outward-scrapper/internal/scraper/parse.go

package scraper

import (
	"encoding/json"
	"math/rand"
	"net/url"
	"path"
	"sort"
	"strconv"
	"strings"

	"github.com/PuerkitoBio/goquery"
	"github.com/gocolly/colly/v2"

	"scrappr/internal/model"
)

func (s *Scraper) shouldVisit(raw string) bool {
	if raw == "" {
		return false
	}

	parsed, err := url.Parse(raw)
	if err != nil {
		return false
	}

	if parsed.Host != s.cfg.AllowedDomain {
		return false
	}

	if !strings.HasPrefix(parsed.Path, "/wiki/") {
		return false
	}

	if s.cfg.IgnoredExact[parsed.Path] {
		return false
	}

	for _, prefix := range s.cfg.IgnoredPrefixes {
		if strings.HasPrefix(parsed.Path, prefix) {
			return false
		}
	}

	return true
}

func (s *Scraper) applyBrowserHeaders(r *colly.Request) {
	r.Headers.Set("User-Agent", s.randomUserAgent())
	r.Headers.Set("Accept", "text/html,application/xhtml+xml,application/xml;q=0.9,image/avif,image/webp,image/apng,*/*;q=0.8")
	r.Headers.Set("Accept-Language", "en-US,en;q=0.9")
	r.Headers.Set("Cache-Control", "max-age=0")
	r.Headers.Set("DNT", "1")
	r.Headers.Set("Sec-Fetch-Dest", "document")
	r.Headers.Set("Sec-Fetch-Mode", "navigate")
	r.Headers.Set("Sec-Fetch-Site", "same-origin")
	r.Headers.Set("Sec-Fetch-User", "?1")
	r.Headers.Set("Upgrade-Insecure-Requests", "1")

	if r.Headers.Get("Referer") == "" {
		r.Headers.Set("Referer", s.cfg.BrowserReferrer)
	}
}

func (s *Scraper) randomUserAgent() string {
	return s.cfg.BrowserAgents[rand.Intn(len(s.cfg.BrowserAgents))]
}

func (s *Scraper) shouldQueueFromPage(fromPath, toURL string) bool {
	parsed, err := url.Parse(toURL)
	if err != nil {
		return false
	}

	toPath := parsed.Path

	if s.cfg.ItemListPathRe.MatchString(fromPath) {
		return strings.HasPrefix(toPath, "/wiki/") &&
			!s.cfg.ItemListPathRe.MatchString(toPath) &&
			!s.cfg.CraftingPathRe.MatchString(toPath)
	}

	if s.cfg.CraftingPathRe.MatchString(fromPath) {
		return strings.HasPrefix(toPath, "/wiki/") &&
			!s.cfg.CraftingPathRe.MatchString(toPath)
	}

	return false
}

func (s *Scraper) shouldQueueTableLink(e *colly.HTMLElement) bool {
	label := s.clean(e.Text)
	if label == "" {
		return false
	}

	cell := e.DOM.Closest("td, th")
	if cell.Length() == 0 {
		return false
	}

	columnIndex := 0
	for prev := cell.Prev(); prev.Length() > 0; prev = prev.Prev() {
		switch goquery.NodeName(prev) {
		case "td", "th":
			columnIndex++
		}
	}

	return columnIndex <= 1
}

func (s *Scraper) pageKindForPath(pagePath string) string {
	switch {
	case s.cfg.ItemListPathRe.MatchString(pagePath):
		return "item-list"
	case s.cfg.CraftingPathRe.MatchString(pagePath):
		return "crafting"
	default:
		return "article"
	}
}

func (s *Scraper) isLikelyItemPage(pagePath, title string, lcCats []string, doc *goquery.Document) bool {
	if s.cfg.ItemListPathRe.MatchString(pagePath) || s.cfg.CraftingPathRe.MatchString(pagePath) {
		return false
	}
	if strings.Contains(strings.ToLower(title), "/") {
		return false
	}

	for _, category := range lcCats {
		if strings.Contains(category, "items") || strings.Contains(category, "food") ||
			strings.Contains(category, "consumables") || strings.Contains(category, "ingredients") ||
			strings.Contains(category, "equipment") || strings.Contains(category, "weapons") ||
			strings.Contains(category, "deployables") {
			return true
		}
	}

