Fix to parsing CSV statement with empty fields separated with separators(commas) (#138)

* UT for robo advisor crashing implemented

* - Fix to crashing scenario

* Bumped version to 0.7.1

* - clarification of warning message

Author: jczaja

Cargo.lock

@@ -1,6 +1,6 @@
 [package]
 name = "etradeTaxReturnHelper"
-version = "0.7.0"
+version = "0.7.1"
 edition = "2021"
 description = "Parses etrade and revolut financial documents for transaction details  (income, tax paid, cost basis) and compute total income and total tax paid according to chosen tax residency (currency)"
 license = "BSD-3-Clause"

Cargo.toml

@@ -0,0 +1,14 @@
+Income from Sells,,,,,,,,,,
+Date acquired,Date sold,Symbol,Security name,ISIN,Country,Quantity,Cost basis,Gross proceeds,Gross PnL,Currency
+,,,,,,,,,,
+Other income & fees,,,,,,,,,,
+Date,Symbol,Security name,ISIN,Country,Gross amount,Withholding tax,Net Amount,Currency,,
+2024-06-04,QDVY,iShares Floating Rate Bond ETF (Dist),IE00BZ048462,IE,2.8,0.68 PLN,2.12 PLN,PLN,,
+2024-06-20,EXI2,iShares Dow Jones Global Titans 50 ETF (Dist),DE0006289382,DE,0.34,0.08 PLN,0.26 PLN,PLN,,
+2024-06-28,IS3K,iShares Short Duration High Yield Corporate Bond ETF (Dist),IE00BCRY6003,IE,3.79,0.94 PLN,2.85 PLN,PLN,,
+2024-07-01,IBCD,iShares Corporate Bond ETF (Dist),IE0032895942,IE,1.07,0.25 PLN,0.82 PLN,PLN,,
+2024-09-27,IBCD,iShares Corporate Bond ETF (Dist),IE0032895942,IE,1.02,0.25 PLN,0.77 PLN,PLN,,
+2024-09-27,IUSU,iShares USD Treasury Bond 1-3yr ETF (Dist),IE00B14X4S71,IE,1.71,0.42 PLN,1.29 PLN,PLN,,
+2024-11-29,QDVY,iShares Floating Rate Bond ETF (Dist),IE00BZ048462,IE,2.92,0.73 PLN,2.19 PLN,PLN,,
+2024-12-17,EXI2,iShares Dow Jones Global Titans 50 ETF (Dist),DE0006289382,DE,0.04,0 PLN,0.04 PLN,PLN,,
+2024-12-31,IBCD,iShares Corporate Bond ETF (Dist),IE0032895942,IE,1.07,0.25 PLN,0.82 PLN,PLN,,

revolut_data/trading-pnl-statement_2024-01-robo-2.csv

@@ -359,6 +359,7 @@ pub fn parse_revolut_transactions(
             .map_err(|e| format!("Error reading CSV: {e}"))?;
 
         let others = CsvReader::new(std::io::Cursor::new(content2.as_bytes()))
+            .truncate_ragged_lines(true)
             .finish()
             .map_err(|e| format!("Error reading CSV: {e}"))?;
 
