tui: render inline images inside tool boxes

Two fixes for image-block rendering after the box refactor.

clipBottomClippedImages decides whether to suppress an image escape
this frame by scanning forward for the matching "image - ..." info
line, treating any non-blank intervening row as a blocker. The image
block layout is: escape row + N blank reservation rows + info row.
With tool boxes, the reservation rows are wrapped as "\u2502    \u2026    \u2502"
and the naive whitespace check now sees them as non-blank \u2014 the scan
broke before reaching the info line, decided the image was clipped,
and zeroed out the escape sequence. Result: a tall blank gap inside
the box where the image should be, with only the metadata caption
visible at the bottom (visible on iTerm2 / Kitty / WezTerm).

Add isBoxBlankLine() that strips ANSI escapes, surrounding whitespace,
and \u2502 box edges before checking for emptiness. Use it in both
clipBottomClippedImages and snapViewportStartToImageBlock so the
viewport snap-back also recognises wrapped reservation rows.

Also add one extra blank reservation row between the image footprint
and the muted "image - mime - WxH - sizeKB" caption, so the caption
doesn't sit flush against the last pixel row of the image.

commit: 998a062a20a1f726654f7383d858482530137237
author: patriceckhart · 5mo ago · 2026-04-28 06:11 UTC
parent: 6513559
signature: unsigned
refs: tag: v0.1.33

2 files · +57 -6

M internal/agent/modes/interactive.go +51 -3


@@ -985,7 +985,7 @@ func snapViewportStartToImageBlock(chat []string, start int) int {
985	985		if start <= 0 \|\| start >= len(chat) {
986	986		return start
987	987		}
988		-	if hasImageEscape(chat[start]) \|\| strings.TrimSpace(chat[start]) != "" {
	988	+	if hasImageEscape(chat[start]) \|\| !isBoxBlankLine(chat[start]) {
989	989		return start
990	990		}
991	991		for k := start - 1; k >= 0; k-- {

@@ -993,7 +993,7 @@ func snapViewportStartToImageBlock(chat []string, start int) int {
993	993		if hasImageEscape(line) {
994	994		return k
995	995		}
996		-	if strings.TrimSpace(line) != "" {
	996	+	if !isBoxBlankLine(line) {
997	997		break
998	998		}
999	999		}

@@ -1041,13 +1041,18 @@ func clipBottomClippedImages(lines []string) []string {
1041	1041		// then one trailing blank. If the info line isn't visible in
1042	1042		// the current chat slice, the image would paint down into the
1043	1043		// fixed status bar area. Suppress that image for this frame.
	1044	+	//
	1045	+	// When the image lives inside a tool box, the reservation rows
	1046	+	// are wrapped in vertical box edges ("│ … │"); those rows
	1047	+	// look non-blank under a naive whitespace check but are still
	1048	+	// reservation rows for this scan, so treat them as blank.
1044	1049		foundInfo := false
1045	1050		for j := i + 1; j < len(out); j++ {
1046	1051		if strings.Contains(out[j], "image - ") {
1047	1052		foundInfo = true
1048	1053		break
1049	1054		}
1050		-	if strings.TrimSpace(out[j]) != "" {
	1055	+	if !isBoxBlankLine(out[j]) {
1051	1056		break
1052	1057		}
1053	1058		}

