summercms/internal/docsite/load.go

package docsite

import (
	"bytes"
	"cmp"
	"fmt"
	"io/fs"
	"os"
	"path"
	"path/filepath"
	"regexp"
	"slices"
	"strings"
	"unicode/utf8"

	"github.com/goccy/go-yaml"
)

// Site is docs/site.yaml, decoded strictly.
type Site struct {
	Title       string `yaml:"title"`
	Description string `yaml:"description"`
	BaseURL     string `yaml:"base_url"`
	// EditURL and SourceURL carry a {path} token replaced with a
	// repository-relative path.
	EditURL   string    `yaml:"edit_url"`
	SourceURL string    `yaml:"source_url"`
	LLMSNotes []string  `yaml:"llms_notes"`
	Sections  []Section `yaml:"sections"`
}

// Section is one sidebar group, listed in sidebar order in site.yaml.
type Section struct {
	Name  string `yaml:"name"`
	Title string `yaml:"title"`
}

// Frontmatter is the YAML block at the top of every docs page. Every field
// is required.
type Frontmatter struct {
	Title       string `yaml:"title"`
	Description string `yaml:"description"`
	Section     string `yaml:"section"`
	Order       int    `yaml:"order"`
}

// Page is one page of the site: a docs/ page or an ingested module README.
type Page struct {
	// Source is the repository-relative source path (forward slashes).
	Source string
	// URL is the extension-less output path, such as "setup/installation",
	// "api/lagoon" or "index".
	URL         string
	Section     string
	Title       string
	Description string
	Order       int
	// Module is the module name for API reference pages.
	Module string
	// Body is the Markdown body, starting with the "# Title" line.
	Body []byte
	// BodyLine is the 1-based line of Body's first line in Source.
	BodyLine int

	abs string
}

const (
	indexSection = "index"
	apiSection   = "api"
	maxDescLen   = 160
)

var frontmatterFields = []string{"title", "description", "section", "order"}

// LoadSite reads and parses a site.yaml file.
func LoadSite(path string) (Site, error) {
	raw, err := os.ReadFile(path)
	if err != nil {
		return Site{}, fmt.Errorf("docsite: read site config %s: %w", path, err)
	}
	return ParseSite(raw)
}

// ParseSite decodes site.yaml, rejecting unknown fields, and validates the
// section list.
func ParseSite(raw []byte) (Site, error) {
	var s Site
	dec := yaml.NewDecoder(bytes.NewReader(raw), yaml.DisallowUnknownField())
	if err := dec.Decode(&s); err != nil {
		return Site{}, fmt.Errorf("docsite: parse site config: %s", firstLine(err.Error()))
	}
	if s.Title == "" {
		return Site{}, fmt.Errorf("docsite: site config: title is required")
	}
	if s.Description == "" {
		return Site{}, fmt.Errorf("docsite: site config: description is required")
	}
	if len(s.Sections) == 0 {
		return Site{}, fmt.Errorf("docsite: site config: sections is required")
	}
	seen := map[string]bool{}
	for _, sec := range s.Sections {
		switch {
		case sec.Name == "" || sec.Title == "":
			return Site{}, fmt.Errorf("docsite: site config: every section needs a name and a title")
		case sec.Name == indexSection:
			return Site{}, fmt.Errorf("docsite: site config: section name %q is reserved for docs/index.md", indexSection)
		case !slugName.MatchString(sec.Name):
			return Site{}, fmt.Errorf("docsite: site config: section name %q must be lowercase letters, digits and dashes", sec.Name)
		case seen[sec.Name]:
			return Site{}, fmt.Errorf("docsite: site config: section %q is listed twice", sec.Name)
		}
		seen[sec.Name] = true
	}
	return s, nil
}

var slugName = regexp.MustCompile(`^[a-z0-9][a-z0-9-]*$`)

func firstLine(s string) string {
	s, _, _ = strings.Cut(strings.TrimSpace(s), "\n")
	return strings.TrimSpace(s)
}

// sectionTitle returns the display title of a section.
func (s Site) sectionTitle(name string) string {
	for _, sec := range s.Sections {
		if sec.Name == name {
			return sec.Title
		}
	}
	return name
}

func (s Site) hasSection(name string) bool {
	return slices.ContainsFunc(s.Sections, func(sec Section) bool { return sec.Name == name })
}

// site is one assembled documentation site.
type site struct {
	opts    Options
	cfg     Site
	cfgRaw  []byte
	base    string
	pages   []*Page // reading order
	outputs map[string][]byte
}

// rel returns the display path of an absolute path: repository-relative
// with forward slashes when inside Root, otherwise the absolute path.
func (s *site) rel(abs string) string {
	if r, err := filepath.Rel(s.opts.Root, abs); err == nil && within(abs, s.opts.Root) {
		return filepath.ToSlash(r)
	}
	return filepath.ToSlash(abs)
}

