From: Geoffrey Allott <geoffrey@allott.email>
Date: Sun, 25 Sep 2022 16:53:54 +0000 (+0100)
Subject: output aux data with a flag
X-Git-Url: https://git.pointlesshacks.com/?a=commitdiff_plain;h=c38e580616434546aa79b765e386d5ba9b47b7d6;p=tANS.git

output aux data with a flag
---

diff --git a/src/stree.c b/src/stree.c
index 77aa7f1..66116aa 100644
--- a/src/stree.c
+++ b/src/stree.c
@@ -65,6 +65,17 @@ static void node_split_son(struct node *self, struct node *edge, struct node *br
         self->son = split;
 }
 
+static size_t node_count_sons(const struct node *self)
+{
+    size_t count = 0;
+    const struct node *node;
+
+    for (node = self->son; node; node = node->brother)
+        ++count;
+
+    return count;
+}
+
 static void node_sort_sons(struct node *self)
 {
     struct node *node, *brother, *tmp;
@@ -156,18 +167,6 @@ static size_t stree_max_size(size_t len)
     return len * 2 + 1;
 }
 
-static uint8_t stree_aux(size_t rem)
-{
-    if (rem <= 0) return 0;
-    if (rem <= 1) return 1;
-    if (rem <= 2) return 2;
-    if (rem <= 3) return 3;
-    if (rem <= 4) return 4;
-    if (rem <= 5) return 5;
-    if (rem <= 6) return 6;
-    return 7;
-}
-
 int stree_encode(size_t len, const uint8_t *in, uint8_t *out, uint8_t *aux)
 {
     struct node *nodes;
@@ -193,7 +192,7 @@ int stree_encode(size_t len, const uint8_t *in, uint8_t *out, uint8_t *aux)
     active_node = root;
 
     for (i = 0; i < len; ++i) {
-        aux[i] = stree_aux(rem);
+        aux[i] = active_len == 0 ? 1 : 0;
         prev = (struct node *) 0;
         ++rem;
         memset(seen, 0, sizeof seen);
@@ -285,12 +284,11 @@ int stree_decode(size_t len, const uint8_t *in, uint8_t *out, uint8_t *aux)
     active_node = root;
 
     for (i = 0; i < len; ++i) {
-        aux[i] = stree_aux(rem);
+        aux[i] = active_len == 0 ? 1 : 0;
         prev = (struct node *) 0;
         ++rem;
         memset(seen, 0, sizeof seen);
         code = in[i];
-
         if (active_len == 0)
             active_edge = node_edge_decode(active_node, out, &brother, seen, &code);
         else
diff --git a/src/tANS.c b/src/tANS.c
index 00a0999..db759f4 100644
--- a/src/tANS.c
+++ b/src/tANS.c
@@ -25,7 +25,7 @@ ref: https://arxiv.org/abs/1311.2540
 static void usage(void)
 {
     printf(
-        "usage: tANS [-hcdstz] [-S .suf] [file...]\n"
+        "usage: tANS [-hcdstxz] [-S .suf] [file...]\n"
         "\n"
         "Compress the given files using suffix trees and tabled Asymmetric Numeral Systems\n"
         "\n"
@@ -33,6 +33,7 @@ static void usage(void)
         "    -c - Send output to stdout\n"
         "    -d - Decompress\n"
         "    -s - Perform suffix tree transformation only\n"
+        "    -x - Output the suffix tree auxilliary data\n"
         "    -t - Perform tANS-rl (tabled Asymmetric Numeral Systems with run-length encoding) compression only\n"
         "    -z - Compress (default)\n"
         "\n"
@@ -447,10 +448,45 @@ fail:
     return -1;
 }
 
+static int stree_aux_data(FILE *input, FILE *output)
+{
+    uint32_t len;
+    uint8_t *read_buf;
+    uint8_t *enc_buf;
+    uint8_t *aux_buf;
+    uint32_t read_sz = STREE_READSZ;
+
+    read_buf = malloc(read_sz);
+    enc_buf = malloc(read_sz);
+    aux_buf = malloc(read_sz);
+
+    if (!read_buf || !enc_buf || !aux_buf) goto fail;
+
+    do {
+        len = (uint32_t) fread(read_buf, 1, read_sz, input);
+        if (stree_encode(len, read_buf, enc_buf, aux_buf) != 0) goto fail;
+        if (len > 0 && fwrite(aux_buf, len, 1, output) != 1) goto fail;
+    } while (len > 0);
+
+    if (ferror(input)) goto fail;
+
+    free(read_buf);
+    free(enc_buf);
+    free(aux_buf);
+    return 0;
+
+fail:
+    free(read_buf);
+    free(enc_buf);
+    free(aux_buf);
+    return -1;
+}
+
 enum mode {
     mode_both,
     mode_stree_only,
     mode_tans_only,
+    mode_aux_data,
 };
 
 static int compress_file(FILE *input, FILE *output, enum mode mode)
@@ -462,6 +498,8 @@ static int compress_file(FILE *input, FILE *output, enum mode mode)
         return stree_compress_file(input, output);
     case mode_tans_only:
         return tANS_compress_file(input, output);
+    case mode_aux_data:
+        return stree_aux_data(input, output);
     default:
         return -1;
     }
@@ -476,6 +514,9 @@ static int decompress_file(FILE *input, FILE *output, enum mode mode)
         return stree_decompress_file(input, output);
     case mode_tans_only:
         return tANS_decompress_file(input, output);
+    case mode_aux_data:
+        fprintf(stderr, "tANS: aux_data cannot be decompressed\n");
+        return -1;
     default:
         return -1;
     }
@@ -490,7 +531,7 @@ int main(int argc, char *argv[])
     const char *suffix = (const char *) 0;
     char outpath[1024];
 
-    while ((opt = getopt(argc, argv, "hcdstzS:")) != -1) {
+    while ((opt = getopt(argc, argv, "hcdstxzS:")) != -1) {
         switch (opt) {
         case 'h':
             usage();
@@ -509,6 +550,10 @@ int main(int argc, char *argv[])
             mode = mode_tans_only;
             default_suffix = ".tans";
             break;
+        case 'x':
+            mode = mode_aux_data;
+            default_suffix = ".aux";
+            break;
         case 'z':
             compress = 1;
             break;
diff --git a/test/test_stree.c b/test/test_stree.c
index 5ba1575..b55e5fa 100644
--- a/test/test_stree.c
+++ b/test/test_stree.c
@@ -22,10 +22,6 @@ static enum test_result test_stree_encode_simple(void)
     ASSERT_EQ('b', out[1]);
     ASSERT_EQ('c', out[2]);
 
-    ASSERT_EQ(0, aux[0]);
-    ASSERT_EQ(0, aux[1]);
-    ASSERT_EQ(0, aux[2]);
-
     return TEST_SUCCESS;
 }
 
@@ -1275,10 +1271,6 @@ static enum test_result test_stree_decode_simple(void)
     ASSERT_EQ('b', out[1]);
     ASSERT_EQ('c', out[2]);
 
-    ASSERT_EQ(0, aux[0]);
-    ASSERT_EQ(0, aux[1]);
-    ASSERT_EQ(0, aux[2]);
-
     return TEST_SUCCESS;
 }