myxce/src/frontend/apiserver/backend/convert.go

//go:build cgo

package backend

/*
#include <stdlib.h>
#include "client.h"
#include "val_helpers.h"

// Helper to index into xce_host_alloc_t** (CGO cannot dereference C pointer-to-pointer arrays).
static xce_host_alloc_t *job_alloc_host(xce_job_alloc_t *a, int i) { return a->allocs[i]; }

// Helpers to index into the reason pointer arrays of xce_job_preason_t.
static xce_host_reason_t *job_host_reason(xce_job_preason_t *p, int i) { return p->host_based_reasons[i]; }
static xce_res_reason_t *job_res_reason(xce_job_preason_t *p, int i) { return p->res_based_reasons[i]; }

// Helper to index into xce_alloc_t** (same CGO limitation as above) -- one
// host's per-resource allocation list, scanned for the "slots" entry.
static xce_alloc_t *host_alloc_res(xce_host_alloc_t *h, int i) { return h->allocs[i]; }

// Helpers to index into job_pidinfo_t's pid/pgid arrays (same CGO limitation).
static unsigned int pidinfo_pid(job_pidinfo_t *p, int i) { return p->pids[i]; }
static unsigned int pidinfo_pgid(job_pidinfo_t *p, int i) { return p->pgids[i]; }
*/
import "C"

import (
	"apiserver/models"
	logger "apiserver/utils"
	"unsafe"

	"go.uber.org/zap"
)

// goString safely converts a C string to a Go string, handling nil.
func goString(s *C.char) string {
	if s == nil {
		return ""
	}
	return C.GoString(s)
}

// cStringOrNil returns a C copy of s, or nil for an empty string so an unset
// field stays NULL in the C struct. The caller frees the result.
func cStringOrNil(s string) *C.char {
	if s == "" {
		return nil
	}
	return C.CString(s)
}

// cJobArrayToModels decodes a compact Data.jobArray push descriptor
// (*C.xce_job_submit_t: base jobid, array_size, one shared job_submit_data_t
// spec — the same JobSubmit shape apiserver_push_jobarray() sends,
// job-array-single-wal-record task 5.2) and expands it locally into N
// models.JobInfo entries via rawJobArrayToModels, mirroring cJobToModel's
// single-job field extraction below but for the array-shaped descriptor
// (task 5.4, Req 6.2, 6.5). Decodes the shared submit spec exactly once
// (same field-by-field extraction cJobToModel uses for job.submit) since
// every element shares it by construction — never once per element.
func cJobArrayToModels(jobArray *C.xce_job_submit_t) []*models.JobInfo {
	if jobArray == nil || jobArray.submit == nil {
		return nil
	}

	s := jobArray.submit
	rd := &rawSubmitData{
		queue:       goString(s.queue),
		user:        goString(s.user),
		jobName:     goString(s.job_name),
		projectName: goString(s.project_name),
		userGroup:   goString(s.user_group),
		jobDesc:     goString(s.job_description),
		resReq:      goString(s.resreq),
		minTasks:    int(s.mintasks),
		maxTasks:    int(s.maxtasks),
		fromHost:    goString(s.from_host),
	}
	if s.start_info != nil {
		si := s.start_info
		rd.startInfo = &rawStartInfo{
			cmdline:    goString(si.cmdline),
			launchType: goString(si.launch_type),
			preExec:    goString(si.job_pre_exec),
			postExec:   goString(si.job_post_exec),
			execHome:   goString(si.exechome),
			execCWD:    goString(si.execcwd),
			execUser:   goString(si.execuser),
			exdata:     goString(si.exdata),
		}
	}

	return rawJobArrayToModels(uint64(jobArray.jobid), uint64(jobArray.array_size), rd)
}

