feat: improvements to pdfreader

Refactored pdfreader to be more extensible and added a mercurycards importer which uses the pdfreader to read credit card transactions.

beancount_reds_importers/importers/genericpdfpaycheck/__init__.py

@@ -25,6 +25,7 @@ def custom_init(self):
         self.pdf_table_extraction_settings = {"join_tolerance": 4, "snap_tolerance": 4}
         self.pdf_table_extraction_crop = (0, 0, 0, 0)
         self.pdf_table_title_height = 0
+        self.pdf_page_break_top = 0
         # Set this true as you play with the extraction settings and crop to view images of what the pdf parser detects
         self.debug = True
 

beancount_reds_importers/importers/mercurycards/__init__.py

@@ -0,0 +1,163 @@
+"""Mercury Cards pdf importer for beancount."""
+
+import re
+from datetime import datetime
+
+import petl as etl
+
+from beancount_reds_importers.libreader import pdfreader
+from beancount_reds_importers.libtransactionbuilder import banking
+
+
+class Importer(banking.Importer, pdfreader.Importer):
+    IMPORTER_NAME = "Mercury Cards"
+
+    def custom_init(self):
+        if not self.custom_init_run:
+            self.max_rounding_error = 0.04
+            self.filename_pattern_def = "Mercury Statement *.pdf"
+            self.pdf_table_extraction_settings = {
+                "vertical_strategy": "text",
+                "horizontal_strategy": "text",
+            }
+            self.pdf_table_extraction_crop = (0, 0, 0, 0)
+            self.pdf_table_title_height = 0
+            self.pdf_page_break_top = 0
+            self.date_format = "%m/%d/%Y"
+            self.transaction_table_section = "table_1"
+            self.meta_text = ""
+            self.skip_transaction_types = {}
+            self.header_map = {
+                "Post Date": "settleDate",
+                "Trans Date": "date",
+                "Description": "memo",
+                "Reference": "reference",
+                "Amount": "amount",
+            }
+
+            # payee and narration are swapped
+            # We need to swap them back. See banking.py
+            self.get_payee = lambda ot: ot.memo
+            self.get_narration = lambda ot: None  # setting to none to use smart importer
+
+            self.debug = True
+            self.custom_init_run = True
+
+    def file_date(self, file):
+        if not self.file_read_done:
+            self.read_file(file)
+
+        return self.get_closing_date()
+
+    def get_closing_date(self):
+        if self.meta_text == "":
+            raise ValueError("No meta_text has been found")
+
+        # Pattern to match "Closing Date" followed by a date in mm/dd/yyyy format
+        pattern = r"Closing Date\s+(\d{2}/\d{2}/\d{4})"
+
+        # Search for all matches in self.meta_text
+        matches = re.findall(pattern, self.meta_text)
+
+        date_string = matches[0]
+        date_format = "%m/%d/%Y"
+        datetime_object = datetime.strptime(date_string, date_format)
+
+        return datetime_object
+
+    def get_adjusted_crop(self, page_idx, page):
+        """Dynamically find the crop positon based on the position of text found on the page."""
+        adjusted_crop = (0, 0, 1, 1)
+        table_start_search_text = "TRANSACTIONS"
+        table_start_search_results = page.search(table_start_search_text)
+        if table_start_search_results:
+            table_start = table_start_search_results[0]
+            table_start_x = table_start["x0"] - 30
+            table_start_y = table_start["bottom"] + 50
+
+            table_end_search_text = "YEAR-TO-DATE"
+            table_end_search_results = page.search(table_end_search_text)
+
+            if table_end_search_results:
+                table_end = table_end_search_results[0]
+                table_end_y = table_end["top"] - 10
+            else:
+                table_end_y = page.bbox[3]  # if no end text is found use the whole page
+
+            adjusted_crop = (
+                (table_start_x),
+                (table_start_y),
+                (page.bbox[2]),
+                (table_end_y),
+            )
+        return adjusted_crop
+
+    def fix_years(self, table):
+        """
+        Determine the correct year for the given date string (MM/DD format).
+        """
+
+        def get_year(d):
+            # Get the current year
+            current_year = self.get_closing_date().year
+
+            return f"{d}/{current_year}"
+
+        date_headers = ["Post Date", "Trans Date"]
+        for i in date_headers:
+            if i in table.header():
+                table = table.convert(i, lambda d: get_year(d))
+
+        return table
+
+    def prepare_tables(self):
+        """Make final adjustments to tables before processing by the transaction builder."""
+        for section, table in self.alltables.items():
+            # set table headers. table was goofy, so they had to be croped out
+            headers = [
+                "Post Date",
+                "Trans Date",
+                "Description",
+                "City",
+                "State",
+                "Reference",
+                "Amount",
+            ]
+            table = etl.wrap(etl.pushheader(table, headers))
+
+            # add year to mm/dd formatted date
+            table = self.fix_years(table)
+
+            table = table.rename(self.header_map)
+            table = self.convert_columns(table)
+
+            # the amounts should be negative since they're charges
+            table = etl.convert(table, "amount", lambda a: a * -1)
+
+            table = self.fix_column_names(table)
+            table = self.prepare_processed_table(
+                table
+            )  # override this to make additonal adjustments
+
+            self.alltables[section] = table
+
+        self.combine_tables()
+        return
+
+    def combine_tables(self):
+        # Initialize an empty table
+        combined_table = None
+
+        for section, table in self.alltables.items():
+            # Convert each table to a petl table
+            petl_table = etl.wrap(table)
+
+            # Combine tables
+            if combined_table is None:
+                combined_table = petl_table  # First table initializes the combined table
+            else:
+                combined_table = etl.cat(
+                    combined_table, petl_table
+                )  # Concatenate additional tables
+
+        return combined_table

