zstd-ruby 2.0.0 → 2.0.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: c217e3eae21252b6af27b109769b412d7f2e527366b7f197c16ceb1a5c9b7374
-  data.tar.gz: 67f593fa170baaf3b1a87e0683794da2d2bbd7b89d6eb58c2ac8e513f18aba84
+  metadata.gz: 91a2e54ff93b0c602d6b2d064f6f8c615a59f0ef963c20b24e4db6d17643163e
+  data.tar.gz: cbeb4c7c21187336a284cb714765fb80b02b982c6b4fafa60784d8f7950fe528
 SHA512:
-  metadata.gz: cdea0867d22baff1f4d1992103b0afd851495f7cb1ab6cbf6fcbaec9fc90f427e4c856fc7cbce47eea70a17165b43e3ad5410199c9a7c7abd71a6c62b69042ee
-  data.tar.gz: b2a4087ac5eabdde754f4b9eb2f33e5409022856d617787f4d2f920b9143de930f5a6ed5a91e52a205743c18031e90bc859679fd70584215fedf454140dfd12b
+  metadata.gz: 70190b1a1233782dc440e5ef913f4a83b6bc936ea96d744f5a44b3dbceb41f8a1ef9c3264746eb251356b95c13c5503b2b2567076bc45c28d628ee278b9c2522
+  data.tar.gz: 600bc166bef41126b93a0e78190bbf6b5698259797e166ee4d787a113d6c4e27e7c6c927135e54b33a7e4cc25a74b2df4cb19413f74618892191863cc4c9b4fb

data/ext/zstdruby/streaming_compress.c

@@ -105,13 +105,13 @@ static VALUE
 no_compress(struct streaming_compress_t* sc, ZSTD_EndDirective endOp)
 {
   ZSTD_inBuffer input = { NULL, 0, 0 };
-  const char* output_data = RSTRING_PTR(sc->buf);
   VALUE result = rb_str_new(0, 0);
   size_t ret;
   do {
+    const char* output_data = RSTRING_PTR(sc->buf);
     ZSTD_outBuffer output = { (void*)output_data, sc->buf_size, 0 };
 
-    size_t const ret = zstd_stream_compress(sc->ctx, &output, &input, endOp, false);
+    ret = zstd_stream_compress(sc->ctx, &output, &input, endOp, false);
     if (ZSTD_isError(ret)) {
       rb_raise(rb_eRuntimeError, "flush error error code: %s", ZSTD_getErrorName(ret));
     }
@@ -131,9 +131,9 @@ rb_streaming_compress_compress(VALUE obj, VALUE src)
   struct streaming_compress_t* sc;
   TypedData_Get_Struct(obj, struct streaming_compress_t, &streaming_compress_type, sc);
 
-  const char* output_data = RSTRING_PTR(sc->buf);
   VALUE result = rb_str_new(0, 0);
   while (input.pos < input.size) {
+    const char* output_data = RSTRING_PTR(sc->buf);
     ZSTD_outBuffer output = { (void*)output_data, sc->buf_size, 0 };
     size_t const ret = zstd_stream_compress(sc->ctx, &output, &input, ZSTD_e_continue, false);
     if (ZSTD_isError(ret)) {
@@ -150,7 +150,6 @@ rb_streaming_compress_write(int argc, VALUE *argv, VALUE obj)
   size_t total = 0;
   struct streaming_compress_t* sc;
   TypedData_Get_Struct(obj, struct streaming_compress_t, &streaming_compress_type, sc);
-  const char* output_data = RSTRING_PTR(sc->buf);
 
   while (argc-- > 0) {
     VALUE str = *argv++;
@@ -160,18 +159,20 @@ rb_streaming_compress_write(int argc, VALUE *argv, VALUE obj)
     ZSTD_inBuffer input = { input_data, input_size, 0 };
 
     while (input.pos < input.size) {
+      const char* output_data = RSTRING_PTR(sc->buf);
       ZSTD_outBuffer output = { (void*)output_data, sc->buf_size, 0 };
       size_t const ret = zstd_stream_compress(sc->ctx, &output, &input, ZSTD_e_continue, false);
       if (ZSTD_isError(ret)) {
         rb_raise(rb_eRuntimeError, "compress error error code: %s", ZSTD_getErrorName(ret));
       }
-      /* collect produced bytes */
+      /* Directly append to the pending buffer */
       if (output.pos > 0) {
         rb_str_cat(sc->pending, output.dst, output.pos);
       }
-      total += RSTRING_LEN(str);
     }
+    total += RSTRING_LEN(str);
   }
+
   return SIZET2NUM(total);
 }
 