@@ -1058,6 +1063,49 @@ func clipBottomClippedImages(lines []string) []string {
1058	1063		return out
1059	1064		}
1060	1065
	1066	+	// isBoxBlankLine reports whether line is visually empty after
	1067	+	// stripping ANSI escape sequences, surrounding whitespace, and the
	1068	+	// vertical box edges drawn by the tool-box renderer. Used by
	1069	+	// clipBottomClippedImages so an image's reservation rows still count
	1070	+	// as blank when those rows are wrapped in "│ … │" inside a tool box.
	1071	+	func isBoxBlankLine(line string) bool {
	1072	+	stripped := stripANSIBytes(line)
	1073	+	stripped = strings.TrimSpace(stripped)
	1074	+	stripped = strings.Trim(stripped, "│")
	1075	+	stripped = strings.TrimSpace(stripped)
	1076	+	return stripped == ""
	1077	+	}
	1078	+
	1079	+	// stripANSIBytes removes ANSI CSI escape sequences (ESC '[' … final
	1080	+	// byte) from s without pulling in the regexp package. Mirrors the
	1081	+	// internal helper in package tui; the duplicated copy avoids exporting
	1082	+	// it just for one caller.
	1083	+	func stripANSIBytes(s string) string {
	1084	+	if !strings.Contains(s, "\x1b") {
	1085	+	return s
	1086	+	}
	1087	+	var b strings.Builder
	1088	+	b.Grow(len(s))
	1089	+	i := 0
	1090	+	for i < len(s) {
	1091	+	if s[i] == 0x1b && i+1 < len(s) && s[i+1] == '[' {
	1092	+	end := i + 2
	1093	+	for end < len(s) {
	1094	+	c := s[end]
	1095	+	end++
	1096	+	if c >= 0x40 && c <= 0x7e {
	1097	+	break
	1098	+	}
	1099	+	}
	1100	+	i = end
	1101	+	continue
	1102	+	}
	1103	+	b.WriteByte(s[i])
	1104	+	i++
	1105	+	}
	1106	+	return b.String()
	1107	+	}
	1108	+
1061	1109		// truncateLine shortens s so it fits within n display cells, with an
1062	1110		// ellipsis if trimmed. Used by the "sliding in" chips so a pasted
1063	1111		// novel doesn't blow past the status line.


@@ -985,7 +985,7 @@ func snapViewportStartToImageBlock(chat []string, start int) int {
985		if start <= 0 \|\| start >= len(chat) {	985		if start <= 0 \|\| start >= len(chat) {
986		return start	986		return start
987		}	987		}
988	-	if hasImageEscape(chat[start]) \|\| strings.TrimSpace(chat[start]) != "" {	988	+	if hasImageEscape(chat[start]) \|\| !isBoxBlankLine(chat[start]) {
989		return start	989		return start
990		}	990		}
991		for k := start - 1; k >= 0; k-- {	991		for k := start - 1; k >= 0; k-- {

@@ -993,7 +993,7 @@ func snapViewportStartToImageBlock(chat []string, start int) int {
993		if hasImageEscape(line) {	993		if hasImageEscape(line) {
994		return k	994		return k
995		}	995		}
996	-	if strings.TrimSpace(line) != "" {	996	+	if !isBoxBlankLine(line) {
997		break	997		break
998		}	998		}
999		}	999		}

@@ -1041,13 +1041,18 @@ func clipBottomClippedImages(lines []string) []string {
1041		// then one trailing blank. If the info line isn't visible in	1041		// then one trailing blank. If the info line isn't visible in
1042		// the current chat slice, the image would paint down into the	1042		// the current chat slice, the image would paint down into the
1043		// fixed status bar area. Suppress that image for this frame.	1043		// fixed status bar area. Suppress that image for this frame.
			1044	+	//
			1045	+	// When the image lives inside a tool box, the reservation rows
			1046	+	// are wrapped in vertical box edges ("│ … │"); those rows
			1047	+	// look non-blank under a naive whitespace check but are still
			1048	+	// reservation rows for this scan, so treat them as blank.
1044		foundInfo := false	1049		foundInfo := false
1045		for j := i + 1; j < len(out); j++ {	1050		for j := i + 1; j < len(out); j++ {
1046		if strings.Contains(out[j], "image - ") {	1051		if strings.Contains(out[j], "image - ") {
1047		foundInfo = true	1052		foundInfo = true
1048		break	1053		break
1049		}	1054		}
1050	-	if strings.TrimSpace(out[j]) != "" {	1055	+	if !isBoxBlankLine(out[j]) {
1051		break	1056		break
1052		}	1057		}
1053		}	1058		}

