Sub2API v1.0 - AI API 网关（二开初始版本，基于上游 Wei-Shaw/sub2api）

backend/internal/pkg/anthropicfp/dateline.go

@@ -0,0 +1,288 @@
+// Package anthropicfp provides pure helpers for suppressing client-side
+// fingerprints that would otherwise be visible to upstream Anthropic when a
+// forwarding gateway sits between the client and api.anthropic.com.
+//
+// Currently exposes NormalizeDateline: it rewrites the "Today's date is
+// YYYY-MM-DD." sentence inside a request body back to a canonical ASCII form,
+// erasing three bits of steganographic signal (four apostrophe code points and
+// a date-separator variant) that some clients embed in that sentence when
+// they detect a non-official base URL.
+package anthropicfp
+
+import (
+	"fmt"
+	"regexp"
+	"sort"
+	"strings"
+
+	"github.com/tidwall/gjson"
+	"github.com/tidwall/sjson"
+)
+
+// datelineRegexes matches the fingerprinted sentence with any of the four
+// apostrophe code points seen in the wild and either separator. Two regexes
+// are used because Go's RE2-based regexp package does not support
+// backreferences: matching `-` and `/` in two passes keeps the two separators
+// inside YYYY?MM?DD forced to agree, so mixed-separator strings like
+// "Today's date is 2026-07/01." never match. This is what filters out
+// user-authored prose like "Today is foo." or "His date is 2026-06-30." from
+// being touched.
+var (
+	datelineRegexHyphen = regexp.MustCompile(`Today(['’ʼʹ])s date is (\d{4})-(\d{2})-(\d{2})\.`)
+	datelineRegexSlash  = regexp.MustCompile(`Today(['’ʼʹ])s date is (\d{4})/(\d{2})/(\d{2})\.`)
+)
+
+// systemReminderRegex matches a <system-reminder> block. The dateline lives in
+// this block once the conversation has advanced past the first turn (system
+// prompt caching hides the top-level system block for subsequent turns), so
+// the messages[].content[] scan is confined to what lives inside these tags.
+var systemReminderRegex = regexp.MustCompile(`(?s)<system-reminder>.*?</system-reminder>`)
+
+// DatelineHit records what a single rewrite normalized, for observability.
+type DatelineHit struct {
+	// ApostropheVariant is one of "ascii" (U+0027), "u2019", "u02bc", "u02b9".
+	ApostropheVariant string
+	// DateSeparator is either "-" or "/" as seen before normalization.
+	DateSeparator string
+}
+
+// canonicalize returns the canonical form of a matched dateline sentence.
+// The output always uses ASCII apostrophe and hyphen separators.
+func canonicalize(year, month, day string) string {
+	return fmt.Sprintf("Today's date is %s-%s-%s.", year, month, day)
+}
+
+func apostropheVariant(r rune) string {
+	switch r {
+	case '’':
+		return "u2019"
+	case 'ʼ':
+		return "u02bc"
+	case 'ʹ':
+		return "u02b9"
+	default:
+		return "ascii"
+	}
+}
+
+type datelineMatch struct {
+	start, end       int
+	apoRune          rune
+	sep              string
+	year, month, day string
+}
+
+func collectMatches(text string, re *regexp.Regexp, sep string) []datelineMatch {
+	locs := re.FindAllStringSubmatchIndex(text, -1)
+	if len(locs) == 0 {
+		return nil
+	}
+	out := make([]datelineMatch, 0, len(locs))
+	for _, m := range locs {
+		var apoRune rune
+		for _, r := range text[m[2]:m[3]] {
+			apoRune = r
+			break
+		}
+		out = append(out, datelineMatch{
+			start:   m[0],
+			end:     m[1],
+			apoRune: apoRune,
+			sep:     sep,
+			year:    text[m[4]:m[5]],
+			month:   text[m[6]:m[7]],
+			day:     text[m[8]:m[9]],
+		})
+	}
+	return out
+}
+
+// NormalizeText replaces every fingerprinted dateline sentence in text with
+// its canonical form. It returns the possibly-rewritten text and the list of
+// hits observed. When no match is found the original string is returned
+// verbatim (byte-identical), and the returned hit slice is nil.
+func NormalizeText(text string) (string, []DatelineHit) {
+	if !strings.Contains(text, "date is ") {
+		return text, nil
+	}
+	matches := collectMatches(text, datelineRegexHyphen, "-")
+	matches = append(matches, collectMatches(text, datelineRegexSlash, "/")...)
+	if len(matches) == 0 {
+		return text, nil
+	}
+	sort.Slice(matches, func(i, j int) bool { return matches[i].start < matches[j].start })
+
+	var b strings.Builder
+	b.Grow(len(text))
+	prev := 0
+	hits := make([]DatelineHit, 0, len(matches))
+	changed := false
+	for _, m := range matches {
+		full := text[m.start:m.end]
+		canonical := canonicalize(m.year, m.month, m.day)
+		if canonical == full {
+			// Already canonical: no rewrite, no hit.
+			continue
+		}
+		_, _ = b.WriteString(text[prev:m.start])
+		_, _ = b.WriteString(canonical)
+		prev = m.end
+		changed = true
+		hits = append(hits, DatelineHit{
+			ApostropheVariant: apostropheVariant(m.apoRune),
+			DateSeparator:     m.sep,
+		})
+	}
+	if !changed {
+		return text, nil
+	}
+	_, _ = b.WriteString(text[prev:])
+	return b.String(), hits
+}
+
+// normalizeSystemReminderScopedText scans only the <system-reminder> blocks
+// inside text and normalizes datelines inside them. Text outside the blocks is
+// preserved byte-for-byte, so user prose, tool_result content, code blocks,
+// or shell commands that happen to contain an apostrophe or a slash date are
+// never touched.
+func normalizeSystemReminderScopedText(text string) (string, []DatelineHit) {
+	if !strings.Contains(text, "<system-reminder>") {
+		return text, nil
+	}
+	locs := systemReminderRegex.FindAllStringIndex(text, -1)
+	if len(locs) == 0 {
+		return text, nil
+	}
+	var b strings.Builder
+	b.Grow(len(text))
+	prev := 0
+	var hits []DatelineHit
+	changed := false
+	for _, loc := range locs {
+		_, _ = b.WriteString(text[prev:loc[0]])
+		block := text[loc[0]:loc[1]]
+		normalized, blockHits := NormalizeText(block)
+		if normalized != block {
+			changed = true
+		}
+		_, _ = b.WriteString(normalized)
+		hits = append(hits, blockHits...)
+		prev = loc[1]
+	}
+	if !changed {
+		return text, nil
+	}
+	_, _ = b.WriteString(text[prev:])
+	return b.String(), hits
+}
+
+// NormalizeDateline scans an Anthropic /v1/messages request body and rewrites
+// every fingerprinted dateline sentence back to its canonical ASCII form.
+//
+// Scope (mirroring where genuine clients place the sentence):
+//   - `system` string, or `.text` field of each text-typed block in `system`.
+//   - Text bodies inside `messages[i].content` — but ONLY the substrings that
+//     appear inside `<system-reminder>...</system-reminder>` tags. Free user
+//     prose, tool_use.input, tool_result.content, and other block types are
+//     never scanned, guaranteeing that legitimate text like a code block, a
+//     shell command, or a chat message that mentions today's date is never
+//     accidentally rewritten.
+//
+// The function is a pure transform: it never modifies the input slice, and if
+// no rewrite is needed it returns the original slice (identity), a nil hit
+// slice, and changed=false.
+func NormalizeDateline(body []byte) ([]byte, []DatelineHit, bool) {
+	if len(body) == 0 {
+		return body, nil, false
+	}
+	out := body
+	var hits []DatelineHit
+	changed := false
+
+	sys := gjson.GetBytes(out, "system")
+	if sys.Exists() {
+		switch {
+		case sys.Type == gjson.String:
+			normalized, sysHits := NormalizeText(sys.String())
+			if normalized != sys.String() {
+				if next, err := sjson.SetBytes(out, "system", normalized); err == nil {
+					out = next
+					changed = true
+					hits = append(hits, sysHits...)
+				}
+			}
+		case sys.IsArray():
+			idx := 0
+			sys.ForEach(func(_, item gjson.Result) bool {
+				if item.Get("type").String() == "text" {
+					t := item.Get("text")
+					if t.Exists() && t.Type == gjson.String {
+						normalized, textHits := NormalizeText(t.String())
+						if normalized != t.String() {
+							path := fmt.Sprintf("system.%d.text", idx)
+							if next, err := sjson.SetBytes(out, path, normalized); err == nil {
+								out = next
+								changed = true
+								hits = append(hits, textHits...)
+							}
+						}
+					}
+				}
+				idx++
+				return true
+			})
+		}
+	}
+
+	messages := gjson.GetBytes(out, "messages")
+	if messages.IsArray() {
+		msgIdx := -1
+		messages.ForEach(func(_, msg gjson.Result) bool {
+			msgIdx++
+			content := msg.Get("content")
+			if !content.Exists() {
+				return true
+			}
+			switch {
+			case content.Type == gjson.String:
+				normalized, contentHits := normalizeSystemReminderScopedText(content.String())
+				if normalized != content.String() {
+					path := fmt.Sprintf("messages.%d.content", msgIdx)
+					if next, err := sjson.SetBytes(out, path, normalized); err == nil {
+						out = next
+						changed = true
+						hits = append(hits, contentHits...)
+					}
+				}
+			case content.IsArray():
+				contentIdx := -1
+				content.ForEach(func(_, block gjson.Result) bool {
+					contentIdx++
+					if block.Get("type").String() != "text" {
+						return true
+					}
+					t := block.Get("text")
+					if !t.Exists() || t.Type != gjson.String {
+						return true
+					}
+					normalized, textHits := normalizeSystemReminderScopedText(t.String())
+					if normalized != t.String() {
+						path := fmt.Sprintf("messages.%d.content.%d.text", msgIdx, contentIdx)
+						if next, err := sjson.SetBytes(out, path, normalized); err == nil {
+							out = next
+							changed = true
+							hits = append(hits, textHits...)
+						}
+					}
+					return true
+				})
+			}
+			return true
+		})
+	}
+
+	if !changed {
+		return body, nil, false
+	}
+	return out, hits, true
+}

backend/internal/pkg/anthropicfp/dateline_test.go

@@ -0,0 +1,248 @@
+package anthropicfp
+
+import (
+	"bytes"
+	"strings"
+	"testing"
+)
+
+func TestNormalizeText_ASCIIHyphenIsIdentity(t *testing.T) {
+	in := "Today's date is 2026-07-01."
+	out, hits := NormalizeText(in)
+	if out != in {
+		t.Fatalf("canonical form should be returned identity, got %q", out)
+	}
+	if len(hits) != 0 {
+		t.Fatalf("no hits expected on canonical input, got %d", len(hits))
+	}
+}
+
+func TestNormalizeText_SlashSeparatorASCIIApostrophe(t *testing.T) {
+	in := "Today's date is 2026/07/01."
+	out, hits := NormalizeText(in)
+	want := "Today's date is 2026-07-01."
+	if out != want {
+		t.Fatalf("want %q, got %q", want, out)
+	}
+	if len(hits) != 1 || hits[0].ApostropheVariant != "ascii" || hits[0].DateSeparator != "/" {
+		t.Fatalf("unexpected hit: %+v", hits)
+	}
+}
+
+func TestNormalizeText_U2019Apostrophe(t *testing.T) {
+	in := "Today’s date is 2026-07-01."
+	out, hits := NormalizeText(in)
+	want := "Today's date is 2026-07-01."
+	if out != want {
+		t.Fatalf("want %q, got %q", want, out)
+	}
+	if len(hits) != 1 || hits[0].ApostropheVariant != "u2019" || hits[0].DateSeparator != "-" {
+		t.Fatalf("unexpected hit: %+v", hits)
+	}
+}
+
+func TestNormalizeText_U02BCApostropheWithSlash(t *testing.T) {
+	in := "Todayʼs date is 2026/07/01."
+	out, hits := NormalizeText(in)
+	want := "Today's date is 2026-07-01."
+	if out != want {
+		t.Fatalf("want %q, got %q", want, out)
+	}
+	if len(hits) != 1 || hits[0].ApostropheVariant != "u02bc" || hits[0].DateSeparator != "/" {
+		t.Fatalf("unexpected hit: %+v", hits)
+	}
+}
+
+func TestNormalizeText_U02B9Apostrophe(t *testing.T) {
+	in := "Todayʹs date is 2026/07/01."
+	out, hits := NormalizeText(in)
+	want := "Today's date is 2026-07-01."
