sm-seo-plugin/classes/pagination.go

package classes

import (
	"context"
	"net/url"
	"regexp"
	"strconv"
	"strings"

	"git.golem15.com/golem15/sm-seo-plugin/models"
	"gorm.io/gorm"
)

// PageParam is the query parameter of the query form (?page=N).
const PageParam = "page"

// The placeholders of the page suffixes (D-41). %page% is filled by the
// resolver; %pages% is left for the layer, which knows the page count
// (P33-24).
const (
	PlaceholderPage  = "%page%"
	PlaceholderPages = "%pages%"
)

// DefaultPageSegments are the path segments of the path form when the
// settings name none for a locale; other locales use "page".
var DefaultPageSegments = map[string]string{"pl": "strona", "en": "page", "de": "seite"}

// segmentPattern is a valid path segment (lowercase letters, digits,
// hyphens).
var segmentPattern = regexp.MustCompile(`^[a-z0-9-]+$`)

// ValidPageSegment reports whether s may be a page segment.
func ValidPageSegment(s string) bool { return segmentPattern.MatchString(s) }

// Pagination is the resolve answer's pagination block (P33-21), present
// only when the request carries a page: the site's form, the parameter of
// the query form, the request locale's segment of the path form, the page
// number (0 when the value is not a number) and whether the value is
// invalid (the layer answers 404).
type Pagination struct {
	Form    string `json:"form"`
	Param   string `json:"param"`
	Segment string `json:"segment"`
	Page    int    `json:"page"`
	Invalid bool   `json:"invalid"`
}

// PageConfig is a site's pagination form (D-40): query or path, and the
// path segment per locale.
type PageConfig struct {
	Form     string
	Segments map[string]string
}

// Segment is the path segment of locale.
func (c PageConfig) Segment(locale string) string {
	if s := c.Segments[locale]; s != "" {
		return s
	}
	if s := DefaultPageSegments[locale]; s != "" {
		return s
	}
	return "page"
}

// PageConfigOf reads the pagination form of the settings: the segment of
// each site locale is its translated setting when valid, else the default.
func (s *Service) PageConfigOf(ctx context.Context, db *gorm.DB, settings models.Settings) PageConfig {
	cfg := PageConfig{Form: models.PageFormQuery, Segments: map[string]string{}}
	if settings.PageForm == models.PageFormPath {
		cfg.Form = models.PageFormPath
	}
	for _, locale := range s.Locales(ctx, db) {
		seg := strings.TrimSpace(s.translatedSetting(ctx, db, &settings, "page_segment", settings.PageSegment, locale))
		if !ValidPageSegment(seg) {
			seg = DefaultPageSegments[locale]
			if seg == "" {
				seg = "page"
			}
		}
		cfg.Segments[locale] = seg
	}
	return cfg
}

// pageRequest is what a URL says about its page.
type pageRequest struct {
	// carried is true when the URL carries a page value.
	carried bool
	invalid bool
	page    int
	// form and segment are the form (and path segment) the URL used.
	form    string
	segment string
	// base is the path without the path-form suffix; rest the query
	// without the page parameter.
	base string
	rest url.Values
	// exact is true when the number is written canonically (no leading
	// zero, one value).
	exact bool
}

// pageNumber reads a page value: digits only, at least 1.
func pageNumber(v string) (int, bool) {
	if v == "" || len(v) > 9 {
		return 0, false
	}
	for _, r := range v {
		if r < '0' || r > '9' {
			return 0, false
		}
	}
	n, err := strconv.Atoi(v)
	if err != nil || n < 1 {
		return 0, false
	}
	return n, true
}

