Implemented upb_stringsink, upb_msgtotext, and exposed the latter to Lua.

12 files changed, 247 insertions, 39 deletions

diff --git a/lang_ext/lua/test.lua b/lang_ext/lua/test.lua
index 072f646..b7f980e 100644
--- a/lang_ext/lua/test.lua
+++ b/lang_ext/lua/test.lua
@@ -20,10 +20,22 @@ for _, def in ipairs(symtab:getdefs(-1)) do
 end
 
 SpeedMessage1 = symtab:lookup("benchmarks.SpeedMessage1")
+SpeedMessage2 = symtab:lookup("benchmarks.SpeedMessage2")
 print(SpeedMessage1:name())
 
 msg = SpeedMessage1()
-print(msg.foo)
+f = assert(io.open("../../benchmarks/google_message1.dat"))
+msg:Parse(f:read("*all"))
+print(msg:ToText())
+
+msg = SpeedMessage2()
+f = assert(io.open("../../benchmarks/google_message2.dat"))
+msg:Parse(f:read("*all"))
+msg.field2 = ""
+print(msg.field2)
+print(msg:ToText())
+--msg:Serialize()
+--msg:FromText(str)
 -- print(msg.field129)
 -- print(msg.field271)
 --print(msg.field15.field15)
diff --git a/lang_ext/lua/upb.c b/lang_ext/lua/upb.c
index 2e0102e..aa5f4a1 100644
--- a/lang_ext/lua/upb.c
+++ b/lang_ext/lua/upb.c
@@ -23,6 +23,7 @@ typedef struct {
 } lupb_def;
 
 void lupb_pushstring(lua_State *L, upb_string *str) {
+  assert(str);
   lua_pushlstring(L, upb_string_getrobuf(str), upb_string_len(str));
 }
 
@@ -155,9 +156,7 @@ static void lupb_pushvalue(lua_State *L, upb_value val, upb_fielddef *f) {
       lua_pushboolean(L, upb_value_getbool(val)); break;
     case UPB_TYPE(STRING):
     case UPB_TYPE(BYTES): {
-      upb_string *str = upb_value_getstr(val);
-      assert(str);
-      lua_pushlstring(L, upb_string_getrobuf(str), upb_string_len(str)); break;
+      lupb_pushstring(L, upb_value_getstr(val)); break;
     }
     case UPB_TYPE(MESSAGE):
     case UPB_TYPE(GROUP): {
@@ -252,6 +251,9 @@ static int lupb_msg_index(lua_State *L) {
   upb_string namestr = UPB_STACK_STRING_LEN(name, len);
   upb_fielddef *f = upb_msgdef_ntof(m->msgdef, &namestr);
   if (f) {
+    if (upb_isarray(f)) {
+      luaL_error(L, "Access of repeated fields not yet implemented.");
+    }
     lupb_pushvalue(L, upb_msg_get(m->msg, f), f);
   } else {
     // It wasn't a field, perhaps it's a method?
@@ -290,12 +292,34 @@ static int lupb_msg_clear(lua_State *L) {
   return 0;
 }
 
+static int lupb_msg_parse(lua_State *L) {
+  lupb_msg *m = lupb_msg_check(L, 1);
+  size_t len;
+  const char *strbuf = luaL_checklstring(L, 2, &len);
+  upb_string str = UPB_STACK_STRING_LEN(strbuf, len);
+  upb_status status = UPB_STATUS_INIT;
+  upb_strtomsg(&str, m->msg, m->msgdef, &status);
+  lupb_checkstatus(L, &status);
+  return 0;
+}
+
+static int lupb_msg_totext(lua_State *L) {
+  lupb_msg *m = lupb_msg_check(L, 1);
+  upb_string *str = upb_string_new();
+  upb_msgtotext(str, m->msg, m->msgdef, false);
+  lupb_pushstring(L, str);
+  upb_string_unref(str);
+  return 1;
+}
+
 static const struct luaL_Reg lupb_msg_mm[] = {
   {"__gc", lupb_msg_gc},
   {"__index", lupb_msg_index},
   {"__newindex", lupb_msg_newindex},
   // Our __index mm will look up methods if the index isn't a field name.
   {"Clear", lupb_msg_clear},
+  {"Parse", lupb_msg_parse},
+  {"ToText", lupb_msg_totext},
   {NULL, NULL}
 };
 
diff --git a/src/upb_glue.c b/src/upb_glue.c
index a3d4e7d..86022d8 100644
--- a/src/upb_glue.c
+++ b/src/upb_glue.c
@@ -8,6 +8,7 @@
 #include "upb_msg.h"
 #include "upb_decoder.h"
 #include "upb_strstream.h"
+#include "upb_textprinter.h"
 
 void upb_strtomsg(upb_string *str, upb_msg *msg, upb_msgdef *md,
                   upb_status *status) {
@@ -37,6 +38,29 @@ void upb_strtomsg(upb_string *str, upb_msg *msg, upb_msgdef *md,
   upb_handlers_uninit(&h);
 }
 
