mesh-catalog/modules/gitea/cmd/forge/copy_test.go

package main

import (
	"context"
	"encoding/base64"
	"encoding/json"
	"fmt"
	"net/http"
	"sort"
	"strings"
	"testing"
)

// sourceRepo is another repository of the fake forge, answering the contents, branch, repository and commit
// reads in the shapes the module already parses elsewhere (forge.go, reads.go, commit.go); a file's content
// is the forge's base64 "content" with its "encoding".
type sourceRepo struct {
	branches map[string]string
	trees    map[string]map[string]string // commit -> path -> content
	symlinks map[string]bool              // paths that are symlinks, in every tree
	moveTo   map[string]string            // branch -> commit it moves to right after it is first read
	tags     map[string]string
	reads    []string
}

func (f *fakeForge) addSource(name string, s *sourceRepo) {
	if f.sources == nil {
		f.sources = map[string]*sourceRepo{}
	}
	f.sources[name] = s
}

func (f *fakeForge) handleSource(w http.ResponseWriter, r *http.Request, path string) {
	rest := strings.TrimPrefix(path, "/api/v1/repos/acme/")
	name, rest, _ := strings.Cut(rest, "/")
	s := f.sources[name]
	if s == nil {
		http.NotFound(w, r)
		return
	}
	reply := func(v any) { _ = json.NewEncoder(w).Encode(v) }
	s.reads = append(s.reads, r.Method+" "+path+"?"+r.URL.RawQuery)
	switch {
	case rest == "":
		reply(map[string]any{"full_name": "acme/" + name, "default_branch": "main"})
	case strings.HasPrefix(rest, "branches/"):
		b := strings.TrimPrefix(rest, "branches/")
		c, ok := s.branches[b]
		if !ok {
			http.NotFound(w, r)
			return
		}
		reply(map[string]any{"name": b, "commit": map[string]any{"id": c}})
		if to, ok := s.moveTo[b]; ok {
			delete(s.moveTo, b)
			s.branches[b] = to
		}
	case rest == "commits":
		if c, ok := s.tags[r.URL.Query().Get("sha")]; ok {
			reply([]any{map[string]any{"sha": c}})
			return
		}
		reply([]any{})
	case strings.HasPrefix(rest, "contents/"):
		p := strings.TrimPrefix(rest, "contents/")
		tree, ok := s.trees[r.URL.Query().Get("ref")]
		if !ok {
			http.NotFound(w, r)
			return
		}
		if content, ok := tree[p]; ok {
			typ := "file"
			if s.symlinks[p] {
				typ = "symlink"
			}
			reply(map[string]any{"type": typ, "path": p, "name": p[strings.LastIndex(p, "/")+1:], "sha": blob(content),
				"size": len(content), "encoding": "base64", "content": base64.StdEncoding.EncodeToString([]byte(content))})
			return
		}
		// A directory: its entries, each with its full path.
		entries := map[string]map[string]any{}
		for q, content := range tree {
			if !strings.HasPrefix(q, p+"/") {
				continue
			}
			sub := strings.TrimPrefix(q, p+"/")
			first, more, nested := strings.Cut(sub, "/")
			_ = more
			if nested {
				entries[first] = map[string]any{"name": first, "path": p + "/" + first, "type": "dir"}
				continue
			}
			typ := "file"
			if s.symlinks[q] {
				typ = "symlink"
			}
			entries[first] = map[string]any{"name": first, "path": q, "type": typ, "size": len(content), "sha": blob(content)}
		}
		if len(entries) == 0 {
			http.NotFound(w, r)
			return
		}
		var names []string
		for n := range entries {
			names = append(names, n)
		}
		sort.Strings(names)
		list := []any{}
		for _, n := range names {
			list = append(list, entries[n])
		}
		reply(list)
	default:
		http.NotFound(w, r)
	}
}

func newSource() *sourceRepo {
	return &sourceRepo{
		branches: map[string]string{"main": "s0"},
		tags:     map[string]string{"v1": "s0"},
		trees: map[string]map[string]string{
			"s0": {"lib/a.go": "package a\n", "lib/deep/b.go": "package b\n", "lib/deep/er/c.txt": "c\n", "top.md": "top\n", "other/x": "x\n"},
		},
	}
}

func copyAsk(t *testing.T, files ...map[string]any) CommitAsk {
	return ask(t, map[string]any{"branch": "fix/1365-x", "files": filesArg(files...)})
}

func cp(path, repo, ref, src string) map[string]any {
	return map[string]any{"path": path, "copy_from": map[string]any{"owner": "acme", "repo": repo, "ref": ref, "path": src}}
}