@@ -202,9 +203,9 @@ rb_streaming_compress_flush(VALUE obj)
   struct streaming_compress_t* sc;
   TypedData_Get_Struct(obj, struct streaming_compress_t, &streaming_compress_type, sc);
   VALUE drained = no_compress(sc, ZSTD_e_flush);
-  rb_str_cat(sc->pending, RSTRING_PTR(drained), RSTRING_LEN(drained));
-  VALUE out = sc->pending;
-  sc->pending = rb_str_new(0, 0);
+  VALUE out = rb_str_dup(sc->pending);
+  rb_str_cat(out, RSTRING_PTR(drained), RSTRING_LEN(drained));
+  rb_str_resize(sc->pending, 0);
   return out;
 }
 
@@ -214,9 +215,9 @@ rb_streaming_compress_finish(VALUE obj)
   struct streaming_compress_t* sc;
   TypedData_Get_Struct(obj, struct streaming_compress_t, &streaming_compress_type, sc);
   VALUE drained = no_compress(sc, ZSTD_e_end);
-  rb_str_cat(sc->pending, RSTRING_PTR(drained), RSTRING_LEN(drained));
-  VALUE out = sc->pending;
-  sc->pending = rb_str_new(0, 0);
+  VALUE out = rb_str_dup(sc->pending);
+  rb_str_cat(out, RSTRING_PTR(drained), RSTRING_LEN(drained));
+  rb_str_resize(sc->pending, 0);
   return out;
 }
 

data/ext/zstdruby/streaming_decompress.c

@@ -100,15 +100,22 @@ rb_streaming_decompress_decompress(VALUE obj, VALUE src)
 
   struct streaming_decompress_t* sd;
   TypedData_Get_Struct(obj, struct streaming_decompress_t, &streaming_decompress_type, sd);
-  const char* output_data = RSTRING_PTR(sd->buf);
   VALUE result = rb_str_new(0, 0);
+
   while (input.pos < input.size) {
+    const char* output_data = RSTRING_PTR(sd->buf);
     ZSTD_outBuffer output = { (void*)output_data, sd->buf_size, 0 };
     size_t const ret = zstd_stream_decompress(sd->dctx, &output, &input, false);
+
     if (ZSTD_isError(ret)) {
       rb_raise(rb_eRuntimeError, "decompress error error code: %s", ZSTD_getErrorName(ret));
     }
-    rb_str_cat(result, output.dst, output.pos);
+    if (output.pos > 0) {
+        rb_str_cat(result, output.dst, output.pos);
+    }
+    if (ret == 0 && output.pos == 0) {
+        break;
+    }
   }
   return result;
 }

data/ext/zstdruby/zstdruby.c

@@ -39,61 +39,91 @@ static VALUE rb_compress(int argc, VALUE *argv, VALUE self)
   return output;
 }
 
-static VALUE decompress_buffered(ZSTD_DCtx* dctx, const char* input_data, size_t input_size)
-{
-  ZSTD_inBuffer input = { input_data, input_size, 0 };
-  VALUE result = rb_str_new(0, 0);
+static VALUE decode_one_frame(ZSTD_DCtx* dctx, const unsigned char* src, size_t size, VALUE kwargs) {
+  VALUE out = rb_str_buf_new(0);
+  size_t cap = ZSTD_DStreamOutSize();
+  char *buf = ALLOC_N(char, cap);
+  ZSTD_inBuffer in = (ZSTD_inBuffer){ src, size, 0 };
 
-  while (input.pos < input.size) {
-    ZSTD_outBuffer output = { NULL, 0, 0 };
-    output.size += ZSTD_DStreamOutSize();
-    VALUE output_string = rb_str_new(NULL, output.size);
-    output.dst = RSTRING_PTR(output_string);
+  ZSTD_DCtx_reset(dctx, ZSTD_reset_session_only);
+  set_decompress_params(dctx, kwargs);
 
-    size_t ret = zstd_stream_decompress(dctx, &output, &input, false);
+  for (;;) {
+    ZSTD_outBuffer o = (ZSTD_outBuffer){ buf, cap, 0 };
+    size_t ret = ZSTD_decompressStream(dctx, &o, &in);
     if (ZSTD_isError(ret)) {
-      ZSTD_freeDCtx(dctx);
-      rb_raise(rb_eRuntimeError, "%s: %s", "ZSTD_decompressStream failed", ZSTD_getErrorName(ret));
+      xfree(buf);
+      rb_raise(rb_eRuntimeError, "ZSTD_decompressStream failed: %s", ZSTD_getErrorName(ret));
+    }
+    if (o.pos) {
+      rb_str_cat(out, buf, o.pos);
+    }
+    if (ret == 0) {
+      break;
     }
-    rb_str_cat(result, output.dst, output.pos);
   }
-  ZSTD_freeDCtx(dctx);
-  return result;
+  xfree(buf);
+  return out;
+}
+
+static VALUE decompress_buffered(ZSTD_DCtx* dctx, const char* data, size_t len) {
+  return decode_one_frame(dctx, (const unsigned char*)data, len, Qnil);
 }
 