@@ -1540,6 +1541,67 @@ mod tests {
         Ok(())
     }
 
+    #[test]
+    fn test_parse_revolut_investment_with_commas_gain_and_losses_dividends() -> Result<(), String> {
+        let expected_result = Ok((
+            vec![
+                (
+                    "06/04/24".to_owned(),
+                    crate::Currency::PLN(2.80),
+                    crate::Currency::PLN(0.68),
+                ),
+                (
+                    "06/20/24".to_owned(),
+                    crate::Currency::PLN(0.34),
+                    crate::Currency::PLN(0.08),
+                ),
+                (
+                    "06/28/24".to_owned(),
+                    crate::Currency::PLN(3.79),
+                    crate::Currency::PLN(0.94),
+                ),
+                (
+                    "07/01/24".to_owned(),
+                    crate::Currency::PLN(1.07),
+                    crate::Currency::PLN(0.25),
+                ),
+                (
+                    "09/27/24".to_owned(),
+                    crate::Currency::PLN(1.02),
+                    crate::Currency::PLN(0.25),
+                ),
+                (
+                    "09/27/24".to_owned(),
+                    crate::Currency::PLN(1.71),
+                    crate::Currency::PLN(0.42),
+                ),
+                (
+                    "11/29/24".to_owned(),
+                    crate::Currency::PLN(2.92),
+                    crate::Currency::PLN(0.73),
+                ),
+                (
+                    "12/17/24".to_owned(),
+                    crate::Currency::PLN(0.04),
+                    crate::Currency::PLN(0.0),
+                ),
+                (
+                    "12/31/24".to_owned(),
+                    crate::Currency::PLN(1.07),
+                    crate::Currency::PLN(0.25),
+                ),
+            ],
+            vec![],
+        ));
+
+        assert_eq!(
+            parse_revolut_transactions("revolut_data/trading-pnl-statement_2024-01-robo-2.csv"),
+            expected_result
+        );
+
+        Ok(())
+    }
+
     #[test]
     fn test_parse_revolut_investment_gain_and_losses_sells_and_dividends() -> Result<(), String> {
         let expected_result = Ok((

src/csvparser.rs

@@ -1222,9 +1222,12 @@ mod tests {
     #[ignore]
     fn test_combined_account_statement() -> Result<(), String> {
         assert_eq!(
-            parse_statement("etrade_data_2024/ClientStatements_110324.pdf"),
+            parse_statement("etrade_data_2024/ClientStatements_010325.pdf"),
             (Ok((
                 vec![
+                    ("12/2/24".to_owned(), 4.88),
+                    ("10/1/24".to_owned(), 24.91),
+                    ("11/1/24".to_owned(), 25.09),
                     ("9/3/24".to_owned(), 23.65), // Interest rates
                     ("8/1/24".to_owned(), 4.34),
                     ("7/1/24".to_owned(), 3.72),
@@ -1238,6 +1241,27 @@ mod tests {
                     ("3/1/24".to_owned(), 380.25, 57.04)
                 ],
                 vec![
+                    (
+                        "12/4/24".to_owned(),
+                        "12/5/24".to_owned(),
+                        30.0,
+                        22.5,
+                        674.98
+                    ),
+                    (
+                        "12/5/24".to_owned(),
+                        "12/6/24".to_owned(),
+                        55.0,
+                        21.96,
+                        1207.76
+                    ),
+                    (
+                        "11/1/24".to_owned(),
+                        "11/4/24".to_owned(),
+                        15.0,
+                        23.32,
+                        349.79
+                    ),
                     (
                         "9/3/24".to_owned(),
                         "9/4/24".to_owned(),

src/pdfparser.rs

@@ -257,7 +257,7 @@ impl etradeTaxReturnHelper::Residency for PL {
             cost_sold
         ));
         if tax_div > tax_pl {
-            (presentation,Some(format!("Warning: Tax paid in US({tax_div} PLN) is higher than the tax that you are to pay in Poland({tax_pl} PLN). This usually means that there was a problem with declaration of your residency to avoid double taxation")))
+            (presentation,Some(format!("Warning: Tax paid in US({tax_div} PLN) is higher than the tax that you are to pay in Poland({tax_pl} PLN). This either means that there was a problem with declaration of your residency to avoid double taxation or you are having income from countries that are having higher tax at source than the one used in Poland(19%)")))
         } else {
             (presentation, None)
         }
@@ -362,7 +362,7 @@ mod tests {
             .zip(&ref_results)
             .for_each(|(a, b)| assert_eq!(a, b));
 
-        let ref_msg = "Warning: Tax paid in US(30 PLN) is higher than the tax that you are to pay in Poland(19 PLN). This usually means that there was a problem with declaration of your residency to avoid double taxation".to_string();
+        let ref_msg = "Warning: Tax paid in US(30 PLN) is higher than the tax that you are to pay in Poland(19 PLN). This either means that there was a problem with declaration of your residency to avoid double taxation or you are having income from countries that are having higher tax at source than the one used in Poland(19%)".to_string();
 
         match warning {
             Some(msg) => assert_eq!(msg, ref_msg),