apply copyurl-multiline patch

5 files changed, 95 insertions, 246 deletions

diff --git a/config.def.h b/config.def.h
index fabcdfa..6525f90 100644
--- a/config.def.h
+++ b/config.def.h
@@ -220,6 +220,7 @@ 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} },
 };
 
 /*
diff --git a/patches/st-copyurl-20190202-0.8.1.diff b/patches/st-copyurl-20190202-0.8.1.diff
deleted file mode 100644
index 8d6782b..0000000
--- a/patches/st-copyurl-20190202-0.8.1.diff
+++ /dev/null
@@ -1,109 +0,0 @@
-From be408247f1c1ff8ccf7ab128b126f54d19bd4392 Mon Sep 17 00:00:00 2001
-From: Michael Buch <michaelbuch12@gmail.com>
-Date: Sat, 2 Feb 2019 14:20:52 +0000
-Subject: [PATCH] Port the copyurl patch to the 0.8.1 st release. Mainly fix 
- usage of depracted selcopy
-
----
- config.def.h |  1 +
- st.c         | 62 ++++++++++++++++++++++++++++++++++++++++++++++++++++
- st.h         |  1 +
- 3 files changed, 64 insertions(+)
-
-diff --git a/config.def.h b/config.def.h
-index 82b1b09..cbe923e 100644
---- a/config.def.h
-+++ b/config.def.h
-@@ -178,6 +178,7 @@ static Shortcut shortcuts[] = {
- 	{ TERMMOD,              XK_Y,           selpaste,       {.i =  0} },
- 	{ TERMMOD,              XK_Num_Lock,    numlock,        {.i =  0} },
- 	{ TERMMOD,              XK_I,           iso14755,       {.i =  0} },
-+	{ MODKEY,               XK_l,           copyurl,        {.i =  0} },
- };
- 
- /*
-diff --git a/st.c b/st.c
-index 46c954b..476eb31 100644
---- a/st.c
-+++ b/st.c
-@@ -2616,3 +2616,65 @@ redraw(void)
- 	tfulldirt();
- 	draw();
- }
-+
-+/* select and copy the previous url on screen (do nothing if there's no url).
-+ * known bug: doesn't handle urls that span multiple lines (wontfix)
-+ * known bug: only finds first url on line (mightfix)
-+ */
-+void
-+copyurl(const Arg *arg) {
-+	/* () and [] can appear in urls, but excluding them here will reduce false
-+	 * positives when figuring out where a given url ends.
-+	 */
-+	static char URLCHARS[] = "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
-+		"abcdefghijklmnopqrstuvwxyz"
-+		"0123456789-._~:/?#@!$&'*+,;=%";
-+
-+	int i, row, startrow;
-+	char *linestr = calloc(sizeof(char), term.col+1); /* assume ascii */
-+	char *c, *match = NULL;
-+
-+	row = (sel.ob.x >= 0 && sel.nb.y > 0) ? sel.nb.y-1 : term.bot;
-+	LIMIT(row, term.top, term.bot);
-+	startrow = row;
-+
-+	/* find the start of the last url before selection */
-+	do {
-+		for (i = 0; i < term.col; ++i) {
-+			if (term.line[row][i].u > 127) /* assume ascii */
-+				continue;
-+			linestr[i] = term.line[row][i].u;
-+		}
-+		linestr[term.col] = '\0';
-+		if ((match = strstr(linestr, "http://"))
-+				|| (match = strstr(linestr, "https://")))
-+			break;
-+		if (--row < term.top)
-+			row = term.bot;
-+	} while (row != startrow);
-+
-+	if (match) {
-+		/* must happen before trim */
-+		selclear();
-+		sel.ob.x = strlen(linestr) - strlen(match);
-+
-+		/* trim the rest of the line from the url match */
-+		for (c = match; *c != '\0'; ++c)
-+			if (!strchr(URLCHARS, *c)) {
-+				*c = '\0';
-+				break;
-+			}
-+
-+		/* select and copy */
-+		sel.mode = 1;
-+		sel.type = SEL_REGULAR;
-+		sel.oe.x = sel.ob.x + strlen(match)-1;
-+		sel.ob.y = sel.oe.y = row;
-+		selnormalize();
-+		tsetdirt(sel.nb.y, sel.ne.y);
-+		xsetsel(getsel());
-+		xclipcopy();
-+	}
-+
-+	free(linestr);
-+}
-diff --git a/st.h b/st.h
-index dac64d8..5a58f8f 100644
---- a/st.h
-+++ b/st.h
-@@ -85,6 +85,7 @@ 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);
--- 
-2.20.1