// Seen failing first: with the expansion switched off the commit carried no copy and the answer named no source.
func TestACopyFromAFileCommitsTheSourcesContentAndNamesItsCommit(t *testing.T) {
	f := newFake(t)
	f.addSource("src", newSource())
	got, err := commit(context.Background(), f.serve(), copyAsk(t, cp("vendor/top.md", "src", "main", "top.md")))
	if err != nil {
		t.Fatal(err)
	}
	if f.trees[got.Commit]["vendor/top.md"] != "top\n" {
		t.Fatalf("tree %v", f.trees[got.Commit])
	}
	if len(got.Copies) != 1 || got.Copies[0].Source != "s0" || got.Copies[0].Path != "vendor/top.md" || got.Copies[0].Files != 1 {
		t.Fatalf("copies %+v", got.Copies)
	}
}

// Seen failing first, as above.
func TestACopyFromADirectoryCopiesEveryFileUnderItToTheSameRelativePaths(t *testing.T) {
	f := newFake(t)
	f.addSource("src", newSource())
	got, err := commit(context.Background(), f.serve(), copyAsk(t, cp("pkg/lib", "src", "v1", "lib"),
		map[string]any{"path": "own.md", "content": "own\n"}))
	if err != nil {
		t.Fatal(err)
	}
	tree := f.trees[got.Commit]
	want := map[string]string{"pkg/lib/a.go": "package a\n", "pkg/lib/deep/b.go": "package b\n", "pkg/lib/deep/er/c.txt": "c\n", "own.md": "own\n"}
	for p, c := range want {
		if tree[p] != c {
			t.Errorf("%s is %q, want %q", p, tree[p], c)
		}
	}
	if tree["pkg/lib/top.md"] != "" || tree["pkg/other/x"] != "" {
		t.Errorf("files outside the directory were copied: %v", tree)
	}
	if len(got.Copies) != 1 || got.Copies[0].Source != "s0" || got.Copies[0].Files != 3 || got.Copies[0].Bytes != len("package a\npackage b\nc\n") {
		t.Fatalf("copies %+v", got.Copies)
	}
	if len(got.Files) != 4 {
		t.Errorf("files %+v", got.Files)
	}
}

// Seen failing first, as above: the files were read at the branch's new head.
func TestEveryCopiedFileIsReadAtTheOneShaTheRefResolvedTo(t *testing.T) {
	f := newFake(t)
	s := newSource()
	s.trees["s1"] = map[string]string{"lib/a.go": "package MOVED\n", "lib/deep/b.go": "package MOVED\n", "lib/deep/er/c.txt": "MOVED\n"}
	s.moveTo = map[string]string{"main": "s1"}
	f.addSource("src", s)
	got, err := commit(context.Background(), f.serve(), copyAsk(t, cp("pkg", "src", "main", "lib")))
	if err != nil {
		t.Fatal(err)
	}
	if f.trees[got.Commit]["pkg/a.go"] != "package a\n" || got.Copies[0].Source != "s0" {
		t.Fatalf("tree %v, copies %+v", f.trees[got.Commit], got.Copies)
	}
	for _, r := range s.reads {
		if strings.Contains(r, "/contents/") && !strings.Contains(r, "ref=s0") {
			t.Errorf("a file was read at another ref: %s", r)
		}
	}
}

