perf: Replace expensive calls in Table Plugin (#5)

* perf: Replace expensive calls in Table Plugin

* add test

Author: abimaelmartell

perf_big_html_test.go

@@ -7,6 +7,7 @@ import (
 	"testing"
 
 	md "github.com/firecrawl/html-to-markdown"
+	"github.com/firecrawl/html-to-markdown/plugin"
 )
 
 // TestPerfBigHTML_Smoke runs a large HTML file (random-ish content),
@@ -61,3 +62,31 @@ func TestPerfBigList_Smoke(t *testing.T) {
 		t.Fatalf("expected non-empty markdown output")
 	}
 }
+
+// TestPerfBigTable_GitHubFlavored runs a large table with randomized data,
+// converts it once using GitHubFlavored plugin, and asserts we get non-empty output.
+//
+// This is intentionally a "smoke" perf test: no golden output, just "it renders".
+// Run it alone to track timings over time:
+//
+//	go test -run '^TestPerfBigTable_GitHubFlavored$' -count=1
+//
+// The input file is a large table with 5,000 rows and 12 columns.
+func TestPerfBigTable_GitHubFlavored(t *testing.T) {
+	p := filepath.Join("testdata", "Perf", "table.html")
+	b, err := os.ReadFile(p)
+	if err != nil {
+		t.Fatalf("read %s: %v", p, err)
+	}
+	html := string(b)
+
+	conv := md.NewConverter("", true, nil)
+	conv.Use(plugin.GitHubFlavored())
+	out, err := conv.ConvertString(html)
+	if err != nil {
+		t.Fatalf("convert: %v", err)
+	}
+	if strings.TrimSpace(out) == "" {
+		t.Fatalf("expected non-empty markdown output")
+	}
+}

plugin/table.go

@@ -6,6 +6,7 @@ import (
 
 	md "github.com/firecrawl/html-to-markdown"
 	"github.com/PuerkitoBio/goquery"
+	"golang.org/x/net/html"
 )
 
 // TableCompat is a compatibility plugin for environments where
@@ -145,17 +146,37 @@ func isHeadingRow(s *goquery.Selection) bool {
 
 	isTableOrBody := parent.Is("table") || isFirstTbody(parent)
 
+	// Check if every cell is a TH - break early if we find a non-TH
 	everyTH := true
-	s.Children().Each(func(i int, s *goquery.Selection) {
-		if goquery.NodeName(s) != "th" {
+	children := s.Children()
+	for i := 0; i < children.Length(); i++ {
+		if goquery.NodeName(children.Eq(i)) != "th" {
 			everyTH = false
+			break
 		}
-	})
+	}
 
-	if parent.Children().First().IsSelection(s) && isTableOrBody && everyTH {
-		return true
+	// Optimize: Check if this is the first child by comparing node pointers directly
+	// instead of creating a new Selection with parent.Children().First()
+	if !everyTH || !isTableOrBody {
+		return false
 	}
 
+	// Check if s is the first element child by comparing nodes directly
+	if len(s.Nodes) == 0 || len(parent.Nodes) == 0 {
+		return false
+	}
+	
+	parentNode := parent.Nodes[0]
+	sNode := s.Nodes[0]
+	
+	// Find the first element child (skip text nodes)
+	for child := parentNode.FirstChild; child != nil; child = child.NextSibling {
+		if child.Type == html.ElementNode {
+			return child == sNode
+		}
+	}
+	
 	return false
 }
 func isFirstTbody(s *goquery.Selection) bool {
@@ -175,15 +196,26 @@ func getCellContent(content string, s *goquery.Selection) string {
 		// nested tables not found
 		content = newLineRe.ReplaceAllString(content, "<br>")
 	}
-	index := -1
-	for i, node := range s.Parent().Children().Nodes {
-		if s.IsNodes(node) {
-			index = i
-			break
+	
+	// Optimize: Check if this is the first element child by comparing node pointers directly
+	// instead of linear search through all children
+	parent := s.Parent()
+	isFirst := false
+	if len(s.Nodes) > 0 && len(parent.Nodes) > 0 {
+		parentNode := parent.Nodes[0]
+		sNode := s.Nodes[0]
+		
+		// Find the first element child (skip text nodes)
+		for child := parentNode.FirstChild; child != nil; child = child.NextSibling {
+			if child.Type == html.ElementNode {
+				isFirst = (child == sNode)
+				break
+			}
 		}
 	}
+	
 	prefix := " "
-	if index == 0 {
+	if isFirst {
 		prefix = "| "
 	}
 	return prefix + content + " |"