Add first working version

README.md

@@ -1,5 +1,5 @@
-GCIMPORT
-========
+GCCONV
+======
 
 GnuCash import helper.
 

gcconv

@@ -0,0 +1,58 @@
+#!/usr/bin/python
+# coding=utf-8
+"""Convert bank statement CSV to GnuCash-importable CSV
+
+usage: gcconv [options] <format> <file>
+       gcconv --list
+
+Options:
+    -e, --encoding ENCODING     input file encoding as understood by Python [default: utf-8]
+    -l, --list                  list known formats
+    --help                      show this screen
+    --version                   show version
+"""
+
+__VERSION__ = "gcconv 0.0.1"
+
+import sys
+
+import docopt
+
+import gcconv
+
+class App(object):
+    """
+    Main application
+    """
+
+    class UsageError(ValueError):
+        """Usage error"""
+        pass
+
+    def __init__(self, args):
+        self.args = args
+
+    @classmethod
+    def main(cls, args):
+        """
+        Main entry point
+        """
+        app = cls(args)
+        app.run()
+
+    def run(self):
+        """
+        Run the app (main entry point)
+        """
+        if self.args['--list']:
+            print "\n".join(gcconv.valid_formats())
+            sys.exit(0)
+        print gcconv.convert_file(
+            self.args['<file>'],
+            self.args['<format>'],
+            encoding=self.args['--encoding']
+        )
+
+
+if __name__ == "__main__":
+    App.main(docopt.docopt(__doc__, version=__VERSION__))

