urbit · pkova · Jul 16, 2025 · Feb 14, 2024 · Feb 14, 2024 · Apr 11, 2024
diff --git a/pkg/noun/build.zig b/pkg/noun/build.zig
@@ -325,6 +325,7 @@ const c_source_files = [_][]const u8{
     "jets/e/tape.c",
     "jets/e/trip.c",
     "jets/e/urwasm.c",
+    "jets/e/zlib.c",
     "jets/f/cell.c",
     "jets/f/comb.c",
     "jets/f/cons.c",

diff --git a/pkg/noun/jets/136/tree.c b/pkg/noun/jets/136/tree.c
@@ -125,6 +125,14 @@ static u3j_core _136_hex_checksum_d[] =
   };
 
 
+static u3j_harm _136_hex__decompress_zlib_a[] = {{".2", u3we_decompress_zlib}, {}};
+static u3j_harm _136_hex__decompress_gzip_a[] = {{".2", u3we_decompress_gzip}, {}};
+static u3j_core _136_hex__zlib_d[] = {
+  {"decompress-zlib", 7, _136_hex__decompress_zlib_a, 0, no_hashes },
+  {"decompress-gzip", 7, _136_hex__decompress_gzip_a, 0, no_hashes },
+  {}};
+
+
 static u3j_harm _136_hex_coed__ed_scad_a[] = {{".2", u3wee_scad}, {}};
 static u3j_harm _136_hex_coed__ed_scas_a[] = {{".2", u3wee_scas}, {}};
 static u3j_harm _136_hex_coed__ed_scap_a[] = {{".2", u3wee_scap}, {}};
@@ -1016,9 +1024,10 @@ static u3j_core _136_hex_d[] =
     { "secp",    6, 0, _136_hex_secp_d,   no_hashes },
     { "mimes",  31, 0, _136_hex_mimes_d,  no_hashes },
     { "json",   31, 0, _136_hex_json_d,   no_hashes },
+    { "checksum", 15, 0, _136_hex_checksum_d, no_hashes},
     { "wasm-sur-v0", 3, 0, _136_hex_wasm_sur_d, no_hashes },
     { "bytestream-v0", 31, 0, _136_hex_bytestream_d, no_hashes},
-    { "checksum", 15, 0, _136_hex_checksum_d, no_hashes},
+    { "zlib-v0", 31, 0, _136_hex__zlib_d, no_hashes },
     {}
   };
 

