From 457d4afaece821a5881e91694db78ec6fa310240 Mon Sep 17 00:00:00 2001
From: Joel Kronqvist <joel@jorl.fi>
Date: Sat, 31 Jan 2026 20:12:54 +0200
Subject: feat: menu parsing

---
 src/__init__.py      |   0
 src/menu/__init__.py |   0
 src/menu/parse.py    | 104 +++++++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 104 insertions(+)
 create mode 100755 src/__init__.py
 create mode 100755 src/menu/__init__.py
 create mode 100755 src/menu/parse.py

(limited to 'src')

diff --git a/src/__init__.py b/src/__init__.py
new file mode 100755
index 0000000..e69de29
diff --git a/src/menu/__init__.py b/src/menu/__init__.py
new file mode 100755
index 0000000..e69de29
diff --git a/src/menu/parse.py b/src/menu/parse.py
new file mode 100755
index 0000000..fb45099
--- /dev/null
+++ b/src/menu/parse.py
@@ -0,0 +1,104 @@
+
+import datetime
+import xml.etree.ElementTree as ET
+import re
+
+
+
+def parse(xml_menu: str) -> dict[datetime.date, str]:
+    """Parse the given xml menu and return a (date -> entry) map."""
+
+    root = ET.fromstring(xml_menu)
+
+    item_nodes: list[ET.Element] = root.findall("./channel/item")
+    
+
+    def getchildren(node: ET.Element) -> tuple[ET.Element, ET.Element]:
+        title = node.find("title")
+        desc  = node.find("description")
+        if title is None or desc is None:
+            raise ValueError("title and description of item should be defined")
+        return (title, desc)
+
+    item_children: map[tuple[ET.Element, ET.Element]] = map(getchildren, item_nodes)
+
+    def gettext(node: ET.Element) -> str:
+        text = node.text
+        if text is None:
+            raise ValueError(f"{node.tag} should contain text")
+        return text
+    items = map(lambda item: (parse_weekday_date(gettext(item[0])), gettext(item[1])), item_children)
+
+    return dict(items)
+
+
+def parse_from_file(xml_file_path: str) -> dict[datetime.date, str]:
+    """Parse the menu from the given xml file and return a (date -> entry) map.
+
+    Throws an exception if file reading is not succesful."""
+
+    with open(xml_file_path, 'r', encoding="utf-8") as f:
+        xml_menu = f.read()
+
+    return parse(xml_menu)
+
+        
+def parse_weekday_date(date: str) -> datetime.date:
+    """Parse the given date of the form 'WD DD.MM.YYYY' and return the corresponding date object.
+
+    The weekday should be in finnish format.
+
+    >>> parse_weekday_date("su 26.1.2026")
+    datetime.date(2026, 1, 26)
+
+    Returns a string indicating an error if no weekday
+    is present or if the date is of the wrong format.
+
+    >>> parse_weekday_date("mon 16.2.2026")
+    'päiväyksen ensimmäinen sanan pitäisi olla viikonpäivän lyhenne'
+
+    Parsing the date after the weekday is done as in `parse_date`.
+    """
+
+    (head, tail) = date.split(maxsplit = 1)
+
+    if head not in ["ma", "ti", "ke", "to", "pe", "la", "su"]:
+        raise ValueError("päiväyksen ensimmäinen sanan pitäisi olla viikonpäivän lyhenne")
+
+    return parse_date(tail)
+
+    
+def parse_date(date: str) -> datetime.date:
+    """Parse the given date of the form 'DD.MM.YYYY' and return the corresponding date object.
+
+    >>> parse_date("16.2.2026")
+    datetime.date(2026, 2, 16)
+
+    Returns a string indicating an error if the date is of the wrong format.
+    >>> parse_date("bogus")
+    'päiväyksen tulee olla muotoa PP.KK.VVVV'
+
+    Nonexistent dates are handled:
+    >>> parse_date("31.2.2024")
+    'päivän tulee olla olemassa ollut tai oleva päivä väliltä 01.01.0001-31.12.9999'
+    """
+
+    date_pat = re.compile("([0-9]{1,2})\\.([0-9]{1,2})\\.([0-9]{4})")
+    date_match = date_pat.fullmatch(date)
+    if date_match is None:
+        raise ValueError("päiväyksen tulee olla muotoa PP.KK.VVVV")
+
+    (day, month, year) = date_match.groups()
+
+    try:
+        date_ = datetime.date(int(year), int(month), int(day))
+        return date_
+    except:
+        raise ValueError(f"päivän tulee olla olemassa ollut tai oleva päivä väliltä 01.01.{datetime.MINYEAR:04d}-31.12.{datetime.MAXYEAR}")
+
+
+if __name__ == "__main__":
+    import doctest
+    doctest.testmod()
+
+
-- 
cgit v1.2.3