Loop through urls on screen and copy to clipboard

Replace url detection heuristics with a DFA, enabling urls that span
multiple lines. Also fix the selection not to use snapping so that urls
are selected exactly.

config.def.h

@@ -216,6 +216,7 @@ static Shortcut shortcuts[] = {
 	{ TERMMOD,              XK_Y,           selpaste,       {.i =  0} },
 	{ ShiftMask,            XK_Insert,      selpaste,       {.i =  0} },
 	{ TERMMOD,              XK_Num_Lock,    numlock,        {.i =  0} },
+	{ MODKEY,               XK_l,           copyurl,        {.i =  0} },
 	{ ShiftMask,            XK_Page_Up,     kscrollup,      {.i = -1} },
 	{ ShiftMask,            XK_Page_Down,   kscrolldown,    {.i = -1} },
 };

st.c

@@ -159,6 +159,11 @@ typedef struct {
 	int narg;              /* nb of args */
 } STREscape;
 
+typedef struct {
+	int state;
+	size_t length;
+} URLdfa;
+
 static void execsh(char *, char **);
 static void stty(char **);
 static void sigchld(int);
@@ -208,6 +213,7 @@ static void tdefutf8(char);
 static int32_t tdefcolor(const int *, int *, int);
 static void tdeftran(char);
 static void tstrsequence(uchar);
+static int daddch(URLdfa *, char);
 
 static void drawregion(int, int, int, int);
 
@@ -2737,3 +2743,90 @@ redraw(void)
 	tfulldirt();
 	draw();
 }
+
+int
+daddch(URLdfa *dfa, char c)
+{
+	/* () and [] can appear in urls, but excluding them here will reduce false
+	 * positives when figuring out where a given url ends.
+	 */
+	static const char URLCHARS[] = "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
+		"abcdefghijklmnopqrstuvwxyz"
+		"0123456789-._~:/?#@!$&'*+,;=%";
+	static const char RPFX[] = "//:sptth";
+
+	if (!strchr(URLCHARS, c)) {
+		dfa->length = 0;
+		dfa->state = 0;
+
+		return 0;
+	}
+
+	dfa->length++;
+
+	if (dfa->state == 2 && c == '/') {
+		dfa->state = 0;
+	} else if (dfa->state == 3 && c == 'p') {
+		dfa->state++;
+	} else if (c != RPFX[dfa->state]) {
+		dfa->state = 0;
+		return 0;
+	}
+
+	if (dfa->state++ == 7) {
+		dfa->state = 0;
+		return 1;
+	}
+
+	return 0;
+}
+
+/*
+** Select and copy the previous url on screen (do nothing if there's no url).
+*/
+void
+copyurl(const Arg *arg) {
+	int row = 0, /* row of current URL */
+		col = 0, /* column of current URL start */
+		colend = 0, /* column of last occurrence */
+		passes = 0; /* how many rows have been scanned */
+
+	const char *c = NULL,
+		 *match = NULL;
+	URLdfa dfa = { 0 };
+
+	row = (sel.ob.x >= 0 && sel.nb.y > 0) ? sel.nb.y : term.bot;
+	LIMIT(row, term.top, term.bot);
+
+	colend = (sel.ob.x >= 0 && sel.nb.y > 0) ? sel.nb.x : term.col;
+	LIMIT(colend, 0, term.col);
+
+	/*
+	** Scan from (term.row - 1,term.col - 1) to (0,0) and find
+	** next occurrance of a URL
+	*/
+	for (passes = 0; passes < term.row; passes++) {
+		/* Read in each column of every row until
+		** we hit previous occurrence of URL
+		*/
+		for (col = colend; col--;)
+			if (daddch(&dfa, term.line[row][col].u < 128 ? term.line[row][col].u : ' '))
+				break;
+
+		if (col >= 0)
+			break;
+
+		if (--row < 0)
+			row = term.row - 1;
+
+		colend = term.col;
+	}
+
+	if (passes < term.row) {
+		selstart(col, row, 0);
+		selextend((col + dfa.length - 1) % term.col, row + (col + dfa.length - 1) / term.col, SEL_REGULAR, 0);
+		selextend((col + dfa.length - 1) % term.col, row + (col + dfa.length - 1) / term.col, SEL_REGULAR, 1);
+		xsetsel(getsel());
+		xclipcopy();
+	}
+}

st.h

@@ -87,6 +87,7 @@ void printscreen(const Arg *);
 void printsel(const Arg *);
 void sendbreak(const Arg *);
 void toggleprinter(const Arg *);
+void copyurl(const Arg *);
 
 int tattrset(int);
 void tnew(int, int);