Code sync, security fixes

internal/search/edge_test.go

@@ -0,0 +1,78 @@
+package search
+
+import (
+	"testing"
+	"time"
+	"unicode/utf8"
+
+	"github.com/andydixon/relata/internal/domain"
+)
+
+// Probe the pure helpers with adversarial input.
+func TestExcerptHelpersOnAdversarialInput(t *testing.T) {
+	inputs := []string{"", " ", "\x00", "é", "ééééééééééé", "a\nb\tc", "🌍🌍🌍"}
+	for _, in := range inputs {
+		for _, n := range []int{0, 1, 2, 5, 1000} {
+			got := truncateWords(in, n)
+			for _, r := range got {
+				if r == '�' {
+					t.Fatalf("truncateWords(%q,%d) produced a replacement char: %q", in, n, got)
+				}
+			}
+		}
+	}
+}
+
+func TestFindTermWindowOnAdversarialInput(t *testing.T) {
+	texts := []string{"", "short", "ééééé match ééééé", "🌍 match 🌍"}
+	for _, txt := range texts {
+		for _, terms := range [][]string{nil, {}, {""}, {"match"}, {"zzz"}, {"é"}} {
+			got := findTermWindow(txt, terms, 20)
+			for _, r := range got {
+				if r == '�' {
+					t.Fatalf("findTermWindow(%q,%v) produced a replacement char: %q", txt, terms, got)
+				}
+			}
+		}
+	}
+}
+
+func TestMatchesFilterHandlesZeroValues(t *testing.T) {
+	it := &domain.Item{}
+	if !matchesFilter(it, domain.ItemFilter{}, Query{}) {
+		t.Fatal("an empty filter rejected an item")
+	}
+	future := time.Now().Add(time.Hour)
+	if matchesFilter(it, domain.ItemFilter{CreatedFrom: &future}, Query{}) {
+		t.Fatal("a zero-time item passed a future CreatedFrom filter")
+	}
+}
+
+// Excerpt building slices UTF-8 text at computed offsets, so it is exactly the
+// kind of code that produces mangled output on unusual input.
+func FuzzExcerptHelpers(f *testing.F) {
+	f.Add("The revised delivery date is 14 March 2025.", "delivery", 40)
+	f.Add("ééééé", "é", 3)
+	f.Add("🌍🌍🌍", "🌍", 2)
+	f.Add("", "", 0)
+
+	f.Fuzz(func(t *testing.T, text, term string, width int) {
+		if width < 0 || width > 1<<16 {
+			return
+		}
+		// Slicing valid UTF-8 must not produce invalid UTF-8.
+		if !utf8.ValidString(text) {
+			return
+		}
+		for name, got := range map[string]string{
+			"truncateWords":  truncateWords(text, width),
+			"findTermWindow": findTermWindow(text, []string{term}, width),
+			"collapse":       collapse(text),
+		} {
+			if !utf8.ValidString(got) {
+				t.Fatalf("%s produced invalid UTF-8 from text=%q term=%q width=%d -> %q",
+					name, text, term, width, got)
+			}
+		}
+	})
+}