gh-156230: Bound a curses window read by the window, not by 2047 (GH-156282)

instr(), in_wstr() and in_wchstr() clamped the count to 2047 and silently
truncated a longer line, which a pad can have.  A window read cannot return
more than the columns left on the line, in the unit each method counts: cells,
characters, or bytes at CCHARW_MAX characters of MB_CUR_MAX bytes per cell.
Cap the count by that, and read the rest of the line when the count is
omitted, which it now can be.

getstr() and get_wstr() read the keyboard rather than the window, so their
limit stays.

Author: serhiy-storchaka

Doc/library/curses.rst

@@ -1404,26 +1404,29 @@ Reading window contents
             window.instr(y, x[, n])
 
    Read the text of the window from the current cursor position,
-   or from *y*, *x* if specified, to the end of the line,
+   or from *y*, *x* if specified, to the end of the line
+   or at most *n* bytes if *n* is specified,
    and return it as a bytes object, in the encoding of the current locale.
    Attributes and color pairs are stripped;
    use :meth:`in_wchstr` to read them too.
-   At most *n* bytes are read; *n* defaults to and cannot exceed 2047.
    A character not representable in the encoding cannot be returned;
    use :meth:`in_wstr` for those.
 
    .. versionchanged:: 3.14
       The maximum value for *n* was increased from 1023 to 2047.
 
+   .. versionchanged:: next
+      *n* is no longer limited to 2047.
+
 .. method:: window.in_wstr([n])
             window.in_wstr(y, x[, n])
 
    Read the text of the window from the current cursor position,
-   or from *y*, *x* if specified, to the end of the line,
+   or from *y*, *x* if specified, to the end of the line
+   or at most *n* characters if *n* is specified,
    and return it as a :class:`str`.
    Attributes and color pairs are stripped;
    use :meth:`in_wchstr` to read them too.
-   At most *n* characters are read; *n* defaults to and cannot exceed 2047.
 
    This is the wide-character variant of :meth:`instr`.
 
@@ -1433,12 +1436,12 @@ Reading window contents
             window.in_wchstr(y, x[, n])
 
    Read the styled cells of the window from the current cursor position,
-   or from *y*, *x* if specified, to the end of the line,
+   or from *y*, *x* if specified, to the end of the line
+   or at most *n* cells if *n* is specified,
    and return them as a :class:`complexstr`.
    Unlike :meth:`instr` and :meth:`in_wstr`, each cell keeps its attributes
    and color pair, so the result can be written back unchanged
    with :meth:`addstr`.
-   At most *n* cells are read; *n* defaults to and cannot exceed 2047.
 
    .. versionadded:: next
 

Lib/test/test_curses.py

@@ -672,6 +672,11 @@ def test_in_wstr(self):
         stdscr.addstr(0, 0, 'abz')
         self.assertEqual(stdscr.in_wstr(0, 0, 0), '')
         self.assertEqual(stdscr.in_wstr(0), '')
+        self.assertEqual(stdscr.in_wstr(0, 0, 2**31), stdscr.in_wstr(0, 0))
+        self.assertRaises(OverflowError, stdscr.in_wstr, 2**1000)
+        self.assertRaises(ValueError, stdscr.in_wstr, -2)
+        self.assertRaises(ValueError, stdscr.in_wstr, 0, 2, -2)
+        self.assertRaises(ValueError, stdscr.in_wstr, -2**1000)
 
     def test_complexchar(self):
         # A complexchar is a styled wide-character cell: str() is its text,
@@ -871,6 +876,11 @@ def test_in_wchstr(self):
         # The count is optional and reads to the end of the line by default.
         stdscr.move(0, 0)
         self.assertEqual(str(stdscr.in_wchstr())[:3], 'AbC')
+        self.assertEqual(stdscr.in_wchstr(0, 0, 2**31), stdscr.in_wchstr(0, 0))
+        self.assertRaises(OverflowError, stdscr.in_wchstr, 2**1000)
+        self.assertRaises(ValueError, stdscr.in_wchstr, -2)
+        self.assertRaises(ValueError, stdscr.in_wchstr, 0, 2, -2)
+        self.assertRaises(ValueError, stdscr.in_wchstr, -2**1000)
 
     def test_complexstr_in_write_methods(self):
         # addstr/addnstr/insstr/insnstr also accept a complexstr, written via
@@ -1188,8 +1198,13 @@ def test_read_from_window(self):
         self.assertEqual(stdscr.instr(3)[:6], b' AB')
         self.assertEqual(stdscr.instr(0, 2)[:4], b'BCD ')
         self.assertEqual(stdscr.instr(0, 2, 4), b'BCD ')
+        # A huge count is bounded by the line, and is not used to size the
+        # read buffer.
+        self.assertEqual(stdscr.instr(0, 0, 2**31), stdscr.instr(0, 0))
+        self.assertRaises(OverflowError, stdscr.instr, 2**1000)
         self.assertRaises(ValueError, stdscr.instr, -2)
         self.assertRaises(ValueError, stdscr.instr, 0, 2, -2)
+        self.assertRaises(ValueError, stdscr.instr, -2**1000)
         # instr(y, x, 1) reads a single cell byte, so only a character that the
         # window encoding maps to one byte is checked.  inch() returns the cell
         # value, which is the locale byte.
@@ -1206,6 +1221,25 @@ def test_read_from_window(self):
                 self.assertEqual(stdscr.instr(2, 0, 1), b)
                 self.assertEqual(stdscr.inch(2, 0), v)
 
+    def test_read_long_line(self):
+        # A pad line can be longer than a window, and a character can be
+        # encoded with several bytes, so instr() can read more bytes than
+        # there are cells.  See _encodable for the character set.
+        width = 3000
+        pad = curses.newpad(1, width)
+        for ch in ['z', '\u00e9', '\u20ac', '\u0434', '\uff71']:
+            if not self._storable(ch):
+                continue
+            pad.addstr(0, 0, ch)
+            if pad.getyx()[1] != 1:
+                continue        # a wide character occupies two cells
+            with self.subTest(ch=ch):
+                line = ch * (width - 1) + ' '   # the last cell is left blank
+                pad.addstr(0, 0, line[:-1])
+                self.assertEqual(pad.instr(0, 0), line.encode(pad.encoding))
+                self.assertEqual(pad.in_wstr(0, 0), line)
+                self.assertEqual(str(pad.in_wchstr(0, 0)), line)
+
     def test_coordinate_errors(self):
         # Addressing a cell outside the window raises curses.error.
         win = curses.newwin(5, 10, 0, 0)

Misc/NEWS.d/next/Library/2026-08-22-14-12-40.gh-issue-156230.Lt9wQd.rst

@@ -0,0 +1,3 @@
+:meth:`curses.window.instr`, :meth:`~curses.window.in_wstr` and
+:meth:`~curses.window.in_wchstr` no longer limit the count to 2047, which
+silently truncated a longer line.