Merge pull request #43 from MHoroszowski/feature/modernization-phase2

feat: Font.color non-mutation + UTC-aware datetimes + Shapes.by_name (Modernization Phase 2)

Author: MHoroszowski

features/modernization-phase2.feature

@@ -0,0 +1,33 @@
+Feature: Modernization & Ergonomics Phase 2 — bug fixes + by_name
+  In order to inspect fonts without polluting the document, round-trip datetimes faithfully, and look up shapes by name
+  As a developer using python-pptx
+  I need a non-mutating Font.color getter, tz-aware core-property datetimes, and Shapes.by_name(name)
+
+
+  Scenario: Reading font.color does not insert a:solidFill
+    Given a fresh slide with a title placeholder
+     When I read run.font.color.rgb on an unstyled run
+     Then the underlying rPr XML is unchanged from before the access
+
+
+  Scenario: Setting font.color.rgb materializes a:solidFill lazily
+    Given a fresh slide with a title placeholder
+     When I set run.font.color.rgb to RGBColor(0xFF, 0x00, 0x00)
+     Then the underlying rPr now contains an a:solidFill child
+      And run.font.color.rgb reads back as FF0000
+
+
+  Scenario: Tz-aware core_properties.created round-trips faithfully
+    Given a fresh presentation for core-property datetimes
+     When I set core_properties.created to a tz-aware UTC datetime
+     Then the reloaded core_properties.created is tz-aware
+
+
+  Scenario: shapes.by_name returns the matching shape
+    Given a fresh slide with a title placeholder
+     Then shapes.by_name("Title 1") returns the title shape
+
+
+  Scenario: shapes.by_name raises KeyError on miss
+    Given a fresh slide with a title placeholder
+     Then shapes.by_name("Bogus") raises KeyError

features/steps/coreprops.py

@@ -2,7 +2,7 @@
 
 from __future__ import annotations
 
-from datetime import datetime, timedelta
+from datetime import datetime, timedelta, timezone
 
 from behave import given, then, when
 from helpers import no_core_props_pptx_path, saved_pptx_path
@@ -28,18 +28,21 @@ def step_when_open_presentation_with_no_core_props_part(context):
 
 @when("I set the core properties to valid values")
 def step_when_set_core_doc_props_to_valid_values(context):
+    # ---issue #29 Phase 2: datetimes round-trip as tz-aware UTC values, so
+    # ---the input fixture is already tz-aware UTC for an apples-to-apples
+    # ---comparison with the reloaded value.
     context.propvals = (
         ("author", "Creator"),
         ("category", "Category"),
         ("comments", "Description"),
         ("content_status", "Content Status"),
-        ("created", datetime(2013, 6, 15, 12, 34, 56)),
+        ("created", datetime(2013, 6, 15, 12, 34, 56, tzinfo=timezone.utc)),
         ("identifier", "Identifier"),
         ("keywords", "key; word; keyword"),
         ("language", "Language"),
         ("last_modified_by", "Last Modified By"),
-        ("last_printed", datetime(2013, 6, 15, 12, 34, 56)),
-        ("modified", datetime(2013, 6, 15, 12, 34, 56)),
+        ("last_printed", datetime(2013, 6, 15, 12, 34, 56, tzinfo=timezone.utc)),
+        ("modified", datetime(2013, 6, 15, 12, 34, 56, tzinfo=timezone.utc)),
         ("revision", 9),
         ("subject", "Subject"),
         # --- exercise unicode-text case for Python 2.7 ---
@@ -60,8 +63,10 @@ def step_then_a_core_props_part_with_def_vals_is_added(context):
     assert core_props.last_modified_by == "python-pptx"
     assert core_props.revision == 1
     # core_props.modified only stores time with seconds resolution, so
-    # comparison needs to be a little loose (within two seconds)
-    modified_timedelta = datetime.utcnow() - core_props.modified
+    # comparison needs to be a little loose (within two seconds). Issue #29
+    # Phase 2 makes the parser return a tz-aware datetime; use a tz-aware
+    # `now()` here so the subtraction is well-typed.
+    modified_timedelta = datetime.now(timezone.utc) - core_props.modified
     max_expected_timedelta = timedelta(seconds=2)
     assert modified_timedelta < max_expected_timedelta
 