	if !s.hasItemInfobox(doc) {
		return false
	}

	infoboxText := strings.ToLower(s.clean(s.infoboxText(doc)))
	descriptionText := strings.ToLower(s.parseDescription(doc))
	return strings.Contains(infoboxText, "item details") ||
		strings.Contains(infoboxText, "consumable details") ||
		strings.Contains(infoboxText, "equipment details") ||
		strings.Contains(infoboxText, "weapon") ||
		strings.Contains(descriptionText, " is an item in outward") ||
		strings.Contains(descriptionText, " is a weapon in outward") ||
		strings.Contains(descriptionText, " is a consumable in outward") ||
		strings.Contains(descriptionText, " is an ingredient in outward") ||
		strings.Contains(descriptionText, " is a shield in outward")
}

func (s *Scraper) isEffectPage(pagePath, title string, lcCats []string) bool {
	if strings.Contains(strings.ToLower(title), "effect") {
		return true
	}

	for _, category := range lcCats {
		if strings.Contains(category, "effects") {
			return true
		}
	}

	return strings.Contains(strings.ToLower(pagePath), "/effect")
}

func (s *Scraper) parseItemPage(doc *goquery.Document, title, pageURL string, categories []string) model.Item {
	item := model.Item{
		Name:       title,
		URL:        pageURL,
		Categories: categories,
		Infobox:    s.parseInfobox(doc),
		ImageURL:   s.parseImageURL(doc),
	}

	item.Description = s.parseDescription(doc)
	item.Effects, item.EffectLinks = s.parseEffectsSection(doc)
	item.Recipes = s.parseRecipesFromPage(doc, title)
	item.Tables = s.parseContentTables(doc)

	return item
}

func (s *Scraper) parseEffectPage(doc *goquery.Document, title, pageURL string, categories []string) model.Effect {
	return model.Effect{
		Name:        title,
		URL:         pageURL,
		Categories:  categories,
		Infobox:     s.parseInfobox(doc),
		Description: s.parseDescription(doc),
	}
}

func (s *Scraper) hasItemInfobox(doc *goquery.Document) bool {
	return doc.Find("aside.portable-infobox").Length() > 0 || doc.Find("table.infoboxtable").Length() > 0
}

func (s *Scraper) infoboxText(doc *goquery.Document) string {
	return s.clean(doc.Find("aside.portable-infobox, table.infoboxtable").First().Text())
}

func (s *Scraper) parseInfobox(doc *goquery.Document) map[string]string {
	out := map[string]string{}

	doc.Find("aside.portable-infobox .pi-item").Each(func(_ int, section *goquery.Selection) {
		label := s.clean(section.Find(".pi-data-label").First().Text())
		value := s.clean(section.Find(".pi-data-value").First().Text())

		if label == "" {
			label = s.clean(section.Find("h3").First().Text())
		}
		if value == "" {
			value = s.clean(section.Find("div").Last().Text())
		}

		if label != "" && value != "" {
			out[label] = value
		}
	})

	doc.Find("table.infoboxtable tr").Each(func(_ int, row *goquery.Selection) {
		cells := row.Find("th, td")
		if cells.Length() < 2 {
			return
		}

		label := s.clean(cells.First().Text())
		if label == "" {
			return
		}

		var values []string
		cells.Slice(1, cells.Length()).Each(func(_ int, cell *goquery.Selection) {
			text := s.clean(cell.Text())
			if text != "" {
				values = append(values, text)
			}
		})

		if len(values) == 0 {
			return
		}

		out[label] = strings.Join(values, " | ")
	})

	return out
}

func (s *Scraper) parseCategories(doc *goquery.Document) []string {
	var categories []string
	seen := map[string]bool{}

	doc.Find(".page-header__categories a, .category-page__member-link, .wds-tabs__tab-label a").Each(func(_ int, node *goquery.Selection) {
		text := s.clean(node.Text())
		if text != "" && !seen[text] {
			seen[text] = true
			categories = append(categories, text)
		}
	})

	if len(categories) == 0 {
		doc.Find("script").EachWithBreak(func(_ int, node *goquery.Selection) bool {
			text := node.Text()
			index := strings.Index(text, "wgCategories")
			if index < 0 {
				return true
			}