beancount_reds_importers/importers/mercurycards/tests/mercury_statement_20241105.pdf.extract

@@ -0,0 +1,120 @@
+
+2024-10-05 * "Whole Foods"
+  Liabilities:Credit-Cards:Mercury  -15.01 USD
+
+2024-10-06 * "Car Wash"
+  Liabilities:Credit-Cards:Mercury  -35.30 USD
+
+2024-10-07 * "Taco Bell"
+  Liabilities:Credit-Cards:Mercury  -22.76 USD
+
+2024-10-07 * "Mcdonald's"
+  Liabilities:Credit-Cards:Mercury  -10.11 USD
+
+2024-10-08 * "Papa John's"
+  Liabilities:Credit-Cards:Mercury  -30.16 USD
+
+2024-10-10 * "Paypal"
+  Liabilities:Credit-Cards:Mercury  -33.97 USD
+
+2024-10-11 * "Amazon"
+  Liabilities:Credit-Cards:Mercury  -101.55 USD
+
+2024-10-11 * "Wm Supercenter"
+  Liabilities:Credit-Cards:Mercury  -53.44 USD
+
+2024-10-12 * "Amazon"
+  Liabilities:Credit-Cards:Mercury  -204.32 USD
+
+2024-10-12 * "Amazon"
+  Liabilities:Credit-Cards:Mercury  -4.90 USD
+
+2024-10-12 * "Target"
+  Liabilities:Credit-Cards:Mercury  -10.90 USD
+
+2024-10-13 * "Bp"
+  Liabilities:Credit-Cards:Mercury  -106.95 USD
+
+2024-10-14 * "Circle K"
+  Liabilities:Credit-Cards:Mercury  -50.69 USD
+
+2024-10-15 * "Amazon"
+  Liabilities:Credit-Cards:Mercury  -10.11 USD
+
+2024-10-15 * "Doordash"
+  Liabilities:Credit-Cards:Mercury  -73.82 USD
+
+2024-10-16 * "Mcdonald's"
+  Liabilities:Credit-Cards:Mercury  -10.68 USD
+
+2024-10-16 * "7-Eleven"
+  Liabilities:Credit-Cards:Mercury  -10.11 USD
+
+2024-10-17 * "Advance Auto Parts"
+  Liabilities:Credit-Cards:Mercury  -13.86 USD
+
+2024-10-18 * "Mcdonald's"
+  Liabilities:Credit-Cards:Mercury  -15.11 USD
+
+2024-10-18 * "Papa John's"
+  Liabilities:Credit-Cards:Mercury  -31.29 USD
+
+2024-10-18 * "Mcdonald's"
+  Liabilities:Credit-Cards:Mercury  -12.62 USD
+
+2024-10-19 * "Amazon Prime"
+  Liabilities:Credit-Cards:Mercury  -23.52 USD
+
+2024-10-19 * "Amazon"
+  Liabilities:Credit-Cards:Mercury  -30.46 USD
+
+2024-10-19 * "Amazon"
+  Liabilities:Credit-Cards:Mercury  -44.21 USD
+
+2024-10-19 * "Amazon"
+  Liabilities:Credit-Cards:Mercury  -23.57 USD
+
+2024-10-19 * "Amazon"
+  Liabilities:Credit-Cards:Mercury  -22.24 USD
+
+2024-10-19 * "Amazon"
+  Liabilities:Credit-Cards:Mercury  -69.23 USD
+
+2024-10-19 * "Amazon"
+  Liabilities:Credit-Cards:Mercury  -60.00 USD
+
+2024-10-19 * "Amazon"
+  Liabilities:Credit-Cards:Mercury  -57.07 USD
+
+2024-10-19 * "Amazon"
+  Liabilities:Credit-Cards:Mercury  -224.05 USD
