From 486331d1a845c50d92836271b0682ff739058a0f Mon Sep 17 00:00:00 2001
From: Whiteknight <wknight8111@gmail.com>
Date: Mon, 9 Apr 2012 19:34:33 -0400
Subject: [PATCH] Implement StringBuilder.push_integer vtable, so we can
 assemble strings by pushing raw codepoints instead of needing to first create
 a 1-character string (chr_s_i) and then pushing that string to the
 StringBuilder

---
 src/pmc/stringbuilder.pmc  | 75 +++++++++++++++++++++++++++++---------
 src/string/encoding/utf8.c |  4 +-
 2 files changed, 60 insertions(+), 19 deletions(-)

diff --git a/src/pmc/stringbuilder.pmc b/src/pmc/stringbuilder.pmc
index 3878c5f55b..dbf927cabc 100644
--- a/src/pmc/stringbuilder.pmc
+++ b/src/pmc/stringbuilder.pmc
@@ -27,7 +27,15 @@ PARROT_WARN_UNUSED_RESULT
 PARROT_CONST_FUNCTION
 static size_t calculate_capacity(PARROT_INTERP, size_t needed);
 
+static void convert_encoding(PARROT_INTERP,
+    STR_VTABLE * dest_encoding,
+    STRING * buffer,
+    size_t size_to_add)
+        __attribute__nonnull__(1);
+
 #define ASSERT_ARGS_calculate_capacity __attribute__unused__ int _ASSERT_ARGS_CHECK = (0)
+#define ASSERT_ARGS_convert_encoding __attribute__unused__ int _ASSERT_ARGS_CHECK = (\
+       PARROT_ASSERT_ARG(interp))
 /* Don't modify between HEADERIZER BEGIN / HEADERIZER END.  Your changes will be lost. */
 /* HEADERIZER END: static */
 
@@ -183,24 +191,8 @@ Append string to current buffer.
                 if (s->encoding != Parrot_utf8_encoding_ptr)
                     s = Parrot_utf8_encoding_ptr->to_encoding(interp, s);
 
-                if (buffer->encoding != Parrot_utf8_encoding_ptr) {
-                    /* Create new temporary string */
-                    STRING * new_buffer;
-
-                    new_buffer = Parrot_utf8_encoding_ptr->to_encoding(interp, buffer);
-                    total_size = new_buffer->bufused + s->bufused;
-
-                    if (total_size > buffer->_buflen) {
-                        /* Reallocate */
-                        total_size = calculate_capacity(INTERP, total_size);
-                        Parrot_gc_reallocate_string_storage(INTERP, buffer, total_size);
-                    }
-                    buffer->bufused  = new_buffer->bufused;
-                    buffer->encoding = new_buffer->encoding;
-
-                    memcpy(buffer->strstart, new_buffer->strstart,
-                            new_buffer->bufused);
-                }
+                if (buffer->encoding != Parrot_utf8_encoding_ptr)
+                    convert_encoding(INTERP, Parrot_utf8_encoding_ptr, buffer, s->bufused);
             }
         }
 
@@ -228,6 +220,32 @@ Append string to current buffer.
         PARROT_ASSERT(buffer->bufused <= Buffer_buflen(buffer));
     }
 
+    VTABLE void push_integer(INTVAL value) {
+        STRING * s = PARROT_STRINGBUILDER(SELF)->buffer;
+        String_iter iter;
+        size_t total_size;
+
+        if (s->encoding != Parrot_utf8_encoding_ptr && value > 0x7F) {
+            if (s->strlen == 0)
+                s->encoding = Parrot_utf8_encoding_ptr;
+            else
+                convert_encoding(INTERP, Parrot_utf8_encoding_ptr, s, sizeof (INTVAL));
+        }
+
+        total_size = s->bufused + sizeof (INTVAL);
+        if (total_size > s->_buflen) {
+            total_size = calculate_capacity(INTERP, total_size);
+            Parrot_gc_reallocate_string_storage(INTERP, s, total_size);
+        }
+
+        STRING_ITER_INIT(INTERP, &iter);
+        iter.charpos = s->strlen;
+        iter.bytepos = s->bufused;
+        STRING_iter_set_and_advance(INTERP, s, &iter, value);
+        s->strlen = iter.charpos;
+        s->bufused = iter.bytepos;
+    }
+
 /*
 
 =item C<VTABLE i_concatenate_str()>
@@ -484,6 +502,27 @@ calculate_capacity(SHIM_INTERP, size_t needed)
     return needed;
 }
 
+static void
+convert_encoding(PARROT_INTERP, STR_VTABLE * dest_encoding, STRING * buffer, size_t size_to_add)
+{
+    STRING * new_buffer;
+    size_t total_size;
+
+    new_buffer = dest_encoding->to_encoding(interp, buffer);
+    total_size = new_buffer->bufused + size_to_add;
+
+    if (total_size > buffer->_buflen) {
+        /* Reallocate */
+        total_size = calculate_capacity(interp, total_size);
+        Parrot_gc_reallocate_string_storage(interp, buffer, total_size);
+    }
+    buffer->bufused  = new_buffer->bufused;
+    buffer->encoding = new_buffer->encoding;
+
+    memcpy(buffer->strstart, new_buffer->strstart,
+            new_buffer->bufused);
+}
+
 /*
  * Local variables:
  *   c-file-style: "parrot"
diff --git a/src/string/encoding/utf8.c b/src/string/encoding/utf8.c
index 2e52a2cfd4..9ae7e30f12 100644
--- a/src/string/encoding/utf8.c
+++ b/src/string/encoding/utf8.c
@@ -577,7 +577,9 @@ utf8_iter_set_and_advance(PARROT_INTERP,
     i->charpos += 1;
     i->bytepos += end - ptr;
 
-    PARROT_ASSERT(i->bytepos <= str->bufused);
+    /* Commenting this out allows us to iterate and assemble strings in empty
+       (but pre-allocated) buffers. */
+    /* PARROT_ASSERT(i->bytepos <= str->bufused); */
 }