Kittens learn about bracketed paste

kittens/tui/loop.py

@@ -56,7 +56,7 @@ def sanitize_term(output_fd):
 
 class Loop:
 
-    def __init__(self, input_fd=None, output_fd=None):
+    def __init__(self, input_fd=None, output_fd=None, sanitize_bracketed_paste='[\x0e\x0f\r\x07\x7f\x8d\x8e\x8f\x90\x9b\x9d\x9e\x9f]'):
         self.input_fd = input_fd or sys.stdin.fileno()
         self.output_fd = output_fd or sys.stdout.fileno()
         self.wakeup_read_fd, self.wakeup_write_fd = os.pipe()
@@ -76,6 +76,10 @@ class Loop:
             self.iov_limit = 255
         self.parse_input_from_terminal = partial(parse_input_from_terminal, self.on_text, self.on_dcs, self.on_csi, self.on_osc, self.on_pm, self.on_apc)
         self.ebs_pat = re.compile('([\177\r])')
+        self.in_bracketed_paste = False
+        self.sanitize_bracketed_paste = bool(sanitize_bracketed_paste)
+        if self.sanitize_bracketed_paste:
+            self.sanitize_ibp_pat = re.compile(sanitize_bracketed_paste)
 
     def _read_ready(self, handler):
         if not self.read_allowed:
@@ -89,24 +93,34 @@ class Loop:
         self.read_buf = data
         self.handler = handler
         try:
-            self.read_buf = self.parse_input_from_terminal(self.read_buf)
+            self.read_buf = self.parse_input_from_terminal(self.read_buf, self.in_bracketed_paste)
         finally:
             del self.handler
 
     def on_text(self, text):
+        if self.in_bracketed_paste and self.sanitize_bracketed_paste:
+            text = self.sanitize_ibp_pat.sub('', text)
+
         for chunk in self.ebs_pat.split(text):
             if chunk == '\r':
                 self.handler.on_key(enter_key)
             elif chunk == '\177':
                 self.handler.on_key(backspace_key)
             else:
-                self.handler.on_text(chunk)
+                self.handler.on_text(chunk, self.in_bracketed_paste)
 
     def on_dcs(self, dcs):
         pass
 
     def on_csi(self, csi):
-        pass
+        q = csi[-1]
+        if q in 'mM':
+            pass
+        elif q == '~':
+            if csi == '200~':
+                self.in_bracketed_paste = True
+            elif csi == '201~':
+                self.in_bracketed_paste = False
 
     def on_pm(self, pm):
         pass

kitty/kittens.c

@@ -12,14 +12,24 @@ parse_input_from_terminal(PyObject *self UNUSED, PyObject *args) {
     enum State { NORMAL, ESC, CSI, ST, ESC_ST };
     enum State state = NORMAL;
     PyObject *uo, *text_callback, *dcs_callback, *csi_callback, *osc_callback, *pm_callback, *apc_callback, *callback;
-    if (!PyArg_ParseTuple(args, "OOOOOOU", &text_callback, &dcs_callback, &csi_callback, &osc_callback, &pm_callback, &apc_callback, &uo)) return NULL;
+    int inbp = 0;
+    if (!PyArg_ParseTuple(args, "OOOOOOUp", &text_callback, &dcs_callback, &csi_callback, &osc_callback, &pm_callback, &apc_callback, &uo, &inbp)) return NULL;
     Py_ssize_t sz = PyUnicode_GET_LENGTH(uo), pos = 0, start = 0, count = 0, consumed = 0;
     callback = text_callback;
     int kind = PyUnicode_KIND(uo);
     void *data = PyUnicode_DATA(uo);
-#define CALL(cb, s, num) {\
+    bool in_bracketed_paste_mode = inbp != 0;
-    if (num > 0) PyObject_CallFunction(cb, "N", PyUnicode_Substring(uo, s, s + num));  \
+#define CALL(cb, s_, num_) {\
-    consumed = s + num; \
+    PyObject *fcb = cb; \
+    Py_ssize_t s = s_, num = num_; \
+    if (in_bracketed_paste_mode && fcb != text_callback) { \
+        fcb = text_callback; num += 2; s -= 2; \
+    } \
+    if (num > 0) { \
+        PyObject *ret = PyObject_CallFunction(fcb, "N", PyUnicode_Substring(uo, s, s + num));  \
+        Py_XDECREF(ret); \
+    } \
+    consumed = s_ + num_; \
     count = 0; \
 }
     START_ALLOW_CASE_RANGE;
@@ -30,6 +40,7 @@ parse_input_from_terminal(PyObject *self UNUSED, PyObject *args) {
                 if (ch == 0x1b) {
                     state = ESC;
                     CALL(text_callback, start, count);
+                    start = pos;
                 } else count++;
                 break;
             case ESC:
@@ -55,7 +66,17 @@ parse_input_from_terminal(PyObject *self UNUSED, PyObject *args) {
                 switch (ch) {
                     case 'a' ... 'z':
                     case 'A' ... 'Z':
+                    case '@':
+                    case '`':
+                    case '{':
+                    case '|':
+                    case '}':
+                    case '~':
+#define IBP(w)  ch == '~' && PyUnicode_READ(kind, data, start + 1) == '2' && PyUnicode_READ(kind, data, start + 2) == '0' && PyUnicode_READ(kind, data, start + 3) == w
+                        if (IBP('1')) in_bracketed_paste_mode = false;
                         CALL(callback, start + 1, count);
+                        if (IBP('0')) in_bracketed_paste_mode = true;
+#undef IBP
                         state = NORMAL;
                         start = pos + 1;
                         break;

kitty_tests/datatypes.py

@@ -339,17 +339,28 @@ class TestDataTypes(BaseTest):
         self.ae(wcswidth('\u2716\u2716\ufe0f\U0001f337'), 5)
         self.ae(sanitize_title('a\0\01 \t\n\f\rb'), 'a b')
 
-        def tp(*data, leftover='', text='', csi='', apc=''):
+        def tp(*data, leftover='', text='', csi='', apc='', ibp=False):
             text_r, csi_r, apc_r, rest = [], [], [], []
             left = ''
+            in_bp = ibp
+
+            def on_csi(x):
+                nonlocal in_bp
+                if x == '200~':
+                    in_bp = True
+                elif x == '201~':
+                    in_bp = False
+                csi_r.append(x)
+
             for d in data:
-                left = parse_input_from_terminal(text_r.append, rest.append, csi_r.append, rest.append, rest.append, apc_r.append, left + d)
+                left = parse_input_from_terminal(text_r.append, rest.append, on_csi, rest.append, rest.append, apc_r.append, left + d, in_bp)
             self.ae(left, leftover)
             self.ae(text, ' '.join(text_r))
             self.ae(csi, ' '.join(csi_r))
             self.ae(apc, ' '.join(apc_r))
             self.assertFalse(rest)
 
+        tp('a\033[200~\033[32mxy\033[201~\033[33ma', text='a \033[32m xy a', csi='200~ 201~ 33m')
         tp('abc', text='abc')
         tp('a\033[38:5:12:32mb', text='a b', csi='38:5:12:32m')
         tp('a\033_x,;(\033\\b', text='a b', apc='x,;(')