Phase 3: PDF/DOCX extraction, chunking, LLM client with mock interface

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

internal/parse/pdf_test.go

@@ -0,0 +1,37 @@
+package parse
+
+import (
+	"strings"
+	"testing"
+)
+
+func TestIsGibberish(t *testing.T) {
+	tests := []struct {
+		name string
+		text string
+		want bool
+	}{
+		{"empty", "", true},
+		{"too short", "hello", true},
+		{"exactly 50 letters", strings.Repeat("a", 50), false},
+		{"49 letters", strings.Repeat("a", 49), true},
+		{"all punctuation", strings.Repeat(".", 100), true},
+		{"1% alpha", strings.Repeat(".", 99) + "a", true},
+		{"2% alpha exactly", strings.Repeat(".", 49) + "a" + strings.Repeat(".", 49) + "a", false},
+		{"normal text", "The quick brown fox jumps over the lazy dog. " + strings.Repeat("word ", 10), false},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			if got := isGibberish(tt.text); got != tt.want {
+				t.Errorf("isGibberish(%q…) = %v, want %v", tt.text[:min(len(tt.text), 20)], got, tt.want)
+			}
+		})
+	}
+}
+
+func min(a, b int) int {
+	if a < b {
+		return a
+	}
+	return b
+}