// cPidInfoToRaw decodes a C job_pidinfo_t — the worker's start report
// (pids/pgids plus exdata, whose {"url": ...} JSON is the interactive
// session URL for ttyd/XJOB jobs) — into the rawPidInfo the pure layer
// consumes. Shared by cJobToModel (the full xce_job_t conversion, where the
// capnp Job wire carries pidinfo at Job.pidinfo@9) and the standalone
// jobPidInfo notify path (NotifyData_data_jobPidInfo), which delivers the
// same struct the moment WAL_JOB_START commits.
func cPidInfoToRaw(pi *C.job_pidinfo_t) *rawPidInfo {
	if pi == nil {
		return nil
	}
	p := &rawPidInfo{exdata: goString(pi.exdata)}
	for i := 0; i < int(pi.npids); i++ {
		p.pids = append(p.pids, int(C.pidinfo_pid(pi, C.int(i))))
	}
	for i := 0; i < int(pi.npgids); i++ {
		p.pgids = append(p.pgids, int(C.pidinfo_pgid(pi, C.int(i))))
	}
	return p
}

// cJobToModel extracts C xce_job_t fields into a rawJobData and maps it to JobInfo.
func cJobToModel(job *C.xce_job_t) *models.JobInfo {
	r := rawJobData{
		jobID:      int64(job.jobid),
		state:      int(job.state),
		runTime:    int(job.run_time),
		subTime:    int64(job.sub_time),
		queueTime:  int64(job.queue_time),
		startTime:  int64(job.start_time),
		finishTime: int64(job.finish_time),
	}
	if job.submit != nil {
		s := job.submit
		rd := &rawSubmitData{
			queue:       goString(s.queue),
			user:        goString(s.user),
			jobName:     goString(s.job_name),
			projectName: goString(s.project_name),
			userGroup:   goString(s.user_group),
			jobDesc:     goString(s.job_description),
			resReq:      goString(s.resreq),
			minTasks:    int(s.mintasks),
			maxTasks:    int(s.maxtasks),
			fromHost:    goString(s.from_host),
		}
		if s.start_info != nil {
			si := s.start_info
			rd.startInfo = &rawStartInfo{
				cmdline:    goString(si.cmdline),
				launchType: goString(si.launch_type),
				preExec:    goString(si.job_pre_exec),
				postExec:   goString(si.job_post_exec),
				execHome:   goString(si.exechome),
				execCWD:    goString(si.execcwd),
				execUser:   goString(si.execuser),
				exdata:     goString(si.exdata),
			}
		}
		r.submit = rd
	}
	if u := job.run_usage; u != nil {
		r.runUsage = &rawRunUsage{
			maxMem:            float64(u.max_mem),
			avgMem:            float64(u.avg_mem),
			usedMem:           float64(u.used_mem),
			cpuTime:           int(u.cpu_time),
			peakCPUTime:       int(u.peak_cpu_time),
			peakCPUDuration:   int(u.peak_cpu_duration),
			memEfficiency:     float64(u.mem_efficiency),
			avgCPUEfficiency:  float64(u.avg_cpu_efficiency),
			peakCPUEfficiency: float64(u.peak_cpu_efficiency),
			peakCPU:           float64(u.peak_cpu),
		}
	}
	if job.allocs != nil {
		n := int(job.allocs.nhosts)
		for i := 0; i < n; i++ {
			h := C.job_alloc_host(job.allocs, C.int(i))
			if h == nil || h.host == nil {
				continue
			}
			// bug-1743-followup: this host's own "slots" resource
			// allocation (xce_host_alloc_t.allocs[]), mirroring the C
			// scheduler's own get_nslots_in_alloc()/repl_worker_alloc_
			// nslots() resname scan -- defaults to 1 when absent, same as
			// those.
			nslots := 1
			nallocs := int(h.nallocs)
			for j := 0; j < nallocs; j++ {
				res := C.host_alloc_res(h, C.int(j))
				if res != nil && res.resname != nil && C.GoString(res.resname) == "slots" {
					nslots = int(res.value)
					break
				}
			}
			r.execHosts = append(r.execHosts, rawHostSlot{host: C.GoString(h.host), slots: nslots})
		}
	}
	// PEND reason: the scheduler pushes these via setpreason; the controller
	// stores them on job->pend_reasons. Maps to models.JobInfo.PendingReason.
	// A detail-only record (main_reason 0) gets its overall-reason NAME from
	// the first detail row at the pure layer (rawJobToModel); the C record is
	// carried through unmutated (the scheduler-framework trace/verify stack
	// counts main_reason as its own occurrence and its committed .replay
	// baselines encode main_reason=0 detail-only PENDs).
	if job.pend_reasons != nil {
		pr := job.pend_reasons
		// main_reason 0 is detail-only (the allocator's ordinary shortfall
		// shape): leave the NAME empty so rawJobToModel's promotion fills it
		// from the first detail row. Pre-filling pendReasonName(0) here
		// ("unknown(0)") made the promotion unreachable on the live C path —
		// the advisor P0 verification caught jobs whose res_reasons named the
		// real cause while main_reason_name read unknown(0).
		mainName := ""
		if pr.main_reason != 0 {
			mainName = pendReasonName(int(pr.main_reason))
		}
		r.pendReason = &rawPendReason{
			MainReason:     int(pr.main_reason),
			MainReasonName: mainName,
		}
		for i := 0; i < int(pr.nhost_reasons); i++ {
			hr := C.job_host_reason(pr, C.int(i))
			if hr != nil && hr.host != nil {
				r.pendReason.HostReasons = append(r.pendReason.HostReasons,
					rawHostReason{Host: goString(hr.host), Reason: int(hr.reason),
						ReasonName: pendReasonName(int(hr.reason))})
			}
		}
		for i := 0; i < int(pr.nres_reasons); i++ {
			rr := C.job_res_reason(pr, C.int(i))
			if rr != nil && rr.resource != nil {
				r.pendReason.ResReasons = append(r.pendReason.ResReasons,
					rawResReason{Resource: goString(rr.resource),
						Host: goString(rr.host), Reason: int(rr.reason),
						ReasonName: pendReasonName(int(rr.reason))})
			}
		}
	}
	// Pidinfo: the worker's start report — pids/pgids plus exdata, whose
	// {"url": ...} JSON is the interactive session URL (ttyd/XJOB). The
	// capnp Job wire carries pidinfo (Job.pidinfo@9 / JobPIDInfo.exdata@3)
	// but the conversion dropped it, so interactive_job_url never reached
	// the API and xcecli's browser flow could never fire.
	r.pidInfo = cPidInfoToRaw(job.pidinfo)
	return rawJobToModel(r)
}