// Seen failing first: each case was taken or refused for another reason (or sent).
func TestEveryRefusalOfACopySendsNothingAndSaysWhy(t *testing.T) {
	big := strings.Repeat("z", maxFileBytes+1)
	many := map[string]string{}
	for i := 0; i < maxFiles+1; i++ {
		many[fmt.Sprintf("many/f%03d", i)] = "m"
	}
	for name, c := range map[string]struct {
		file map[string]any
		want string
	}{
		"unknown repository": {cp("t", "nope", "main", "top.md"), "no repository acme/nope"},
		"unknown ref":        {cp("t", "src", "nothing", "top.md"), `no branch, tag or commit "nothing"`},
		"unknown path":       {cp("t", "src", "main", "missing/dir"), `no "missing/dir"`},
		"symlink in source":  {cp("t", "src", "main", "link"), "link is a symlink"},
		"symlink under dir":  {cp("t", "src", "main", "withlink"), "withlink/l is a symlink"},
		"file over 4 MiB":    {cp("t", "src", "main", "big"), "more than the 4.0 MiB"},
		"over 200 files":     {cp("t", "src", "main", "many"), "201 files"},
		"with content":       {map[string]any{"path": "t", "content": "x", "copy_from": cp("t", "src", "main", "top.md")["copy_from"]}, "copy_from together with content"},
		"with base64":        {map[string]any{"path": "t", "content_base64": "eA==", "copy_from": cp("t", "src", "main", "top.md")["copy_from"]}, "copy_from together with content"},
		"with delete":        {map[string]any{"path": "t", "delete": true, "copy_from": cp("t", "src", "main", "top.md")["copy_from"]}, "copy_from together with content"},
		"meets a given file": {cp("pkg", "src", "main", "lib"), "pkg/a.go is named twice"},
	} {
		f := newFake(t)
		s := newSource()
		s.trees["s0"]["link"] = "top.md"
		s.symlinks = map[string]bool{"link": true, "withlink/l": true}
		s.trees["s0"]["withlink/l"] = "top.md"
		s.trees["s0"]["withlink/ok"] = "ok"
		s.trees["s0"]["big/f"] = big
		for p, v := range many {
			s.trees["s0"][p] = v
		}
		f.addSource("src", s)
		files := []map[string]any{c.file}
		if name == "meets a given file" {
			files = append(files, map[string]any{"path": "pkg/a.go", "content": "mine"})
		}
		a, err := readCommitAsk(map[string]any{"owner": "acme", "repo": "app", "message": "m", "branch": "fix/1365-x", "files": filesArg(files...)})
		if err == nil {
			_, err = commit(context.Background(), f.serve(), a)
		}
		if err == nil || !strings.Contains(err.Error(), c.want) || !strings.Contains(err.Error(), "Nothing was sent") {
			t.Errorf("%s: %v (want %q)", name, err, c.want)
		}
		if len(f.writes) != 0 || len(f.patches) != 0 || f.branches["fix/1365-x"] != "" {
			t.Errorf("%s: something was sent: %v %v", name, f.writes, f.branches)
		}
	}
}

// Seen failing first: the copy was not counted, so the commit was sent.
func TestACopyOverTheTotalNamesItsCountAndSizeSoItCanBeSplit(t *testing.T) {
	f := newFake(t)
	s := newSource()
	chunk := strings.Repeat("q", maxFileBytes)
	for i := 0; i < 5; i++ {
		s.trees["s0"][fmt.Sprintf("heavy/f%d", i)] = chunk
	}
	f.addSource("src", s)
	_, err := commit(context.Background(), f.serve(), copyAsk(t, cp("h", "src", "main", "heavy")))
	if err == nil || !strings.Contains(err.Error(), "5 files, 20.0 MiB together") || !strings.Contains(err.Error(), "by subdirectory") {
		t.Fatalf("%v", err)
	}
	if len(f.writes) != 0 {
		t.Fatal("sent")
	}
}

func TestCopyFromIsInTheToolsSchemaAndTheReadme(t *testing.T) {
	var schema map[string]any
	for _, tl := range Tools(func() (*Forge, error) { return nil, fmt.Errorf("unset") }) {
		if tl.Name == "gitea_commit" {
			schema = tl.Input
		}
	}
	item := schema["properties"].(map[string]any)["files"].(map[string]any)["items"].(map[string]any)["properties"].(map[string]any)
	cf, ok := item["copy_from"].(map[string]any)
	if !ok {
		t.Fatalf("copy_from is not in the schema: %v", item)
	}
	props := cf["properties"].(map[string]any)
	for _, k := range []string{"owner", "repo", "ref", "path"} {
		if _, ok := props[k]; !ok {
			t.Errorf("copy_from has no %s", k)
		}
	}
}

func TestACopyFromThatIsNotAnObjectIsRefused(t *testing.T) {
	_, err := readCommitAsk(map[string]any{"owner": "acme", "repo": "app", "message": "m", "branch": "b",
		"files": filesArg(map[string]any{"path": "x", "copy_from": "acme/src:top.md"})})
	if err == nil || !strings.Contains(err.Error(), "copy_from is an object") {
		t.Fatalf("%v", err)
	}
}