// parsePage reads the page of a normalised path and query in either form:
// ?page=N, or a trailing /<segment>/N with any locale's segment. A path
// suffix whose value is not a number counts only in the path form.
func parsePage(path string, query url.Values, cfg PageConfig) pageRequest {
	pr := pageRequest{base: path, rest: url.Values{}, exact: true}
	for k, vs := range query {
		if k != PageParam {
			pr.rest[k] = vs
		}
	}
	if vs, ok := query[PageParam]; ok {
		pr.carried, pr.form = true, models.PageFormQuery
		v := ""
		if len(vs) > 0 {
			v = vs[0]
		}
		n, ok := pageNumber(v)
		if !ok {
			pr.invalid = true
			return pr
		}
		pr.page, pr.exact = n, len(vs) == 1 && v == strconv.Itoa(n)
	}
	trimmed := strings.TrimSuffix(path, "/")
	i := strings.LastIndex(trimmed, "/")
	if i <= 0 {
		return pr
	}
	head, value := trimmed[:i], trimmed[i+1:]
	j := strings.LastIndex(head, "/")
	if j < 0 {
		return pr
	}
	prefix, segment := head[:j], head[j+1:]
	known := false
	for _, s := range cfg.Segments {
		if s == segment {
			known = true
		}
	}
	if !known {
		return pr
	}
	n, ok := pageNumber(value)
	if !ok {
		if cfg.Form == models.PageFormPath {
			pr.carried, pr.invalid, pr.form, pr.segment = true, true, models.PageFormPath, segment
		}
		return pr
	}
	if prefix == "" {
		prefix = "/"
	}
	exact := value == strconv.Itoa(n) && !pr.carried
	pr.carried, pr.invalid, pr.form, pr.segment = true, false, models.PageFormPath, segment
	pr.page, pr.base, pr.exact = n, prefix, exact
	return pr
}

// pagedPath is base with the path-form suffix of page.
func pagedPath(base, segment string, page int) string {
	b := strings.TrimSuffix(base, "/")
	return b + "/" + segment + "/" + strconv.Itoa(page)
}

// pageLocation is the address of page of base in the site's form, with
// the other query parameters; page 1 (or none) is the address without a
// number (D-41, D-42). It is escaped for a Location.
func pageLocation(base string, rest url.Values, page int, cfg PageConfig, locale string, opts NormalizeOptions) string {
	path := base
	query := url.Values{}
	for k, vs := range rest {
		query[k] = vs
	}
	if page >= 2 {
		if cfg.Form == models.PageFormPath {
			path = pagedPath(base, cfg.Segment(locale), page)
			if n, _ := Normalize(&url.URL{Path: path}, opts); n != nil {
				path = n.Path
			}
		} else {
			query.Set(PageParam, strconv.Itoa(page))
		}
	}
	return joinQuery(escapePath(path), encodeQuery(query))
}

// pageAt is the page of a resolved URL (page 2 and later): its form, the
// segment of the path form and the number.
type pageAt struct {
	form    string
	segment string
	page    int
}

// fillPage fills %page% of a suffix and keeps %pages%.
func fillPage(suffix string, page int) string {
	return strings.ReplaceAll(suffix, PlaceholderPage, strconv.Itoa(page))
}

// pageSuffix is the title or description suffix in locale: the translated
// setting, else the default copy.
func (s *Service) pageSuffix(ctx context.Context, db *gorm.DB, settings models.Settings, locale string, title bool) string {
	field, own, key := "page_description_suffix", settings.PageDescriptionSuffix, "golem15.seo::lang.settings.page_description_suffix_default"
	if title {
		field, own, key = "page_title_suffix", settings.PageTitleSuffix, "golem15.seo::lang.settings.page_title_suffix_default"
	}
	v := strings.TrimSpace(s.translatedSetting(ctx, db, &settings, field, own, locale))
	if v == "" {
		v = s.PhraseIn(locale, key, nil)
	}
	return v
}

// localeOfPath guesses the locale of a path without a record: the site
// locale whose /<code> prefix it carries, else the default locale.
func (s *Service) localeOfPath(ctx context.Context, db *gorm.DB, path string) string {
	lower := strings.ToLower(path)
	for _, l := range s.Locales(ctx, db) {
		prefix := "/" + strings.ToLower(l)
		if lower == prefix || strings.HasPrefix(lower, prefix+"/") {
			return l
		}
	}
	return s.DefaultLocale(ctx, db)
}

// localeOfTag is the site locale whose language tag is tag, else tag.
func (s *Service) localeOfTag(ctx context.Context, db *gorm.DB, tag string) string {
	for _, l := range s.Locales(ctx, db) {
		if LangTag(l) == tag {
			return l
		}
	}
	return tag
}