// assemble loads, verifies and renders a site in memory.
func assemble(opts Options) (*site, []Problem, error) {
	opts, err := opts.normalize()
	if err != nil {
		return nil, nil, err
	}
	s := &site{opts: opts, outputs: map[string][]byte{}}
	cfgPath := filepath.Join(opts.Src, "site.yaml")
	raw, err := os.ReadFile(cfgPath)
	if err != nil {
		return nil, nil, fmt.Errorf("docsite: read site config %s: %w", cfgPath, err)
	}
	cfg, err := ParseSite(raw)
	if err != nil {
		return nil, []Problem{{File: s.rel(cfgPath), Line: 1, Rule: "site", Message: strings.TrimPrefix(err.Error(), "docsite: ")}}, nil
	}
	s.cfg, s.cfgRaw = cfg, raw
	s.base = strings.TrimRight(cfg.BaseURL, "/")
	if opts.BaseURL != "" {
		s.base = strings.TrimRight(opts.BaseURL, "/")
	}

	guides, problems, err := s.loadGuides()
	if err != nil {
		return nil, nil, err
	}
	s.order(guides)
	problems = append(problems, s.emptySections()...)
	if len(problems) == 0 {
		rp, err := s.render()
		if err != nil {
			return nil, nil, err
		}
		problems = append(problems, rp...)
	}
	sortProblems(problems)
	return s, problems, nil
}

// loadGuides walks Src and loads every page with its frontmatter.
func (s *site) loadGuides() ([]*Page, []Problem, error) {
	files, err := walkPages(s.opts.Src)
	if err != nil {
		return nil, nil, err
	}
	var pages []*Page
	var problems []Problem
	orders := map[string]map[int]string{}
	hasIndex := false
	for _, abs := range files {
		raw, err := os.ReadFile(abs)
		if err != nil {
			return nil, nil, fmt.Errorf("docsite: read %s: %w", abs, err)
		}
		relSrc, err := filepath.Rel(s.opts.Src, abs)
		if err != nil {
			return nil, nil, fmt.Errorf("docsite: %w", err)
		}
		relSrc = filepath.ToSlash(relSrc)
		file := s.rel(abs)
		fail := func(detail string) {
			problems = append(problems, Problem{File: file, Line: 1, Rule: "frontmatter", Message: detail})
		}
		fmRaw, body, bodyLine, ok := splitFrontmatter(raw)
		if !ok {
			fail(`the file must start with a "---" frontmatter block closed by a "---" line`)
			continue
		}
		fm, details := parseFrontmatter(fmRaw)
		if len(details) > 0 {
			for _, d := range details {
				fail(d)
			}
			continue
		}
		dir := path.Dir(relSrc)
		switch {
		case relSrc == "index.md":
			hasIndex = true
			if fm.Section != indexSection {
				fail(fmt.Sprintf("section %q does not match directory %q (the landing page uses section %q)", fm.Section, ".", indexSection))
			}
		case dir == ".":
			fail(fmt.Sprintf("section %q does not match directory %q (only index.md sits at the docs root)", fm.Section, "."))
		case fm.Section != dir:
			fail(fmt.Sprintf("section %q does not match directory %q", fm.Section, dir))
		case fm.Section == apiSection:
			fail(fmt.Sprintf("section %q is reserved for the ingested module READMEs", apiSection))
		case !s.cfg.hasSection(fm.Section):
			fail(fmt.Sprintf("section %q is not listed in %s", fm.Section, s.rel(filepath.Join(s.opts.Src, "site.yaml"))))
		}
		if first, _, _ := bytes.Cut(body, []byte("\n")); string(first) != "# "+fm.Title {
			fail(fmt.Sprintf("first line must be %q", "# "+fm.Title))
		}
		if utf8.RuneCountInString(fm.Description) > maxDescLen {
			fail(fmt.Sprintf("description is longer than %d characters", maxDescLen))
		}
		if orders[fm.Section] == nil {
			orders[fm.Section] = map[int]string{}
		}
		if other, dup := orders[fm.Section][fm.Order]; dup {
			fail(fmt.Sprintf("order %d already used by %s", fm.Order, other))
		} else {
			orders[fm.Section][fm.Order] = file
		}
		pages = append(pages, &Page{
			Source:      file,
			URL:         strings.TrimSuffix(relSrc, ".md"),
			Section:     fm.Section,
			Title:       fm.Title,
			Description: fm.Description,
			Order:       fm.Order,
			Body:        body,
			BodyLine:    bodyLine,
			abs:         abs,
		})
	}
	if !hasIndex {
		problems = append(problems, Problem{File: s.rel(filepath.Join(s.opts.Src, "index.md")), Rule: "page", Message: "the landing page is missing"})
	}
	return pages, problems, nil
}

