librus-python-api 1.0.0rc1__py3-none-any.whl

librus_python_api/timetable.py

@@ -0,0 +1,261 @@
+"""Independently authored bounded timetable parser with explicit week identity."""
+
+import re
+from dataclasses import replace
+from datetime import date, time, timedelta
+
+from lxml import html
+
+from librus_python_api import markup
+from librus_python_api.config import (
+    TIMETABLE_DATE_ATTRIBUTE,
+    TIMETABLE_END_ATTRIBUTE,
+    TIMETABLE_MAX_CHANGES_PER_SLOT,
+    TIMETABLE_MAX_LESSONS_PER_SLOT,
+    TIMETABLE_MAX_PERIODS,
+    TIMETABLE_START_ATTRIBUTE,
+    timetable_form,
+)
+from librus_python_api.exceptions import ErrorKind, LibrusError
+from librus_python_api.models import (
+    TimetableChange,
+    TimetableDay,
+    TimetableInterval,
+    TimetableLesson,
+    TimetablePeriod,
+)
+from librus_python_api.parsers import parse_page
+
+
+def _clock(value: str | None) -> time:
+    if not re.fullmatch(r"[0-9]{2}:[0-9]{2}", value or ""):
+        raise LibrusError(ErrorKind.PARSE)
+    failed = False
+    try:
+        return time.fromisoformat(value or "")
+    except ValueError:
+        failed = True
+    assert failed
+    raise LibrusError(ErrorKind.PARSE)
+
+
+def _interval(start: str | None, end: str | None) -> TimetableInterval:
+    first, last = _clock(start), _clock(end)
+    if first >= last:
+        raise LibrusError(ErrorKind.PARSE)
+    return TimetableInterval(first, last)
+
+
+def _lesson(element: html.HtmlElement) -> TimetableLesson:
+    subjects = list(element.iter("b"))
+    if len(subjects) != 1:
+        raise LibrusError(ErrorKind.PARSE)
+    subject, rendered = markup.text(subjects[0]), markup.text(element)
+    if not subject or not rendered.startswith(subject):
+        raise LibrusError(ErrorKind.PARSE)
+    remainder = rendered[len(subject) :].strip()
+    if remainder and not remainder.startswith("-"):
+        raise LibrusError(ErrorKind.UNSUPPORTED_CAPABILITY)
+    teacher = remainder.removeprefix("-").strip() or None
+    return TimetableLesson(subject, teacher)
+
+
+def _change(
+    element: html.HtmlElement, wrapper: html.HtmlElement | None
+) -> TimetableChange:
+    """A notice; its tooltip anchor sits inside it or wraps it."""
+    label = markup.text(element)
+    anchors = list(element.iter("a")) + ([wrapper] if wrapper is not None else [])
+    if not label or len(anchors) > 1:
+        raise LibrusError(ErrorKind.PARSE)
+    if anchors and anchors[0].get("title") is None:
+        raise LibrusError(ErrorKind.PARSE)
+    metadata = markup.tooltip_fields(anchors[0]) if anchors else {}
+    if any(len(key) > 1024 or len(value) > 1024 for key, value in metadata.items()):
+        raise LibrusError(ErrorKind.LIMIT)
+    return TimetableChange(label, tuple(metadata.items()))
+
+
+def _content(
+    cell: html.HtmlElement,
+) -> tuple[tuple[TimetableLesson, ...], tuple[TimetableChange, ...]]:
+    if list(cell.iter("table")):
+        raise LibrusError(ErrorKind.UNSUPPORTED_CAPABILITY)
+    lessons: list[TimetableLesson] = []
+    changes: list[TimetableChange] = []
+    consumed: set[html.HtmlElement] = set()
+    for node in cell.iter("div"):
+        classes = set(node.get("class", "").split())
+        if "text" in classes or {"center", "plan-lekcji-info"} <= classes:
+            parent = node.getparent()
+            wrapper = (
+                parent
+                if "text" not in classes
+                and parent is not None
+                and parent is not cell
+                and parent.tag == "a"
+                and len(parent) == 1
+                # A pure wrapper only: text beside the notice must not vanish.
