use git VLQ varint encoding

analysis/code_analysis.py

@@ -1,8 +1,6 @@
 import csv
-import io
 import json
 import sys
-import leb128
 from collections import defaultdict
 from dataclasses import dataclass, field
 from pathlib import Path
@@ -52,6 +50,36 @@ def test_decode_varint():
     assert decode_varint(bytes([0xFF, 0xFF, 0x7f])) == (2113663, 3)
 
 
+# int encode_varint(uintmax_t value, unsigned char *buf)
+# {
+# 	unsigned char varint[16];
+# 	unsigned pos = sizeof(varint) - 1;
+# 	varint[pos] = value & 127;
+# 	while (value >>= 7)
+# 		varint[--pos] = 128 | (--value & 127);
+# 	if (buf)
+# 		memcpy(buf, varint + pos, sizeof(varint) - pos);
+# 	return sizeof(varint) - pos;
+# }
+
+def encode_varint(value) -> bytes:
+    varint = bytearray()
+    varint.append(value & 127)
+    while value := value >> 7:
+        value -= 1
+        varint.append(128 | (value & 127))
+    ret = bytes(reversed(varint))
+    return ret
+
+
+def test_encode_varint():
+    assert encode_varint(0) == bytes([0])
+    assert encode_varint(1) == bytes([1])
+    assert encode_varint(127) == bytes([127])
+    assert encode_varint(128) == bytes([0x80, 0])
+    assert encode_varint(2113663) == bytes([0xFF, 0xFF, 0x7f])
+
+
 @dataclass
 class Chunk:
     first_instruction_offset: int
@@ -257,20 +285,16 @@ def encode(self, chunks: dict[int, int]) -> tuple[bytes, int]:
             assert 0 <= value < self.VALUE_MOD
             delta = i - last_chunk_index
             e = delta * self.VALUE_MOD + value
-            ops += leb128.u.encode(e)
+            ops += encode_varint(e)
             last_chunk_index = i + 1
         return ops, 8 * len(ops)
 
     def decode(self, ops: bytes) -> dict[int, int]:
-        stream = io.BytesIO(ops)
-        stream.seek(0, 2)
-        end = stream.tell()
-        stream.seek(0, 0)
-
         m = {}
         index = 0
-        while stream.tell() != end:
-            e, _ = leb128.u.decode_reader(stream)
+        while ops:
+            e, move = decode_varint(ops)
+            ops = ops[move:]
             delta = e // self.VALUE_MOD
             value = e % self.VALUE_MOD
             index += delta

analysis/top_bytecodes_analysis.csv

@@ -1,5 +1,5 @@
 example address,earliest block,latest block,gas used,code length,code chunks,push bytes,PUSH1 zeros,jumpdests,invalid jumpdests,scheme f11,scheme f10,scheme f9,scheme f8,VLQM33
-total,52029,19759709,179861783901613,10635333,332355,40.42%,1.82%,3.11%,0.07%,12820,12663,12477,11526,11303
+total,52029,19759709,179861783901613,10635333,332355,40.42%,1.82%,3.11%,0.07%,12820,12663,12477,11526,11302
 0xc02aaa39b223fe8d0a0e5c4f27ead9083c756cc2,4753408,19759709,14946073640548,3124,98,49.52%,2.21%,2.05%,0.16%,9,8,8,7,7
 0x00001bea43608c5ee487f82b773af8bd7cb20a6f,10008355,19759709,13147711796412,11293,353,52.43%,1.30%,2.33%,0.03%,5,4,4,3,3
 0x7a250d5630b4cf539739df2c5dacb4c659f2488d,10208463,19759709,11636690698027,21943,686,49.76%,1.59%,2.13%,0.00%,2,2,3,2,2
@@ -669,7 +669,7 @@ total,52029,19759709,179861783901613,10635333,332355,40.42%,1.82%,3.11%,0.07%,12
 0x9ebfb53fa8526906738856848a27cb11b0285c3f,15223734,18295281,21043686342,14538,455,39.31%,2.27%,3.52%,0.01%,5,4,4,4,4
 0xcca06cd29c61123d9d65b904b18174382380ca64,12297849,19433031,20981375879,6803,213,28.03%,1.93%,3.07%,0.04%,7,7,6,6,5
 0xd8de6af55f618a7bc69835d55ddc6582220c36c0,13181841,19755829,20979207171,22142,692,33.21%,1.43%,2.96%,0.04%,17,15,16,14,14
-0x79a8c46dea5ada233abaffd40f3a0a2b1e5a4f27,9567303,19744897,20871789593,21590,675,36.84%,3.15%,3.37%,0.01%,9,8,7,7,7
+0x79a8c46dea5ada233abaffd40f3a0a2b1e5a4f27,9567303,19744897,20871789593,21590,675,36.84%,3.15%,3.37%,0.01%,9,8,7,7,6
 0x41e5560054824ea6b0732e656e3ad64e20e94e45,4009734,19759699,20848852723,4160,130,31.01%,1.13%,3.51%,0.10%,7,7,6,5,5
 0x84a0856b038eaad1cc7e297cf34a7e72685a8693,12216864,19758870,20755603594,5340,167,36.93%,1.33%,3.45%,0.07%,9,8,8,8,8
 0xf8a95b2409c27678a6d18d950c5d913d5c38ab03,12379095,19271682,20721735531,22142,692,33.21%,1.43%,2.96%,0.05%,18,18,18,16,16