Replace memmove with pointer arithmetic

warc-dump-responses.c

@@ -27,6 +27,7 @@ int main(int argc, char* argv[]) {
 	char buf[2 * BUFSIZE];
 	size_t n;
 	int state = STATE_BEFORE_RECORD;
+	char* bufp;
 	char* m0;
 	char* m1;
 	size_t record_bytes_read;
@@ -34,8 +35,9 @@ int main(int argc, char* argv[]) {
 	size_t nscan;
 
 	while ((n = fread(buf, 1, BUFSIZE, stdin)) > 0) {
+		bufp = buf;
 checkstate:
-		DEBUG_PRINTF("Have %zu bytes of buffer (at %p)\n", n, (void*)&buf);
+		DEBUG_PRINTF("Have %zu bytes of buffer (at %p)\n", n, (void*)bufp);
 		if (n == 0) {
 			break;
 		}
@@ -45,50 +47,50 @@ checkstate:
 				fprintf(stderr, "Error: too little data before WARC headers\n");
 				return 1;
 			}
-			if (memcmp(buf, "WARC/1.0\r\n", 10) == 0 || memcmp(buf, "WARC/1.1\r\n", 10) == 0) {
+			if (memcmp(bufp, "WARC/1.0\r\n", 10) == 0 || memcmp(bufp, "WARC/1.1\r\n", 10) == 0) {
 				// Got some headers; find the record type, content length, and end of headers
-				m0 = memmem(buf, n, "\r\nContent-Length:", 17);
+				m0 = memmem(bufp, n, "\r\nContent-Length:", 17);
 				if (!m0) {
 					fprintf(stderr, "Error: Content-Length missing\n");
 					return 1;
 				}
-				DEBUG_PRINTF("Found Content-Length header at %p (offset %zu)\n", (void*)(m0 + 2), m0 + 2 - buf);
-				m1 = memmem(m0 + 1, n - (m0 + 1 - buf), "\r\n", 2);
+				DEBUG_PRINTF("Found Content-Length header at %p (offset %zu)\n", (void*)(m0 + 2), m0 + 2 - bufp);
+				m1 = memmem(m0 + 1, n - (m0 + 1 - bufp), "\r\n", 2);
 				if (!m1) {
 					fprintf(stderr, "Error: CRLF after Content-Length missing\n");
 					return 1;
 				}
 				m0 += 17;
-				while (m0 < buf + n && (*m0 == ' ' || *m0 == '\t')) ++m0;
+				while (m0 < bufp + n && (*m0 == ' ' || *m0 == '\t')) ++m0;
 				if (!sscanf(m0, "%zu%n", &record_length, &nscan)) {
 					fprintf(stderr, "Error: invalid Content-Length\n");
 					return 1;
 				}
-				if (nscan > n - (m0 - buf)) {
+				if (nscan > n - (m0 - bufp)) {
 					fprintf(stderr, "Error: buffer overread\n");
 					return 1;
 				}
 				m0 += nscan;
-				while (m0 < buf + n && (*m0 == ' ' || *m0 == '\t')) ++m0;
+				while (m0 < bufp + n && (*m0 == ' ' || *m0 == '\t')) ++m0;
 				if (m0 != m1) {
 					fprintf(stderr, "Error: invalid Content-Length (noise before EOL)\n");
 					return 1;
 				}
 				DEBUG_PRINTF("Record body length: %zu\n", record_length);
 
