path: root/doStats.go

// Copyright 2017-2025 WIT.COM Inc. All rights reserved.
// Use of this source code is governed by the GPL 3.0

package main

import (
	"errors"
	"fmt"
	"os"
	"path/filepath"
	"slices"
	"strings"

	"go.wit.com/lib/cobol"
	"go.wit.com/lib/config"
	"go.wit.com/lib/env"
	"go.wit.com/lib/protobuf/argvpb"
	"go.wit.com/lib/protobuf/gitpb"
	"go.wit.com/log"
	"google.golang.org/protobuf/types/known/timestamppb"
)

func doStats(cmd *StatsCmd) (string, error) {
	var allerr error
	if argv.All {
		for r := range me.forge.Repos.IterByFullPath() {
			if r.Stats().Len() == 0 {
				_, err := doStatsRepo(r)
				allerr = errors.Join(allerr, err)
			}
		}
		return "verify ran everywhere", nil
	}

	repo := workingDirToRepo()
	if repo == nil {
		return "no repo", errors.New("working dir isn't a repo I know about")
	}

	if cmd.List {
		s := repo.Stats()
		s.SortPatchId()
		// s.SortCtime()
		footer := s.PrintTableLimit(-1)
		return "stats table: " + footer, nil
	}

	s, err := doStatsRepo(repo)
	allerr = errors.Join(allerr, err)

	user, err := refHash(repo, "heads/"+repo.GetUserBranchName())
	allerr = errors.Join(allerr, err)
	master, err := refHash(repo, "remotes/origin/master")
	allerr = errors.Join(allerr, err)
	HEAD, err := refHash(repo, "remotes/origin/HEAD")
	allerr = errors.Join(allerr, err)
	log.Printf("user=%10.10s master=%10.10s HEAD=%10.10s\n", user, master, HEAD)

	safeDelete(repo, "refs/heads/"+repo.GetUserBranchName(), "refs/remotes/origin/HEAD") // delete user if safely contained in HEAD
	safeDelete(repo, user, HEAD)                                                         // checkes by hash ID                                   // delete user if safely contained in HEAD

	return s, allerr
}

func doStatsRepo(r *gitpb.Repo) (string, error) {
	var allerr error
	pb, err := r.LoadStats("origin")
	if err == nil {
		log.Info("LoadStats() ok", pb.Filename)
	} else {
		log.Info("LoadStats() err", err)
	}
	allerr = errors.Join(allerr, err)

	if hasOrigin(r) {
		log.Info("todo: detect origin")
	}

	// collect the stats
	counter, err := last100(r, pb)
	allerr = errors.Join(allerr, err)
	s := log.Sprintf("found %d new hashes", counter)

	for stat := range pb.IterAll() {
		if stat.PatchId == "" {
			stat.PatchId, err = r.FindPatchIdByHash(stat.Hash)
			allerr = errors.Join(allerr, err)
			log.Info("patchid for hash", stat.Hash, "is", stat.PatchId)
			counter += 1
		}
	}

	if counter > 0 {
		pb.SaveVerbose()
	}
	return s, nil
}

func collectStats(r *gitpb.Repo, pb *gitpb.Stats) error {
	return nil
}

// git show-ref refs/heads/devel
func refHash(r *gitpb.Repo, name string) (string, error) {
	var hash string
	refname := "refs/" + name
	cmd := []string{"git", "show-ref", refname}
	cmdout := r.Run(cmd)
	for i, line := range cmdout.Stdout {
		parts := strings.Fields(line)
		if env.If("stats") {
			log.Info(parts[0], "LINE:", i, line)
		}
		hash = parts[0]
	}
	if len(cmdout.Stdout) != 1 {
		return "", errors.New("no refname:" + name)
	}
	return hash, nil
}

// git show-ref --verify refs/heads/devel
func hasOrigin(r *gitpb.Repo) bool {
	// git show-ref refs/heads/devel
	return true
}

var standardFmts []string = []string{"H", "T", "at", "ct", "f"}
var standardSeperator string = "___FORGE___"

func makeFmts() string {
	// fmts := strings.Fields(config.GetPanic("standardFmts"))
	// fmts := strings.Fields(config.GetPanic("standardSeperator"))
	var all []string
	for _, fmtvar := range standardFmts {
		all = append(all, "%"+fmtvar)
	}
	return "--format=" + strings.Join(all, standardSeperator)
}