@@ -1058,6 +1063,49 @@ func clipBottomClippedImages(lines []string) []string {
1058		return out	1063		return out
1059		}	1064		}
1060			1065
			1066	+	// isBoxBlankLine reports whether line is visually empty after
			1067	+	// stripping ANSI escape sequences, surrounding whitespace, and the
			1068	+	// vertical box edges drawn by the tool-box renderer. Used by
			1069	+	// clipBottomClippedImages so an image's reservation rows still count
			1070	+	// as blank when those rows are wrapped in "│ … │" inside a tool box.
			1071	+	func isBoxBlankLine(line string) bool {
			1072	+	stripped := stripANSIBytes(line)
			1073	+	stripped = strings.TrimSpace(stripped)
			1074	+	stripped = strings.Trim(stripped, "│")
			1075	+	stripped = strings.TrimSpace(stripped)
			1076	+	return stripped == ""
			1077	+	}
			1078	+
			1079	+	// stripANSIBytes removes ANSI CSI escape sequences (ESC '[' … final
			1080	+	// byte) from s without pulling in the regexp package. Mirrors the
			1081	+	// internal helper in package tui; the duplicated copy avoids exporting
			1082	+	// it just for one caller.
			1083	+	func stripANSIBytes(s string) string {
			1084	+	if !strings.Contains(s, "\x1b") {
			1085	+	return s
			1086	+	}
			1087	+	var b strings.Builder
			1088	+	b.Grow(len(s))
			1089	+	i := 0
			1090	+	for i < len(s) {
			1091	+	if s[i] == 0x1b && i+1 < len(s) && s[i+1] == '[' {
			1092	+	end := i + 2
			1093	+	for end < len(s) {
			1094	+	c := s[end]
			1095	+	end++
			1096	+	if c >= 0x40 && c <= 0x7e {
			1097	+	break
			1098	+	}
			1099	+	}
			1100	+	i = end
			1101	+	continue
			1102	+	}
			1103	+	b.WriteByte(s[i])
			1104	+	i++
			1105	+	}
			1106	+	return b.String()
			1107	+	}
			1108	+
1061		// truncateLine shortens s so it fits within n display cells, with an	1109		// truncateLine shortens s so it fits within n display cells, with an
1062		// ellipsis if trimmed. Used by the "sliding in" chips so a pasted	1110		// ellipsis if trimmed. Used by the "sliding in" chips so a pasted
1063		// novel doesn't blow past the status line.	1111		// novel doesn't blow past the status line.

M internal/tui/view.go +6 -3


@@ -1149,14 +1149,17 @@ func (v *View) renderImageBlock(b provider.ImageBlock, width int) []string {
1149	1149		}
1150	1150		// Reserve the image footprint first, then place the escape
1151	1151		// in the top-left of that blank rectangle. Keeping the
1152		-	// metadata above the image made some terminals blend text and
1153		-	// graphics layers during full-screen repaints; metadata below
1154		-	// the reserved rectangle is more stable and easier to read.
	1152	+	// metadata above the image made some terminals blend text
	1153	+	// and graphics layers during full-screen repaints; metadata
	1154	+	// below the reserved rectangle is more stable and easier to
	1155	+	// read. One extra blank row before the metadata gives the
	1156	+	// caption breathing room from the image's last pixel row.
1155	1157		out := make([]string, 0, rows+3)
1156	1158		out = append(out, " "+seq)
1157	1159		for i := 1; i < rows; i++ {
1158	1160		out = append(out, "")
1159	1161		}
	1162	+	out = append(out, "")
1160	1163		out = append(out, v.Theme.FG256(v.Theme.Muted, info))
1161	1164		return out
1162	1165		}

 			}
 			// Reserve the image footprint first, then place the escape
 			// in the top-left of that blank rectangle. Keeping the
-			// metadata above the image made some terminals blend text and
+			// metadata above the image made some terminals blend text
-			// graphics layers during full-screen repaints; metadata below
+			// and graphics layers during full-screen repaints; metadata
-			// the reserved rectangle is more stable and easier to read.
+			// below the reserved rectangle is more stable and easier to
+			// read. One extra blank row before the metadata gives the
+			// caption breathing room from the image's last pixel row.
 			out := make([]string, 0, rows+3)
 			out = append(out, "    "+seq)
 			for i := 1; i < rows; i++ {
 				out = append(out, "")
 			}
+			out = append(out, "")
 			out = append(out, v.Theme.FG256(v.Theme.Muted, info))
 			return out
 		}