gcconv.py

@@ -0,0 +1,268 @@
+# coding=utf-8
+"""Convert Bank statement CSV to Gnucash-acceptable format"""
+
+
+def valid_formats():
+    return _parsers.keys()
+
+
+def convert_file(fpath, fmt, encoding='utf-8'):
+    try:
+        parser_class = _parsers[fmt]
+    except KeyError:
+        raise ValueError('unsupported format: %s' % fmt)
+    else:
+        parser = parser_class()
+        parser.parse_file(fpath, encoding)
+        return parser.format_csv()
+
+
+def _unquote(value):
+    if len(value) < 2:
+        return value
+    if value[0] == value[-1] == "'":
+        return value[1:-1]
+    if value[0] == value[-1] == '"':
+        return value[1:-1]
+    return value
+
+
+class BaseTransaction(object):
+    """A CSV-imported transaction"""
+    pass
+
+
+class GCTransaction(object):
+    """A Gnucash-acceptable transaction"""
+
+    def __init__(self, date, deposit, withdrawal, description):
+        self.date = date
+        self.deposit = deposit
+        self.withdrawal = withdrawal
+        self.description = description
+
+    def __str__(self):
+        """Naive CSV line printer"""
+        values = [self.date, self.deposit, self.withdrawal, self.description]
+        return ";".join(['"%s"' % v for v in values])
+
+
+class BaseTransactionParser(object):
+    """A statement parser"""
+    name = "__NONE__"
+
+    def __init__(self):
+        self.transactions = []
+        self.state = False
+        self.next_state = False
+
+    def is_open(self):
+        return self.state
+
+    def is_closed(self):
+        return not self.state
+
+    def parse_file(self, fpath, encoding='utf-8'):
+        lines_read = 0
+        with open(fpath) as f:
+            for line in f.readlines():
+                line = line.decode(encoding).strip()
+                lines_read += 1
+                self.state = self.pick_state(line, lineno=lines_read)
+                if self.is_open():
+                    self.transactions.append(self.parse_line(line).to_gc())
+
+    def format_csv(self):
+        return "\n".join([
+            unicode(t).encode('utf-8')
+            for t in self.transactions
+        ])
+
+
+class MbankTransaction(BaseTransaction):
+    """Item in an Mbank CSV"""
+
+    # Datum uskutečnění transakce
+    # Datum zaúčtování transakce
+    # Popis transakce
+    # Zpráva pro příjemce
+    # Plátce/Příjemce
+    # Číslo účtu plátce/příjemce
+    # KS
+    # VS
+    # SS
+    # Částka transakce
+    # Účetní zůstatek po transakci
+
+    def __init__(self, line):
+        fields = [self._cleanup_field(f) for f in line.split(";")]
+        self.date_r = self._convert_date(fields.pop(0))
+        self.date_b = self._convert_date(fields.pop(0))
+        __ = fields.pop()
+        self.new_balance = self._parse_currency(fields.pop())
+        amount = self._parse_currency(fields.pop())
+        if amount >= 0:
+            self.amountd = amount
+            self.amountw = 0
+        else:
+            self.amountd = 0
+            self.amountw = -amount
+        self._scrap = fields
+
+    def _cleanup_field(self, field):
+        x = ' '.join(_unquote(field).strip().split())
+        return '' if x == '-' else x
+
+    def _convert_date(self, text):
+        day, month, year = text.split('-')
+        return '-'.join([year, month, day])
+
+    def _description(self):
+        type_, message, party, number, ks, vs, ss = self._scrap
+        message = message.replace(u' DATUM PROVEDENÍ TRANSAKCE: ', '; DATUM: ')
+        out = []
+        if type_:
+            out.append(type_)
+        if message:
+            out.append(message)
+        if party:
+            out.append(party)
+        return " / ".join(out)
+
+    def _parse_currency(self, text):
+        """Read Mbank currency format"""
+        num = text.replace(' ', '')
+        num = num.replace(',', '.')
+        return float(num)
+
+    def to_gc(self):
+        """Convert to GCTransaction"""
+        return GCTransaction(
+            date=self.date_r,
+            deposit=self.amountd,
+            withdrawal=self.amountw,
+            description=self._description()
+        )
+
+
+class MbankTransactionParser(BaseTransactionParser):
+
+    def parse_line(self, line):
+        return MbankTransaction(line)
+
+    def pick_state(self, line, lineno=None):
+        """Choose parser state according to current line and line no."""
+        if self.next_state:
+            self.next_state = False
+            return True
+        if self.is_closed() and line.startswith(u'#Datum uskute'):
+            self.next_state = True
+            return False
+        if self.is_open() and not line:
+            return False
+        return self.state
+
+
+class FioTransaction(BaseTransaction):
+
+    # ID operace
+    # Datum
+    # Objem
+    # Měna
+    # Protiúčet
+    # Název protiúčtu
+    # Kód banky
+    # Název banky
+    # KS
+    # VS
+    # SS
+    # Poznámka
+    # Zpráva pro příjemce
+    # Typ
+    # Provedl
+    # Upřesnění
+    # Poznámka
+    # BIC
+    # ID pokynu
+
+    def __init__(self, line):
+        fields = [self._cleanup_field(f) for f in line.split(";")]
+        __ = fields.pop(0)
+        self.date = self._convert_date(fields.pop(0))
+        amount = self._parse_currency(fields.pop(0))
+        if amount >= 0:
+            self.amountd = amount
+            self.amountw = 0
+        else:
+            self.amountd = 0
+            self.amountw = -amount
+        self.currency = fields.pop(0)
+        assert self.currency == 'CZK'
+        self._scrap = [self._cleanup_field(f) for f in fields]
+
+    def _cleanup_field(self, field):
+        x = ' '.join(_unquote(field).strip().split())
+        return '' if x == '-' else x
+
+    def _description(self):
+        (
+            party_acc,
+            party_accnname,
+            party_bankcode,
+            party_bankname,
+            ks,
+            vs,
+            ss,
+            note1,
+            message,
+            type_,
+            who,
+            explanation,
+            note2,
+            bic,
+            comm_id
+            ) = self._scrap
+        out = []
+        if message:
+            out.append(message)
+        return " / ".join(out)
+
+    def _convert_date(self, text):
+        day, month, year = text.split('.')
+        return '-'.join([year, month, day])
+
+    def _parse_currency(self, text):
+        return float(text)
+
+    def to_gc(self):
+        """Convert to GCTransaction"""
+        return GCTransaction(
+            date=self.date,
+            deposit=self.amountd,
+            withdrawal=self.amountw,
+            description=self._description()
+        )
+
+
+class FioTransactionParser(BaseTransactionParser):
+
+    def parse_line(self, line):
+        return FioTransaction(line)
+
+    def pick_state(self, line, lineno=None):
+        """Choose parser state according to current line and line no."""
+        if self.next_state:
+            self.next_state = False
+            return True
+        if self.is_closed() and line.startswith(u'"ID operace";"Datum";'):
+            self.next_state = True
+            return False
+        if self.is_open() and line.startswith(u'"bankId";"2010"'):
+            return False
+        return self.state
+
+
+_parsers = {
+    'CzMbank': MbankTransactionParser,
+    'CzFio': FioTransactionParser,
+}