+void upb_msgtotext(upb_string *str, upb_msg *msg, upb_msgdef *md,
+                   bool single_line) {
+  upb_stringsink strsink;
+  upb_stringsink_init(&strsink);
+  upb_stringsink_reset(&strsink, str);
+
+  upb_textprinter *p = upb_textprinter_new();
+  upb_handlers h;
+  upb_handlers_init(&h);
+  upb_textprinter_reset(p, &h, upb_stringsink_bytesink(&strsink), single_line);
+
+  upb_status status = UPB_STATUS_INIT;
+  upb_msg_runhandlers(msg, md, &h, &status);
+  // None of {upb_msg_runhandlers, upb_textprinter, upb_stringsink} should be
+  // capable of returning an error.
+  assert(upb_ok(&status));
+  upb_status_uninit(&status);
+
+  upb_stringsink_uninit(&strsink);
+  upb_textprinter_free(p);
+  upb_handlers_uninit(&h);
+}
+
 void upb_parsedesc(upb_symtab *symtab, upb_string *str, upb_status *status) {
   upb_stringsrc strsrc;
   upb_stringsrc_init(&strsrc);
diff --git a/src/upb_glue.h b/src/upb_glue.h
index ca32436..afb7b59 100644
--- a/src/upb_glue.h
+++ b/src/upb_glue.h
@@ -4,8 +4,8 @@
  * upb's core components like upb_decoder and upb_msg are carefully designed to
  * avoid depending on each other for maximum orthogonality.  In other words,
  * you can use a upb_decoder to decode into *any* kind of structure; upb_msg is
- * just one such structure.  You can use upb_decoder without having to link in
- * upb_msg.
+ * just one such structure.  A upb_msg can be serialized/deserialized into any
+ * format, protobuf binary format is just one such format.
  *
  * However, for convenience we provide functions here for doing common
  * operations like deserializing protobuf binary format into a upb_msg.  The
@@ -13,12 +13,20 @@
  * which could be undesirable if you're trying to use a trimmed-down build of
  * upb.
  *
+ * While these routines are convenient, they do not reuse any encoding/decoding
+ * state.  For example, if a decoder is JIT-based, it will be re-JITted every
+ * time these functions are called.  For this reason, if you are parsing lots
+ * of data and efficiency is an issue, these may not be the best functions to
+ * use (though they are useful for prototyping, before optimizing).
+ *
  * Copyright (c) 2011 Joshua Haberman.  See LICENSE for details.
  */
 
 #ifndef UPB_GLUE_H
 #define UPB_GLUE_H
 
+#include <stdbool.h>
+
 #ifdef __cplusplus
 extern "C" {
 #endif
@@ -36,6 +44,9 @@ struct _upb_symtab;
 void upb_strtomsg(struct _upb_string *str, struct _upb_msg *msg,
                   struct _upb_msgdef *md, struct _upb_status *s);
 
+void upb_msgtotext(struct _upb_string *str, struct _upb_msg *msg,
+                   struct _upb_msgdef *md, bool single_line);
+
 void upb_parsedesc(struct _upb_symtab *symtab, struct _upb_string *str,
                    struct _upb_status *status);
 
diff --git a/src/upb_msg.c b/src/upb_msg.c
index 89913dd..a3fd825 100644
--- a/src/upb_msg.c
+++ b/src/upb_msg.c
@@ -180,6 +180,77 @@ upb_value upb_msg_get(upb_msg *msg, upb_fielddef *f) {
   }
 }
 