// cWorkerToModel extracts C xce_worker_t fields into a rawWorkerData and maps it to HostInfo.
func cWorkerToModel(worker *C.xce_worker_t) *models.HostInfo {
	workerName := goString(worker.name)
	r := rawWorkerData{
		name:   workerName,
		status: int(worker.status),
	}
	if worker.attrs != nil {
		a := worker.attrs
		r.hasAttrs = true
		r.hostType = goString(a.hosttype)
		r.hostModel = goString(a.hostmodel)
		r.cpuFactor = float64(a.cpufactor)
		r.maxCPUs = int(a.maxcpus)
		r.maxMem = int(a.maxmem)
		r.maxSwap = int(a.maxswap)
		r.maxTmp = int(a.maxtmp)
		r.nDisks = int(a.ndisks)
	}
	n := int(worker.nresvals_to_sched)
	if n > 0 {
		entries := unsafe.Slice(worker.resvals_to_sched, n)
		r.resvals = make([]models.ResourceDistributionInfo, 0, n)
		for _, nv := range entries {
			if rd := cNameValToWorkerResval(workerName, nv); rd != nil {
				r.resvals = append(r.resvals, *rd)
			}
		}
	}
	return rawWorkerToModel(r)
}

// cQueueToModel extracts C xce_queue_t fields and maps them to QueueInfo.
func cQueueToModel(queue *C.xce_queue_t) *models.QueueInfo {
	return rawQueueToModel(rawQueueData{
		name:     goString(queue.queue),
		status:   int(queue.status),
		priority: int(queue.priority),
	})
}

// cResourceToModel extracts C xce_resource_t fields and maps them to ResourceDefinitionInfo.
func cResourceToModel(res *C.xce_resource_t) *models.ResourceDefinitionInfo {
	return rawResourceToModel(rawResourceData{
		name:       goString(res.name),
		desc:       goString(res.desc),
		valType:    int(res.valtype),
		ordType:    int(res.ordtype),
		flags:      int(res.flags),
		interval:   int(res.interval),
		lastReport: int64(res.last_report),
	})
}

