oc-monitord/workflow_builder/argo_builder.go

// A class that translates the informations held in the graph object
// via its lists of components into an argo file, using the a list of
// link ID to build the dag

package workflow_builder

import (
	"fmt"
	. "oc-monitord/models"
	"os"
	"regexp"
	"strings"
	"time"

	oclib "cloud.o-forge.io/core/oc-lib"
	"cloud.o-forge.io/core/oc-lib/models/resources/compute"
	"cloud.o-forge.io/core/oc-lib/models/resources/processing"
	"cloud.o-forge.io/core/oc-lib/models/resources/workflow/graph"
	w "cloud.o-forge.io/core/oc-lib/models/workflow"
	"github.com/nwtgck/go-fakelish"
	"github.com/rs/zerolog"
	"gopkg.in/yaml.v3"
)

var logger zerolog.Logger

type ArgoBuilder struct {
	OriginWorkflow *w.Workflow
	Workflow       Workflow
	Services       []*Service
	Timeout        int
}

type Workflow struct {
	ApiVersion string `yaml:"apiVersion"`
	Kind       string `yaml:"kind"`
	Metadata   struct {
		Name string `yaml:"name"`
	} `yaml:"metadata"`
	Spec Spec `yaml:"spec,omitempty"`
}

func (b *Workflow) setDag(dag *Dag) {
	for _, t := range b.Spec.Templates {
		if t.Name == "dag" {
			t.Dag = dag
		}
	}
}

func (b *Workflow) getDag() *Dag {
	for _, t := range b.Spec.Templates {
		if t.Name == "dag" {
			return t.Dag
		}
	}
	b.Spec.Templates = append(b.Spec.Templates, Template{Name: "dag", Dag: &Dag{}})
	return b.Spec.Templates[len(b.Spec.Templates)-1].Dag
}

type Spec struct {
	Entrypoint string                `yaml:"entrypoint"`
	Arguments  []Parameter           `yaml:"arguments,omitempty"`
	Volumes    []VolumeClaimTemplate `yaml:"volumeClaimTemplates,omitempty"`
	Templates  []Template            `yaml:"templates"`
	Timeout    int                   `yaml:"activeDeadlineSeconds,omitempty"`
}

func (b *ArgoBuilder) CreateDAG(write bool) (string, int, []string, []string, error) {

	// handle services by checking if there is only one processing with hostname and port
	firstItems, lastItems, volumes := b.createTemplates()
	b.createVolumes(volumes)

	if b.Timeout > 0 {
		b.Workflow.Spec.Timeout = b.Timeout
	}
	b.Workflow.Spec.Entrypoint = "dag"
	b.Workflow.ApiVersion = "argoproj.io/v1alpha1"
	b.Workflow.Kind = "Workflow"
	if !write {
		return "", len(b.Workflow.getDag().Tasks), firstItems, lastItems, nil
	}
	random_name := fakelish.GenerateFakeWord(5, 8) + "-" + fakelish.GenerateFakeWord(5, 8)
	b.Workflow.Metadata.Name = "oc-monitor-" + random_name
	logger = oclib.GetLogger()
	yamlified, err := yaml.Marshal(b.Workflow)
	if err != nil {
		logger.Error().Msg("Could not transform object to yaml file")
		return "", 0, firstItems, lastItems, err
	}
	// Give a unique name to each argo file with its timestamp DD:MM:YYYY_hhmmss
	current_timestamp := time.Now().Format("02_01_2006_150405")
	file_name := random_name + "_" + current_timestamp + ".yml"
	workflows_dir := "./argo_workflows/"
	err = os.WriteFile(workflows_dir+file_name, []byte(yamlified), 0660)
	if err != nil {
		logger.Error().Msg("Could not write the yaml file")
		return "", 0, firstItems, lastItems, err
	}
	return file_name, len(b.Workflow.getDag().Tasks), firstItems, lastItems, nil
}