+                and not (parent.text or "").strip()
+                and not (node.tail or "").strip()
+                else None
+            )
+            descendants = set(node.iter()) | (
+                {wrapper} if wrapper is not None else set()
+            )
+            if descendants.intersection(consumed):
+                raise LibrusError(ErrorKind.UNSUPPORTED_CAPABILITY)
+            consumed.update(descendants)
+            if "text" in classes:
+                lessons.append(_lesson(node))
+            else:
+                changes.append(_change(node, wrapper))
+            if (
+                len(lessons) > TIMETABLE_MAX_LESSONS_PER_SLOT
+                or len(changes) > TIMETABLE_MAX_CHANGES_PER_SLOT
+            ):
+                raise LibrusError(ErrorKind.LIMIT)
+    for node in cell.iter():
+        if not isinstance(node.tag, str):
+            continue
+        if node not in consumed and (
+            (node.text or "").strip() or (node.tag == "a" and node.get("title"))
+        ):
+            raise LibrusError(ErrorKind.UNSUPPORTED_CAPABILITY)
+        if (node.tail or "").strip() and node.getparent() not in consumed:
+            raise LibrusError(ErrorKind.UNSUPPORTED_CAPABILITY)
+    return tuple(lessons), tuple(changes)
+
+
+def _slot(
+    cell: html.HtmlElement, number: int, monday: date
+) -> tuple[int, TimetablePeriod]:
+    if cell.get("colspan", "1") != "1" or cell.get("rowspan", "1") != "1":
+        raise LibrusError(ErrorKind.PARSE)
+    value = cell.get(TIMETABLE_DATE_ATTRIBUTE, "")
+    if not re.fullmatch(r"[0-9]{4}-[0-9]{2}-[0-9]{2}", value):
+        raise LibrusError(ErrorKind.PARSE)
+    failed = False
+    day = None
+    try:
+        day = date.fromisoformat(value)
+    except ValueError:
+        failed = True
+    if failed or day is None or not 0 <= (day - monday).days < 7:
+        raise LibrusError(ErrorKind.PARSE)
+    interval = _interval(
+        cell.get(TIMETABLE_START_ATTRIBUTE), cell.get(TIMETABLE_END_ATTRIBUTE)
+    )
+    lessons, changes = _content(cell)
+    return (day - monday).days, TimetablePeriod(
+        number, interval, lessons, changes, None
+    )
+
+
+def _period(row: html.HtmlElement, monday: date) -> tuple[TimetablePeriod, ...]:
+    cells = markup.cells(row)
+    slots = [cell for cell in cells if cell.get("id") == "timetableEntryBox"]
+    numbers = [
+        cell
+        for cell in cells
+        if cell.tag == "td"
+        and cell not in slots
+        and "center" in cell.get("class", "").split()
+        and re.fullmatch(r"[0-9]{1,2}", markup.text(cell))
+    ]
+    if len(slots) != 7 or not 1 <= len(numbers) <= 2:
+        raise LibrusError(ErrorKind.PARSE)
+    values = {markup.text(cell) for cell in numbers}
+    if len(values) != 1:
+        raise LibrusError(ErrorKind.PARSE)
+    raw_number = values.pop()
+    selected: dict[int, TimetablePeriod] = {}
+    for cell in slots:
+        offset, period = _slot(cell, int(raw_number), monday)
+        if offset in selected:
+            raise LibrusError(ErrorKind.PARSE)
+        selected[offset] = period
+    return tuple(selected[index] for index in range(7))
+
+
+def _recess(
+    row: html.HtmlElement, previous: tuple[TimetablePeriod, ...]
+) -> tuple[TimetablePeriod, ...]:
+    cells = [
+        cell
+        for cell in markup.cells(row)
+        if "center" in cell.get("class", "").split()
+        and re.fullmatch(
+            r"[0-9]{2}:[0-9]{2}\s*-\s*[0-9]{2}:[0-9]{2}", markup.text(cell)
+        )
+    ]
+    if (
+        len(cells) != 1
+        or not previous
+        or any(p.next_recess is not None for p in previous)
+    ):
+        raise LibrusError(ErrorKind.PARSE)
+    start, separator, end = markup.text(cells[0]).partition("-")
+    if not separator:
+        raise LibrusError(ErrorKind.PARSE)
+    # Upstream recess clocks are reported values, not guaranteed positive gaps.
+    # Preserve valid clocks without inventing a duration or rejecting a week.
+    interval = TimetableInterval(_clock(start.strip()), _clock(end.strip()))
+    return tuple(replace(period, next_recess=interval) for period in previous)
+
+
+def _grid(document: html.HtmlElement) -> html.HtmlElement:
+    tables = [
+        t
+        for t in document.iter("table")
+        if {"decorated", "plan-lekcji"} <= set(t.get("class", "").split())
+    ]
+    if len(tables) != 1:
+        raise LibrusError(ErrorKind.PARSE)
+    table = tables[0]
+    for node in document.iter():
+        if not isinstance(node.tag, str):
+            continue
+        known = node.get("id") == "timetableEntryBox" or (
+            node.tag == "div" and "plan-lekcji-info" in node.get("class", "").split()
+        )
+        if known and next(node.iterancestors("table"), None) is not table:
+            raise LibrusError(ErrorKind.UNSUPPORTED_CAPABILITY)
+    return table
+
+
+def parse_timetable(body: bytes, monday: date) -> tuple[TimetableDay, ...]:
+    timetable_form(monday)
+    # The upstream repeats this particular slot ID. Other duplicate IDs and
+    # parser repairs remain errors; semantic slot uniqueness is checked below.
+    table = _grid(parse_page(body, repeatable_id="timetableEntryBox"))
+    periods: list[tuple[TimetablePeriod, ...]] = []
+    for row in markup.rows(table):
+        classes = set(row.get("class", "").split())
+        if "line1" in classes:
+            current = _period(row, monday)
+            if periods:
+                for previous, following in zip(periods[-1], current, strict=True):
+                    earliest = previous.interval.ends_at
+                    if (
+                        following.number <= previous.number
+                        or following.interval.starts_at < earliest
+                    ):
+                        raise LibrusError(ErrorKind.PARSE)
+            periods.append(current)
+            if len(periods) > TIMETABLE_MAX_PERIODS:
+                raise LibrusError(ErrorKind.LIMIT)
+        elif "line0" in classes:
+            if not periods:
+                raise LibrusError(ErrorKind.PARSE)
+            periods[-1] = _recess(row, periods[-1])
+        elif any(cell.get("id") == "timetableEntryBox" for cell in markup.cells(row)):
+            raise LibrusError(ErrorKind.UNSUPPORTED_CAPABILITY)
+    if not periods:
+        raise LibrusError(ErrorKind.PARSE)
+    return tuple(
+        TimetableDay(
+            monday + timedelta(days=index), tuple(row[index] for row in periods)
+        )
+        for index in range(7)
+    )