// cClusterConfQueues extracts queue models from a cluster_config_t.
func cClusterConfQueues(conf *C.cluster_config_t) []models.QueueInfo {
	n := int(conf.nqueues)
	if n == 0 || conf.queues == nil {
		return nil
	}
	qps := unsafe.Slice(conf.queues, n)
	out := make([]models.QueueInfo, 0, n)
	for _, qp := range qps {
		if qp != nil {
			out = append(out, *cQueueToModel(qp))
		}
	}
	return out
}

// cClusterConfResources extracts resource definition models from a cluster_config_t.
func cClusterConfResources(conf *C.cluster_config_t) []models.ResourceDefinitionInfo {
	n := int(conf.nres)
	if n == 0 || conf.resources == nil {
		return nil
	}
	rps := unsafe.Slice(conf.resources, n)
	out := make([]models.ResourceDefinitionInfo, 0, n)
	for _, rp := range rps {
		if rp != nil {
			out = append(out, *cResourceToModel(rp))
		}
	}
	return out
}

// cHostConfigToModel extracts C xce_hostconfig_t fields and maps them to HostInfo.
func cHostConfigToModel(hc *C.xce_hostconfig_t) *models.HostInfo {
	return rawHostConfigToModel(rawHostConfigData{
		name:      goString(hc.name),
		hostmodel: goString(hc.hostmodel),
		hosttype:  goString(hc.hosttype),
		mxj:       int(hc.mxj),
	})
}

// cStrList converts an xce_strlist_t to a Go slice.
func cStrList(l *C.xce_strlist_t) []string {
	n := int(l.count)
	if n == 0 || l.data == nil {
		return nil
	}
	out := make([]string, 0, n)
	for _, p := range unsafe.Slice(l.data, n) {
		if p != nil {
			out = append(out, C.GoString(p))
		}
	}
	return out
}

// cClusterConfUserGroups extracts the user groups from a cluster_config_t
// (bug-1934: the self_ug job scope resolves them).
func cClusterConfUserGroups(conf *C.cluster_config_t) []models.UserGroupInfo {
	n := int(conf.nugroups)
	if n == 0 || conf.ugroups == nil {
		return nil
	}
	out := make([]models.UserGroupInfo, 0, n)
	for _, gp := range unsafe.Slice(conf.ugroups, n) {
		if gp == nil || gp.name == nil {
			continue
		}
		out = append(out, models.UserGroupInfo{
			Name:      C.GoString(gp.name),
			Members:   cStrList(&gp.members),
			ExMembers: cStrList(&gp.ex_members),
		})
	}
	return out
}

// cClusterConfHostConfigs extracts host config models from a cluster_config_t.
func cClusterConfHostConfigs(conf *C.cluster_config_t) []models.HostInfo {
	n := int(conf.nhconfigs)
	if n == 0 || conf.hconfigs == nil {
		return nil
	}
	hps := unsafe.Slice(conf.hconfigs, n)
	out := make([]models.HostInfo, 0, n)
	for _, hp := range hps {
		if hp != nil {
			out = append(out, *cHostConfigToModel(hp))
		}
	}
	return out
}

// cClusterConfToClusterInfo extracts C cluster_config_t cluster-level fields and maps them to ClusterInfo.
// Returns nil when clustername is empty so dispatchConfigPush skips the OnClusterUpdated call.
// C struct field names are snake_case (clustername, heartbeat_interval, etc.) — not Cap'n Proto names.
func cClusterConfToClusterInfo(conf *C.cluster_config_t) *models.ClusterInfo {
	name := goString(conf.clustername)
	if name == "" {
		return nil
	}
	return rawClusterToModel(rawClusterData{
		name:                   name,
		heartbeatInterval:      int(conf.heartbeat_interval),
		vcpus:                  0, // vcpus moved to myconf_t (--vcpus CLI flag); not in cluster_config_t
		inactiveHeartbeatCount: int(conf.inactive_heartbeat_count),
	})
}

