Stream decoding benchmark.

5 files changed, 152 insertions, 49 deletions

diff --git a/Makefile b/Makefile
index 749c5a7..203bed6 100644
--- a/Makefile
+++ b/Makefile
@@ -54,7 +54,7 @@ clean:
 
 # The core library (core/libupb.a)
 SRC=core/upb.c stream/upb_decoder.c core/upb_table.c core/upb_def.c core/upb_string.c \
-    core/upb_stream.c stream/upb_stdio.c stream/upb_textprinter.c \
+    core/upb_stream.c stream/upb_stdio.c stream/upb_strstream.c stream/upb_textprinter.c \
     descriptor/descriptor.c
 $(SRC): perf-cppflags
 # Parts of core that are yet to be converted.
@@ -154,10 +154,10 @@ tests/tests: core/libupb.a
 tools/upbc: core/libupb.a
 
 # Benchmarks
-UPB_BENCHMARKS=benchmarks/b.parsetostruct_googlemessage1.upb_table_byval \
-               benchmarks/b.parsetostruct_googlemessage1.upb_table_byref \
-               benchmarks/b.parsetostruct_googlemessage2.upb_table_byval \
-               benchmarks/b.parsetostruct_googlemessage2.upb_table_byref
+#UPB_BENCHMARKS=benchmarks/b.parsetostruct_googlemessage1.upb_table \
+#               benchmarks/b.parsetostruct_googlemessage2.upb_table
+UPB_BENCHMARKS=benchmarks/b.parsestream_googlemessage1.upb_table \
+               benchmarks/b.parsestream_googlemessage2.upb_table
 
 BENCHMARKS=$(UPB_BENCHMARKS) \
            benchmarks/b.parsetostruct_googlemessage1.proto2_table \
@@ -204,6 +204,20 @@ benchmarks/b.parsetostruct_googlemessage2.upb_table_byref: \
 	  -DMESSAGE_FILE=\"google_message2.dat\" \
 	  -DBYREF=true $(LIBUPB)
 
+benchmarks/b.parsestream_googlemessage1.upb_table \
+benchmarks/b.parsestream_googlemessage2.upb_table: \
+    benchmarks/parsestream.upb_table.c $(LIBUPB) benchmarks/google_messages.proto.pb
+	$(CC) $(CFLAGS) $(CPPFLAGS) -o benchmarks/b.parsestream_googlemessage1.upb_table $< \
+	  -DMESSAGE_NAME=\"benchmarks.SpeedMessage1\" \
+	  -DMESSAGE_DESCRIPTOR_FILE=\"google_messages.proto.pb\" \
+	  -DMESSAGE_FILE=\"google_message1.dat\" \
+	  $(LIBUPB)
+	$(CC) $(CFLAGS) $(CPPFLAGS) -o benchmarks/b.parsestream_googlemessage2.upb_table $< \
+	  -DMESSAGE_NAME=\"benchmarks.SpeedMessage2\" \
+	  -DMESSAGE_DESCRIPTOR_FILE=\"google_messages.proto.pb\" \
+	  -DMESSAGE_FILE=\"google_message2.dat\" \
+	  $(LIBUPB)
+
 benchmarks/b.parsetostruct_googlemessage1.proto2_table \
 benchmarks/b.parsetostruct_googlemessage2.proto2_table: \
     benchmarks/parsetostruct.proto2_table.cc benchmarks/google_messages.pb.cc
diff --git a/benchmarks/parsestream.upb_table.c b/benchmarks/parsestream.upb_table.c
new file mode 100644
index 0000000..c6acad9
--- /dev/null
+++ b/benchmarks/parsestream.upb_table.c
@@ -0,0 +1,113 @@
+
+#include "main.c"
+
+#include "upb_def.h"
+#include "upb_decoder.h"
+#include "upb_strstream.h"
+
+static upb_stringsrc *stringsrc;
+static upb_string *input_str;
+static upb_string *tmp_str;
+static upb_msgdef *def;
+static upb_decoder *decoder;
+
+static bool initialize()
+{
+  // Initialize upb state, decode descriptor.
+  upb_status status = UPB_STATUS_INIT;
+  upb_symtab *s = upb_symtab_new();
+  upb_symtab_add_descriptorproto(s);
+  upb_string *fds_str = upb_strreadfile(MESSAGE_DESCRIPTOR_FILE);
+  if(fds_str == NULL) {
+    fprintf(stderr, "Couldn't read " MESSAGE_DESCRIPTOR_FILE ":"),
+    upb_printerr(&status);
+    return false;
+  }
+
+  upb_stringsrc *ssrc = upb_stringsrc_new();
+  upb_stringsrc_reset(ssrc, fds_str);
+  upb_def *fds_def = upb_symtab_lookup(
+      s, UPB_STRLIT("google.protobuf.FileDescriptorSet"));
+  upb_decoder *d = upb_decoder_new(upb_downcast_msgdef(fds_def));
+  upb_decoder_reset(d, upb_stringsrc_bytesrc(ssrc));
+
+  upb_symtab_addfds(s, upb_decoder_src(d), &status);
+
+  if(!upb_ok(&status)) {
+    fprintf(stderr, "Error importing " MESSAGE_DESCRIPTOR_FILE ":");
+    upb_printerr(&status);
+    return false;
+  }
+
+  upb_string_unref(fds_str);
+  upb_decoder_free(d);
+  upb_stringsrc_free(ssrc);
+  upb_def_unref(fds_def);
+
+  def = upb_downcast_msgdef(upb_symtab_lookup(s, UPB_STRLIT(MESSAGE_NAME)));
+  if(!def) {
+    fprintf(stderr, "Error finding symbol '" UPB_STRFMT "'.\n",
+            UPB_STRARG(UPB_STRLIT(MESSAGE_NAME)));
+    return false;
+  }
+  upb_symtab_unref(s);
+
+  // Read the message data itself.
+  input_str = upb_strreadfile(MESSAGE_FILE);
+  if(input_str == NULL) {
+    fprintf(stderr, "Error reading " MESSAGE_FILE "\n");
+    return false;
+  }
+  tmp_str = NULL;
+  decoder = upb_decoder_new(def);
+  stringsrc = upb_stringsrc_new();
+  return true;
+}
+
+static void cleanup()
+{
+  upb_string_unref(input_str);
+  upb_string_unref(tmp_str);
+  upb_def_unref(UPB_UPCAST(def));
+  upb_decoder_free(decoder);
+  upb_stringsrc_free(stringsrc);
+}
+
+static size_t run(int i)
+{
+  (void)i;
+  upb_status status = UPB_STATUS_INIT;
+  upb_stringsrc_reset(stringsrc, input_str);
+  upb_decoder_reset(decoder, upb_stringsrc_bytesrc(stringsrc));
+  upb_src *src = upb_decoder_src(decoder);
+  upb_fielddef *f;
+  upb_string *str = NULL;
+  int depth = 0;
+  while(1) {
+    while((f = upb_src_getdef(src)) != NULL) {
+      if(upb_issubmsg(f)) {
+        upb_src_startmsg(src);
+        ++depth;
+      } else if(upb_isstring(f)) {
+        tmp_str = upb_string_tryrecycle(str);
+        upb_src_getstr(src, tmp_str);
+      } else {
+        // Primitive type.
+        upb_value val;
+        upb_src_getval(src, upb_value_addrof(&val));
+      }
+    }
+    // If we're not EOF now, the loop terminated due to an error.
+    if (!upb_src_eof(src)) goto err;
+    if (depth == 0) break;
+    --depth;
+    upb_src_endmsg(src);
+  }
+  if(!upb_ok(&status)) goto err;
+  return upb_string_len(input_str);
+
+err:
+  fprintf(stderr, "Decode error");
+  upb_printerr(&status);
+  return 0;
+}
diff --git a/core/upb_stream.h b/core/upb_stream.h
index b7400c5..861bd1c 100644
--- a/core/upb_stream.h
+++ b/core/upb_stream.h
@@ -128,7 +128,8 @@ bool upb_bytesrc_get(upb_bytesrc *src, upb_string *str, upb_strlen_t minlen);
 