+
+2024-10-20 * "Whole Foods"
+  Liabilities:Credit-Cards:Mercury  -6.39 USD
+
+2024-10-21 * "Doordash"
+  Liabilities:Credit-Cards:Mercury  -79.56 USD
+
+2024-10-21 * "Amazon"
+  Liabilities:Credit-Cards:Mercury  -20.24 USD
+
+2024-10-22 * "Papa John's"
+  Liabilities:Credit-Cards:Mercury  -37.85 USD
+
+2024-10-22 * "Mcdonald's"
+  Liabilities:Credit-Cards:Mercury  -10.11 USD
+
+2024-10-22 * "Racetrac"
+  Liabilities:Credit-Cards:Mercury  -32.90 USD
+
+2024-10-23 * "Mcdonald's"
+  Liabilities:Credit-Cards:Mercury  -10.11 USD
+
+2024-10-24 * "Doordash"
+  Liabilities:Credit-Cards:Mercury  -28.41 USD
+
+2024-10-24 * "Doordash"
+  Liabilities:Credit-Cards:Mercury  -40.83 USD
+
+2024-10-25 * "Doordash"
+  Liabilities:Credit-Cards:Mercury  -68.35 USD

beancount_reds_importers/importers/mercurycards/tests/mercury_statement_20241105.pdf.file_account

@@ -0,0 +1 @@
+Liabilities:Credit-Cards:Mercury

beancount_reds_importers/importers/mercurycards/tests/mercury_statement_20241105.pdf.file_date

@@ -0,0 +1 @@
+2024-11-05T00:00:00

beancount_reds_importers/importers/mercurycards/tests/mercury_statement_20241105.pdf.file_name

@@ -0,0 +1 @@
+mercury_statement_20241105.pdf

beancount_reds_importers/importers/mercurycards/tests/mercurycards_test.py

@@ -0,0 +1,21 @@
+from os import path
+
+from beancount.ingest import regression_pytest as regtest
+
+from beancount_reds_importers.importers import mercurycards
+
+
+@regtest.with_importer(
+    mercurycards.Importer(
+        {
+            "main_account": "Liabilities:Credit-Cards:Mercury",
+            "emit_filing_account_metadata": False,
+            "filename_pattern": "mercury_statement_20241105.pdf",
+            "skip_transaction_types": {},
+            "currency": "USD",
+        }
+    )
+)
+@regtest.with_testdir(path.dirname(__file__))
+class TestMercuryCards(regtest.ImporterTestBase):
+    pass

beancount_reds_importers/libreader/csv_multitable_reader.py

@@ -42,7 +42,8 @@ def file_date(self, file):
         pass
 
     def convert_columns(self, rdr):
-        pass
+        # Convert columns for a single table. Can be called from prepare_tables for each table.
+        return super().convert_columns(rdr)
 
     def is_section_title(self, row):
         # Match against rows that contain section titles. Eg: 'section1', 'section2', ...