// modelJobToSubmit converts a Go JobInfo to C xce_job_submit_t heap structs.
// The caller must free all returned C memory by calling freeJobSubmit.
func modelJobToSubmit(job *models.JobInfo) (*C.xce_job_submit_t, *C.job_submit_data_t, *C.job_startinfo_t) {
	rsi := submitStartInfo(job)
	si := (*C.job_startinfo_t)(C.calloc(1, C.sizeof_job_startinfo_t))
	si.cmdline = C.CString(rsi.cmdline)
	// bug-1500/bug-1780 (same fix both sides): carry the exec user on the
	// start_info. Without it the worker receives a NULL/empty execuser and
	// runs every job as "nobody".
	si.execuser = cStringOrNil(rsi.execUser)
	si.launch_type = cStringOrNil(rsi.launchType)
	si.job_pre_exec = cStringOrNil(rsi.preExec)
	si.job_post_exec = cStringOrNil(rsi.postExec)
	si.exechome = cStringOrNil(rsi.execHome)
	si.execcwd = cStringOrNil(rsi.execCWD)
	si.exdata = cStringOrNil(rsi.exdata)

	sd := (*C.job_submit_data_t)(C.calloc(1, C.sizeof_job_submit_data_t))
	sd.queue = C.CString(job.QueueName)
	sd.user = C.CString(job.SubmitUser)
	sd.job_name = C.CString(job.JobName)
	if job.ProjectName != "" {
		sd.project_name = C.CString(job.ProjectName)
	}
	if job.UserGroup != "" {
		sd.user_group = C.CString(job.UserGroup)
	}
	if job.JobDescription != "" {
		sd.job_description = C.CString(job.JobDescription)
	}
	if job.ResReq != "" {
		sd.resreq = C.CString(job.ResReq)
	}
	if job.FromHost != "" {
		sd.from_host = C.CString(job.FromHost)
	}
	sd.mintasks = C.uint(job.NumProcessors)
	if sd.mintasks == 0 {
		sd.mintasks = 1
	}
	sd.maxtasks = C.uint(job.MaxNumProcessors)
	if sd.maxtasks < sd.mintasks {
		sd.maxtasks = sd.mintasks
	}
	sd.start_info = si

	js := (*C.xce_job_submit_t)(C.calloc(1, C.sizeof_xce_job_submit_t))
	js.jobid = C.uint64_t(job.Unique_JobID)
	js.array_size = 1
	js.submit = sd

	return js, sd, si
}

// freeJobSubmit releases C memory allocated by modelJobToSubmit.
func freeJobSubmit(js *C.xce_job_submit_t, sd *C.job_submit_data_t, si *C.job_startinfo_t) {
	if si != nil {
		C.free(unsafe.Pointer(si.cmdline))
		C.free(unsafe.Pointer(si.execuser))
		C.free(unsafe.Pointer(si.launch_type))
		C.free(unsafe.Pointer(si.job_pre_exec))
		C.free(unsafe.Pointer(si.job_post_exec))
		C.free(unsafe.Pointer(si.exechome))
		C.free(unsafe.Pointer(si.execcwd))
		C.free(unsafe.Pointer(si.exdata))
		C.free(unsafe.Pointer(si))
	}
	if sd != nil {
		C.free(unsafe.Pointer(sd.queue))
		C.free(unsafe.Pointer(sd.user))
		C.free(unsafe.Pointer(sd.job_name))
		C.free(unsafe.Pointer(sd.project_name))
		C.free(unsafe.Pointer(sd.user_group))
		C.free(unsafe.Pointer(sd.job_description))
		C.free(unsafe.Pointer(sd.resreq))
		C.free(unsafe.Pointer(sd.from_host))
		C.free(unsafe.Pointer(sd))
	}
	if js != nil {
		C.free(unsafe.Pointer(js))
	}
}

