Dont strip trailing punctuation from end of URL if next line startswith url characters

Author: kovidgoyal

kitty/line.c

@@ -133,16 +133,24 @@ line_url_start_at(Line *self, index_type x) {
 }
 
 index_type
-line_url_end_at(Line *self, index_type x, bool check_short, char_type sentinel) {
+line_url_end_at(Line *self, index_type x, bool check_short, char_type sentinel, bool next_line_starts_with_url_chars) {
     index_type ans = x;
     if (x >= self->xnum || (check_short && self->xnum <= MIN_URL_LEN + 3)) return 0;
     if (sentinel) { while (ans < self->xnum && self->cpu_cells[ans].ch != sentinel && is_url_char(self->cpu_cells[ans].ch)) ans++; }
     else { while (ans < self->xnum && is_url_char(self->cpu_cells[ans].ch)) ans++; }
     if (ans) ans--;
-    while (ans > x && can_strip_from_end_of_url(self->cpu_cells[ans].ch)) ans--;
+    if (ans < self->xnum - 1 || !next_line_starts_with_url_chars) {
+        while (ans > x && can_strip_from_end_of_url(self->cpu_cells[ans].ch)) ans--;
+    }
     return ans;
 }
 
+bool
+line_startswith_url_chars(Line *self) {
+    return is_url_char(self->cpu_cells[0].ch);
+}
+
+
 static PyObject*
 url_start_at(Line *self, PyObject *x) {
 #define url_start_at_doc "url_start_at(x) -> Return the start cell number for a URL containing x or self->xnum if not found"
@@ -153,8 +161,9 @@ static PyObject*
 url_end_at(Line *self, PyObject *args) {
 #define url_end_at_doc "url_end_at(x) -> Return the end cell number for a URL containing x or 0 if not found"
     unsigned int x, sentinel = 0;
-    if (!PyArg_ParseTuple(args, "I|I", &x, &sentinel)) return NULL;
-    return PyLong_FromUnsignedLong((unsigned long)line_url_end_at(self, x, true, sentinel));
+    int next_line_starts_with_url_chars = 0;
+    if (!PyArg_ParseTuple(args, "I|Ip", &x, &sentinel, &next_line_starts_with_url_chars)) return NULL;
+    return PyLong_FromUnsignedLong((unsigned long)line_url_end_at(self, x, true, sentinel, next_line_starts_with_url_chars));
 }
 
 // }}}

kitty/lineops.h

@@ -61,7 +61,8 @@ void line_set_char(Line *, unsigned int , uint32_t , unsigned int , Cursor *, bo
 void line_right_shift(Line *, unsigned int , unsigned int );
 void line_add_combining_char(Line *, uint32_t , unsigned int );
 index_type line_url_start_at(Line *self, index_type x);
-index_type line_url_end_at(Line *self, index_type x, bool, char_type);
+index_type line_url_end_at(Line *self, index_type x, bool, char_type, bool);
+bool line_startswith_url_chars(Line*);
 index_type line_as_ansi(Line *self, Py_UCS4 *buf, index_type buflen, bool*, const GPUCell**) __attribute__((nonnull));
 unsigned int line_length(Line *self);
 size_t cell_as_unicode(CPUCell *cell, bool include_cc, Py_UCS4 *buf, char_type);

kitty/mouse.c

@@ -227,11 +227,14 @@ extend_url(Screen *screen, Line *line, index_type *x, index_type *y, char_type s
     unsigned int count = 0;
     while(count++ < 10) {
         if (*x != line->xnum - 1) break;
+        bool next_line_starts_with_url_chars = false;
+        line = screen_visual_line(screen, *y + 2);
+        if (line) next_line_starts_with_url_chars = line_startswith_url_chars(line);
         line = screen_visual_line(screen, *y + 1);
         if (!line) break;
         // we deliberately allow non-continued lines as some programs, like
         // mutt split URLs with newlines at line boundaries
-        index_type new_x = line_url_end_at(line, 0, false, sentinel);
+        index_type new_x = line_url_end_at(line, 0, false, sentinel, next_line_starts_with_url_chars);
         if (!new_x) break;
         *y += 1; *x = new_x;
     }
@@ -274,7 +277,15 @@ detect_url(Screen *screen, unsigned int x, unsigned int y) {
     if (line) {
         url_start = line_url_start_at(line, x);
         sentinel = get_url_sentinel(line, url_start);
-        if (url_start < line->xnum) url_end = line_url_end_at(line, x, true, sentinel);
+        if (url_start < line->xnum) {
+            bool next_line_starts_with_url_chars = false;
+            if (y < screen->lines - 1) {
+                line = screen_visual_line(screen, y+1);
+                next_line_starts_with_url_chars = line_startswith_url_chars(line);
+                line = screen_visual_line(screen, y);
+            }
+            url_end = line_url_end_at(line, x, true, sentinel, next_line_starts_with_url_chars);
+        }
         has_url = url_end > url_start;
     }
     if (has_url) {

kitty_tests/datatypes.py

@@ -285,6 +285,10 @@ def no_url(t):
             l4 = create('http://a.b?q=1' + trail)
             self.ae(l4.url_end_at(1), len(l4) - 1)
 
+        l4 = create('http://a.b.')
+        self.ae(l4.url_end_at(0), len(l4) - 2)
+        self.ae(l4.url_end_at(0, 0, True), len(l4) - 1)
+
     def rewrap(self, lb, lb2):
         hb = HistoryBuf(lb2.ynum, lb2.xnum)
         cy = lb.rewrap(lb2, hb)