Merge bc19ed6 into 9eb4361

beancount_import/source/generic_importer_source.py

@@ -0,0 +1,164 @@
+"""This module implements a Source Subclass for wrapping
+`beancount.ingest.importer.ImporterProtocol` subclasses importers.
+The importers are considered athoritative of the account they represent.
+
+The Transaction.narration set by each importer is copied to Posting.meta[source_desc]
+This helps in predicting postings for similar transaction while allowing the
+user to change the Transaction description and payee from UI
+(see readme.md for more on source_desc)
+This `source_desc` meta is also used for check cleared postings and should not be
+changed manually
+
+Author: Sufiyan Adhikari(github.com/dumbPy)
+"""
+
+import os
+import hashlib
+from glob import glob
+from typing import List
+from collections import defaultdict
+import itertools
+import datetime
+
+from beancount.core.data import Transaction, Posting,  Directive
+from beancount.core.amount import Amount
+from beancount.ingest.importer import ImporterProtocol
+from beancount.core.compare import hash_entry
+from beancount.ingest.cache import get_file
+
+from ..matching import FIXME_ACCOUNT, SimpleInventory
+from . import ImportResult, SourceResults
+from ..journal_editor import JournalEditor
+from .description_based_source import DescriptionBasedSource, get_pending_and_invalid_entries
+from .mint import _get_key_from_posting
+
+
+class ImporterSource(DescriptionBasedSource):
+    def __init__(self,
+                 directory: str,
+                 account: str,
+                 importer: ImporterProtocol,
+                 **kwargs) -> None:
+        super().__init__(**kwargs)
+        self.directory = os.path.expanduser(directory)
+        self.importer = importer
+        self.account = account
+
+        # get _FileMemo object for each file
+        files = [get_file(f) for f in
+                    filter(os.path.isfile,
+                 glob(os.path.join(directory, '**', '*'), recursive=True)
+                           )
+        ]
+        # filter the valid files for this importer
+        self.files = [f for f in files if self.importer.identify(f)]
+
+    @property
+    def name(self):
+        return self.importer.name()
+
+    def prepare(self, journal: 'JournalEditor', results: SourceResults) -> None:
+        results.add_account(self.account)
+        entries = defaultdict(list)
+        for f in self.files:
+            f_entries = self.importer.extract(f, existing_entries=journal.entries)
+            # collect  all entries in current statement, grouped by hash
+            hashed_entries = defaultdict(list)
+            for entry in f_entries:
+                key_ = self._get_key_from_imported_entry(entry)
+                self._add_description(entry)
+                hashed_entries[key_].append(entry)
+            # deduplicate across statements
+            for key_ in hashed_entries:
+                # skip the existing entries from other statements. add remaining
+                n = len(entries[key_])
+                entries[key_].extend(hashed_entries[key_][n:])
+
+        get_pending_and_invalid_entries(
+            raw_entries=list(itertools.chain.from_iterable(entries.values())),
+            journal_entries=journal.all_entries,
+            account_set=set([self.account]),
+            get_key_from_posting=_get_key_from_posting,
+            get_key_from_raw_entry=self._get_key_from_imported_entry,
+            make_import_result=self._make_import_result,
+            results=results)
+
+    def _add_description(self, entry: Transaction):
+        if not isinstance(entry, Transaction): return None
+        postings = entry.postings #type: ignore
+        to_mutate = []
+        for i, posting in enumerate(postings):
+            if posting.account != self.account: continue
+            if isinstance(posting.meta, dict):
+                posting.meta["source_desc"] = entry.narration
+                posting.meta["date"] = entry.date
+                break
+            else:
+                to_mutate.append(i)
+                break
+        for i in to_mutate:
+            p = postings.pop(i)
+            p = Posting(p.account, p.units, p.cost, p.price, p.flag,
+                        {"source_desc":entry.narration, "date": entry.date})
+            postings.insert(i, p)
+
+    def _get_source_posting(self, entry:Transaction):
+        for posting in entry.postings:
+            if posting.account == self.account: return posting
+
+    def is_posting_cleared(self, posting: Posting) -> bool:
+        """Given than this source is athoritative of the account of a particular posting,
+        return if that posting is cleared.
+        All postings which have `source_desc` meta key are considered cleared
+        """
+        if posting.account != self.account: return False
+        return super().is_posting_cleared(posting)
+
+    def _get_key_from_imported_entry(self, entry:Transaction):
+        return (self.account,
+                entry.date,
+                self._get_source_posting(entry).units,
+                entry.narration)
+
+    def _make_import_result(self, imported_entry:Directive):
+        if isinstance(imported_entry, Transaction): balance_amounts(imported_entry)
+        result = ImportResult(
+            date=imported_entry.date, info=get_info(imported_entry), entries=[imported_entry])
+        # delete filename since it is used by beancount-import to determine if the
+        # entry is from journal.
+        imported_entry.meta.pop('filename')
+        return result
+
+def _get_key_from_posting(entry: Transaction, posting: Posting,
+                          source_postings: List[Posting], source_desc: str,
+                          posting_date: datetime.date):
+    del entry
+    del source_postings
+    return (posting.account, posting_date, posting.units, source_desc)
+
+def get_info(raw_entry: Directive) -> dict:
+    return dict(
+        type=get_file(raw_entry.meta['filename']).mimetype(),
+        filename=raw_entry.meta['filename'],
+        line=raw_entry.meta['lineno'],
+    )
+
+def balance_amounts(txn:Transaction)-> None:
+    """Add FIXME account for the remaing amount to balance accounts"""
+    inventory = SimpleInventory()
+    for posting in txn.postings:
+        inventory += posting.units
+    for currency in inventory:
+        txn.postings.append(
+            Posting(
+                account=FIXME_ACCOUNT,
+                units=Amount(currency=currency, number=-inventory[currency]),
+                cost=None,
+                price=None,
+                flag=None,
+                meta={},
+            ))
+
+
+def load(spec, log_status):
+    return ImporterSource(log_status=log_status, **spec)