-
diff --git a/patches/st-copyurl-20220221-0.8.5.diff b/patches/st-copyurl-20220221-0.8.5.diff
deleted file mode 100644
index 5d914c4..0000000
--- a/patches/st-copyurl-20220221-0.8.5.diff
+++ /dev/null
@@ -1,137 +0,0 @@
-From 897c3958d01d8df80ebf1666b972b8a658b419ba Mon Sep 17 00:00:00 2001
-From: Santtu Lakkala <inz@inz.fi>
-Date: Wed, 16 Feb 2022 20:34:20 +0200
-Subject: [PATCH] Loop through urls on screen and copy to clipboard
-
-Based on the previous highlighting patches, slightly simplified and
-fixes graphical issues with mixed copyurl and selection.
----
- config.def.h |  1 +
- st.c         | 82 ++++++++++++++++++++++++++++++++++++++++++++++++++++
- st.h         |  1 +
- 3 files changed, 84 insertions(+)
-
-diff --git a/config.def.h b/config.def.h
-index 91ab8ca..3f365c7 100644
---- a/config.def.h
-+++ b/config.def.h
-@@ -201,6 +201,7 @@ 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} },
- };
- 
- /*
-diff --git a/st.c b/st.c
-index 51049ba..bf3d81a 100644
---- a/st.c
-+++ b/st.c
-@@ -200,6 +200,7 @@ static void tdefutf8(char);
- static int32_t tdefcolor(const int *, int *, int);
- static void tdeftran(char);
- static void tstrsequence(uchar);
-+static const char *findlastany(const char *, const char**, size_t);
- 
- static void drawregion(int, int, int, int);
- 
-@@ -2688,3 +2689,84 @@ redraw(void)
- 	tfulldirt();
- 	draw();
- }
-+
-+const char *
-+findlastany(const char *str, const char**find, size_t len)
-+{
-+	const char *found = NULL;
-+	int i = 0;
-+
-+	for (found = str + strlen(str) - 1; found >= str; --found) {
-+		for(i = 0; i < len; i++) {
-+			if (strncmp(found, find[i], strlen(find[i])) == 0) {
-+				return found;
-+			}
-+		}
-+	}
-+
-+	return NULL;
-+}
-+
-+/*
-+** Select and copy the previous url on screen (do nothing if there's no url).
-+**
-+** FIXME: doesn't handle urls that span multiple lines; will need to add support
-+**        for multiline "getsel()" first
-+*/
-+void
-+copyurl(const Arg *arg) {
-+	/* () 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* URLSTRINGS[] = {"http://", "https://"};
-+
-+	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 */
-+
-+	char linestr[term.col + 1];
-+	const char *c = NULL,
-+		 *match = NULL;
-+
-+	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 = 0; col < colend; ++col)
-+			linestr[col] = term.line[row][col].u < 128 ? term.line[row][col].u : ' ';
-+		linestr[col] = '\0';
-+
-+		if ((match = findlastany(linestr, URLSTRINGS,
-+						sizeof(URLSTRINGS)/sizeof(URLSTRINGS[0]))))
-+			break;
-+
-+		if (--row < 0)
-+			row = term.row - 1;
-+
-+		colend = term.col;
-+	};
-+
-+	if (match) {
-+		size_t l = strspn(match, URLCHARS);
-+		selstart(match - linestr, row, 0);
-+		selextend(match - linestr + l - 1, row, SEL_REGULAR, 0);
-+		selextend(match - linestr + l - 1, row, SEL_REGULAR, 1);
-+		xsetsel(getsel());
-+		xclipcopy();
-+	}
-+}
-diff --git a/st.h b/st.h
-index 519b9bd..0458005 100644
---- a/st.h
-+++ b/st.h
-@@ -85,6 +85,7 @@ 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);
--- 
-2.32.0
-
diff --git a/st.c b/st.c
index 2ed0094..130b4eb 100644
--- a/st.c
+++ b/st.c
@@ -152,6 +152,11 @@ 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);
@@ -200,6 +205,7 @@ 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);
 
@@ -2656,3 +2662,90 @@ 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();
+	}
+}
diff --git a/st.h b/st.h
index 9ed9f4e..f085e62 100644
--- a/st.h
+++ b/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);