Fix BEL-terminated OSC 8 hyperlink widths

README.md

@@ -1118,7 +1118,8 @@ Some terminals support a special grouping of ANSI escape sequences that are inte
 much in the same way they are shown in browsers. These are handled just as mentioned before: non-printable
 ANSI escape sequences are removed prior to string length calculation. The only difference with escaped
 hyperlinks is that column width will be based on the length of the URL _text_ rather than the URL
-itself (terminals would show this text). For example:
+itself (terminals would show this text). Both ST (`ESC \\`) and BEL (`\x07`) terminators
+are supported for OSC 8 hyperlinks. For example:
 
     >>> len('\x1b]8;;https://example.com\x1b\\example\x1b]8;;\x1b\\')  # display length is 7, showing 'example'
     40

tabulate/__init__.py

@@ -771,7 +771,7 @@ def escape_empty(val):
 # Also include the terminal hyperlink sequences as described here:
 # https://gist.github.com/egmontkob/eb114294efbcd5adb1944c9f3cb5feda
 #
-# OSC 8 ; params ; uri ST display_text OSC 8 ;; ST
+# OSC 8 ; params ; uri (ST or BEL) display_text OSC 8 ;; (ST or BEL)
 #
 # Example: \x1b]8;;https://example.com\x5ctext to show\x1b]8;;\x5c
 #
@@ -780,10 +780,12 @@ def escape_empty(val):
 # params: 0..n optional key value pairs separated by ':' (e.g. foo=bar:baz=qux:abc=123)
 # URI: the actual URI with protocol scheme (e.g. https://, file://, ftp://)
 # ST: ESC followed by the '\' character (0x5c)
+# BEL: the bell character (0x07), also accepted as an OSC terminator
 _esc = r"\x1b"
 _csi = rf"{_esc}\["
 _osc = rf"{_esc}\]"
 _st = rf"{_esc}\\"
+_osc_end = rf"(?:{_st}|\x07)"
 
 _ansi_escape_pat = rf"""
     (
@@ -797,10 +799,10 @@ def escape_empty(val):
         {_osc}8;        # OSC opening
         (\w+=\w+:?)*    # key=value params list (submatch 2)
         ;               # delimiter
-        ([^{_esc}]+)    # URI - anything but ESC (submatch 3)
-        {_st}           # ST
+        ([^{_esc}\x07]+) # URI - anything but ESC or BEL (submatch 3)
+        {_osc_end}      # ST or BEL
         ([^{_esc}]+)    # link text - anything but ESC (submatch 4)
-        {_osc}8;;{_st}  # "closing" OSC sequence
+        {_osc}8;;{_osc_end}  # "closing" OSC sequence
     )
 """
 _ansi_codes = re.compile(_ansi_escape_pat, re.VERBOSE)
@@ -1091,7 +1093,7 @@ def _strip_ansi(s):
     if isinstance(s, str):
         return _ansi_codes.sub(r"\4", s)
     else:  # a bytestring
-        return _ansi_codes_bytes.sub(r"\4", s)
+        return _ansi_codes_bytes.sub(rb"\4", s)
 
 
 def _visible_width(s):

test/test_hyperlinks.py

@@ -0,0 +1,40 @@
+"""OSC 8 hyperlinks must not affect column widths."""
+
+from pytest import mark
+
+import tabulate as T
+
+
+@mark.parametrize("opening_end", ["\x1b\\", "\x07"])
+@mark.parametrize("closing_end", ["\x1b\\", "\x07"])
+@mark.parametrize("as_bytes", [False, True])
+def test_strip_hyperlink_terminators(opening_end, closing_end, as_bytes):
+    link = f"\x1b]8;id=docs;https://example.com/a;b{opening_end}docs\x1b]8;;{closing_end}"
+    text = "before " + link + " after"
+    expected = "before docs after"
+    if as_bytes:
+        text = text.encode()
+        expected = expected.encode()
+    assert T._strip_ansi(text) == expected
+
+
+@mark.parametrize("opening_end", ["\x1b\\", "\x07"])
+@mark.parametrize("closing_end", ["\x1b\\", "\x07"])
+@mark.parametrize("wide_chars_mode", [False, True])
+@mark.parametrize("as_bytes", [False, True])
+def test_hyperlink_table_alignment(
+    monkeypatch, opening_end, closing_end, wide_chars_mode, as_bytes
+):
+    monkeypatch.setattr(T, "WIDE_CHARS_MODE", wide_chars_mode)
+    link = f"\x1b]8;;https://example.com{opening_end}docs\x1b]8;;{closing_end}"
+    cell = link.encode() if as_bytes else link
+    result = T.tabulate([[cell, "ok"]], headers=[link, "state"], tablefmt="grid")
+    expected = T.tabulate([["docs", "ok"]], headers=["docs", "state"], tablefmt="grid")
+    assert result.replace(link, "docs") == expected
+    assert result.count(link) == 2
+
+
+def test_adjacent_hyperlinks_with_different_terminators():
+    first = "\x1b]8;;https://example.com/one\x07one\x1b]8;;\x07"
+    second = "\x1b]8;;https://example.com/two\x1b\\two\x1b]8;;\x1b\\"
+    assert T._strip_ansi(first + " " + second) == "one two"