From 123210a989eae73fb60146d707f7ba890f8311ee Mon Sep 17 00:00:00 2001
From: Scott Shawcroft <scott@tannewt.org>
Date: Thu, 19 Nov 2020 17:13:24 -0800
Subject: [PATCH] Speed up JSON parsing with readinto

Get a chunk of data from readinto instead of a single byte. This
speeds up the parsing by reducing the number of function calls.

Fixes #3703
---
 extmod/modujson.c | 35 ++++++++++++++++++++++++++---------
 1 file changed, 26 insertions(+), 9 deletions(-)

diff --git a/extmod/modujson.c b/extmod/modujson.c
index 431b5e0cf1..515606b70a 100644
--- a/extmod/modujson.c
+++ b/extmod/modujson.c
@@ -57,6 +57,8 @@ typedef struct _ujson_stream_t {
     int errcode;
     mp_obj_t python_readinto[2 + 1];
     mp_obj_array_t bytearray_obj;
+    size_t start;
+    size_t end;
     byte cur;
 } ujson_stream_t;
 
@@ -77,28 +79,43 @@ STATIC byte ujson_stream_next(ujson_stream_t *s) {
     return s->cur;
 }
 
+// We read from an object's `readinto` method in chunks larger than the json
+// parser needs to reduce the number of function calls done.
+
+#define CIRCUITPY_JSON_READ_CHUNK_SIZE 64
+
 STATIC mp_uint_t ujson_python_readinto(mp_obj_t obj, void *buf, mp_uint_t size, int *errcode) {
     ujson_stream_t* s = obj;
-    s->bytearray_obj.items = buf;
-    s->bytearray_obj.len = size;
-    *errcode = 0;
-    mp_obj_t ret = mp_call_method_n_kw(1, 0, s->python_readinto);
-    if (ret == mp_const_none) {
-        *errcode = MP_EAGAIN;
-        return MP_STREAM_ERROR;
+
+    if (s->start == s->end) {
+        *errcode = 0;
+        mp_obj_t ret = mp_call_method_n_kw(1, 0, s->python_readinto);
+        if (ret == mp_const_none) {
+            *errcode = MP_EAGAIN;
+            return MP_STREAM_ERROR;
+        }
+        s->start = 0;
+        s->end = mp_obj_get_int(ret);
     }
-    return mp_obj_get_int(ret);
+
+    *((uint8_t *)buf) = ((uint8_t*) s->bytearray_obj.items)[s->start];
+    s->start++;
+    return 1;
 }
 
 STATIC mp_obj_t _mod_ujson_load(mp_obj_t stream_obj, bool return_first_json) {
     const mp_stream_p_t *stream_p = mp_proto_get(MP_QSTR_protocol_stream, stream_obj);
     ujson_stream_t s;
+    uint8_t character_buffer[CIRCUITPY_JSON_READ_CHUNK_SIZE];
     if (stream_p == NULL) {
+        s.start = 0;
+        s.end = 0;
         mp_load_method(stream_obj, MP_QSTR_readinto, s.python_readinto);
         s.bytearray_obj.base.type = &mp_type_bytearray;
         s.bytearray_obj.typecode = BYTEARRAY_TYPECODE;
+        s.bytearray_obj.len = CIRCUITPY_JSON_READ_CHUNK_SIZE;
         s.bytearray_obj.free = 0;
-        // len and items are set at read time
+        s.bytearray_obj.items = character_buffer;
         s.python_readinto[2] = MP_OBJ_FROM_PTR(&s.bytearray_obj);
         s.stream_obj = &s;
         s.read = ujson_python_readinto;