 static VALUE rb_decompress(int argc, VALUE *argv, VALUE self)
 {
-  VALUE input_value;
-  VALUE kwargs;
+  VALUE input_value, kwargs;
   rb_scan_args(argc, argv, "10:", &input_value, &kwargs);
   StringValue(input_value);
-  char* input_data = RSTRING_PTR(input_value);
-  size_t input_size = RSTRING_LEN(input_value);
-  ZSTD_DCtx* const dctx = ZSTD_createDCtx();
-  if (dctx == NULL) {
-    rb_raise(rb_eRuntimeError, "%s", "ZSTD_createDCtx failed");
-  }
-  set_decompress_params(dctx, kwargs);
 
-  unsigned long long const uncompressed_size = ZSTD_getFrameContentSize(input_data, input_size);
-  if (uncompressed_size == ZSTD_CONTENTSIZE_ERROR) {
-    rb_raise(rb_eRuntimeError, "%s: %s", "not compressed by zstd", ZSTD_getErrorName(uncompressed_size));
-  }
-  // ZSTD_decompressStream may be called multiple times when ZSTD_CONTENTSIZE_UNKNOWN, causing slowness.
-  // Therefore, we will not standardize on ZSTD_decompressStream
-  if (uncompressed_size == ZSTD_CONTENTSIZE_UNKNOWN) {
-    return decompress_buffered(dctx, input_data, input_size);
-  }
+  size_t in_size = RSTRING_LEN(input_value);
+  const unsigned char *in_r = (const unsigned char *)RSTRING_PTR(input_value);
+  unsigned char *in = ALLOC_N(unsigned char, in_size);
+  memcpy(in, in_r, in_size);
+
+  size_t off = 0;
+  const uint32_t ZSTD_MAGIC = 0xFD2FB528U;
+  const uint32_t SKIP_LO    = 0x184D2A50U; /* ...5F */
+
+  while (off + 4 <= in_size) {
+    uint32_t magic = (uint32_t)in[off]
+                   | ((uint32_t)in[off+1] << 8)
+                   | ((uint32_t)in[off+2] << 16)
+                   | ((uint32_t)in[off+3] << 24);
+
+    if ((magic & 0xFFFFFFF0U) == (SKIP_LO & 0xFFFFFFF0U)) {
+      if (off + 8 > in_size) break;
+      uint32_t skipLen = (uint32_t)in[off+4]
+                       | ((uint32_t)in[off+5] << 8)
+                       | ((uint32_t)in[off+6] << 16)
+                       | ((uint32_t)in[off+7] << 24);
+      size_t adv = (size_t)8 + (size_t)skipLen;
+      if (off + adv > in_size) break;
+      off += adv;
+      continue;
+    }
 
-  VALUE output = rb_str_new(NULL, uncompressed_size);
-  char* output_data = RSTRING_PTR(output);
+    if (magic == ZSTD_MAGIC) {
+      ZSTD_DCtx *dctx = ZSTD_createDCtx();
+      if (!dctx) {
+        xfree(in);
+        rb_raise(rb_eRuntimeError, "ZSTD_createDCtx failed");
+      }
+
+      VALUE out = decode_one_frame(dctx, in + off, in_size - off, kwargs);
 
-  size_t const decompress_size = zstd_decompress(dctx, output_data, uncompressed_size, input_data, input_size, false);
-  if (ZSTD_isError(decompress_size)) {
-    rb_raise(rb_eRuntimeError, "%s: %s", "decompress error", ZSTD_getErrorName(decompress_size));
+      ZSTD_freeDCtx(dctx);
+      xfree(in);
+      RB_GC_GUARD(input_value);
+      return out;
+    }
+
+    off += 1;
   }
-  ZSTD_freeDCtx(dctx);
-  return output;
+
+  xfree(in);
+  RB_GC_GUARD(input_value);
+  rb_raise(rb_eRuntimeError, "not a zstd frame (magic not found)");
 }
 
 static void free_cdict(void *dict)

data/lib/zstd-ruby/version.rb

@@ -1,3 +1,3 @@
 module Zstd
-  VERSION = "2.0.0"
+  VERSION = "2.0.1"
 end

metadata

@@ -1,13 +1,13 @@
 --- !ruby/object:Gem::Specification
 name: zstd-ruby
 version: !ruby/object:Gem::Version
-  version: 2.0.0
+  version: 2.0.1
 platform: ruby
 authors:
 - SpringMT
 bindir: exe
 cert_chain: []
-date: 2025-09-26 00:00:00.000000000 Z
+date: 2025-09-29 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler