Add url_excluded_characters option to exclude characters from URLs.

This option, like select_by_word_characters, is a set of characters, but
for which to *exclude* from URL parsing. See
https://github.com/kovidgoyal/kitty/issues/3688#issuecomment-862711148.

kitty/options/definition.py

@@ -378,6 +378,15 @@ are still clickable.
 '''
     )
 
+opt('url_excluded_characters', '',
+    ctype='!url_excluded_characters',
+    long_text='''
+Characters not considered part of a URL under the mouse. In addition to these
+characters any character that is marked as the unicode C and Z categories will
+be excluded.
+'''
+    )
+
 opt('copy_on_select', 'no',
     option_type='copy_on_select',
     long_text='''

kitty/options/parse.py

@@ -1206,6 +1206,9 @@ class Parser:
     def url_color(self, val: str, ans: typing.Dict[str, typing.Any]) -> None:
         ans['url_color'] = to_color(val)
 
+    def url_excluded_characters(self, val: str, ans: typing.Dict[str, typing.Any]) -> None:
+        ans['url_excluded_characters'] = str(val)
+
     def url_prefixes(self, val: str, ans: typing.Dict[str, typing.Any]) -> None:
         ans['url_prefixes'] = url_prefixes(val)
 

kitty/options/to-c-generated.h

@@ -265,6 +265,19 @@ convert_from_opts_detect_urls(PyObject *py_opts, Options *opts) {
     Py_DECREF(ret);
 }
 
+static void
+convert_from_python_url_excluded_characters(PyObject *val, Options *opts) {
+    url_excluded_characters(val, opts);
+}
+
+static void
+convert_from_opts_url_excluded_characters(PyObject *py_opts, Options *opts) {
+    PyObject *ret = PyObject_GetAttrString(py_opts, "url_excluded_characters");
+    if (ret == NULL) return;
+    convert_from_python_url_excluded_characters(ret, opts);
+    Py_DECREF(ret);
+}
+
 static void
 convert_from_python_select_by_word_characters(PyObject *val, Options *opts) {
     select_by_word_characters(val, opts);
@@ -931,6 +944,8 @@ convert_opts_from_python_opts(PyObject *py_opts, Options *opts) {
     if (PyErr_Occurred()) return false;
     convert_from_opts_detect_urls(py_opts, opts);
     if (PyErr_Occurred()) return false;
+    convert_from_opts_url_excluded_characters(py_opts, opts);
+    if (PyErr_Occurred()) return false;
     convert_from_opts_select_by_word_characters(py_opts, opts);
     if (PyErr_Occurred()) return false;
     convert_from_opts_click_interval(py_opts, opts);

kitty/options/to-c.h

@@ -124,6 +124,15 @@ url_prefixes(PyObject *up, Options *opts) {
     }
 }
 
+static void
+url_excluded_characters(PyObject *chars, Options *opts) {
+    if (!PyUnicode_Check(chars)) { PyErr_SetString(PyExc_TypeError, "url_excluded_characters must be a string"); return; }
+    for (size_t i = 0; i < MIN((size_t)PyUnicode_GET_LENGTH(chars), sizeof(opts->url_excluded_characters)/sizeof(opts->url_excluded_characters[0])); i++) {
+        opts->url_excluded_characters[i] = PyUnicode_READ(PyUnicode_KIND(chars), PyUnicode_DATA(chars), i);
+    }
+    opts->url_excluded_characters_count = PyUnicode_GET_LENGTH(chars);
+}
+
 static void
 select_by_word_characters(PyObject *chars, Options *opts) {
     if (!PyUnicode_Check(chars)) { PyErr_SetString(PyExc_TypeError, "select_by_word_characters must be a string"); return; }

kitty/options/types.py

@@ -417,6 +417,7 @@ option_names = (  # {{{
  'touch_scroll_multiplier',
  'update_check_interval',
  'url_color',
+ 'url_excluded_characters',
  'url_prefixes',
  'url_style',
  'visual_bell_duration',
@@ -546,6 +547,7 @@ class Options:
     touch_scroll_multiplier: float = 1.0
     update_check_interval: float = 24.0
     url_color: Color = Color(red=0, green=135, blue=189)
+    url_excluded_characters: str = ''
     url_prefixes: typing.Tuple[str, ...] = ('http', 'https', 'file', 'ftp', 'gemini', 'irc', 'gopher', 'mailto', 'news', 'git')
     url_style: int = 3
     visual_bell_duration: float = 0

kitty/state.h

@@ -70,6 +70,7 @@ typedef struct {
         UrlPrefix *values;
         size_t num, max_prefix_len;
     } url_prefixes;
+    char_type url_excluded_characters[256]; size_t url_excluded_characters_count;
     bool detect_urls;
     bool tab_bar_hidden;
     double font_size;

kitty/unicode-data.h

@@ -1,5 +1,6 @@
 #pragma once
 #include "data-types.h"
+#include "state.h"
 #define VS15 1285
 #define VS16 1286
 
@@ -13,6 +14,8 @@ combining_type mark_for_codepoint(char_type c);
 
 static inline bool
 is_url_char(uint32_t ch) {
+    for (size_t i = 0; i < OPT(url_excluded_characters_count); i++)
+        if (ch == OPT(url_excluded_characters)[i]) return false;
     return ch && !is_CZ_category(ch);
 }