			arrayStart := strings.Index(text[index:], "[")
			arrayEnd := strings.Index(text[index:], "]")
			if arrayStart < 0 || arrayEnd < 0 || arrayEnd <= arrayStart {
				return true
			}

			raw := text[index+arrayStart : index+arrayEnd+1]
			var parsed []string
			if err := json.Unmarshal([]byte(raw), &parsed); err != nil {
				return true
			}

			for _, category := range parsed {
				category = s.clean(category)
				if category != "" && !seen[category] {
					seen[category] = true
					categories = append(categories, category)
				}
			}

			return len(categories) == 0
		})
	}

	if len(categories) == 0 {
		headerText := s.clean(doc.Find("body").Text())
		if idx := strings.Index(headerText, "in:"); idx >= 0 {
			line := strings.TrimPrefix(headerText[idx:], "in:")
			for _, part := range strings.Split(line, ",") {
				part = s.clean(part)
				if part != "" && !seen[part] {
					seen[part] = true
					categories = append(categories, part)
				}
			}
		}
	}

	return categories
}

func (s *Scraper) parseDescription(doc *goquery.Document) string {
	var description string

	doc.Find(".mw-parser-output > p").EachWithBreak(func(_ int, paragraph *goquery.Selection) bool {
		text := s.clean(paragraph.Text())
		if text == "" {
			return true
		}

		lower := strings.ToLower(text)
		if strings.Contains(lower, "is a") || strings.Contains(lower, "is an") || len(text) > 30 {
			description = text
			return false
		}

		return true
	})

	return description
}

func (s *Scraper) parseEffectsSection(doc *goquery.Document) ([]string, []string) {
	var effects []string
	var effectLinks []string

	seenText := map[string]bool{}
	seenLink := map[string]bool{}
	section := s.findSection(doc, "Effects")
	if section.Length() == 0 {
		return effects, effectLinks
	}

	s.walkSectionUntilNextHeading(section, func(node *goquery.Selection) {
		node.Find("li").Each(func(_ int, item *goquery.Selection) {
			text := s.clean(item.Text())
			if text != "" && !seenText[text] {
				seenText[text] = true
				effects = append(effects, text)
			}

			item.Find("a[href^='/wiki/']").Each(func(_ int, anchor *goquery.Selection) {
				href, _ := anchor.Attr("href")
				label := s.clean(anchor.Text())
				if href != "" && label != "" && s.looksLikeEffectLink(href, label) && !seenLink[href] {
					seenLink[href] = true
					effectLinks = append(effectLinks, href)
				}
			})
		})
	})

	return effects, effectLinks
}

func (s *Scraper) parseRecipesFromPage(doc *goquery.Document, pageTitle string) []model.Recipe {
	var recipes []model.Recipe

	doc.Find("table").Each(func(_ int, table *goquery.Selection) {
		if table.HasClass("infoboxtable") {
			return
		}

		headerText := strings.ToLower(s.clean(table.Find("tr").First().Text()))
		if !(strings.Contains(headerText, "result") && strings.Contains(headerText, "ingredient")) {
			return
		}

		table.Find("tr").Each(func(i int, row *goquery.Selection) {
			if i == 0 {
				return
			}

			cells := row.Find("th, td")
			if cells.Length() < 2 {
				return
			}

			resultCell := cells.Eq(0)
			ingredientCell := cells.Eq(1)
			stationCell := cells.Eq(2)

			resultText := s.clean(resultCell.Text())
			if resultText == "" {
				return
			}

			resultCount := ""
			if matches := s.cfg.AmountPrefixRe.FindStringSubmatch(resultText); len(matches) > 1 {
				resultCount = matches[1]
				resultText = s.clean(strings.TrimPrefix(resultText, matches[0]))
			}

			var ingredients []string
			ingredientCell.Find("li").Each(func(_ int, ingredient *goquery.Selection) {
				text := s.clean(ingredient.Text())
				if text != "" {
					ingredients = append(ingredients, text)
				}
			})

			if len(ingredients) == 0 {
				for _, line := range s.splitLines(ingredientCell.Text()) {
					if line != "" {
						ingredients = append(ingredients, line)
					}
				}
			}

