sm-seo-plugin/classes/robots_value.go

package classes

import (
	"errors"
	"fmt"
	"regexp"
	"strconv"
	"strings"
	"time"
)

// ErrRobotsValue is the refusal of a malformed robots value.
var ErrRobotsValue = errors.New("golem15.seo: malformed robots value")

// Robots is a parsed robots value (UI-SPEC C5, D-08): comma-separated
// directives in the fixed order index|noindex, follow|nofollow, noarchive,
// nosnippet, noimageindex, max-snippet:N, max-image-preview:none|standard|
// large, max-video-preview:N, unavailable_after:YYYY-MM-DD. An empty string
// inherits everything; an unset directive inherits too.
type Robots struct {
	Index            string // "index", "noindex" or "" (inherit)
	Follow           string // "follow", "nofollow" or "" (inherit)
	NoArchive        bool
	NoSnippet        bool
	NoImageIndex     bool
	MaxSnippet       *int
	MaxImagePreview  string
	MaxVideoPreview  *int
	UnavailableAfter string
}

var robotsDate = regexp.MustCompile(`^\d{4}-\d{2}-\d{2}$`)

// ParseRobots reads a robots value; spaces around directives and their
// case are ignored. A directive outside the grammar, a repeated directive,
// two contradicting directives or a malformed number or date is refused.
func ParseRobots(value string) (Robots, error) {
	var r Robots
	value = strings.TrimSpace(value)
	if value == "" {
		return r, nil
	}
	seen := map[string]bool{}
	for _, part := range strings.Split(value, ",") {
		d := strings.ToLower(strings.TrimSpace(part))
		if d == "" {
			return Robots{}, fmt.Errorf("%w: empty directive", ErrRobotsValue)
		}
		name, arg, hasArg := strings.Cut(d, ":")
		name, arg = strings.TrimSpace(name), strings.TrimSpace(arg)
		group := name
		switch name {
		case "index", "noindex":
			group = "index"
		case "follow", "nofollow":
			group = "follow"
		}
		if seen[group] {
			return Robots{}, fmt.Errorf("%w: %s given twice", ErrRobotsValue, group)
		}
		seen[group] = true
		switch name {
		case "index", "noindex":
			if hasArg {
				return Robots{}, fmt.Errorf("%w: %s", ErrRobotsValue, d)
			}
			r.Index = name
		case "follow", "nofollow":
			if hasArg {
				return Robots{}, fmt.Errorf("%w: %s", ErrRobotsValue, d)
			}
			r.Follow = name
		case "noarchive", "nosnippet", "noimageindex":
			if hasArg {
				return Robots{}, fmt.Errorf("%w: %s", ErrRobotsValue, d)
			}
			switch name {
			case "noarchive":
				r.NoArchive = true
			case "nosnippet":
				r.NoSnippet = true
			default:
				r.NoImageIndex = true
			}
		case "max-snippet", "max-video-preview":
			n, err := strconv.Atoi(arg)
			if !hasArg || err != nil || n < -1 {
				return Robots{}, fmt.Errorf("%w: %s", ErrRobotsValue, d)
			}
			if name == "max-snippet" {
				r.MaxSnippet = &n
			} else {
				r.MaxVideoPreview = &n
			}
		case "max-image-preview":
			if arg != "none" && arg != "standard" && arg != "large" {
				return Robots{}, fmt.Errorf("%w: %s", ErrRobotsValue, d)
			}
			r.MaxImagePreview = arg
		case "unavailable_after":
			if !robotsDate.MatchString(arg) {
				return Robots{}, fmt.Errorf("%w: %s", ErrRobotsValue, d)
			}
			if _, err := time.Parse("2006-01-02", arg); err != nil {
				return Robots{}, fmt.Errorf("%w: %s", ErrRobotsValue, d)
			}
			r.UnavailableAfter = arg
		default:
			return Robots{}, fmt.Errorf("%w: unknown directive %s", ErrRobotsValue, name)
		}
	}
	return r, nil
}

// directives lists the set directives in the fixed order.
func (r Robots) directives() []string {
	var out []string
	if r.Index != "" {
		out = append(out, r.Index)
	}
	if r.Follow != "" {
		out = append(out, r.Follow)
	}
	if r.NoArchive {
		out = append(out, "noarchive")
	}
	if r.NoSnippet {
		out = append(out, "nosnippet")
	}
	if r.NoImageIndex {
		out = append(out, "noimageindex")
	}
	if r.MaxSnippet != nil {
		out = append(out, "max-snippet:"+strconv.Itoa(*r.MaxSnippet))
	}
	if r.MaxImagePreview != "" {
		out = append(out, "max-image-preview:"+r.MaxImagePreview)
	}
	if r.MaxVideoPreview != nil {
		out = append(out, "max-video-preview:"+strconv.Itoa(*r.MaxVideoPreview))
	}
	if r.UnavailableAfter != "" {
		out = append(out, "unavailable_after:"+r.UnavailableAfter)
	}
	return out
}

// String serialises the stored form: directives in the fixed order joined
// by ",". The empty value inherits everything.
func (r Robots) String() string { return strings.Join(r.directives(), ",") }

// Header serialises the value for a meta robots tag and the X-Robots-Tag
// header: the same directives joined by ", ".
func (r Robots) Header() string { return strings.Join(r.directives(), ", ") }

// Empty reports whether no directive is set.
func (r Robots) Empty() bool { return len(r.directives()) == 0 }

// Inherit fills every directive r leaves unset from lower (the next layer
// of the precedence chain); a directive r sets wins.
func (r Robots) Inherit(lower Robots) Robots {
	out := r
	if out.Index == "" {
		out.Index = lower.Index
	}
	if out.Follow == "" {
		out.Follow = lower.Follow
	}
	out.NoArchive = out.NoArchive || lower.NoArchive
	out.NoSnippet = out.NoSnippet || lower.NoSnippet
	out.NoImageIndex = out.NoImageIndex || lower.NoImageIndex
	if out.MaxSnippet == nil {
		out.MaxSnippet = lower.MaxSnippet
	}
	if out.MaxImagePreview == "" {
		out.MaxImagePreview = lower.MaxImagePreview
	}
	if out.MaxVideoPreview == nil {
		out.MaxVideoPreview = lower.MaxVideoPreview
	}
	if out.UnavailableAfter == "" {
		out.UnavailableAfter = lower.UnavailableAfter
	}
	return out
}

// NormalizeRobots parses and re-serialises a robots value in the fixed
// order; a malformed value is an error.
func NormalizeRobots(value string) (string, error) {
	r, err := ParseRobots(value)
	if err != nil {
		return "", err
	}
	return r.String(), nil
}