// the correct syntax for
// git log -n 8 --format=%H%00%ae%00%as%00%s origin/HEAD
func last100(r *gitpb.Repo, pb *gitpb.Stats) (int, error) {
	var allerr error
	var counter int
	cmd := []string{"git", "log", "-n", "100", makeFmts(), "origin/" + r.GetMasterBranchName()} // must use 'master' as queried from the git server
	// cmd := []string{"git", "log", "-n", "100", makeFmts(), "origin/HEAD"} // HEAD is _NOT_ always set
	if env.If("stats") {
		log.Info("Run:", cmd)
	}
	cmdout := r.Run(cmd)
	for i, line := range cmdout.Stdout {
		parts := strings.Split(line, standardSeperator)
		hash := parts[0]
		if env.If("stats") {
			log.Printf("LINE:%8.8s %2d %v\n", hash, i, parts[1:])
		}
		found := pb.FindByHash(hash)
		if found != nil {
			// already have this hash
			continue
		}
		counter += 1
		astat := new(gitpb.Stat)
		astat.Hash = hash
		ctime, err := cobol.GetTime(parts[2])
		allerr = errors.Join(allerr, err)
		astat.Ctime = timestamppb.New(*ctime)
		// astat.Subject = parts[4]
		// astat.Type = gitpb.Stat_REMOTE
		pb.Append(astat)
	}
	return counter, allerr
}

func findPatchIdInStats(pb *gitpb.Stats, patchId string) *gitpb.Stat {
	for stat := range pb.IterAll() {
		if stat.PatchId == patchId {
			return stat
		}
	}
	// log.Info("findPatchId searched in", pb.Len(), "stats")
	return nil
}

// delete localRef if it's completely contained in the masterRef
func safeDelete(r *gitpb.Repo, deleteHash string, keepHash string) error {
	// compare the branches
	hashok, hashbad, cmd1, cmd2, err := r.CompareHashes(keepHash, deleteHash)

	if err != nil {
		// things are really really messed up. might be 'branchless' at this point (?)
		log.Printf("%-13.13s %-55.55s err='%v' %s %v\n", "CMD ERR", r.FullPath, err, "NOT SAFE TO DELETE. Reload()?", cmd1)
		log.Printf("%-13.13s %-55.55s err='%v' %s %v\n", "CMD ERR", r.FullPath, err, "NOT SAFE TO DELETE. Reload()?", cmd2)
		return err
	}

	// things only in the master branch (safe to ignore)
	for _, line := range hashok {
		parts := strings.Split(line, "%00") // git log doesn't actually convert %00 to NULL
		patchId, err := r.FindPatchIdByHash(parts[0])
		_ = err
		log.Printf("%-13.13s %-55.55s hId %10.10s pId %10.10s %v\n", "OK delete", r.FullPath, parts[0], patchId, parts[2:])
	}

	if len(hashbad) > 0 {
		log.Printf("%-13.13s %v\n", "BAD cmd", cmd1)
	}
	var ACTUALLYOK bool = true
	// things still only in the local branch (bad to delete)
	for _, line := range hashbad {
		parts := strings.Split(line, "%00") // git log doesn't actually convert %00 to NULL
		patchId, err := r.FindPatchIdByHash(parts[0])
		_ = err
		searchResult := log.Sprintf("NOPE(%d)", r.Stats().Len())
		stat := findPatchIdInStats(r.Stats(), patchId)
		if stat != nil {
			searchResult = log.Sprintf("FOUND %10.10s %s", stat.PatchId, "todo: []slice")
		} else {
			ACTUALLYOK = false
		}

		log.Printf("%-13.13s %-55.55s hId %10.10s pId %10.10s %s %v\n", "BAD keep", r.FullPath, parts[0], patchId, searchResult, parts[2:])
	}

