feat: initial commit

internal/scraper/parse.go

@@ -0,0 +1,124 @@
+package scraper
+
+import (
+	"fmt"
+	"regexp"
+	"strconv"
+	"strings"
+
+	"github.com/PuerkitoBio/goquery"
+
+	"git.mkz.me/mycroft/root-me-api/internal/scoreboard"
+)
+
+// The rankings ajax fragment is a single <table> whose rows each have 6 <td>:
+//
+//	[0] "# <rank>"
+//	[1] avatar <img alt="<username>">
+//	[2] <a href="/<user>?lang=en" title="profil of <user>"><user></a>
+//	[3] country flag  <img src=".../pays/<cc>.svg">
+//	[4] grade icon    <img src=".../rang/<grade>.svg" alt="<grade>">
+//	[5] score         <a href="<user>?inc=score" title="<big>"><visible-score></a>
+//
+// Selectors are intentionally lenient (match on structure/attributes rather
+// than volatile CSS classes) so minor template changes don't break parsing.
+
+var (
+	rankRe    = regexp.MustCompile(`#?\s*(\d+)`)
+	flagRe    = regexp.MustCompile(`/pays/([a-zA-Z]{2})\.svg`)
+	gradeRe   = regexp.MustCompile(`/rang/([a-zA-Z0-9_-]+)\.svg`)
+	nonDigits = regexp.MustCompile(`\D+`)
+)
+
+// parseRankings parses one rankings fragment into entries.
+func parseRankings(html string) ([]scoreboard.Entry, error) {
+	doc, err := goquery.NewDocumentFromReader(strings.NewReader(html))
+	if err != nil {
+		return nil, fmt.Errorf("scraper: parse html: %w", err)
+	}
+
+	var entries []scoreboard.Entry
+	var rowErr error
+
+	doc.Find("table tr").Each(func(_ int, tr *goquery.Selection) {
+		tds := tr.Find("td")
+		if tds.Length() < 6 {
+			return // header row (uses <td>Position</td>... but lacks the score link) or layout row
+		}
+		e, ok, err := parseRow(tds)
+		if err != nil {
+			rowErr = err
+			return
+		}
+		if ok {
+			entries = append(entries, e)
+		}
+	})
+	if rowErr != nil {
+		return nil, rowErr
+	}
+	if len(entries) == 0 {
+		return nil, fmt.Errorf("scraper: no rows parsed (fragment layout may have changed)")
+	}
+	return entries, nil
+}
+
+func parseRow(tds *goquery.Selection) (scoreboard.Entry, bool, error) {
+	var e scoreboard.Entry
+
+	// [0] rank — header row's first cell is "Position" (no digits) → skip.
+	rankText := strings.TrimSpace(tds.Eq(0).Text())
+	m := rankRe.FindStringSubmatch(rankText)
+	if m == nil {
+		return e, false, nil // not a data row
+	}
+	rank, err := strconv.Atoi(m[1])
+	if err != nil {
+		return e, false, fmt.Errorf("scraper: bad rank %q: %w", rankText, err)
+	}
+	e.Rank = rank
+
+	// [2] username + profile path.
+	link := tds.Eq(2).Find("a").First()
+	e.Username = strings.TrimSpace(link.Text())
+	if e.Username == "" {
+		// Fall back to avatar alt in [1].
+		e.Username = strings.TrimSpace(tds.Eq(1).Find("img").AttrOr("alt", ""))
+	}
+	if href, ok := link.Attr("href"); ok {
+		if p := strings.SplitN(href, "?", 2)[0]; p != "" {
+			e.ProfilePath = p
+		}
+	}
+	if e.Username == "" {
+		return e, false, fmt.Errorf("scraper: row rank %d has no username", rank)
+	}
+
+	// [3] country flag.
+	if src, ok := tds.Eq(3).Find("img").Attr("src"); ok {
+		if fm := flagRe.FindStringSubmatch(src); fm != nil {
+			e.Country = strings.ToLower(fm[1])
+		}
+	}
+
+	// [4] grade — prefer the icon's alt, else derive from the svg filename.
+	grade := tds.Eq(4).Find("img").AttrOr("alt", "")
+	if grade == "" {
+		if src, ok := tds.Eq(4).Find("img").Attr("src"); ok {
+			if gm := gradeRe.FindStringSubmatch(src); gm != nil {
+				grade = gm[1]
+			}
+		}
+	}
+	e.Grade = strings.TrimSpace(grade)
+
+	// [5] score — visible text of the link.
+	scoreText := nonDigits.ReplaceAllString(tds.Eq(5).Text(), "")
+	if scoreText != "" {
+		if e.Score, err = strconv.Atoi(scoreText); err != nil {
+			return e, false, fmt.Errorf("scraper: bad score %q for rank %d: %w", scoreText, rank, err)
+		}
+	}
+
+	return e, true, nil
+}