beancount_import/source/generic_importer_source_test.py

@@ -0,0 +1,38 @@
+import os
+
+import pytest
+
+from .source_test import check_source_example
+from beancount.ingest.importers.csv import Importer as CSVImporter, Col
+
+testdata_dir = os.path.realpath(
+    os.path.join(
+        os.path.dirname(__file__), '..', '..', 'testdata', 'source', 'generic_importer'))
+
+examples = [
+    'test_basic',
+    'test_invalid',
+    'test_training_examples'
+]
+
+importer = CSVImporter({Col.DATE: 'Date',
+                        Col.NARRATION1: 'Description',
+                        Col.AMOUNT: 'Amount',
+                        },
+                       'Assets:Bank',
+                       'USD',
+                       '"Date","Description","Amount"',
+                       )
+
+
+@pytest.mark.parametrize('name', examples)
+def test_source(name: str):
+    check_source_example(
+        example_dir=os.path.join(testdata_dir, name),
+        source_spec={
+            'module': 'beancount_import.source.generic_importer_source',
+            'directory': testdata_dir,
+            'account': 'Assets:Bank',
+            'importer': importer,
+        },
+        replacements=[(testdata_dir, '<testdata>')])

testdata/source/generic_importer/generic_statement.csv

@@ -0,0 +1,7 @@
+"Date","Description","Amount"
+2020-01-01,by debit card-OTHPG 063441 GOOGLE CLOUD INDIA PVTTHANE-,-1
+2020-01-01,by debit card-OTHPG 063444 GOOGLE CLOUD INDIA PVTTHANE-,-1
+2020-01-02,BULK POSTING- 00000008237 250120 GOOGLE CLOUD INDIA PVT-,1
+2020-01-02,ATM-WD Some Random ATM Machine,500
+2020-01-02,BULK POSTING- 00000008237 250120 GOOGLE CLOUD INDIA PVT-,1
+2020-01-05,Transfer to 1234567890123,300

