A terminal input parse helper for the kittens

kitty/data-types.c

@@ -173,6 +173,7 @@ extern bool init_keys(PyObject *module);
 extern bool init_graphics(PyObject *module);
 extern bool init_shaders(PyObject *module);
 extern bool init_mouse(PyObject *module);
+extern bool init_kittens(PyObject *module);
 #ifdef __APPLE__
 extern int init_CoreText(PyObject *);
 extern bool init_cocoa(PyObject *module);
@@ -206,6 +207,7 @@ PyInit_fast_data_types(void) {
         if (!init_graphics(m)) return NULL;
         if (!init_shaders(m)) return NULL;
         if (!init_mouse(m)) return NULL;
+        if (!init_kittens(m)) return NULL;
 #ifdef __APPLE__
         if (!init_CoreText(m)) return NULL;
         if (!init_cocoa(m)) return NULL;

kitty/kittens.c

@@ -0,0 +1,93 @@
+/*
+ * kittens.c
+ * Copyright (C) 2018 Kovid Goyal <kovid at kovidgoyal.net>
+ *
+ * Distributed under terms of the GPL3 license.
+ */
+
+#include "data-types.h"
+
+static PyObject*
+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, "UOOOOOO", &uo, &text_callback, &dcs_callback, &csi_callback, &osc_callback, &pm_callback, &apc_callback)) 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) {\
+    if (num > 0) PyObject_CallFunction(cb, "N", PyUnicode_Substring(uo, s, s + num));  \
+    consumed = s + num; \
+    count = 0; \
+}
+    START_ALLOW_CASE_RANGE;
+    while (pos < sz) {
+        Py_UCS4 ch = PyUnicode_READ(kind, data, pos);
+        switch(state) {
+            case NORMAL:
+                if (ch == 0x1b) {
+                    state = ESC;
+                    CALL(text_callback, start, count);
+                } else count++;
+                break;
+            case ESC:
+                start = pos;
+                count = 0;
+                switch(ch) {
+                    case 'P':
+                        state = ST; callback = dcs_callback; break;
+                    case '[':
+                        state = CSI; callback = csi_callback; break;
+                    case ']':
+                        state = ST; callback = osc_callback; break;
+                    case '^':
+                        state = ST; callback = pm_callback; break;
+                    case '_':
+                        state = ST; callback = apc_callback; break;
+                    default:
+                        state = NORMAL; break;
+                }
+                break;
+            case CSI:
+                count++;
+                switch (ch) {
+                    case 'a' ... 'z':
+                    case 'A' ... 'Z':
+                        CALL(callback, start + 1, count);
+                        state = NORMAL;
+                        start = pos + 1;
+                        break;
+                }
+                break;
+            case ESC_ST:
+                if (ch == '\\') {
+                    CALL(callback, start + 1, count);
+                    state = NORMAL; start = pos + 1;
+                    consumed += 2;
+                } else count += 2;
+                break;
+            case ST:
+                if (ch == 0x1b) { state = ESC_ST; }
+                else count++;
+                break;
+        }
+        pos++;
+    }
+    if (state == NORMAL && count > 0) CALL(text_callback, start, count);
+    return PyUnicode_Substring(uo, consumed, sz);
+    END_ALLOW_CASE_RANGE;
+#undef CALL
+}
+
+static PyMethodDef module_methods[] = {
+    METHODB(parse_input_from_terminal, METH_VARARGS),
+    {NULL, NULL, 0, NULL}        /* Sentinel */
+};
+
+bool
+init_kittens(PyObject *module) {
+    if (PyModule_AddFunctions(module, module_methods) != 0) return false;
+    return true;
+}

kitty_tests/datatypes.py

@@ -4,7 +4,8 @@
 
 from kitty.config import build_ansi_color_table, defaults
 from kitty.fast_data_types import (
-    REVERSE, ColorProfile, Cursor as C, HistoryBuf, LineBuf, wcswidth, wcwidth
+    REVERSE, ColorProfile, Cursor as C, HistoryBuf, LineBuf,
+    parse_input_from_terminal, wcswidth, wcwidth
 )
 from kitty.rgb import to_color
 from kitty.utils import sanitize_title
@@ -338,6 +339,25 @@ 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=''):
+            text_r, csi_r, apc_r, rest = [], [], [], []
+            left = ''
+            for d in data:
+                left = parse_input_from_terminal(left + d, text_r.append, rest.append, csi_r.append, rest.append, rest.append, apc_r.append)
+            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('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,;(')
+        tp('a\033', '[', 'mb', text='a b', csi='m')
+        tp('a\033[', 'mb', text='a b', csi='m')
+        tp('a\033', '_', 'x\033', '\\b', text='a b', apc='x')
+        tp('a\033_', 'x', '\033', '\\', 'b', text='a b', apc='x')
+
     def test_color_profile(self):
         c = ColorProfile()
         c.update_ansi_color_table(build_ansi_color_table())