diff --git a/pkg/noun/jets/e/zlib.c b/pkg/noun/jets/e/zlib.c
@@ -0,0 +1,225 @@
+/// @file
+
+#include <allocate.h>
+#include <stdio.h>
+#include "zlib.h"
+
+#include "jets/q.h"
+#include "jets/w.h"
+
+#include "noun.h"
+
+static void*
+zlib_malloc(voidpf opaque, uInt items, uInt size)
+{
+  size_t len = items * size;
+  void* result = u3a_malloc(len);
+  return result;
+}
+
+static void
+zlib_free(voidpf opaque, voidpf address)
+{
+  u3a_free(address);
+}
+
+u3_noun
+_decompress(u3_atom pos, u3_noun octs, int window_bits)
+{
+  u3_atom p_octs = u3h(octs);
+  u3_atom q_octs = u3t(octs);
+
+  c3_w p_octs_w, pos_w;
+
+  if ( c3n == u3r_safe_word(p_octs, &p_octs_w) ) {
+    return u3_none;
+  }
+  if (c3n == u3r_safe_word(pos, &pos_w)) {
+    return u3_none;
+  }
+
+  c3_w len_w = u3r_met(3, q_octs);
+
+  int leading_zeros = 0;
+
+  if (p_octs_w > len_w) {
+    leading_zeros = p_octs_w - len_w;
+  }
+  else {
+    len_w = p_octs_w;
+  }
+
+  // Bytestream exhausted
+  //
+  if (pos_w >= len_w) {
+    return u3_none;
+  }
+
+  c3_y* input;
+
+  if (c3y == u3a_is_cat(q_octs)) {
+    input = (c3_y*)&q_octs + pos_w;
+  }
+  else {
+    u3a_atom* vat_u = u3a_to_ptr(q_octs);
+    input = (c3_y*)vat_u->buf_w + pos_w;
+  }
+
+  int ret;
+  z_stream strm;
+
+  if (pos_w < len_w) {
+    strm.avail_in = (len_w - pos_w);
+  }
+  else {
+    strm.avail_in = 0;
+  }
+
+  strm.zalloc = zlib_malloc;
+  strm.zfree = zlib_free;
+  strm.opaque = Z_NULL;
+  strm.next_in = input;
+
+  ret = inflateInit2(&strm, window_bits);
+
+  if (ret != Z_OK) {
+    u3l_log("%i", ret);
+    u3l_log("%s", strm.msg);
+    return u3m_bail(c3__exit);
+  }
+
+  c3_w chunk_w = len_w / 10;
+  u3i_slab sab_u;
+
+#define INIT_SZ 16384
+  strm.avail_out = INIT_SZ;
+  u3i_slab_init(&sab_u, 3, INIT_SZ);
+  strm.next_out = sab_u.buf_y;
+
+  void* this_address = strm.next_out;
+
+#define ZEROS_SZ 256
+  c3_y zeros[ZEROS_SZ];
+
+  if (leading_zeros) {
+    memset(zeros, 0, ZEROS_SZ);
+  }
+
+  while ((ret = inflate(&strm, Z_FINISH)) == Z_BUF_ERROR) {
+
+    // Output exhausted: reallocate
+    //
+    if (strm.avail_out == 0) {
+      strm.avail_out = chunk_w;
+
+      u3i_slab_grow(&sab_u, 3, strm.total_out + chunk_w);
+      strm.next_out = sab_u.buf_y + strm.total_out;
+    }
+
+    // Input exhausted: input leading zeros?
+    //
+    if (strm.avail_in == 0) {
+
+      if (leading_zeros) {
+        // Position in the stream exceeded atom bytes,
+        // but is still below stream length
+        //
+        if (strm.total_in + pos_w >= len_w
+            && strm.total_in + pos_w < p_octs_w) {
+
+          c3_w rem_w = p_octs_w - (strm.total_in + pos_w);
+          strm.next_in = zeros;
+
+          if (rem_w > ZEROS_SZ) {
+            strm.avail_in = ZEROS_SZ;
+          }
+          else {
+            strm.avail_in = rem_w;
+          }
+        }
+        else {
+          u3l_log("%i", ret);
+          u3l_log("%s", strm.msg);
+          inflateEnd(&strm);
+          u3i_slab_free(&sab_u);
+          return u3m_bail(c3__exit);
+        }
+      }
+      else {
+        u3l_log("%i", ret);
+        u3l_log("%s", strm.msg);
+        inflateEnd(&strm);
+        u3i_slab_free(&sab_u);
+        return u3m_bail(c3__exit);
+      }
+    }
+  }
+  if (ret != Z_STREAM_END) {
+    u3l_log("%i", ret);
+    u3l_log("%s", strm.msg);
+    inflateEnd(&strm);
+    u3i_slab_free(&sab_u);
+    return u3m_bail(c3__exit);
+  }
+  ret = inflateEnd(&strm);
+
+  if (ret != Z_OK) {
+    u3l_log("%i", ret);
+    u3l_log("%s", strm.msg);
+    u3i_slab_free(&sab_u);
+    return u3m_bail(c3__exit);
+  }
+
+  u3_noun decompressed_octs = u3nc(strm.total_out, u3i_slab_mint(&sab_u));
+  u3_noun new_pos = pos_w + strm.total_in;
+  u3_noun new_stream = u3nc(u3i_word(new_pos), u3k(octs));
+
+  return u3nc(decompressed_octs, new_stream);
+}
+
+u3_noun
+u3qe_decompress_gzip(u3_atom pos, u3_noun octs)
+{
+  return _decompress(pos, octs, 31);
+}
+u3_noun
+u3qe_decompress_zlib(u3_atom pos, u3_noun octs)
+{
+  return _decompress(pos, octs, 15);
+}
+
+u3_noun
+u3we_decompress_gzip(u3_noun cor)
+{
+  u3_atom pos;
+  u3_noun octs;
+
+  u3_noun a = u3r_at(u3x_sam, cor);
+  u3x_cell(a, &pos, &octs);
+
+  if(_(u3a_is_atom(pos)) && _(u3a_is_cell(octs))) {
+    return u3qe_decompress_gzip(pos, octs);
+  }
+
+  else {
+    return u3m_bail(c3__exit);
+  }
+}
+
+u3_noun
+u3we_decompress_zlib(u3_noun cor)
+{
+  u3_atom pos;
+  u3_noun octs;
+
+  u3_noun a = u3r_at(u3x_sam, cor);
+  u3x_cell(a, &pos, &octs);
+
+  if(_(u3a_is_atom(pos)) && _(u3a_is_cell(octs))) {
+    return u3qe_decompress_zlib(pos, octs);
+  }
+
+  else {
+    return u3m_bail(c3__exit);
+  }
+}
diff --git a/pkg/noun/jets/q.h b/pkg/noun/jets/q.h
@@ -225,6 +225,9 @@
     u3_noun u3qes_gte(u3_atom, u3_atom);
     u3_noun u3qes_gth(u3_atom, u3_atom);
 
+    u3_noun  u3qe_decompress_zlib(u3_atom, u3_noun);
+    u3_noun  u3qe_decompress_gzip(u3_atom, u3_noun);
+
   /** Tier 6.
   **/
     u3_noun u3qf_bull(u3_noun, u3_noun);

diff --git a/pkg/noun/jets/w.h b/pkg/noun/jets/w.h
@@ -319,6 +319,8 @@
     u3_noun u3wes_gth(u3_noun);
 
     u3_noun u3we_crc32(u3_noun);
+    u3_noun u3we_decompress_zlib(u3_noun);
+    u3_noun u3we_decompress_gzip(u3_noun);
 
     u3_noun u3we_lia_run(u3_noun);
     u3_noun u3we_lia_run_once(u3_noun);