	if ACTUALLYOK {
		// todo: force checkout to local master branch
		// before doing this
		cmd := []string{"git", "update-ref", "-d", deleteHash}
		// log.Printf("%-13.13s %-55.55s %v %s\n", "CMD OK", r.FullPath, cmd1, "")
		// log.Printf("%-13.13s %-55.55s %v %s\n", "CMD OK", r.FullPath, cmd2, "")
		if r.GetCurrentBranchName() == r.GetMasterBranchName() {
			log.Printf("%-55.55s %v %s\n", r.FullPath, cmd, "SAFE TO DELETE add --fix")
		} else {
			log.Printf("%-55.55s %v %s\n", r.FullPath, cmd, "SAFE TO DELETE BUT NOT ON MASTER BRANCH add --fix")
		}
		if argv.Fix {
			err := r.RunVerbose(cmd)
			if err != nil {
				log.Info(deleteHash, r.FullPath)
				s := "local user branch could not be deleted"
				argvpb.BadExit(s, err)
			}
		}
		return ErrorNeedArgvFix
	}

	return log.Errorf("NOT SAFE")
}

//	/tmp/go-nsupdate$ git ls-remote
//	From [email protected]:SpComb/go-nsupdate.git
//	d65f28d10991354b3af0168001a4beea6b8242f9	HEAD
//	d65f28d10991354b3af0168001a4beea6b8242f9	refs/heads/master
//	fb04ebe13a16c01e1a7eb3892a0aca8c6db96fa0	refs/pull/3/head
//	d774220311b2d44c770e7431ec663c8875488a1e	refs/pull/6/head
//	fefb9ea80745893a2203576bdc2872d33e539064	refs/pull/7/head
//	c09c90d0a931888862262a6ed3753eed0af4ddca	refs/pull/8/head

// the correct syntax for
// git log -n 8 --format=%H%00%ae%00%as%00%s origin/HEAD
func updateStats(r *gitpb.Repo, pb *gitpb.Stats, remoteName string) (int, error) {
	var allerr error
	var counter int
	cmd := []string{"git", "log", "-n", "100", makeFmts(), remoteName} // must use 'master' as queried from the git server
	// cmd := []string{"git", "log", "-n", "100", makeFmts(), "origin/HEAD"} // HEAD is _NOT_ always set
	if env.True("stats") {
		log.Info("Run:", cmd)
	}
	cmdout := r.Run(cmd)
	for i, line := range cmdout.Stdout {
		parts := strings.Split(line, standardSeperator)
		hash := parts[0]
		if len(parts) < 2 {
			log.Printf("Repo: %s\n", r.FullPath)
			log.Printf("CMD: %v\n", cmd)
			log.Printf("LINE:%8.8s %2d %v\n", hash, i, parts[1:])
		}
		if env.True("stats") {
			log.Printf("LINE:%8.8s %2d %v\n", hash, i, parts[1:])
		}
		found := pb.FindByHash(hash)
		if found != nil {
			// already have this hash
			continue
		}
		counter += 1
		astat := new(gitpb.Stat)
		astat.Hash = hash
		ctime, err := cobol.GetTime(parts[2])
		allerr = errors.Join(allerr, err)
		astat.Ctime = timestamppb.New(*ctime)
		// astat.Subject = parts[4]
		// astat.Type = gitpb.Stat_REMOTE
		pb.Append(astat)
	}
	return counter, allerr
}

/*
func lookupRefs(r *gitpb.Repo, remoteRefs *gitpb.Stats) error {
	rname := remoteRefs.Name()
	for stat := range remoteRefs.IterAll() {
		log.Info("remote", rname, stat.Name)
	}
}
*/

// returns err if anything changes or anything is wrong (todo: should these be different?)
func makeRefs(r *gitpb.Repo, remoteName string) error {
	var counter int // inc if anything changes
	fullname := filepath.Join(r.FullPath, ".git", remoteName+".refs.pb")
	stats := gitpb.NewStats()
	stats.Filename = fullname
	err := config.ForceCreatePB(stats)
	if err != nil {
		return err
	}

	var last *gitpb.Stat
	for stat := range stats.IterAll() {
		if last == nil {
			last = stat
			continue
		}
		if strings.Compare(stat.Hash, last.Hash) > 0 {
			log.Info("Compare worked", stat.Hash, last.Hash, r.FullPath)
		} else {
			log.Info("Compare failed", stat.Hash, last.Hash, r.FullPath)
			os.Remove(stats.Filename)
			return errors.New("out of order")
		}
		if strings.Compare(stat.Hash, last.Hash) == 0 {
			panic("was match")
		}
		last = stat
	}