features/steps/modernization_phase2.py

@@ -0,0 +1,106 @@
+"""Gherkin steps for Modernization Phase 2 (issue #29)."""
+
+from __future__ import annotations
+
+import datetime as dt
+import io
+
+import pytest
+from behave import given, then, when
+from lxml import etree
+
+from pptx import Presentation
+from pptx.dml.color import RGBColor
+
+
+# given ===================================================
+
+
+@given("a fresh slide with a title placeholder")
+def given_a_fresh_slide_with_title(context):
+    prs = Presentation()
+    slide = prs.slides.add_slide(prs.slide_layouts[1])
+    context.prs = prs
+    context.slide = slide
+
+
+@given("a fresh presentation for core-property datetimes")
+def given_a_fresh_presentation_for_core_props(context):
+    context.prs = Presentation()
+
+
+# when ====================================================
+
+
+@when("I read run.font.color.rgb on an unstyled run")
+def when_read_font_color_unstyled(context):
+    tf = context.slide.shapes.title.text_frame
+    run = tf.paragraphs[0].add_run()
+    run.text = "x"
+    rPr = run._r.get_or_add_rPr()
+    context.run = run
+    context.rPr = rPr
+    context.before_xml = etree.tostring(rPr)
+    _ = run.font.color.rgb
+
+
+@when("I set run.font.color.rgb to RGBColor(0xFF, 0x00, 0x00)")
+def when_set_font_color_red(context):
+    tf = context.slide.shapes.title.text_frame
+    run = tf.paragraphs[0].add_run()
+    run.text = "x"
+    context.run = run
+    context.rPr = run._r.get_or_add_rPr()
+    run.font.color.rgb = RGBColor(0xFF, 0x00, 0x00)
+
+
+@when("I set core_properties.created to a tz-aware UTC datetime")
+def when_set_created_utc(context):
+    context.target_dt = dt.datetime(2024, 7, 4, 17, 0, 0, tzinfo=dt.timezone.utc)
+    context.prs.core_properties.created = context.target_dt
+
+
+# the "save and reload via stream" step is shared with tbl_styles.py
+
+
+# then ====================================================
+
+
+@then("the underlying rPr XML is unchanged from before the access")
+def then_rPr_unchanged(context):
+    after = etree.tostring(context.rPr)
+    assert context.before_xml == after, (context.before_xml, after)
+
+
+@then("the underlying rPr now contains an a:solidFill child")
+def then_rPr_has_solidFill(context):
+    ns = "{http://schemas.openxmlformats.org/drawingml/2006/main}"
+    assert context.rPr.find("%ssolidFill" % ns) is not None
+
+
+@then("run.font.color.rgb reads back as FF0000")
+def then_font_color_rgb_reads_FF0000(context):
+    assert context.run.font.color.rgb == RGBColor(0xFF, 0x00, 0x00)
+
+
+@then("the reloaded core_properties.created is tz-aware")
+def then_reloaded_created_is_tzaware(context):
+    buf = io.BytesIO()
+    context.prs.save(buf)
+    buf.seek(0)
+    prs2 = Presentation(buf)
+    reloaded = prs2.core_properties.created
+    assert reloaded.tzinfo is not None
+    assert reloaded == context.target_dt
+
+
+@then('shapes.by_name("Title 1") returns the title shape')
+def then_by_name_returns_title(context):
+    sh = context.slide.shapes.by_name("Title 1")
+    assert sh.name == "Title 1"
+
+
+@then('shapes.by_name("Bogus") raises KeyError')
+def then_by_name_raises_keyerror(context):
+    with pytest.raises(KeyError):
+        context.slide.shapes.by_name("Bogus")

src/pptx/oxml/coreprops.py

@@ -210,34 +210,44 @@ def _get_or_add(self, prop_name: str):
         return element
 
     @classmethod
-    def _offset_dt(cls, datetime: dt.datetime, offset_str: str):
-        """Return |datetime| instance offset from `datetime` by offset specified in `offset_str`.
-
-        `offset_str` is a string like `'-07:00'`.
-        """
+    def _tzinfo_from_offset_str(cls, offset_str: str) -> dt.timezone:
+        """Return a :class:`datetime.timezone` parsed from a W3CDTF offset like '-08:00'."""
         match = cls._offset_pattern.match(offset_str)
         if match is None:
             raise ValueError(f"{repr(offset_str)} is not a valid offset string")
         sign, hours_str, minutes_str = match.groups()
-        sign_factor = -1 if sign == "+" else 1
+        sign_factor = 1 if sign == "+" else -1
         hours = int(hours_str) * sign_factor
         minutes = int(minutes_str) * sign_factor
-        td = dt.timedelta(hours=hours, minutes=minutes)
-        return datetime + td
+        return dt.timezone(dt.timedelta(hours=hours, minutes=minutes))
 
     _offset_pattern = re.compile(r"([+-])(\d\d):(\d\d)")
 
