Fix inconsistent whitespace detection in PDF parser.

Signed-off-by: yzrh <yzrh@noema.org>

1 files changed, 22 insertions, 13 deletions

diff --git a/src/pdf_parser.c b/src/pdf_parser.c
index bb45e63..2585e6f 100644
--- a/src/pdf_parser.c
+++ b/src/pdf_parser.c
@@ -19,26 +19,35 @@ static void *
 _memmem_whitespace(const void *p0, size_t s0, const void *p1, size_t s1)
 {
 	const char whitespace[6] = {
-		' ',
-		'\r',
-		'\n',
-		'\f',
-		'\t',
-		'\0'
+		0x00,
+		0x09,
+		0x0a,
+		0x0c,
+		0x0d,
+		0x20
 	};
 
-	char tmp[s1 + 1];
-	memcpy(tmp, p1, s1);
+	char *ret = NULL;
 
-	char *ret;
+	char str[s1 + 1];
+	memcpy(str, p1, s1);
+
+	size_t tmp_size = 0;
+	char *tmp;
 
 	for (int i = 0; i < 6; i++) {
-		tmp[s1] = whitespace[i];
-		if ((ret = memmem(p0, s0, tmp, s1 + 1)) != NULL)
-			return ret;
+		str[s1] = whitespace[i];
+
+		if ((tmp = memmem(p0, s0, str, s1 + 1)) == NULL)
+			continue;
+
+		if (tmp_size == 0 || (size_t) (tmp - (char *) p0) < tmp_size) {
+			tmp_size = tmp - (char *) p0;
+			ret = tmp;
+		}
 	}
 
-	return NULL;
+	return ret;
 }
 
 static int