	// cmd := []string{"git", "show-ref"} // must use 'master' as queried from the git server
	cmd := []string{"git", "ls-remote", remoteName} // must use 'master' as queried from the git server // GO has 250k remote refs
	// cmd := []string{"git", "log", "-n", "100", makeFmts(), "origin/HEAD"} // HEAD is _NOT_ always set
	if env.True("stats") {
		log.Info("STATS VERBOSE Run:", cmd)
	}
	cmdout := r.Run(cmd)
	if len(cmdout.Stdout) == 0 {
		return errors.New("got nothing back")
	}
	var ticker int
	var done int
	var allnew []*gitpb.Stat
	for _, line := range cmdout.Stdout {
		line = strings.TrimSpace(line)
		parts := strings.Fields(line)
		if len(parts) != 2 {
			log.Printf("Repo: %s\n", r.FullPath)
			log.Printf("CMD: %v\n", cmd)
			log.Printf("LINE:%s\n", line)
			return errors.New(line)
		}
		if env.True("stats") {
			log.Printf("LINE:%v %d %s\n", parts, ticker, r.FullPath)
		}
		// refpath := filepath.Join("refs/remote", remoteName)
		// if strings.HasPrefix(parts[1], refpath) {
		// this ref is not from the remote
		// 	continue
		// }
		counter += 1
		newstat := new(gitpb.Stat)
		// newstat.Type = gitpb.Stat_REMOTE
		newstat.Hash = parts[0]
		// newstat.Name = parts[1]
		if stats.Len() == 0 {
			stats.Append(newstat)
			continue
		}
		n, found := slices.BinarySearchFunc(stats.Stats, newstat, func(a, b *gitpb.Stat) int {
			return strings.Compare(a.Hash, b.Hash)
		})
		_ = n

		if n > stats.Len() {
			log.Info("WTF n,len =", n, stats.Len())
			stats.Append(newstat)
			continue
		}
		if n-1 < 0 {
			log.Info("WTF n,len =", n, stats.Len())
			continue
		}

		testfind := stats.Stats[n-1]
		if testfind.Hash == newstat.Hash {
			done += 1
			log.Info(counter, ticker, "N WAS RIGHT", n, found, newstat.Hash, "HASH", testfind.Hash)
			panic("fucknuts")
		}
		if found {
			done += 1
			teststat := stats.FindByHash(newstat.Hash)
			if teststat == nil {
				// log.Info(counter, ticker, "FOUND TEST STAT:", n, found, "HASH", newstat)
				log.Printf("FOUND:%v %d/%d/%d %s %v\n", parts, done, counter, len(cmdout.Stdout), r.FullPath, teststat)
				panic("fucknuts")
			}
			continue
		} else {
			teststat := stats.FindByHash(newstat.Hash)
			if teststat != nil {
				log.Printf("NOT FOUND:%v %d/%d/%d %s %v\n", parts, done, counter, len(cmdout.Stdout), r.FullPath, teststat)
				// log.Info(counter, ticker,  "NOT FOUND TEST STAT:", n, found, teststat, "HASH", newstat)
				panic("fucknuts")
			}
		}
		allnew = append(allnew, newstat)

		ticker += 1
		if ticker > 1000 {
			log.Printf("TICKER:%v %d/%d/%d %s\n", parts, done, counter, len(cmdout.Stdout), r.FullPath)
			ticker = 0
			break
			/*
				stats.Stats = append(stats.Stats, allnew...)
				stats.SortByHash()
				stats.Save()
				log.Printf("TICKER:%v %d/%d/%d %s\n", parts, done, counter, len(cmdout.Stdout), r.FullPath)
				return nil
			*/
		}
	}
	if counter > 0 {
		stats.Stats = append(stats.Stats, allnew...)
		stats.SortByHash()
		stats.Save()
		return errors.New(fmt.Sprintf("len(%d), ticker(%d) counter(%d) refs changed", stats.Len(), ticker, counter))
	}
	return nil
}