testdata/source/generic_importer/test_basic/accounts.txt

@@ -0,0 +1 @@
+Assets:Bank

testdata/source/generic_importer/test_basic/import_results.beancount

@@ -0,0 +1,113 @@
+;; date: 2020-01-01
+;; info: {"filename": "<testdata>/generic_statement.csv", "line": 1, "type": "text/csv"}
+
+; features: [
+;             {
+;               "amount": "-1 USD",
+;               "date": "2020-01-01",
+;               "key_value_pairs": {
+;                 "desc": "by debit card-OTHPG 063441 GOOGLE CLOUD INDIA PVTTHANE-"
+;               },
+;               "source_account": "Assets:Bank"
+;             }
+;           ]
+2020-01-01 * "by debit card-OTHPG 063441 GOOGLE CLOUD INDIA PVTTHANE-"
+  Assets:Bank     -1 USD
+    date: 2020-01-01
+    source_desc: "by debit card-OTHPG 063441 GOOGLE CLOUD INDIA PVTTHANE-"
+  Expenses:FIXME   1 USD
+
+;; date: 2020-01-01
+;; info: {"filename": "<testdata>/generic_statement.csv", "line": 2, "type": "text/csv"}
+
+; features: [
+;             {
+;               "amount": "-1 USD",
+;               "date": "2020-01-01",
+;               "key_value_pairs": {
+;                 "desc": "by debit card-OTHPG 063444 GOOGLE CLOUD INDIA PVTTHANE-"
+;               },
+;               "source_account": "Assets:Bank"
+;             }
+;           ]
+2020-01-01 * "by debit card-OTHPG 063444 GOOGLE CLOUD INDIA PVTTHANE-"
+  Assets:Bank     -1 USD
+    date: 2020-01-01
+    source_desc: "by debit card-OTHPG 063444 GOOGLE CLOUD INDIA PVTTHANE-"
+  Expenses:FIXME   1 USD
+
+;; date: 2020-01-02
+;; info: {"filename": "<testdata>/generic_statement.csv", "line": 3, "type": "text/csv"}
+
+; features: [
+;             {
+;               "amount": "1 USD",
+;               "date": "2020-01-02",
+;               "key_value_pairs": {
+;                 "desc": "BULK POSTING- 00000008237 250120 GOOGLE CLOUD INDIA PVT-"
+;               },
+;               "source_account": "Assets:Bank"
+;             }
+;           ]
+2020-01-02 * "BULK POSTING- 00000008237 250120 GOOGLE CLOUD INDIA PVT-"
+  Assets:Bank      1 USD
+    date: 2020-01-02
+    source_desc: "BULK POSTING- 00000008237 250120 GOOGLE CLOUD INDIA PVT-"
+  Expenses:FIXME  -1 USD
+
+;; date: 2020-01-02
+;; info: {"filename": "<testdata>/generic_statement.csv", "line": 5, "type": "text/csv"}
+
+; features: [
+;             {
+;               "amount": "1 USD",
+;               "date": "2020-01-02",
+;               "key_value_pairs": {
+;                 "desc": "BULK POSTING- 00000008237 250120 GOOGLE CLOUD INDIA PVT-"
+;               },
+;               "source_account": "Assets:Bank"
+;             }
+;           ]
+2020-01-02 * "BULK POSTING- 00000008237 250120 GOOGLE CLOUD INDIA PVT-"
+  Assets:Bank      1 USD
+    date: 2020-01-02
+    source_desc: "BULK POSTING- 00000008237 250120 GOOGLE CLOUD INDIA PVT-"
+  Expenses:FIXME  -1 USD
+
+;; date: 2020-01-02
+;; info: {"filename": "<testdata>/generic_statement.csv", "line": 4, "type": "text/csv"}
+
+; features: [
+;             {
+;               "amount": "500 USD",
+;               "date": "2020-01-02",
+;               "key_value_pairs": {
+;                 "desc": "ATM-WD Some Random ATM Machine"
+;               },
+;               "source_account": "Assets:Bank"
+;             }
+;           ]
+2020-01-02 * "ATM-WD Some Random ATM Machine"
+  Assets:Bank      500 USD
+    date: 2020-01-02
+    source_desc: "ATM-WD Some Random ATM Machine"
+  Expenses:FIXME  -500 USD
+
+;; date: 2020-01-05
+;; info: {"filename": "<testdata>/generic_statement.csv", "line": 6, "type": "text/csv"}
+
+; features: [
+;             {
+;               "amount": "300 USD",
+;               "date": "2020-01-05",
+;               "key_value_pairs": {
+;                 "desc": "Transfer to 1234567890123"
+;               },
+;               "source_account": "Assets:Bank"
+;             }
+;           ]
+2020-01-05 * "Transfer to 1234567890123"
+  Assets:Bank      300 USD
+    date: 2020-01-05
+    source_desc: "Transfer to 1234567890123"
+  Expenses:FIXME  -300 USD

testdata/source/generic_importer/test_basic/journal.beancount

@@ -0,0 +1 @@
+1900-01-01 open Assets:Bank

testdata/source/generic_importer/test_basic/training_examples.json

@@ -0,0 +1 @@
+[]

testdata/source/generic_importer/test_invalid/accounts.txt

@@ -0,0 +1 @@
+Assets:Bank

testdata/source/generic_importer/test_invalid/import_results.beancount

@@ -0,0 +1 @@
+