// cNameValToWorkerResval converts a C xce_name_val_t pointer to a ResourceDistributionInfo.
// workerName is set as Location. Returns nil if nv or nv.val is nil.
func cNameValToWorkerResval(workerName string, nv *C.xce_name_val_t) *models.ResourceDistributionInfo {
	if nv == nil {
		return nil
	}
	rd := &models.ResourceDistributionInfo{
		ResourceName: C.GoString(nv.name),
		Location:     workerName,
		IsDynamic:    true,
	}
	if nv.val == nil {
		logger.Warn("worker resval with nil value, skipping",
			zap.String("resource", rd.ResourceName),
			zap.String("host", workerName))
		return nil
	}
	switch C.xce_val_which(nv.val) {
	case C.Value_u_ivalue32:
		rd.ValueType = int(models.NUMERIC)
		rd.NumericValue = float64(C.xce_val_i32(nv.val))
	case C.Value_u_ivalue64:
		rd.ValueType = int(models.NUMERIC)
		rd.NumericValue = float64(C.xce_val_i64(nv.val))
	case C.Value_u_uvalue32:
		rd.ValueType = int(models.NUMERIC)
		rd.NumericValue = float64(C.xce_val_u32(nv.val))
	case C.Value_u_uvalue64:
		rd.ValueType = int(models.NUMERIC)
		rd.NumericValue = float64(C.xce_val_u64(nv.val))
	case C.Value_u_fvalue:
		rd.ValueType = int(models.NUMERIC)
		rd.NumericValue = float64(C.xce_val_fval(nv.val))
	case C.Value_u_svalue:
		rd.ValueType = int(models.STRING)
		rd.StringValue = C.GoString(C.xce_val_str(nv.val))
	case C.Value_u_bvalue:
		rd.ValueType = int(models.BOOLEAN)
		rd.BooleanValue = C.xce_val_bool(nv.val) != 0
	default:
		logger.Warn("unknown xce_val_t discriminant in worker resval, skipping",
			zap.Int("which", int(C.xce_val_which(nv.val))),
			zap.String("resource", rd.ResourceName),
			zap.String("host", workerName))
		return nil
	}
	return rd
}

// cResvalToDistribution converts a C xce_resval_t pointer to a Go ResourceDistributionInfo.
// Returns nil if vp.val is NULL (nil guard prevents CGO panic).
// vp.val is a *C.xce_val_t pointer field — pass directly, no &.
func cResvalToDistribution(vp *C.xce_resval_t) *models.ResourceDistributionInfo {
	rd := &models.ResourceDistributionInfo{
		ResourceName: C.GoString(vp.name),
		Location:     C.GoString(vp.host),
		IsDynamic:    true, // resource values pushed from workers are always dynamic
	}
	// Guard: vp.val may be NULL if the controller sends a resval with no value set.
	// Dereferencing NULL in xce_val_which() would segfault in C.
	if vp.val == nil {
		logger.Warn("resval with nil value, skipping",
			zap.String("resource", rd.ResourceName),
			zap.String("host", rd.Location))
		return nil
	}
	switch C.xce_val_which(vp.val) {
	case C.Value_u_ivalue32:
		rd.ValueType = int(models.NUMERIC)
		rd.NumericValue = float64(C.xce_val_i32(vp.val))
	case C.Value_u_ivalue64:
		rd.ValueType = int(models.NUMERIC)
		rd.NumericValue = float64(C.xce_val_i64(vp.val))
	case C.Value_u_uvalue32:
		rd.ValueType = int(models.NUMERIC)
		rd.NumericValue = float64(C.xce_val_u32(vp.val))
	case C.Value_u_uvalue64:
		rd.ValueType = int(models.NUMERIC)
		rd.NumericValue = float64(C.xce_val_u64(vp.val))
	case C.Value_u_fvalue:
		rd.ValueType = int(models.NUMERIC)
		rd.NumericValue = float64(C.xce_val_fval(vp.val))
	case C.Value_u_svalue:
		rd.ValueType = int(models.STRING)
		rd.StringValue = C.GoString(C.xce_val_str(vp.val))
	case C.Value_u_bvalue:
		rd.ValueType = int(models.BOOLEAN)
		rd.BooleanValue = C.xce_val_bool(vp.val) != 0
	default:
		logger.Warn("unknown xce_val_t discriminant, skipping",
			zap.Int("which", int(C.xce_val_which(vp.val))),
			zap.String("resource", rd.ResourceName),
			zap.String("host", rd.Location))
		return nil
	}
	return rd
}