checksrc

commit: 6a113122f8730963d35260e4d92ca7754fff9f2f
parent eb2baa335d05b790185a8d8752804454a4d07fd9
Author: Haelwenn (lanodan) Monnier <contact@hacktivis.me>
Date:   Wed, 24 Sep 2025 05:55:41 +0200

check for dumps

Diffstat:
M
README.md
12
+++++++-----
M
checksrc.c
52
++++++++++++++++++++++++++++++++++++++++++++++------

2 files changed, 53 insertions(+), 11 deletions(-)
diff --git a/README.md b/README.md
@@ -46,18 +46,20 @@ of the file looks like regular code.
 So `checksrc` can take much longer depending on the payloads.
 
 ## Detections
+### Minor
+Throws a warning but doesn't stops reading the file:
+
+- dump: block of 10 consecutive lines with lengths varying by less than 3 bytes (to detect hex dumps, base64, …)
+- (planned) more punctuation, symbols, and numbers than letters [a-zA-Z\x7C-\xFF] (within 4KB blocks)
+
 ### Major
 Throws an error, stops reading the file, exits unsuccessfully:
 
 - minified code: average line length of more than 100 characters (within 4KB blocks)
 - non-printable character (byte under 0x20 other than `\n`, `\r`, `\t`)
+- 3 blocks of dump (see Minor)
 - (planned) string indicating generated code
 
-### Minor
-Throws a warning but doesn't stops reading the file:
-
-- (planned) more than 10 consecutive lines with lengths varying by less than 2 bytes (to detect hex dumps, base64, …)
-- (planned) more punctuation, symbols, and numbers than letters [a-zA-Z\x7C-\xFF] (within 4KB blocks)
 
 ```
 SPDX-FileCopyrightText: 2017 Haelwenn (lanodan) Monnier <contact+checksrc@hacktivis.me>
diff --git a/checksrc.c b/checksrc.c
@@ -41,7 +41,10 @@ checkfile(const char *fname)
 		return -1;
 	}
 
-	size_t chars = 0, newlines = 0;
+	size_t lines_len[10];
+	size_t chars = 0, line = 0;
+	size_t last_dump = 0;
+	int dumps = 0;
 	for(;;)
 	{
 		static char buf[4096];
@@ -63,23 +66,60 @@ checkfile(const char *fname)
 			if(buf[i] == '\t')
 			{
 				chars++;
+				lines_len[line % 10]++;
 				continue;
 			}
 
 			if(buf[i] == '\n')
 			{
-				newlines++;
+				if(line - last_dump > 10)
+				{
+					size_t maxdiff = 0;
+					for(size_t li = 1; li < 10; li++)
+					{
+						size_t diff = 0;
+						if(lines_len[li - 1] > lines_len[li])
+							diff = lines_len[li - 1] - lines_len[li];
+						else
+							diff = lines_len[li] - lines_len[li - 1];
+
+						if(diff > maxdiff) maxdiff = diff;
+					}
+
+					if(maxdiff < 3)
+					{
+						last_dump = line;
+						printf("%s:[%zu - %zu] line length varied only by a maximum of %zu characters\n",
+						       fname,
+						       line - 10,
+						       line,
+						       maxdiff);
+
+						if(++dumps == 3)
+						{
+							printf("%s: Found 3 dumps, stopping there\n", fname);
+							return 1;
+						}
+					}
+				}
+
+				line++;
+				lines_len[line % 10] = 0;
 				continue;
 			}
 
 			if(buf[i] < ' ')
 			{
-				printf("binary (byte 0x%X found at position 0x%zX): %s\n", buf[i], i, fname);
+				printf("%s: binary (byte 0x%X found at position 0x%zX)\n", fname, buf[i], i);
 				return 1;
 			}
 
 			// somewhat UTF-8 centric
-			if(buf[i] < 0x7F) chars++;
+			if(buf[i] < 0x7F)
+			{
+				chars++;
+				lines_len[line % 10]++;
+			}
 		}
 	}
 
@@ -89,10 +129,10 @@ checkfile(const char *fname)
 
 	// Consider that lines are on average shorter than 100 character
 	// One false-positive being formats like SVG, fine for this tool
-	size_t lineavg = chars / newlines;
+	size_t lineavg = chars / line;
 	if(chars > 200 && lineavg > 100)
 	{
-		printf("minified (%zd characters / %zd newlines = %zd): %s\n", chars, newlines, lineavg, fname);
+		printf("%s: minified (%zd characters / %zd newlines = %zd)\n", fname, chars, line, lineavg);
 		return 1;
 	}