+static upb_flow_t upb_msg_dispatch(upb_msg *msg, upb_msgdef *md,
+                                   upb_dispatcher *d, upb_status *s);
+
+static upb_flow_t upb_msg_pushval(upb_value val, upb_fielddef *f,
+                                  upb_dispatcher *d, upb_status *s) {
+#define CHECK_FLOW(x) do { \
+  flow = x; if (flow != UPB_CONTINUE) return flow; \
+  } while(0)
+
+// For when a SKIP can be implemented just through an early return.
+#define CHECK_FLOW_LOCAL(x) do { \
+  flow = x; \
+  if (flow != UPB_CONTINUE) { \
+    if (flow == UPB_SKIPSUBMSG) flow = UPB_CONTINUE; \
+    goto end; \
+  } \
+} while (0)
+  upb_flow_t flow;
+  if (upb_issubmsg(f)) {
+    upb_msg *msg = upb_value_getmsg(val);
+    CHECK_FLOW_LOCAL(upb_dispatch_startsubmsg(d, f));
+    CHECK_FLOW_LOCAL(upb_msg_dispatch(msg, upb_downcast_msgdef(f->def), d, s));
+    CHECK_FLOW(upb_dispatch_endsubmsg(d, f));
+  } else {
+    CHECK_FLOW(upb_dispatch_value(d, f, val));
+  }
+
+end:
+  return flow;
+}
+
+static upb_flow_t upb_msg_dispatch(upb_msg *msg, upb_msgdef *md,
+                                   upb_dispatcher *d, upb_status *s) {
+  upb_msg_iter i;
+  upb_flow_t flow;
+  for(i = upb_msg_begin(md); !upb_msg_done(i); i = upb_msg_next(md, i)) {
+    upb_fielddef *f = upb_msg_iter_field(i);
+    if (!upb_msg_has(msg, f)) continue;
+    upb_value val = upb_msg_get(msg, f);
+    if (upb_isarray(f)) {
+      upb_array *arr = upb_value_getarr(val);
+      for (uint32_t j = 0; j < upb_array_len(arr); ++j) {
+        CHECK_FLOW_LOCAL(upb_msg_pushval(upb_array_get(arr, f, j), f, d, s));
+      }
+    } else {
+      CHECK_FLOW_LOCAL(upb_msg_pushval(val, f, d, s));
+    }
+  }
+  return UPB_CONTINUE;
+
+end:
+  // Need to copy/massage the error.
+  upb_copyerr(s, d->top->handlers.status);
+  if (upb_ok(s)) {
+    upb_seterr(s, UPB_ERROR, "Callback returned UPB_BREAK");
+  }
+  return flow;
+#undef CHECK_FLOW
+#undef CHECK_FLOW_LOCAL
+}
+
+void upb_msg_runhandlers(upb_msg *msg, upb_msgdef *md, upb_handlers *h,
+                         upb_status *status) {
+  upb_dispatcher d;
+  upb_dispatcher_init(&d);
+  upb_dispatcher_reset(&d, h, true);
+
+  if (upb_dispatch_startmsg(&d) != UPB_CONTINUE) return;
+  if (upb_msg_dispatch(msg, md, &d, status) != UPB_CONTINUE) return;
+  if (upb_dispatch_endmsg(&d) != UPB_CONTINUE) return;
+}
 
 static upb_valueptr upb_msg_getappendptr(upb_msg *msg, upb_fielddef *f) {
   upb_valueptr p = _upb_msg_getptr(msg, f);
diff --git a/src/upb_msg.h b/src/upb_msg.h
index 1e0a176..fd5750f 100644
--- a/src/upb_msg.h
+++ b/src/upb_msg.h
@@ -274,6 +274,11 @@ INLINE void upb_msg_clear(upb_msg *msg, upb_msgdef *md) {
   memset(msg->data, 0, md->set_flags_bytes);
 }
 
+// A non-resumable upb_src that pushes the current contents of the message to
+// the given handlers.
+void upb_msg_runhandlers(upb_msg *msg, upb_msgdef *md, upb_handlers *h,
+                         upb_status *status);
+
 typedef struct {
   upb_msg *msg;
   upb_msgdef *msgdef;
diff --git a/src/upb_string.c b/src/upb_string.c
index 29ce7d4..cbc3414 100644
--- a/src/upb_string.c
+++ b/src/upb_string.c
@@ -83,27 +83,30 @@ void upb_string_substr(upb_string *str, upb_string *target_str,
   str->len = len;
 }
 
-void upb_string_vprintf(upb_string *str, const char *format, va_list args) {
+size_t upb_string_vprintf_at(upb_string *str, size_t offset, const char *format,
+                             va_list args) {
   // Try once without reallocating.  We have to va_copy because we might have
   // to call vsnprintf again.
-  uint32_t size = UPB_MAX(upb_string_size(str), 16);
-  char *buf = upb_string_getrwbuf(str, size);
+  uint32_t size = UPB_MAX(upb_string_size(str) - offset, 16);
+  char *buf = upb_string_getrwbuf(str, offset + size) + offset;
   va_list args_copy;
   va_copy(args_copy, args);
   uint32_t true_size = vsnprintf(buf, size, format, args_copy);
   va_end(args_copy);
 
