sm-seo-plugin/classes/resolver.go

package classes

import (
	"context"
	"errors"
	"net/url"
	"strings"

	"git.golem15.com/golem15/summercms/modules/pact"
)

// Config keys of the resolver.
const (
	// SiteURLConfigKey is the absolute origin of the public site (for
	// example https://example.com). Canonical and other absolute URLs are
	// built from it, falling back to app.url; never from a request's Host.
	SiteURLConfigKey = "golem15.seo.site_url"
	// UnprefixedLocaleConfigKey names the locale whose public URLs carry no
	// locale prefix; empty leaves the choice to each content type.
	UnprefixedLocaleConfigKey = "golem15.seo.unprefixed_locale"
)

// MaxURLBytes caps the URL the resolver accepts.
const MaxURLBytes = 2048

// ErrBadURL is returned for an input that is not a path or an http(s) URL.
var ErrBadURL = errors.New("golem15.seo: url must be a path or an absolute http(s) URL")

// The actions of a resolve answer.
const (
	ActionOK       = "ok"
	ActionNotFound = "not_found"
)

// The sources of a resolved value (Result.Sources).
const (
	SourceRule     = "rule"
	SourceRecord   = "record"
	SourceTemplate = "template"
	SourceGlobal   = "global"
	SourceContent  = "content"
	SourceDefault  = "default"
)

// Result is the effective SEO of one URL. Action is ok or not_found; Status
// is the HTTP status the frontend should answer with. URL is the absolute
// URL of the page (path and kept query on the site URL). Sources maps each
// value name (title, description, canonical, robots, og_title,
// og_description, og_image) to where it came from.
type Result struct {
	Action  string            `json:"action"`
	Status  int               `json:"status"`
	URL     string            `json:"url"`
	Locale  string            `json:"locale"`
	Staging bool              `json:"staging"`
	SEO     *SEO              `json:"seo"`
	Sources map[string]string `json:"-"`
	// TypeKey and RecordID name the record behind the URL, "" when none.
	TypeKey  string `json:"-"`
	RecordID string `json:"-"`
}

// SEO is the head metadata of a resolved page.
type SEO struct {
	Title        string     `json:"title"`
	Description  string     `json:"description"`
	Canonical    string     `json:"canonical"`
	Robots       string     `json:"robots"`
	XRobotsTag   string     `json:"x_robots_tag"`
	OG           OpenGraph  `json:"og"`
	Twitter      TwitterTag `json:"twitter"`
	Verification []MetaTag  `json:"verification"`
}

// OpenGraph holds the og: values.
type OpenGraph struct {
	Title       string `json:"title"`
	Description string `json:"description"`
	Image       string `json:"image"`
	URL         string `json:"url"`
}

// TwitterTag holds the twitter: values.
type TwitterTag struct {
	Card string `json:"card"`
	Site string `json:"site"`
}

// MetaTag is one verification meta tag.
type MetaTag struct {
	Name    string `json:"name"`
	Content string `json:"content"`
}

// Resolver answers the effective SEO of a URL.
type Resolver struct {
	Service *Service
}

// Resolver returns the service's resolver.
func (s *Service) Resolver() *Resolver { return &Resolver{Service: s} }

// SiteURL is the configured public origin without a trailing slash:
// golem15.seo.site_url, else app.url, else "".
func (s *Service) SiteURL() string {
	cfg := s.Config()
	if cfg == nil {
		return ""
	}
	for _, key := range []string{SiteURLConfigKey, "app.url"} {
		if u := strings.TrimRight(strings.TrimSpace(cfg.String(key)), "/"); u != "" {
			return u
		}
	}
	return ""
}

// UnprefixedLocale is the configured unprefixed locale, "" when unset.
func (s *Service) UnprefixedLocale() string {
	cfg := s.Config()
	if cfg == nil {
		return ""
	}
	return strings.TrimSpace(cfg.String(UnprefixedLocaleConfigKey))
}

// parseTarget turns the resolver input into a path and its query. A path
// must start with "/"; an absolute URL must be http or https. Fragments are
// dropped.
func parseTarget(raw string) (string, url.Values, error) {
	raw = strings.TrimSpace(raw)
	if raw == "" || len(raw) > MaxURLBytes {
		return "", nil, ErrBadURL
	}
	u, err := url.Parse(raw)
	if err != nil {
		return "", nil, ErrBadURL
	}
	if u.IsAbs() {
		if u.Scheme != "http" && u.Scheme != "https" {
			return "", nil, ErrBadURL
		}
	} else if !strings.HasPrefix(raw, "/") || strings.HasPrefix(raw, "//") {
		return "", nil, ErrBadURL
	}
	path := u.Path
	if path == "" {
		path = "/"
	}
	return path, u.Query(), nil
}

// lookup finds the published record behind path; key is "" when none.
func (r *Resolver) lookup(ctx context.Context, path string) (key string, match *pact.ContentMatchResult, rec *pact.ContentRecord, err error) {
	reg := r.Service.Registry()
	if reg == nil {
		return "", nil, nil, nil
	}
	key, match, err = reg.MatchPath(ctx, path, r.Service.UnprefixedLocale())
	if err != nil || match == nil {
		return "", nil, nil, err
	}
	rec, err = reg.Describe(ctx, key, match.ID, r.Service.UnprefixedLocale())
	if err != nil {
		return "", nil, nil, err
	}
	if rec == nil || !rec.Published {
		return "", nil, nil, nil
	}
	return key, match, rec, nil
}

// contentLocale returns the record's description in locale, else its first
// locale.
func contentLocale(rec *pact.ContentRecord, locale string) pact.ContentLocale {
	if rec == nil || len(rec.Locales) == 0 {
		return pact.ContentLocale{Locale: locale}
	}
	for _, l := range rec.Locales {
		if l.Locale == locale {
			return l
		}
	}
	return rec.Locales[0]
}

// Resolve answers the effective SEO of rawURL, a path or an absolute URL of
// the site. A path no published record owns answers not_found.
func (r *Resolver) Resolve(ctx context.Context, rawURL string) (Result, error) {
	path, _, err := parseTarget(rawURL)
	if err != nil {
		return Result{}, err
	}
	site := r.Service.SiteURL()
	res := Result{Action: ActionNotFound, Status: 404, URL: site + path, Sources: map[string]string{}}
	key, match, rec, err := r.lookup(ctx, path)
	if err != nil {
		return Result{}, err
	}
	if key == "" {
		return res, nil
	}
	typ, _ := r.Service.Registry().ContentType(key)
	db, err := r.Service.DB(ctx)
	if err != nil {
		return Result{}, err
	}
	entry, _, err := LoadEntry(db, typ.Content.Model, match.ID, match.Locale)
	if err != nil {
		return Result{}, err
	}
	content := contentLocale(rec, match.Locale)
	seo := &SEO{Canonical: site + path, Verification: []MetaTag{}}
	if entry.Title != "" {
		seo.Title, res.Sources["title"] = entry.Title, SourceRecord
	} else {
		seo.Title, res.Sources["title"] = content.Title, SourceContent
	}
	res.Action, res.Status, res.Locale, res.SEO = ActionOK, 200, match.Locale, seo
	res.TypeKey, res.RecordID = key, match.ID
	return res, nil
}