func (b *ArgoBuilder) createTemplates() ([]string, []string, []VolumeMount) {
	volumes := []VolumeMount{}
	firstItems := []string{}
	lastItems := []string{}
	for _, comp := range b.OriginWorkflow.GetProcessings() {
		if comp.Processing.Container != nil {
			volumes, firstItems, lastItems = b.createArgoTemplates(
				comp.ID, comp.Processing, volumes, firstItems, lastItems)
		} else {
			logger.Error().Msg("Not enough configuration setup, template can't be created : " + comp.Processing.GetName())
			return firstItems, lastItems, volumes
		}
	}
	firstWfTasks := map[string][]string{}
	latestWfTasks := map[string][]string{}
	relatedWfTasks := map[string][]string{}
	for _, wf := range b.OriginWorkflow.GetWorkflows() {
		realWorkflow, code, err := w.New().LoadOne(wf.Workflow.WorkflowID)
		if code != 200 {
			logger.Error().Msg("Error loading the workflow : " + err.Error())
			continue
		}
		subBuilder := ArgoBuilder{OriginWorkflow: realWorkflow.(*w.Workflow), Timeout: b.Timeout}
		_, _, fi, li, err := subBuilder.CreateDAG(false)
		if err != nil {
			logger.Error().Msg("Error creating the subworkflow : " + err.Error())
			continue
		}
		firstWfTasks[wf.ID] = fi
		if ok, depsOfIds := subBuilder.isArgoDependancy(wf.ID); ok { // IS BEFORE
			latestWfTasks[wf.ID] = li
			relatedWfTasks[wf.ID] = depsOfIds
		}
		subDag := subBuilder.Workflow.getDag()
		d := b.Workflow.getDag()
		d.Tasks = append(d.Tasks, subDag.Tasks...) // add the tasks of the subworkflow to the main workflow
		b.Workflow.Spec.Templates = append(b.Workflow.Spec.Templates, subBuilder.Workflow.Spec.Templates...)
		b.Workflow.Spec.Volumes = append(b.Workflow.Spec.Volumes, subBuilder.Workflow.Spec.Volumes...)
		b.Workflow.Spec.Arguments = append(b.Workflow.Spec.Arguments, subBuilder.Workflow.Spec.Arguments...)
		b.Services = append(b.Services, subBuilder.Services...)
	}
	for wfID, depsOfIds := range relatedWfTasks {
		for _, dep := range depsOfIds {
			for _, task := range b.Workflow.getDag().Tasks {
				if strings.Contains(task.Name, dep) {
					index := -1
					for i, depp := range task.Dependencies {
						if strings.Contains(depp, wfID) {
							index = i
							break
						}
					}
					if index != -1 {
						task.Dependencies = append(task.Dependencies[:index], task.Dependencies[index+1:]...)
					}
					task.Dependencies = append(task.Dependencies, latestWfTasks[wfID]...)
				}
			}
		}
	}
	for wfID, fi := range firstWfTasks {
		deps := b.getArgoDependencies(wfID)
		if len(deps) > 0 {
			for _, dep := range fi {
				for _, task := range b.Workflow.getDag().Tasks {
					if strings.Contains(task.Name, dep) {
						task.Dependencies = append(task.Dependencies, deps...)
					}
				}
			}
		}
	}
	if b.Services != nil {
		dag := b.Workflow.getDag()
		dag.Tasks = append(dag.Tasks, Task{Name: "workflow-service-pod", Template: "workflow-service-pod"})
		b.addServiceToArgo()
	}
	return firstItems, lastItems, volumes
}

func (b *ArgoBuilder) getAllComputeFromProcessing(p *processing.ProcessingResource) []*compute.ComputeResource {
	computeResources := []*compute.ComputeResource{}
	for _, link := range b.OriginWorkflow.Graph.Links {
		if link.Source.ID == p.GetID() && b.OriginWorkflow.Graph.Items[link.Destination.ID].Compute != nil {
			computeResources = append(computeResources, b.OriginWorkflow.Graph.Items[link.Destination.ID].Compute)
		} else if link.Destination.ID == p.GetID() && b.OriginWorkflow.Graph.Items[link.Source.ID].Compute != nil {
			computeResources = append(computeResources, b.OriginWorkflow.Graph.Items[link.Source.ID].Compute)
		}
	}
	return computeResources
}

