From 6fb59b3291a63445499f81b609171db0db0d90ee Mon Sep 17 00:00:00 2001
From: Bryce Kwon <bryce@brycekwon.com>
Date: Sun, 1 Nov 2026 11:29:40 -1000
Subject: Dump only the first 64 KB of a binary file

---
 source/ui-tree.c | 22 ++++++++++++++++------
 1 file changed, 16 insertions(+), 6 deletions(-)

(limited to 'source')

diff --git a/source/ui-tree.c b/source/ui-tree.c
index aff4f0f..8e17cc4 100644
--- a/source/ui-tree.c
+++ b/source/ui-tree.c
@@ -21,6 +21,10 @@
 #define HEXDUMP_ROW_HALF (HEXDUMP_ROW_BYTES / 2)
 #define HEXDUMP_GAP_WIDTH 4
 
+// Every byte dumped costs about seven bytes of markup, so a blob near
+// max-blob-size would make a page several times its own size.
+#define HEXDUMP_MAX_BYTES (64 * 1024)
+
 // How many single-child directories a listing row follows into one path.
 #define MAX_DIR_CHAIN_LEVELS 15
 
@@ -101,9 +105,9 @@ static int is_graphic(unsigned char ch)
 	return isprint(ch) && !isspace(ch);
 }
 
-static void print_binary_buffer(char *buf, size_t size)
+static void print_binary_buffer(char *buf, size_t size, const char *path, const char *rev)
 {
-	size_t offset, idx;
+	size_t offset, idx, shown = size < HEXDUMP_MAX_BYTES ? size : HEXDUMP_MAX_BYTES;
 	char ascii[HEXDUMP_ROW_BYTES + 1];
 	struct strbuf row = STRBUF_INIT;
 
@@ -111,10 +115,10 @@ static void print_binary_buffer(char *buf, size_t size)
 	html("<tr><th>ofs</th><th>hex dump</th><th>ascii</th></tr>\n");
 
 	// A row goes out in one write rather than a write per byte.
-	for (offset = 0; offset < size; offset += HEXDUMP_ROW_BYTES, buf += HEXDUMP_ROW_BYTES) {
+	for (offset = 0; offset < shown; offset += HEXDUMP_ROW_BYTES, buf += HEXDUMP_ROW_BYTES) {
 		strbuf_reset(&row);
 		strbuf_addf(&row, "<tr><td class='right'>%04zx</td><td class='hex'>", offset);
-		for (idx = 0; idx < HEXDUMP_ROW_BYTES && offset + idx < size; idx++) {
+		for (idx = 0; idx < HEXDUMP_ROW_BYTES && offset + idx < shown; idx++) {
 			int gap = idx == HEXDUMP_ROW_HALF ? HEXDUMP_GAP_WIDTH : 1;
 
 			strbuf_addf(&row, "%*s%02x", gap, "", (unsigned char)buf[idx]);
@@ -122,7 +126,7 @@ static void print_binary_buffer(char *buf, size_t size)
 		strbuf_addstr(&row, " </td><td class='hex'>");
 		html_raw(row.buf, row.len);
 
-		for (idx = 0; idx < HEXDUMP_ROW_BYTES && offset + idx < size; idx++)
+		for (idx = 0; idx < HEXDUMP_ROW_BYTES && offset + idx < shown; idx++)
 			ascii[idx] = is_graphic(buf[idx]) ? buf[idx] : '.';
 		ascii[idx] = '\0';
 		html_txt(ascii);
@@ -130,6 +134,12 @@ static void print_binary_buffer(char *buf, size_t size)
 	}
 	strbuf_release(&row);
 	html("</table>\n");
+	if (shown < size) {
+		htmlf("<div class='truncated'>Only the first %d KB of this file are shown. ",
+			HEXDUMP_MAX_BYTES / 1024);
+		cgit_plain_link("View the whole file", NULL, NULL, ctx.qry.head, rev, path);
+		html(".</div>\n");
+	}
 }
 
 /*
@@ -179,7 +189,7 @@ static bool print_object(const struct object_id *oid, const char *path,
 	html(")</div>\n");
 
 	if (is_binary)
-		print_binary_buffer(buf, size);
+		print_binary_buffer(buf, size, path, rev);
 	else
 		print_text_buffer(filename, buf, size);
 
-- 
cgit 2.12.1