+  // Resize to be the correct size.
   if (true_size >= size) {
-    // Need to reallocate.  We reallocate even if the sizes were equal,
-    // because snprintf excludes the terminating NULL from its count.
-    // We don't care about the terminating NULL, but snprintf might
-    // bail out of printing even other characters if it doesn't have
-    // enough space to write the NULL also.
-    upb_string_recycle(&str);
-    buf = upb_string_getrwbuf(str, true_size + 1);
+    // Need to print again, because some characters were truncated.  vsnprintf
+    // has weird behavior (and contrary IMO to what the standard says): it will
+    // not write the entire string unless you give it space to store the NULL
+    // terminator also.  So we can't give it space for the string itself and
+    // let NULL get truncated (after all, we don't care about it): we *must*
+    // give it space for NULL.
+    buf = upb_string_getrwbuf(str, offset + true_size + 1) + offset;
     vsnprintf(buf, true_size + 1, format, args);
   }
-  str->len = true_size;
+  str->len = offset + true_size;
+  return true_size;
 }
 
 upb_string *upb_string_asprintf(const char *format, ...) {
diff --git a/src/upb_string.h b/src/upb_string.h
index efafa44..38c3d55 100644
--- a/src/upb_string.h
+++ b/src/upb_string.h
@@ -192,12 +192,18 @@ INLINE void upb_string_recycle(upb_string **_str) {
 char *upb_string_getrwbuf(upb_string *str, upb_strlen_t len);
 
 // Replaces the contents of str with the contents of the given printf.
-void upb_string_vprintf(upb_string *str, const char *format, va_list args);
-INLINE void upb_string_printf(upb_string *str, const char *format, ...) {
+size_t upb_string_vprintf_at(upb_string *str, size_t offset, const char *format,
+                             va_list args);
+INLINE size_t upb_string_vprintf(upb_string *str, const char *format,
+                                 va_list args) {
+  return upb_string_vprintf_at(str, 0, format, args);
+}
+INLINE size_t upb_string_printf(upb_string *str, const char *format, ...) {
   va_list args;
   va_start(args, format);
-  upb_string_vprintf(str, format, args);
+  size_t written = upb_string_vprintf(str, format, args);
   va_end(args);
+  return written;
 }
 
 // Sets the contents of "str" to be the given substring of "target_str", to
diff --git a/src/upb_strstream.c b/src/upb_strstream.c
index a7967d4..c957648 100644
--- a/src/upb_strstream.c
+++ b/src/upb_strstream.c
@@ -9,6 +9,9 @@
 #include <stdlib.h>
 #include "upb_string.h"
 
+
+/* upb_stringsrc **************************************************************/
+
 static upb_strlen_t upb_stringsrc_read(upb_bytesrc *_src, void *buf,
                                        upb_strlen_t count, upb_status *status) {
   upb_stringsrc *src = (upb_stringsrc*)_src;
@@ -39,12 +42,12 @@ static bool upb_stringsrc_getstr(upb_bytesrc *_src, upb_string *str,
 }
 
 void upb_stringsrc_init(upb_stringsrc *s) {
-  static upb_bytesrc_vtbl bytesrc_vtbl = {
+  static upb_bytesrc_vtbl vtbl = {
     upb_stringsrc_read,
     upb_stringsrc_getstr,
   };
+  upb_bytesrc_init(&s->bytesrc, &vtbl);
   s->str = NULL;
-  upb_bytesrc_init(&s->bytesrc, &bytesrc_vtbl);
 }
 
