go-ethereum – How to Scrape Ethereum Blockchain in Human-Readable Format (CSV) – Step-by-Step Instructions

I wanted to scrape the whole ethereum blockchain in csv format. For this I reverse engineered and extended this code by vitalik: https://github.com/ethereum/research/blob/master/uncle_regressions/block_datadump_generator.py

My version is here:
https://github.com/ankitchiplunkar/analyzeEthereum

The problem is that I can download the blockchain till block 2,675,000 (EIP 155 hardfork) but after that I get the following error:

>>> tempTxDictionary = tempTx.to_dict()
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/******/anaconda2/envs/pyethapp/lib/python2.7/site-packages/ethereum/transactions.py",
line 155, in to_dict
d['sender'] = self.sender
File "/home/******/anaconda2/envs/pyethapp/lib/python2.7/site-packages/ethereum/transactions.py",
line 81, in sender
raise InvalidTransaction("Invalid signature values!")
ethereum.exceptions.InvalidTransaction: Invalid signature values!

On a closer look I can see that

tempTx.v = 37 or 38

The yellow paper says that "v" should be either 27 or 28 (equation 211).
There is something different in how pyethereum/transactions.py is reading the block after EIP155.

What should I do to fix this?

PS: here is a simpler method to recreate the error

import rlp
from ethereum.blocks import BlockHeader
from ethereum.transactions import Transaction
from ethereum import utils
import csv
gethDumpFileName = 'geth.dump'
f = open(gethDumpFileName)
pos = 3561286829 # position of block 2,675,002
f.seek(pos)
prefix = f.read(10)
_typ, _len, _pos = rlp.codec.consume_length_prefix(prefix, 0)
blkdata = prefix + f.read(_pos + _len - 10)
header = rlp.decode(rlp.descend(blkdata, 0), BlockHeader)
headerDictionary = header.to_dict()
i = 0
tempTx = rlp.decode(rlp.descend(blkdata, 1, i), Transaction)
tempTx.v = tempTx.v -10
tempTxDictionary = tempTx.to_dict()