 // Appends the next "len" bytes in the stream in-place to "str".  This should
 // be used when the caller needs to build a contiguous string of the existing
-// data in "str" with more data.
+// data in "str" with more data.  The call fails if fewer than len bytes are
+// available in the stream.
 bool upb_bytesrc_append(upb_bytesrc *src, upb_string *str, upb_strlen_t len);
 
 // Returns the current error status for the stream.
diff --git a/core/upb_string.c b/core/upb_string.c
index 93686f5..847a3ee 100644
--- a/core/upb_string.c
+++ b/core/upb_string.c
@@ -131,3 +131,21 @@ upb_string *upb_strdup(upb_string *s) {
   upb_strcpy(str, s);
   return str;
 }
+
+upb_string *upb_strreadfile(const char *filename) {
+  FILE *f = fopen(filename, "rb");
+  if(!f) return NULL;
+  if(fseek(f, 0, SEEK_END) != 0) goto error;
+  long size = ftell(f);
+  if(size < 0) goto error;
+  if(fseek(f, 0, SEEK_SET) != 0) goto error;
+  upb_string *s = upb_string_new();
+  char *buf = upb_string_getrwbuf(s, size);
+  if(fread(buf, size, 1, f) != 1) goto error;
+  fclose(f);
+  return s;
+
+error:
+  fclose(f);
+  return NULL;
+}
diff --git a/stream/upb_byteio.h b/stream/upb_byteio.h
deleted file mode 100644
index 69a28b3..0000000
--- a/stream/upb_byteio.h
+++ /dev/null
@@ -1,43 +0,0 @@
-/*
- * upb - a minimalist implementation of protocol buffers.
- *
- * This file contains upb_bytesrc and upb_bytesink implementations for common
- * interfaces like strings, UNIX fds, and FILE*.
- *
- * Copyright (c) 2009-2010 Joshua Haberman.  See LICENSE for details.
- */
-
-#ifndef UPB_BYTEIO_H
-#define UPB_BYTEIO_H
-
-#include "upb_srcsink.h"
-
-#ifdef __cplusplus
-extern "C" {
-#endif
-
-/* upb_stringsrc **************************************************************/
-
-struct upb_stringsrc;
-typedef struct upb_stringsrc upb_stringsrc;
-
-// Create/free a stringsrc.
-upb_stringsrc *upb_stringsrc_new();
-void upb_stringsrc_free(upb_stringsrc *s);
-
-// Resets the stringsrc to a state where it will vend the given string.  The
-// stringsrc will take a reference on the string, so the caller need not ensure
-// that it outlives the stringsrc.  A stringsrc can be reset multiple times.
-void upb_stringsrc_reset(upb_stringsrc *s, upb_string *str);
-
-// Returns the upb_bytesrc* for this stringsrc.  Invalidated by reset above.
-upb_bytesrc *upb_stringsrc_bytesrc();
-
-
-/* upb_fdsrc ******************************************************************/
-
-#ifdef __cplusplus
-}  /* extern "C" */
-#endif
-
-#endif