Parse C1 CSI terminal sequences

Author: SqlRush

docs/cc-100-roadmap.md

@@ -1040,6 +1040,8 @@ M7 补充：terminal input parser 和 configurable keybinding name parser 现在
 
 本轮补充：terminal sequence dispatcher/parser 现在把 DCS/APC/PM/SOS string-control 序列分类为 `stringControl` action，保留 payload、terminator 和 incomplete flush 状态，同时 visible text 继续忽略这些不可见控制串。
 
+本轮补充：terminal tokenizer、sequence dispatcher、CSI parser 和 visible-text stripping 现在接受 8-bit C1 CSI (`0x9b`) 序列，覆盖分块 SGR 输入以及 input tokenizer 的 X10 mouse payload 边界。
+
 本轮补充：message renderer 增加 ANSI-aware wrapping/padding，带 SGR 的 message text 会通过 terminal parser 按 grapheme 可见宽度换行，并把 `TextStyle` action 重新渲染为 SGR 序列，避免 escape bytes 参与 layout 宽度计算；普通文本路径保持不变。
 
 本轮补充：基础 wrap/pad/trim 改为按 terminal grapheme 可见宽度计算，普通 message、status/dialog/viewport/prompt 的 CJK/emoji 宽字符不再按单 rune 宽度参与布局，继续向 terminal column parity 收口。

docs/claude-code-go-rewrite-plan.md

@@ -585,6 +585,7 @@ test/parity/                 # golden tests against TS/official behavior
 - 本轮补充：terminal tokenizer 补齐 PM (`ESC ^`) 和 SOS (`ESC X`) string-control 状态，和 OSC/DCS/APC 一样支持 BEL 或 ST terminator，避免这些控制串 payload 泄漏为 text token。
 - 本轮补充：terminal sequence dispatcher/parser 现在把 DCS/APC/PM/SOS string-control 序列分类为 `stringControl` action，保留 payload、terminator 和 incomplete flush 状态，同时 visible text 继续忽略这些不可见控制串。
 - 本轮补充：snapshot/OSC 复用 terminal parser 的 visible-text pipeline，`StripANSI` 不再维护独立手写 scanner；可见文本提取统一覆盖 CSI/OSC/DCS/APC/PM/SOS、flushed partial OSC 和 raw BEL 兼容行为，为后续 ANSI parser 与 renderer/snapshot parity 收口。
+- 本轮补充：terminal tokenizer、sequence dispatcher、CSI parser 和 visible-text stripping 现在接受 8-bit C1 CSI (`0x9b`) 序列，覆盖分块 SGR 输入以及 input tokenizer 的 X10 mouse payload 边界。
 - 本轮补充：message renderer 增加 ANSI-aware wrapping/padding，带 SGR 的 message text 会通过 terminal parser 按 grapheme 可见宽度换行，并把 `TextStyle` action 重新渲染为 SGR 序列，避免 escape bytes 参与 layout 宽度计算；普通文本路径保持不变。
 - 本轮补充：基础 wrap/pad/trim 改为按 terminal grapheme 可见宽度计算，普通 message、status/dialog/viewport/prompt 的 CJK/emoji 宽字符不再按单 rune 宽度参与布局，继续向 terminal column parity 收口。
 - 本轮补充：prompt layout 的 chunking 和 cursor column 映射改为按 terminal grapheme 可见宽度计算，宽字符输入换行和 cursor CSI 定位不再按 rune index 误算列宽。

docs/first-second-parity-audit.md

@@ -386,6 +386,7 @@ M7 progress now includes:
 - `internal/tui`: terminal CSI parsing now marks DEC selective erase `CSI ? Ps J` / `CSI ? Ps K` separately from ordinary ED/EL erase actions.
 - `internal/tui`: terminal CSI parsing now exposes ECMA `CSI Ps N` / `CSI Ps O` erase-in-field and erase-in-area actions with the same to-end/to-start/all regions used by other erase operations.
 - `internal/tui`: terminal CSI parsing now treats explicit zero count/position params as ANSI defaults for cursor movement/position/column, insert/repeat/erase chars, and scroll actions while preserving raw zero selector semantics for mode/report/erase selector sequences.
+- `internal/tui`: terminal tokenizer, sequence dispatcher, CSI parser, and visible-text stripping now accept 8-bit C1 CSI (`0x9b`) sequences, including chunked SGR input and X10 mouse payload boundaries.
 - `internal/tui`: keybinding config, keymap resolution, and interaction script named-key input now accept terminal aliases for `ctrl-h`/`ctrl-i`/`ctrl-j`/`ctrl-m`, `ctrl-[`, and `ctrl-?`, including `control-*` and compact/camel variants.
 - `internal/tui`: keybinding config now accepts page/navigation/editing aliases such as `pgup`, `pg-up`, `prior`, `pageUpKey`, `pgdn`, `pg-down`, `next`, `pageDownKey`, `homeKey`, `endKey`, `deleteForward`, `forwardDelete`, and `deleteBackward`.
 - `internal/tui`: keybinding config and named-key script input now accept DOM/fixture control-key aliases such as `enterKey`, `returnKey`, `numpadEnter`, `escapeKey`, `escKey`, `tabKey`, `shiftEnterKey`, `shiftNumpadEnter`, `shiftTabKey`, and `backtabKey`.

internal/tui/terminal_csi.go

@@ -7,6 +7,7 @@ import (
 
 const (
 	CSIPrefix         = "\x1b["
+	C1CSIPrefix       = "\x9b"
 	CursorLeft        = "\x1b[G"
 	CursorSave        = "\x1b[s"
 	CursorRestore     = "\x1b[u"
@@ -354,10 +355,10 @@ func IsCSIFinal(b byte) bool {
 }
 
 func ParseCSISequence(sequence string) (CSIAction, bool) {
-	if !strings.HasPrefix(sequence, CSIPrefix) {
+	inner, ok := trimCSIPrefix(sequence)
+	if !ok {
 		return CSIAction{}, false
 	}
-	inner := strings.TrimPrefix(sequence, CSIPrefix)
 	if len(inner) == 0 {
 		return CSIAction{}, false
 	}
@@ -526,6 +527,17 @@ func ParseCSISequence(sequence string) (CSIAction, bool) {
 	return CSIAction{Type: CSIActionUnknown, Sequence: sequence}, true
 }
 
+func trimCSIPrefix(sequence string) (string, bool) {
+	switch {
+	case strings.HasPrefix(sequence, CSIPrefix):
+		return strings.TrimPrefix(sequence, CSIPrefix), true
+	case strings.HasPrefix(sequence, C1CSIPrefix):
+		return strings.TrimPrefix(sequence, C1CSIPrefix), true
+	default:
+		return "", false
+	}
+}
+
 func parseCSIParams(paramStr string) []int {
 	if paramStr == "" {
 		return nil

internal/tui/terminal_sequence.go

@@ -34,6 +34,9 @@ type TerminalSequenceAction struct {
 }
 
 func IdentifyTerminalSequence(sequence string) TerminalSequenceType {
+	if strings.HasPrefix(sequence, C1CSIPrefix) {
+		return TerminalSequenceCSI
+	}
 	if len(sequence) < 2 || sequence[0] != ESCPrefix[0] {
 		return TerminalSequenceUnknown
 	}

internal/tui/terminal_tokenizer.go

@@ -3,6 +3,7 @@ package tui
 const (
 	terminalBEL = byte(0x07)
 	terminalESC = byte(0x1b)
+	terminalCSI = byte(0x9b)
 
