Improve error messages when reading XML

src/homebank/Homebank.py

@@ -1,42 +1,141 @@
 from __future__ import annotations
-from typing import Optional, Dict, List
-from lxml import etree
+
 import datetime
+from dataclasses import dataclass
+from typing import Dict, List, Optional, Tuple
+
+from lxml import etree
 
 
+# XML functions
+def get_attribute(node: etree.Element, name: str) -> str:
+    if not name in node.attrib:
+        raise KeyError(f"Node '{node.tag}' in line {node.sourceline} is missing attribute '{name}'")
+
+    return node.attrib[name]
+
+def get_attribute_optional(node: etree.Element, name: str) -> Optional[str]:
+    return node.attrib.get(name, None)
+
+def get_attribute_optional_int(node: etree.Element, name: str) -> Optional[int]:
+    if not name in node.attrib: return None
+    val = node.attrib[name]
+    try:
+        return int(val)
+    except ValueError as ex:
+        raise type(ex)(f"Could not convert attribute '{name}' of node '{node.tag}' in line {node.sourceline} to integer") from ex
+
+def get_attribute_int(node: etree.Element, name: str) -> int:
+    val = get_attribute_optional_int(node, name)
+    if val is None:
+        raise KeyError(f"Node '{node.tag}' in line {node.sourceline} is missing attribute '{name}'")
+    return val
+
+def get_attribute_optional_float(node: etree.Element, name: str) -> Optional[float]:
+    if not name in node.attrib: return None
+    val = node.attrib[name]
+    try:
+        return float(val)
+    except ValueError as ex:
+        raise type(ex)(f"Could not convert attribute '{name}' of node '{node.tag}' in line {node.sourceline} to float") from ex
+
+def get_attribute_float(node: etree.Element, name: str) -> float:
+    val = get_attribute_optional_float(node, name)
+    if val is None:
+        raise KeyError(f"Node '{node.tag}' in line {node.sourceline} is missing attribute '{name}'")
+    return val
+
+
+# Homebank Data classes
+
+@dataclass
 class Payee:
-    def __init__(self, name: str):
-        self.name: str = name
+    name: str
 
     def __repr__(self) -> str:
         return f"Payee(\"{self.name}\")"
 
+    @staticmethod
+    def _from_node(node: etree.Element) -> Tuple[int, "Payee"]:
+        key = get_attribute_int(node, "key")
+        name = get_attribute(node, "name")
+        return key, Payee(name)
+
+
+@dataclass
 class Account:
-    def __init__(self, name: str):
-        self.name: str = name
+    name: str
 
     def __repr__(self) -> str:
         return f"Account(\"{self.name}\")"
 
+    @staticmethod
+    def _from_node(node: etree.Element) -> Tuple[int, "Account"]:
+        key = get_attribute_int(node, "key")
+        name = get_attribute(node, "name")
+        return key, Account(name)
+
+
+@dataclass
 class Category:
-    def __init__(self, name: str, parent: Optional[Category] = None):
-        self.name: str = name
-        self.parent: Optional[Category] = parent
+    name: str
+    parent: Optional[Category] = None
 
     def __repr__(self) -> str:
         if self.parent is not None:
             return f"Category(\"{self.name}\", {repr(self.parent)})"
         else:
             return f"Category(\"{self.name}\", None)"
-    
+
+    @staticmethod
+    def _from_node(node: etree.Element, categories: Dict[int, "Category"]) -> Tuple[int, "Category"]:
+        key: int = get_attribute_int(node, "key")
+        name: str = get_attribute(node, "name")
+        parent: Optional[int] = get_attribute_optional_int(node, "parent")
+        parent_class: Optional["Category"] = None
+        if parent is not None:
+            if parent not in categories:
+                raise KeyError(f"Parent category {parent} for category '{name}' not found")
+            parent_class = categories[parent]
+        return key, Category(name, parent_class)
+
+
+@dataclass
 class Operation:
-    def __init__(self, date: datetime.date, amount: float, account: Account, dst_account: Optional[Account], payee: Optional[Payee], category: Optional[Category], wording: Optional[str]):
-        self.date: datetime.date = date
-        self.amount: float = amount
-        self.account: Account = account
-        self.dst_account: Optional[Account] = dst_account
-        self.payee: Optional[Payee] = payee
-        self.category: Optional[Category] = category
+    date: datetime.date
+    amount: float
+    account: Account
+    dst_account: Optional[Account]
+    payee: Optional[Payee]
+    category: Optional[Category]
+    wording: Optional[str]
+
+    @staticmethod
+    def _from_node(node: etree.Element, accounts: Dict[int, Account], payees: Dict[int, Payee], categories: Dict[int, Category]) -> "Operation":
+        date_i = get_attribute_int(node, "date")
+        date = datetime.date(1,1,1) + datetime.timedelta(days=date_i-1)
+        amount = get_attribute_float(node, "amount")
+        def find_account(account_i: int):
+            if not account_i in accounts:
+                raise KeyError(f"Account {account} for operation ({date}, {amount}) not found")
+            return accounts[account_i]
+        account_i = get_attribute_int(node, "account")
+        account = find_account(account_i)
+        dst_account_i = get_attribute_optional_int(node, "dst_account")
+        dst_account: Optional[Account] = None
+        if dst_account_i is not None:
+            dst_account = find_account(dst_account_i)
+        payee_i = get_attribute_optional_int(node, "payee")
+        payee: Optional[Payee] = None
+        if payee_i is not None:
+            payee = payees[payee_i]
+        category_i = get_attribute_optional_int(node, "category")
+        category: Optional[Category] = None
+        if category_i is not None:
+            category = categories[category_i]
+        wording = get_attribute_optional(node, "wording")
+        return Operation(date, amount, account, dst_account, payee, category, wording)
+
 
 class Homebank:
     def __init__(self, path: str):
@@ -49,20 +148,14 @@ class Homebank:
 
         for node in root:
             if node.tag == "account":
-                self.accounts[int(node.attrib["key"])] = Account(node.attrib["name"])
+                key, account = Account._from_node(node)
+                self.accounts[key] = account
             elif node.tag == "pay":
-                self.payees[int(node.attrib["key"])] = Payee(node.attrib["name"])
+                key, payee = Payee._from_node(node)
+                self.payees[key] = payee
             elif node.tag == "cat":
-                parent = node.attrib.get("parent", None)
-                if parent is not None: parent = self.categories[int(parent)]
-                self.categories[int(node.attrib["key"])] = Category(node.attrib["name"], parent)
+                key, category = Category._from_node(node, self.categories)
+                self.categories[key] = category
             elif node.tag == "ope":
-                dst_account = node.attrib.get("dst_account", None)
-                if dst_account is not None: dst_account = self.accounts[int(dst_account)]
-                payee = node.attrib.get("payee", None)
-                if payee is not None: payee = self.payees[int(payee)]
-                category = node.attrib.get("category", None)
-                if category is not None: category = self.categories[int(category)]
-                date = datetime.date(1,1,1) + datetime.timedelta(days=int(node.attrib["date"])-1)
-                self.operations.append(Operation(date, float(node.attrib["amount"]), self.accounts[int(node.attrib["account"])], dst_account, payee, category, node.attrib.get("wording", None)))
-
+                operation = Operation._from_node(node, self.accounts, self.payees, self.categories)
+                self.operations.append(operation)

src/homebank/__init__.py

@@ -0,0 +1 @@
+from .Homebank import Account, Category, Homebank, Operation, Payee