+	if out != want {
+		t.Fatalf("want %q, got %q", want, out)
+	}
+	if len(hits) != 1 || hits[0].ApostropheVariant != "u02b9" || hits[0].DateSeparator != "/" {
+		t.Fatalf("unexpected hit: %+v", hits)
+	}
+}
+
+func TestNormalizeText_MixedSeparatorsNoMatch(t *testing.T) {
+	// backreference \3 forces the two separators to agree
+	in := "Today's date is 2026-07/01."
+	out, hits := NormalizeText(in)
+	if out != in {
+		t.Fatalf("mixed-separator input should not be matched, got %q", out)
+	}
+	if len(hits) != 0 {
+		t.Fatalf("expected no hits, got %d", len(hits))
+	}
+}
+
+func TestNormalizeText_NegativeLookalike(t *testing.T) {
+	cases := []string{
+		"Today is a great day.",
+		"His date is 2026-07-01.",
+		"Yesterday's date was 2026-06-30.",
+		"'s date is 2026-07-01.",
+	}
+	for _, c := range cases {
+		out, hits := NormalizeText(c)
+		if out != c {
+			t.Fatalf("input %q should not be modified, got %q", c, out)
+		}
+		if len(hits) != 0 {
+			t.Fatalf("input %q should produce no hits", c)
+		}
+	}
+}
+
+func TestNormalizeText_Idempotent(t *testing.T) {
+	in := "Today’s date is 2026/07/01."
+	out1, _ := NormalizeText(in)
+	out2, hits2 := NormalizeText(out1)
+	if out1 != out2 {
+		t.Fatalf("normalization not idempotent: %q vs %q", out1, out2)
+	}
+	if len(hits2) != 0 {
+		t.Fatalf("second pass should produce no hits, got %d", len(hits2))
+	}
+}
+
+func TestNormalizeText_MultipleOccurrences(t *testing.T) {
+	in := "First line.\nToday’s date is 2026/07/01.\nMore text.\nTodayʼs date is 2026-07-01.\nEnd."
+	out, hits := NormalizeText(in)
+	if strings.Count(out, "Today's date is") != 2 {
+		t.Fatalf("expected two canonicalized sentences, got: %q", out)
+	}
+	if strings.Contains(out, "’") || strings.Contains(out, "ʼ") || strings.Contains(out, "2026/07/01") {
+		t.Fatalf("fingerprint characters must be gone, got: %q", out)
+	}
+	if len(hits) != 2 {
+		t.Fatalf("expected 2 hits, got %d", len(hits))
+	}
+}
+
+func TestNormalizeDateline_SystemString(t *testing.T) {
+	body := []byte(`{"system":"You are helpful.\nToday’s date is 2026/07/01.\nBe brief.","messages":[]}`)
+	out, hits, changed := NormalizeDateline(body)
+	if !changed {
+		t.Fatalf("expected changed=true")
+	}
+	if len(hits) != 1 {
+		t.Fatalf("expected 1 hit, got %d", len(hits))
+	}
+	if !bytes.Contains(out, []byte("Today's date is 2026-07-01.")) {
+		t.Fatalf("output missing canonical dateline: %s", string(out))
+	}
+	if bytes.Contains(out, []byte("2026/07/01")) {
+		t.Fatalf("output should not contain slash date: %s", string(out))
+	}
+}
+
+func TestNormalizeDateline_SystemBlocksArray(t *testing.T) {
+	body := []byte(`{"system":[{"type":"text","text":"You are helpful."},{"type":"text","text":"Todayʼs date is 2026/07/01."}],"messages":[]}`)
+	out, hits, changed := NormalizeDateline(body)
+	if !changed {
+		t.Fatalf("expected changed=true")
+	}
+	if len(hits) != 1 || hits[0].ApostropheVariant != "u02bc" {
+		t.Fatalf("unexpected hits: %+v", hits)
+	}
+	if !bytes.Contains(out, []byte("Today's date is 2026-07-01.")) {
+		t.Fatalf("output missing canonical dateline: %s", string(out))
+	}
+}
+
+func TestNormalizeDateline_MessagesContentStringInSystemReminder(t *testing.T) {
+	body := []byte(`{"messages":[{"role":"user","content":"<system-reminder>\n# currentDate\nToday’s date is 2026/07/01.\n</system-reminder>\nHello, please help."}]}`)
+	out, hits, changed := NormalizeDateline(body)
+	if !changed {
+		t.Fatalf("expected changed=true")
+	}
+	if len(hits) != 1 {
+		t.Fatalf("expected 1 hit, got %d", len(hits))
+	}
+	if !bytes.Contains(out, []byte("Today's date is 2026-07-01.")) {
+		t.Fatalf("canonical dateline missing: %s", string(out))
+	}
+}
+
+func TestNormalizeDateline_MessagesContentBlocksInSystemReminder(t *testing.T) {
+	body := []byte(`{"messages":[{"role":"user","content":[{"type":"text","text":"<system-reminder>\nToday’s date is 2026/07/01.\n</system-reminder>"},{"type":"text","text":"do X"}]}]}`)
+	out, hits, changed := NormalizeDateline(body)
+	if !changed {
+		t.Fatalf("expected changed=true")
+	}
+	if len(hits) != 1 {
+		t.Fatalf("expected 1 hit, got %d", len(hits))
+	}
+	if !bytes.Contains(out, []byte("Today's date is 2026-07-01.")) {
+		t.Fatalf("canonical dateline missing: %s", string(out))
+	}
+}
+
+func TestNormalizeDateline_LeavesOutOfScopeUntouched(t *testing.T) {
+	// User prose outside <system-reminder> that mentions today's date must
+	// not be modified. tool_use.input / tool_result.content are never scanned.