 void upb_stringsrc_reset(upb_stringsrc *s, upb_string *str) {
@@ -63,3 +66,52 @@ void upb_stringsrc_uninit(upb_stringsrc *s) {
 upb_bytesrc *upb_stringsrc_bytesrc(upb_stringsrc *s) {
   return &s->bytesrc;
 }
+
+
+/* upb_stringsink *************************************************************/
+
+void upb_stringsink_uninit(upb_stringsink *s) {
+  upb_string_unref(s->str);
+}
+
+// Resets the stringsink to a state where it will append to the given string.
+// The string must be newly created or recycled.  The stringsink will take a
+// reference on the string, so the caller need not ensure that it outlives the
+// stringsink.  A stringsink can be reset multiple times.
+void upb_stringsink_reset(upb_stringsink *s, upb_string *str) {
+  if (str != s->str) {
+    upb_string_unref(s->str);
+    s->str = upb_string_getref(str);
+  }
+  // Resize to 0.
+  upb_string_getrwbuf(s->str, 0);
+}
+
+upb_bytesink *upb_stringsink_bytesink(upb_stringsink *s) {
+  return &s->bytesink;
+}
+
+static upb_strlen_t upb_stringsink_vprintf(upb_bytesink *_sink, upb_status *s,
+                                           const char *fmt, va_list args) {
+  (void)s;  // No errors can occur.
+  upb_stringsink *sink = (upb_stringsink*)_sink;
+  return upb_string_vprintf_at(sink->str, upb_string_len(sink->str), fmt, args);
+}
+
+static upb_strlen_t upb_stringsink_putstr(upb_bytesink *_sink, upb_string *str,
+                                          upb_status *s) {
+  (void)s;  // No errors can occur.
+  upb_stringsink *sink = (upb_stringsink*)_sink;
+  upb_strcat(sink->str, str);
+  return upb_string_len(str);
+}
+
+void upb_stringsink_init(upb_stringsink *s) {
+  static upb_bytesink_vtbl vtbl = {
+    NULL,
+    upb_stringsink_putstr,
+    upb_stringsink_vprintf
+  };
+  upb_bytesink_init(&s->bytesink, &vtbl);
+  s->str = NULL;
+}
diff --git a/src/upb_strstream.h b/src/upb_strstream.h
index 1a8792b..cd8ff01 100644
--- a/src/upb_strstream.h
+++ b/src/upb_strstream.h
@@ -40,22 +40,24 @@ upb_bytesrc *upb_stringsrc_bytesrc(upb_stringsrc *s);
 
 /* upb_stringsink *************************************************************/
 
-struct upb_stringsink;
-typedef struct upb_stringsink upb_stringsink;
+struct _upb_stringsink {
+  upb_bytesink bytesink;
+  upb_string *str;
+};
+typedef struct _upb_stringsink upb_stringsink;
 
 // Create/free a stringsrc.
-upb_stringsink *upb_stringsink_new();
-void upb_stringsink_free(upb_stringsink *s);
+void upb_stringsink_init(upb_stringsink *s);
+void upb_stringsink_uninit(upb_stringsink *s);
 
-// Gets a string containing the data that has been written to this stringsink.
-// The caller does *not* own any references to this string.
-upb_string *upb_stringsink_getstring(upb_stringsink *s);
+// Resets the stringsink to a state where it will append to the given string.
+// The string must be newly created or recycled.  The stringsink will take a
+// reference on the string, so the caller need not ensure that it outlives the
+// stringsink.  A stringsink can be reset multiple times.
+void upb_stringsink_reset(upb_stringsink *s, upb_string *str);
 
-// Clears the internal string of accumulated data, resetting it to empty.
-void upb_stringsink_reset(upb_stringsink *s);
-
-// Returns the upb_bytesrc* for this stringsrc.  Invalidated by reset above.
-upb_bytesink *upb_stringsrc_bytesink();
+// Returns the upb_bytesink* for this stringsrc.  Invalidated by reset above.
+upb_bytesink *upb_stringsink_bytesink();
 
 
 #ifdef __cplusplus
diff --git a/src/upb_table.h b/src/upb_table.h
index c658a6e..a799653 100644
--- a/src/upb_table.h
+++ b/src/upb_table.h
@@ -133,7 +133,7 @@ INLINE void *_upb_inttable_fastlookup(upb_inttable *t, uint32_t key,
   while (1) {
     //DEBUGPRINTF("%d, %d, %d\n", e->val.has_entry, e->hdr.key, key);
     if (e->hdr.key == key) {
-      DEBUGPRINTF("returning val from hash part\n");
+      //DEBUGPRINTF("returning val from hash part\n");
       return &e->val;
     }
     if ((bucket = e->hdr.next) == UPB_END_OF_CHAIN) return NULL;
diff --git a/src/upb_textprinter.c b/src/upb_textprinter.c
index c0fb944..7e99ebd 100644
--- a/src/upb_textprinter.c
+++ b/src/upb_textprinter.c
@@ -20,8 +20,7 @@ struct _upb_textprinter {
 
 #define CHECK(x) if ((x) < 0) goto err;
 
-static int upb_textprinter_indent(upb_textprinter *p)
-{
+static int upb_textprinter_indent(upb_textprinter *p) {
   if(!p->single_line)
     for(int i = 0; i < p->indent_depth; i++)
       CHECK(upb_bytesink_putstr(p->bytesink, UPB_STRLIT("  "), &p->status));
@@ -107,8 +106,7 @@ err:
   return UPB_BREAK;
 }
 
-static upb_flow_t upb_textprinter_endsubmsg(void *_p, upb_fielddef *f)
-{
+static upb_flow_t upb_textprinter_endsubmsg(void *_p, upb_fielddef *f) {
   (void)f;
   upb_textprinter *p = _p;
   p->indent_depth--;