-				m0 = memmem(buf, n, "\r\nWARC-Type:", 12);
+				m0 = memmem(bufp, n, "\r\nWARC-Type:", 12);
 				if (!m0) {
 					fprintf(stderr, "Error: WARC-Type missing\n");
 					return 1;
 				}
-				DEBUG_PRINTF("Found WARC-Type header at %p (offset %zu)\n", (void*)(m0 + 2), m0 + 2 - buf);
-				m1 = memmem(m0, n - (m0 - buf), "\r\n", 2);
+				DEBUG_PRINTF("Found WARC-Type header at %p (offset %zu)\n", (void*)(m0 + 2), m0 + 2 - bufp);
+				m1 = memmem(m0, n - (m0 - bufp), "\r\n", 2);
 				if (!m1) {
 					fprintf(stderr, "Error: CRLF after WARC-Type missing\n");
 					return 1;
 				}
 				m0 += 12;
-				while (m0 < buf + n && (*m0 == ' ' || *m0 == '\t')) ++m0;
+				while (m0 < bufp + n && (*m0 == ' ' || *m0 == '\t')) ++m0;
 				if (memcmp(m0, "response", 8) == 0) {
 					DEBUG_PRINTF("Response record\n");
 					state = STATE_RESPONSE_RECORD;
@@ -97,18 +99,17 @@ checkstate:
 					state = STATE_OTHER_RECORD;
 				}
 
-				m0 = memmem(buf, n, "\r\n\r\n", 4);
+				m0 = memmem(bufp, n, "\r\n\r\n", 4);
 				if (!m0) {
 					fprintf(stderr, "Error: end of headers not found\n");
 					return 1;
 				}
 				m0 += 4;
-				DEBUG_PRINTF("Record body begins at %p (offset %zu)\n", (void*)m0, m0 - buf);
+				DEBUG_PRINTF("Record body begins at %p (offset %zu)\n", (void*)m0, m0 - bufp);
 
-				//TODO Replace all the memmove business with pointer logic to avoid needless constant memory copying (is more wrooom).
-				DEBUG_PRINTF("Moving %zu bytes backwards by %zu\n", n - (m0 - buf), m0 - buf);
-				memmove(buf, m0, n - (m0 - buf));
-				n = n - (m0 - buf);
+				DEBUG_PRINTF("Adjusting buffer pointer and n by %zu\n", m0 - bufp);
+				n = n - (m0 - bufp);
+				bufp = m0;
 				record_bytes_read = 0;
 				goto checkstate;
 			} else {
@@ -121,7 +122,7 @@ checkstate:
 				DEBUG_PRINTF("Partial record\n");
 				if (state == STATE_RESPONSE_RECORD) {
 					DEBUG_PRINTF("Copying %zu bytes to stdout\n", n);
-					fwrite(buf, 1, n, stdout);
+					fwrite(bufp, 1, n, stdout);
 				}
 				record_bytes_read += n;
 				DEBUG_PRINTF("%zu of %zu bytes from this record written\n", record_bytes_read, record_length);
@@ -130,18 +131,20 @@ checkstate:
 				DEBUG_PRINTF("Full record\n");
 				if (state == STATE_RESPONSE_RECORD) {
 					DEBUG_PRINTF("Copying %zu bytes to stdout\n", record_length - record_bytes_read);
-					fwrite(buf, 1, record_length - record_bytes_read, stdout);
+					fwrite(bufp, 1, record_length - record_bytes_read, stdout);
 					fprintf(stdout, "\n");
 				}
-				if (memcmp(buf + record_length - record_bytes_read, "\r\n\r\n", 4) != 0) {
+				if (memcmp(bufp + record_length - record_bytes_read, "\r\n\r\n", 4) != 0) {
 					fprintf(stderr, "Error: end of block not found\n");
 					return 1;
 				}
-				DEBUG_PRINTF("Moving %zu bytes backwards by %zu\n", n - (record_length + 4 - record_bytes_read), record_length + 4 - record_bytes_read);
-				memmove(buf, buf + record_length + 4 - record_bytes_read, n - (record_length + 4 - record_bytes_read));
+				DEBUG_PRINTF("Adjusting buffer pointer and n by %zu\n", record_length + 4 - record_bytes_read);
 				n = n - (record_length + 4 - record_bytes_read);
+				bufp = bufp + record_length + 4 - record_bytes_read;
 				if (n < BUFSIZE) {
-					DEBUG_PRINTF("Refilling buffer\n");
+					DEBUG_PRINTF("Buffer too small (%zu bytes), moving and refilling\n", n);
+					memmove(buf, bufp, n);
+					bufp = buf;
 					n += fread(buf + n, 1, BUFSIZE, stdin);
 				}
 				state = STATE_BEFORE_RECORD;