 	escTypeCSI = byte('[')
 	escTypeOSC = byte(']')
@@ -129,6 +130,11 @@ func tokenizeTerminal(input string, initialState TerminalTokenizerState, initial
 				seqStart = i
 				state = terminalTokenizerEscape
 				i++
+			} else if code == terminalCSI {
+				flushText()
+				seqStart = i
+				state = terminalTokenizerCSI
+				i++
 			} else {
 				i++
 			}
@@ -181,7 +187,7 @@ func tokenizeTerminal(input string, initialState TerminalTokenizerState, initial
 				textStart = seqStart
 			}
 		case terminalTokenizerCSI:
-			if x10Mouse && code == 'M' && i-seqStart == 2 && x10MousePayloadLooksPrintable(data, i) {
+			if x10Mouse && code == 'M' && i-seqStart == csiPayloadOffset(data, seqStart) && x10MousePayloadLooksPrintable(data, i) {
 				if i+4 <= len(data) {
 					i += 4
 					emitSequence(data[seqStart:i])
@@ -252,3 +258,10 @@ func x10MousePayloadLooksPrintable(data string, index int) bool {
 	}
 	return true
 }
+
+func csiPayloadOffset(data string, seqStart int) int {
+	if seqStart >= 0 && seqStart < len(data) && data[seqStart] == terminalCSI {
+		return 1
+	}
+	return 2
+}

internal/tui/terminal_tokenizer_test.go

@@ -23,6 +23,24 @@ func TestTerminalTokenizerFeedsTextAndSequencesAcrossChunks(t *testing.T) {
 	}
 }
 
+func TestTerminalTokenizerFeedsC1CSISequencesAcrossChunks(t *testing.T) {
+	tokenizer := NewTerminalTokenizer(TerminalTokenizerOptions{})
+	tokens := tokenizer.Feed("hi \x9b")
+	want := []TerminalToken{{Type: TerminalTokenText, Value: "hi "}}
+	if !reflect.DeepEqual(tokens, want) || tokenizer.Buffer() != "\x9b" {
+		t.Fatalf("first c1 feed tokens=%#v buffer=%q", tokens, tokenizer.Buffer())
+	}
+
+	tokens = tokenizer.Feed("31mred")
+	want = []TerminalToken{
+		{Type: TerminalTokenSequence, Value: "\x9b31m"},
+		{Type: TerminalTokenText, Value: "red"},
+	}
+	if !reflect.DeepEqual(tokens, want) || tokenizer.Buffer() != "" {
+		t.Fatalf("second c1 feed tokens=%#v buffer=%q", tokens, tokenizer.Buffer())
+	}
+}
+
 func TestTerminalTokenizerFlushesIncompleteSequences(t *testing.T) {
 	tokenizer := NewTerminalTokenizer(TerminalTokenizerOptions{})
 	if tokens := tokenizer.Feed("a\x1b[?"); !reflect.DeepEqual(tokens, []TerminalToken{{Type: TerminalTokenText, Value: "a"}}) {

internal/tui/tui_test.go

@@ -5789,6 +5789,31 @@ func TestParseCSISequenceActions(t *testing.T) {
 	}
 }
 
+func TestParseC1CSISequenceActions(t *testing.T) {
+	const c1CSI = "\x9b"
+
+	erase, ok := ParseCSISequence(c1CSI + "2J")
+	if !ok || erase.Type != CSIActionErase || erase.Erase.Type != CSIEraseActionDisplay || erase.Erase.Region != CSIEraseAll {
+		t.Fatalf("c1 erase action = %#v ok=%v", erase, ok)
+	}
+
+	action, ok := ParseTerminalSequence(c1CSI + "31m")
+	if !ok || action.Type != TerminalSequenceCSI || action.CSI.Type != CSIActionSGR || action.CSI.SGRParams != "31" {
+		t.Fatalf("c1 terminal action = %#v ok=%v", action, ok)
+	}
+
+	parser := NewTerminalParser()
+	actions := parser.Feed("a" + c1CSI)
+	actions = append(actions, parser.Feed("31mb"+c1CSI+"0mc")...)
+	actions = append(actions, parser.Flush()...)
+	if visible := TerminalActionsVisibleText(actions); visible != "abc" {
+		t.Fatalf("c1 visible text = %q actions=%#v", visible, actions)
+	}
+	if visible := StripANSI("a" + c1CSI + "31mb" + c1CSI + "0mc"); visible != "abc" {
+		t.Fatalf("c1 stripped text = %q", visible)
+	}
+}
+
 func TestParseESCSequenceActions(t *testing.T) {
 	if !IsESCFinal('0') || !IsESCFinal('~') || IsESCFinal('/') || IsESCFinal(0x7f) {
 		t.Fatalf("ESC final range mismatch")