnectToNewObject("Chilkat.JsonObject")

loo_PlanJson.EmitCompact = 0

li_Success = loo_HttpCurl.ExaminePlan(ls_TargetCurl,loo_PlanJson)
Write-Debug "Execution plan before first call:"
Write-Debug loo_PlanJson.Emit()

// Expected output:
// 
// {
//   "plan": [{
//     "function": "getSite",
//     "inputs": ["site_name"],
//     "outputs": ["site_id"]
//   },{
//     "function": "getDrives",
//     "inputs": ["site_id"],
//     "outputs": ["drive_id"]
//   },{
//     "function": "targetCurl",
//     "inputs": ["drive_id"],
//     "outputs": []
//   }]
// }

// -----------------------------------------------------------------------------
// FIRST CALL:
// At this point, drive_id and site_id are NOT known.
// The execution plan will include all required steps:
// 
//   1) getSite    → produces site_id
//   2) getDrives  → produces drive_id
//   3) targetCurl → uses drive_id
// -----------------------------------------------------------------------------
li_Success = loo_HttpCurl.DoYourThing(ls_TargetCurl)
if li_Success = 0 then
    Write-Debug loo_HttpCurl.LastErrorText
    destroy loo_HttpCurl
    destroy loo_JsonOAuth2
    destroy loo_PlanJson
    return
end if

Write-Debug "First call completed."
Write-Debug "site_id  = " + loo_HttpCurl.GetVar("site_id")
Write-Debug "drive_id = " + loo_HttpCurl.GetVar("drive_id")

// -----------------------------------------------------------------------------
// SECOND CALL (DIFFERENT TARGET):
// Now that site_id is already known, we can run another curl command that
// depends only on site_id. No need to call getSite again.
// 
// Execution plan:
// 
//   1) targetCurl2 → uses existing site_id
// -----------------------------------------------------------------------------

ls_TargetCurl2 = "curl -X GET https://graph.microsoft.com/v1.0/sites/{{site_id}}"

li_Success = loo_HttpCurl.ExaminePlan(ls_TargetCurl2,loo_PlanJson)
Write-Debug ""
Write-Debug loo_PlanJson.Emit()

// Expected output:
// 
// {
//   "plan": [{
//     "function": "targetCurl",
//     "inputs": ["site_id"],
//     "outputs": []
//   }]
// }

li_Success = loo_HttpCurl.DoYourThing(ls_TargetCurl)
if li_Success = 0 then
    Write-Debug loo_HttpCurl.LastErrorText
    destroy loo_HttpCurl
    destroy loo_JsonOAuth2
    destroy loo_PlanJson
    return
end if

Write-Debug "Second call completed (used existing site_id)."

// -----------------------------------------------------------------------------
// THIRD CALL (ORIGINAL TARGET AGAIN):
// drive_id is already known from the first call.
// Therefore, no helper functions are needed this time.
// 
// Execution plan:
// 
//   1) targetCurl → uses existing drive_id
// -----------------------------------------------------------------------------
li_Success = loo_HttpCurl.ExaminePlan(ls_TargetCurl,loo_PlanJson)
Write-Debug ""
Write-Debug loo_PlanJson.Emit()

// Expected output:
// 
// {
//   "plan": [{
//     "function": "targetCurl",
//     "inputs": ["drive_id"],
//     "outputs": []
//   }]
// }

li_Success = loo_HttpCurl.DoYourThing(ls_TargetCurl)
if li_Success = 0 then
    Write-Debug loo_HttpCurl.LastErrorText
    destroy loo_HttpCurl
    destroy loo_JsonOAuth2
    destroy loo_PlanJson
    return
end if

Write-Debug "Third call completed."
Write-Debug "No dependency resolution was needed because all inputs were already known."


destroy loo_HttpCurl
destroy loo_JsonOAuth2
destroy loo_PlanJson