			recipes = append(recipes, model.Recipe{
				Result:      resultText,
				ResultCount: resultCount,
				Ingredients: ingredients,
				Station:     s.clean(stationCell.Text()),
				SourcePage:  pageTitle,
			})
		})
	})

	return recipes
}

func (s *Scraper) parseImageURL(doc *goquery.Document) string {
	image := doc.Find("aside.portable-infobox img, table.infoboxtable img").First()
	if image.Length() == 0 {
		return ""
	}

	for _, attr := range []string{"src", "data-src"} {
		if raw, ok := image.Attr(attr); ok {
			return s.normalizeImageURL(raw)
		}
	}

	return ""
}

func (s *Scraper) normalizeImageURL(raw string) string {
	raw = strings.TrimSpace(raw)
	switch {
	case raw == "":
		return ""
	case strings.HasPrefix(raw, "//"):
		return "https:" + raw
	default:
		return raw
	}
}

func (s *Scraper) parseContentTables(doc *goquery.Document) []model.Table {
	var tables []model.Table

	doc.Find(".mw-parser-output table").Each(func(_ int, table *goquery.Selection) {
		if table.HasClass("infoboxtable") || table.Find("table").Length() > 0 {
			return
		}

		rows := table.Find("tr")
		if rows.Length() < 2 {
			return
		}

		headerCells := rows.First().Find("th, td")
		if headerCells.Length() == 0 {
			return
		}

		headers := make([]string, 0, headerCells.Length())
		headerKeys := make([]string, 0, headerCells.Length())
		headerCells.Each(func(i int, cell *goquery.Selection) {
			header := s.clean(cell.Text())
			if header == "" {
				header = "Column " + strconv.Itoa(i+1)
			}
			headers = append(headers, header)
			headerKeys = append(headerKeys, s.tableHeaderKey(header, i))
		})

		parsedTable := model.Table{
			Title:   s.tableTitle(table),
			Headers: headers,
		}

		rows.Slice(1, rows.Length()).Each(func(_ int, row *goquery.Selection) {
			cells := row.Find("th, td")
			if cells.Length() == 0 {
				return
			}

			rawRow := make([]string, 0, cells.Length())
			rowData := map[string]string{}

			cells.Each(func(i int, cell *goquery.Selection) {
				text := s.clean(cell.Text())
				rawRow = append(rawRow, text)
				key := s.tableColumnKey(headerKeys, headers, i)
				if text != "" {
					rowData[key] = text
				}
			})

			if s.rowIsEmpty(rawRow) {
				return
			}

			parsedTable.RawRows = append(parsedTable.RawRows, rawRow)
			if len(rowData) > 0 {
				parsedTable.Rows = append(parsedTable.Rows, rowData)
			}
		})

		if len(parsedTable.RawRows) == 0 {
			return
		}

		tables = append(tables, parsedTable)
	})

	return tables
}

func (s *Scraper) tableTitle(table *goquery.Selection) string {
	var parts []string

	for prev := table.Prev(); prev.Length() > 0; prev = prev.Prev() {
		switch goquery.NodeName(prev) {
		case "h4", "h3":
			text := strings.TrimSpace(strings.TrimSuffix(s.clean(prev.Text()), "[]"))
			if text != "" {
				parts = append([]string{text}, parts...)
			}
		case "h2":
			text := strings.TrimSpace(strings.TrimSuffix(s.clean(prev.Text()), "[]"))
			if text != "" {
				parts = append([]string{text}, parts...)
			}
			return strings.Join(parts, " / ")
		}
	}

	return strings.Join(parts, " / ")
}

func (s *Scraper) tableHeaderKey(header string, index int) string {
	header = strings.TrimSpace(header)
	if header == "" {
		return s.fallbackColumnKey(index)
	}

	header = strings.ToLower(header)
	header = strings.ReplaceAll(header, " ", "_")
	header = strings.ReplaceAll(header, "/", "_")
	header = strings.ReplaceAll(header, "-", "_")
	header = s.cfg.WhitespaceRe.ReplaceAllString(header, "_")
	header = strings.Trim(header, "_")
	if header == "" {
		return s.fallbackColumnKey(index)
	}