// walkPages lists the Markdown pages under src in lexical order, skipping
// examples/, dot and underscore entries.
func walkPages(src string) ([]string, error) {
	var files []string
	err := filepath.WalkDir(src, func(p string, d fs.DirEntry, err error) error {
		if err != nil {
			return err
		}
		if p == src {
			return nil
		}
		name := d.Name()
		skip := strings.HasPrefix(name, ".") || strings.HasPrefix(name, "_") ||
			(d.IsDir() && filepath.Dir(p) == src && name == "examples")
		if d.IsDir() {
			if skip {
				return filepath.SkipDir
			}
			return nil
		}
		if !skip && strings.HasSuffix(name, ".md") {
			files = append(files, p)
		}
		return nil
	})
	if err != nil {
		return nil, fmt.Errorf("docsite: walk %s: %w", src, err)
	}
	return files, nil
}

// splitFrontmatter separates a leading "---" block from the body. bodyLine
// is the 1-based line of the body's first line.
func splitFrontmatter(raw []byte) (fm, body []byte, bodyLine int, ok bool) {
	rest, found := bytes.CutPrefix(raw, []byte("---\n"))
	if !found {
		return nil, nil, 0, false
	}
	line := 2
	for off := 0; off <= len(rest); {
		end := bytes.IndexByte(rest[off:], '\n')
		var cur []byte
		next := len(rest) + 1
		if end < 0 {
			cur = rest[off:]
		} else {
			cur = rest[off : off+end]
			next = off + end + 1
		}
		if string(bytes.TrimRight(cur, " \t\r")) == "---" {
			if next > len(rest) {
				return rest[:off], nil, line + 1, true
			}
			return rest[:off], rest[next:], line + 1, true
		}
		off = next
		line++
	}
	return nil, nil, 0, false
}

// parseFrontmatter decodes a frontmatter block strictly. It returns the
// UI-SPEC problem details: unknown and missing fields first, then decode
// errors.
func parseFrontmatter(raw []byte) (Frontmatter, []string) {
	var keys map[string]any
	if err := yaml.Unmarshal(raw, &keys); err != nil {
		return Frontmatter{}, []string{firstLine(err.Error())}
	}
	var details []string
	unknown := make([]string, 0)
	for k := range keys {
		if !slices.Contains(frontmatterFields, k) {
			unknown = append(unknown, k)
		}
	}
	slices.Sort(unknown)
	for _, k := range unknown {
		details = append(details, fmt.Sprintf("unknown field %q", k))
	}
	for _, k := range frontmatterFields {
		v, ok := keys[k]
		if !ok || v == nil || v == "" {
			details = append(details, fmt.Sprintf("missing field %q", k))
		}
	}
	if len(details) > 0 {
		return Frontmatter{}, details
	}
	var fm Frontmatter
	dec := yaml.NewDecoder(bytes.NewReader(raw), yaml.DisallowUnknownField())
	if err := dec.Decode(&fm); err != nil {
		return Frontmatter{}, []string{firstLine(err.Error())}
	}
	return fm, nil
}

// order sorts pages into reading order: index first, then each site.yaml
// section in order with its pages by order (API pages by module name).
func (s *site) order(pages []*Page) {
	bySection := map[string][]*Page{}
	for _, p := range pages {
		bySection[p.Section] = append(bySection[p.Section], p)
	}
	s.pages = s.pages[:0]
	s.pages = append(s.pages, bySection[indexSection]...)
	for _, sec := range s.cfg.Sections {
		list := bySection[sec.Name]
		slices.SortStableFunc(list, func(a, b *Page) int {
			if a.Module != "" || b.Module != "" {
				return strings.Compare(a.Module, b.Module)
			}
			return cmp.Or(cmp.Compare(a.Order, b.Order), strings.Compare(a.URL, b.URL))
		})
		s.pages = append(s.pages, list...)
	}
}

// emptySections reports every site.yaml section without pages.
func (s *site) emptySections() []Problem {
	count := map[string]int{}
	for _, p := range s.pages {
		count[p.Section]++
	}
	var problems []Problem
	for _, sec := range s.cfg.Sections {
		if count[sec.Name] > 0 {
			continue
		}
		problems = append(problems, Problem{
			File:    s.rel(filepath.Join(s.opts.Src, "site.yaml")),
			Line:    sectionLine(s.cfgRaw, sec.Name),
			Rule:    "section",
			Message: fmt.Sprintf("%q has no pages (add the section in the same change as its first page)", sec.Name),
		})
	}
	return problems
}

// sectionLine finds the site.yaml line that names a section.
func sectionLine(raw []byte, name string) int {
	re := regexp.MustCompile(`^\s*(-\s*)?name:\s*["']?` + regexp.QuoteMeta(name) + `["']?\s*$`)
	for i, line := range strings.Split(string(raw), "\n") {
		if re.MatchString(line) {
			return i + 1
		}
	}
	return 1
}