func (b *ArgoBuilder) createArgoTemplates(id string,
	processing *processing.ProcessingResource,
	volumes []VolumeMount,
	firstItems []string,
	lastItems []string) ([]VolumeMount, []string, []string) {
	_, firstItems, lastItems = b.addTaskToArgo(b.Workflow.getDag(), id, processing, firstItems, lastItems)
	template := &Template{Name: getArgoName(processing.GetName(), id)}
	template.CreateContainer(processing, b.Workflow.getDag())
	// get datacenter from the processing
	computes := b.getAllComputeFromProcessing(processing)
	for _, compute := range computes {
		b.CreateService(id, processing)
		template.Metadata.Labels = make(map[string]string)
		template.Metadata.Labels["app"] = "oc-service-" + processing.GetName() + "-" + compute.GetName() // Construct the template for the k8s service and add a link in graph between k8s service and processing
	}
	storages := b.OriginWorkflow.GetStoragesByRelatedProcessing(id)
	for _, storage := range storages {
		if storage.Local {
			volumes = template.Container.AddVolumeMount(VolumeMount{
				Name:      strings.ReplaceAll(strings.ToLower(storage.GetName()), " ", "-"),
				MountPath: storage.Path,
				Storage:   storage,
			}, volumes)
		}
	}
	b.Workflow.Spec.Templates = append(b.Workflow.Spec.Templates, *template)
	return volumes, firstItems, lastItems
}
func (b *ArgoBuilder) addTaskToArgo(dag *Dag, graphItemID string, processing *processing.ProcessingResource,
	firstItems []string, lastItems []string) (*Dag, []string, []string) {
	unique_name := getArgoName(processing.GetName(), graphItemID)
	step := Task{Name: unique_name, Template: unique_name}
	if processing.Container != nil {
		for name, value := range processing.Container.Env {
			step.Arguments.Parameters = append(step.Arguments.Parameters, Parameter{
				Name:  name,
				Value: b.affectVariableEnv(value, b.OriginWorkflow.Graph),
			})
		}
	}
	step.Dependencies = b.getArgoDependencies(graphItemID)
	name := ""
	if b.OriginWorkflow.Graph.Items[graphItemID].Processing != nil {
		name = b.OriginWorkflow.Graph.Items[graphItemID].Processing.GetName()
	}
	if b.OriginWorkflow.Graph.Items[graphItemID].Workflow != nil {
		name = b.OriginWorkflow.Graph.Items[graphItemID].Workflow.GetName()
	}
	if len(step.Dependencies) == 0 && name != "" {
		firstItems = append(firstItems, getArgoName(name, graphItemID))
	}
	if ok, _ := b.isArgoDependancy(graphItemID); !ok && name != "" {
		lastItems = append(lastItems, getArgoName(name, graphItemID))
	}
	dag.Tasks = append(dag.Tasks, step)
	return dag, firstItems, lastItems
}

func (b *ArgoBuilder) affectVariableEnv(envVar string, graph *graph.Graph) string {
	var myExp = regexp.MustCompile(`(\{\{.*\}\})`) // regex to find all the variables in the command
	matches := myExp.FindAllString(envVar, -1)     // find all the variables in the command
	for _, match := range matches {                // for each variable in the command
		splitted := strings.Split( // split the variable to get the inout and the vars only
			strings.ReplaceAll(strings.ReplaceAll(strings.ReplaceAll(match, "{{", ""), "}}", ""), " ", ""), "_")
		if len(splitted) < 3 { // if the variable is not well formatted, we skip it
			logger.Error().Msgf("The variable %v is not well formatted", match)
			continue
		}
		graphItemID := splitted[1] // graphitemid is the id of the object
		vars := splitted[2]        // vars is the name of the variable of the object
		_, obj := graph.GetResource(graphItemID)
		if obj != nil {
			envVar = strings.ReplaceAll(envVar, match, fmt.Sprintf("%v", obj.Serialize()[vars]))
		}
	}
	return envVar
}

func (b *ArgoBuilder) createVolumes(volumes []VolumeMount) { // TODO : one think about remote volume but TG
	for _, volume := range volumes {
		new_volume := VolumeClaimTemplate{}
		new_volume.Metadata.Name = strings.ReplaceAll(strings.ToLower(volume.Name), " ", "-")
		new_volume.Spec.AccessModes = []string{"ReadWriteOnce"}
		new_volume.Spec.Resources.Requests.Storage = fmt.Sprintf("%v", volume.Storage.Size) + volume.Storage.SizeType.ToArgo()
		b.Workflow.Spec.Volumes = append(b.Workflow.Spec.Volumes, new_volume)
	}
}

func (b *ArgoBuilder) isArgoDependancy(id string) (bool, []string) {
	dependancyOfIDs := []string{}
	isDeps := false
	for _, link := range b.OriginWorkflow.Graph.Links {
		source := b.OriginWorkflow.Graph.Items[link.Destination.ID].Processing
		if id == link.Source.ID && source != nil {
			isDeps = true
			dependancyOfIDs = append(dependancyOfIDs, getArgoName(source.GetName(), link.Destination.ID))
		}
		wourceWF := b.OriginWorkflow.Graph.Items[link.Destination.ID].Workflow
		if id == link.Source.ID && wourceWF != nil {
			isDeps = true
			dependancyOfIDs = append(dependancyOfIDs, getArgoName(wourceWF.GetName(), link.Destination.ID))
		}
	}
	return isDeps, dependancyOfIDs
}

func (b *ArgoBuilder) getArgoDependencies(id string) (dependencies []string) {
	for _, link := range b.OriginWorkflow.Graph.Links {
		source := b.OriginWorkflow.Graph.Items[link.Source.ID].Processing
		if id == link.Destination.ID && source != nil {
			dependency_name := getArgoName(source.GetName(), link.Source.ID)
			dependencies = append(dependencies, dependency_name)
			continue
		}
	}
	return
}

func getArgoName(raw_name string, component_id string) (formatedName string) {
	formatedName = strings.ReplaceAll(raw_name, " ", "-")
	formatedName += "-" + component_id
	formatedName = strings.ToLower(formatedName)
	return
}