     @classmethod
     def _parse_W3CDTF_to_datetime(cls, w3cdtf_str: str) -> dt.datetime:
-        # valid W3CDTF date cases:
-        # yyyy e.g. '2003'
-        # yyyy-mm e.g. '2003-12'
-        # yyyy-mm-dd e.g. '2003-12-31'
-        # UTC timezone e.g. '2003-12-31T10:14:55Z'
-        # numeric timezone e.g. '2003-12-31T10:14:55-08:00'
+        """Parse a W3CDTF string into a :class:`datetime.datetime`.
+
+        Returns a tz-aware datetime when the input string carries timezone
+        information — ``Z`` suffix maps to UTC; numeric offsets like
+        ``-08:00`` map to a fixed-offset :class:`datetime.timezone`. When the
+        input has no timezone marker (year-only, year-month, year-month-day,
+        or a bare timestamp), the returned datetime is naive — callers
+        should not assume any specific timezone for naive inputs.
+
+        Closes scanny/python-pptx#957 — the prior implementation always
+        returned a naive datetime, even for strings that explicitly carried
+        timezone information.
+
+        valid W3CDTF date cases:
+            - yyyy e.g. '2003'
+            - yyyy-mm e.g. '2003-12'
+            - yyyy-mm-dd e.g. '2003-12-31'
+            - UTC timezone e.g. '2003-12-31T10:14:55Z'
+            - numeric timezone e.g. '2003-12-31T10:14:55-08:00'
+        """
         templates = ("%Y-%m-%dT%H:%M:%S", "%Y-%m-%d", "%Y-%m", "%Y")
-        # strptime isn't smart enough to parse literal timezone offsets like
-        # '-07:30', so we have to do it ourselves
+        # ---strptime can't parse literal timezone offsets like '-07:30', so
+        # ---we strip the offset and add tzinfo ourselves below.
         parseable_part = w3cdtf_str[:19]
         offset_str = w3cdtf_str[19:]
         timestamp = None
@@ -249,15 +259,31 @@ def _parse_W3CDTF_to_datetime(cls, w3cdtf_str: str) -> dt.datetime:
         if timestamp is None:
             tmpl = "could not parse W3CDTF datetime string '%s'"
             raise ValueError(tmpl % w3cdtf_str)
+        # ---'Z' means UTC---
+        if offset_str == "Z":
+            return timestamp.replace(tzinfo=dt.timezone.utc)
+        # ---numeric offset like '-08:00'---
         if len(offset_str) == 6:
-            return cls._offset_dt(timestamp, offset_str)
+            return timestamp.replace(tzinfo=cls._tzinfo_from_offset_str(offset_str))
+        # ---no timezone marker; return naive (don't assume UTC)---
         return timestamp
 
     def _set_element_datetime(self, prop_name: str, value: dt.datetime) -> None:
-        """Set date/time value of child element having `prop_name` to `value`."""
+        """Set date/time value of child element having `prop_name` to `value`.
+
+        Accepts both naive and tz-aware datetimes. Tz-aware inputs are
+        converted to UTC before serialization so the on-disk W3CDTF form
+        always uses the canonical ``YYYY-MM-DDTHH:MM:SSZ`` shape. Naive
+        inputs are written as-is (with the trailing ``Z`` suffix indicating
+        UTC) — callers passing naive datetimes are responsible for the
+        timezone interpretation.
+        """
         if not isinstance(value, dt.datetime):  # pyright: ignore[reportUnnecessaryIsInstance]
             tmpl = "property requires <type 'datetime.datetime'> object, got %s"
             raise ValueError(tmpl % type(value))
+        # ---tz-aware -> normalize to UTC before serializing---
+        if value.tzinfo is not None:
+            value = value.astimezone(dt.timezone.utc).replace(tzinfo=None)
         element = self._get_or_add(prop_name)
         dt_str = value.strftime("%Y-%m-%dT%H:%M:%SZ")
         element.text = dt_str

src/pptx/shapes/shapetree.py

@@ -108,6 +108,23 @@ def __len__(self) -> int:
         shape_elms = list(self._iter_member_elms())
         return len(shape_elms)
 
+    def by_name(self, name: str) -> BaseShape:
+        """Return the first shape in this collection whose `.name` equals `name`.
+
+        Lookup is case-sensitive, matching PowerPoint's own behavior. When
+        multiple shapes share the same name (uncommon but possible —
+        PowerPoint does not enforce uniqueness), the first match in
+        document order is returned. Raises |KeyError| with a clear message
+        if no match is found.
+
+        Closes scanny/python-pptx#798, scanny/python-pptx#309, and
+        scanny/python-pptx#532.
+        """
+        for shape in self:
+            if shape.name == name:
+                return shape
+        raise KeyError("no shape named %r in this collection" % name)
+
     def clone_placeholder(self, placeholder: LayoutPlaceholder) -> None:
         """Add a new placeholder shape based on `placeholder`."""
         sp = placeholder.element