	return header
}

func (s *Scraper) tableColumnKey(headerKeys, headers []string, index int) string {
	if index < len(headerKeys) && headerKeys[index] != "" {
		return headerKeys[index]
	}
	if index < len(headers) && headers[index] != "" {
		return headers[index]
	}
	return s.fallbackColumnKey(index)
}

func (s *Scraper) fallbackColumnKey(index int) string {
	return "column_" + strconv.Itoa(index+1)
}

func (s *Scraper) rowIsEmpty(row []string) bool {
	for _, value := range row {
		if value != "" {
			return false
		}
	}
	return true
}

func (s *Scraper) findSection(doc *goquery.Document, title string) *goquery.Selection {
	var found *goquery.Selection

	doc.Find(".mw-parser-output h2, .mw-parser-output h3, .mw-parser-output h4").EachWithBreak(func(_ int, section *goquery.Selection) bool {
		text := strings.TrimSpace(strings.TrimSuffix(s.clean(section.Text()), "[]"))
		if strings.EqualFold(text, title) {
			found = section
			return false
		}
		return true
	})

	if found == nil {
		return &goquery.Selection{}
	}

	return found
}

func (s *Scraper) walkSectionUntilNextHeading(start *goquery.Selection, fn func(*goquery.Selection)) {
	for section := start.Next(); section.Length() > 0; section = section.Next() {
		if goquery.NodeName(section) == "h2" || goquery.NodeName(section) == "h3" || goquery.NodeName(section) == "h4" {
			break
		}
		fn(section)
	}
}

func (s *Scraper) looksLikeEffectLink(href, label string) bool {
	lowerHref := strings.ToLower(href)
	lowerLabel := strings.ToLower(label)

	if strings.Contains(lowerHref, "/wiki/") && (strings.Contains(lowerHref, "effect") || strings.Contains(lowerLabel, "effect")) {
		return true
	}

	knownHints := []string{
		"bleeding", "burning", "poisoned", "possessed", "discipline", "rage",
		"sapped", "scorched", "curse", "doomed", "chill", "warm", "cool",
		"barrier", "protection", "imbue", "energized", "shimmer",
	}

	for _, hint := range knownHints {
		if strings.Contains(lowerLabel, hint) {
			return true
		}
	}

	return false
}

func (s *Scraper) absoluteWikiURL(href string) string {
	if href == "" {
		return ""
	}

	if strings.HasPrefix(href, "http://") || strings.HasPrefix(href, "https://") {
		return href
	}

	if strings.HasPrefix(href, "/wiki/") {
		return s.cfg.BaseURL + href
	}

	return ""
}

func (s *Scraper) flattenItems() []model.Item {
	out := make([]model.Item, 0, len(s.items))
	for _, item := range s.items {
		out = append(out, *item)
	}

	sort.Slice(out, func(i, j int) bool {
		return strings.ToLower(out[i].Name) < strings.ToLower(out[j].Name)
	})

	return out
}

func (s *Scraper) flattenEffects() []model.Effect {
	out := make([]model.Effect, 0, len(s.effects))
	for _, effect := range s.effects {
		out = append(out, *effect)
	}

	sort.Slice(out, func(i, j int) bool {
		return strings.ToLower(out[i].Name) < strings.ToLower(out[j].Name)
	})

	return out
}

func (s *Scraper) clean(value string) string {
	value = strings.ReplaceAll(value, "\u00a0", " ")
	value = strings.ReplaceAll(value, "\n", " ")
	value = strings.ReplaceAll(value, "\t", " ")
	value = strings.TrimSpace(value)
	return s.cfg.WhitespaceRe.ReplaceAllString(value, " ")
}

func (s *Scraper) splitLines(value string) []string {
	value = strings.ReplaceAll(value, "\u00a0", " ")
	raw := strings.Split(value, "\n")
	out := make([]string, 0, len(raw))

	for _, line := range raw {
		line = s.clean(line)
		if line != "" {
			out = append(out, line)
		}
	}

	return out
}

func (s *Scraper) lowerSlice(in []string) []string {
	out := make([]string, 0, len(in))
	for _, value := range in {
		out = append(out, strings.ToLower(value))
	}
	return out
}

func (s *Scraper) debugURLName(raw string) string {
	parsed, err := url.Parse(raw)
	if err != nil {
		return raw
	}
	return path.Base(parsed.Path)
}