+	body := []byte(`{"messages":[` +
+		`{"role":"user","content":"Today’s date is 2026/07/01. Please help."},` +
+		`{"role":"assistant","content":[{"type":"tool_use","id":"x","name":"y","input":{"note":"Today’s date is 2026/07/01."}}]},` +
+		`{"role":"user","content":[{"type":"tool_result","tool_use_id":"x","content":"log: Today’s date is 2026/07/01."}]}` +
+		`]}`)
+	out, hits, changed := NormalizeDateline(body)
+	if changed {
+		t.Fatalf("expected changed=false, hits=%v out=%s", hits, string(out))
+	}
+	if !bytes.Equal(out, body) {
+		t.Fatalf("output should equal input byte-for-byte")
+	}
+}
+
+func TestNormalizeDateline_Idempotent(t *testing.T) {
+	body := []byte(`{"messages":[{"role":"user","content":"<system-reminder>\nToday’s date is 2026/07/01.\n</system-reminder>"}]}`)
+	first, _, changed1 := NormalizeDateline(body)
+	if !changed1 {
+		t.Fatalf("expected first pass to change body")
+	}
+	second, _, changed2 := NormalizeDateline(first)
+	if changed2 {
+		t.Fatalf("second pass should not report changes")
+	}
+	if !bytes.Equal(first, second) {
+		t.Fatalf("second pass diverged: %s vs %s", string(first), string(second))
+	}
+}
+
+func TestNormalizeDateline_EmptyBody(t *testing.T) {
+	out, hits, changed := NormalizeDateline(nil)
+	if changed || out != nil || hits != nil {
+		t.Fatalf("empty body should be no-op")
+	}
+}
+
+func TestNormalizeDateline_NoDateline(t *testing.T) {
+	body := []byte(`{"messages":[{"role":"user","content":"hello"}],"system":"just a system prompt"}`)
+	out, hits, changed := NormalizeDateline(body)
+	if changed || len(hits) != 0 {
+		t.Fatalf("expected no changes; changed=%v hits=%v", changed, hits)
+	}
+	if &out[0] != &body[0] {
+		// Identity is a bonus but not strict; verify content equality at minimum
+		if !bytes.Equal(out, body) {
+			t.Fatalf("output must byte-match input when no changes needed")
+		}
+	}
+}
+
+func TestNormalizeDateline_MultipleSystemReminderBlocksInSameText(t *testing.T) {
+	body := []byte(`{"messages":[{"role":"user","content":"<system-reminder>\nToday’s date is 2026/07/01.\n</system-reminder>\nsome prose\n<system-reminder>\nAlso Todayʼs date is 2026/07/01.\n</system-reminder>"}]}`)
+	out, hits, changed := NormalizeDateline(body)
+	if !changed {
+		t.Fatalf("expected changed=true")
+	}
+	if len(hits) != 2 {
+		t.Fatalf("expected 2 hits, got %d", len(hits))
+	}
+	if bytes.Contains(out, []byte("2026/07/01")) {
+		t.Fatalf("slash separator must be gone: %s", string(out))
+	}
+}