python-hwpx 6.4.0__py3-none-any.whl → 6.5.0__py3-none-any.whl

hwpx/equation/authoring.py

@@ -78,6 +78,11 @@ _LATEX_TO_EQEDIT.update(
         "\\forall": "FORALL",
         "\\iint": "dint",
         "\\iiint": "tint",
+        # ``triangle`` is accepted when reading (observed in real documents) but
+        # only the documented uppercase spelling is emitted when authoring: the
+        # lowercase form has not been render-verified, and lowercase ``forall``
+        # already turned out to render as literal text.
+        "\\triangle": "TRIANGLE",
         # Common LaTeX aliases sharing a verified target.
         "\\le": "leq",
         "\\ge": "geq",

hwpx/equation/mathml.py

@@ -9,11 +9,16 @@ code block rather than dropping the equation.
 
 from __future__ import annotations
 
+import re
 from typing import Callable
 
+from .tokens import EQEDIT_MATHML_OPERATOR_COMMANDS
+
 # ``False`` marks a resolved-but-unavailable converter; ``None`` means "not yet
 # probed" so the import is attempted lazily on first use.
 _converter: Callable[[str], str] | bool | None = None
+_LATEX_WORD_COMMAND_RE = re.compile(r"\\[A-Za-z]+")
+_TEXT_COMMAND = "\\text{"
 
 
 class MathMLUnavailableError(RuntimeError):
@@ -62,8 +67,69 @@ def latex_to_mathml(latex: str) -> str:
         raise ValueError(f"latex2mathml could not render fragment: {exc}") from exc
 
 
+def eqedit_latex_to_mathml(latex: str) -> str:
+    """Convert EqEdit-derived LaTeX while preserving known token roles.
+
+    ``latex2mathml`` classifies plain ``\\triangle`` as an identifier. EqEdit's
+    token map identifies it as a mathematical symbol, so the MathML-only input
+    receives an operator wrapper before conversion. The returned/public LaTeX
+    is not rewritten, and serialized MathML is never patched afterward.
+    """
+
+    def annotate_role(match: re.Match[str]) -> str:
+        command = match.group(0)
+        if command in EQEDIT_MATHML_OPERATOR_COMMANDS:
+            return rf"\mathop{{{command}}}"
+        return command
+
+    return latex_to_mathml(
+        "".join(
+            segment if literal else _LATEX_WORD_COMMAND_RE.sub(annotate_role, segment)
+            for segment, literal in _split_text_literals(latex)
+        )
+    )
+
+
+def _split_text_literals(latex: str) -> list[tuple[str, bool]]:
+    """Split ``latex`` into ``(segment, is_text_literal)`` pieces.
+
+    Quoted EqEdit literals become ``\\text{...}``; their content is prose, so
+    the role annotation must not rewrite a ``\\triangle`` typed inside quotes.
+    Braces inside the literal are balanced; an unterminated group is passed
+    through untouched so ``latex2mathml`` reports it instead of this splitter.
+    """
+
+    pieces: list[tuple[str, bool]] = []
+    cursor = 0
+    while True:
+        start = latex.find(_TEXT_COMMAND, cursor)
+        if start < 0:
+            break
+        depth = 0
+        end = -1
+        for index in range(start + len(_TEXT_COMMAND) - 1, len(latex)):
+            char = latex[index]
+            if char == "{":
+                depth += 1
+            elif char == "}":
+                depth -= 1
+                if depth == 0:
+                    end = index + 1
+                    break
+        if end < 0:
+            break
+        if start > cursor:
+            pieces.append((latex[cursor:start], False))
+        pieces.append((latex[start:end], True))
+        cursor = end
+    if cursor < len(latex):
+        pieces.append((latex[cursor:], False))
+    return pieces
+
+
 __all__ = [
     "MathMLUnavailableError",
+    "eqedit_latex_to_mathml",
     "latex2mathml_available",
     "latex_to_mathml",
 ]

hwpx/equation/render.py

@@ -18,7 +18,7 @@ from dataclasses import dataclass
 import html
 
 from .eqedit import EquationConversionError, eqedit_to_latex
-from .mathml import MathMLUnavailableError, latex_to_mathml
+from .mathml import MathMLUnavailableError, eqedit_latex_to_mathml
 
 # Fidelity labels surfaced in the preview (Constitution IX, honest reporting).
 LABEL_MATHML = "수식 MathML 렌더"
@@ -63,7 +63,7 @@ def render_equation(script: str) -> EquationRender:
         return _fallback_block("script", LABEL_SCRIPT, script, latex=None)
 
     try:
-        mathml = latex_to_mathml(latex)
+        mathml = eqedit_latex_to_mathml(latex)
     except MathMLUnavailableError:
         return _fallback_block("latex", LABEL_LATEX_NO_LIB, latex, latex=latex)
     except ValueError:

hwpx/equation/tokens.py

@@ -116,6 +116,11 @@ OPERATORS: dict[str, str] = {
     "partial": r"\partial",
     "nabla": r"\nabla",
     "angle": r"\angle",
+    # Hancom's symbol guide spells this token in uppercase, while legacy
+    # documents also contain the exact lowercase form. Mixed-case variants are
+    # intentionally not accepted because the reader uses exact token lookup.
+    "triangle": r"\triangle",
+    "TRIANGLE": r"\triangle",
     "cdots": r"\cdots",
     "ldots": r"\ldots",
     "vdots": r"\vdots",
@@ -254,11 +259,17 @@ DELIMITERS: dict[str, str] = {
 # Structural keywords handled directly by the parser (not simple substitution).
 STRUCTURAL = frozenset({"over", "atop", "sqrt", "root", "of", "LEFT", "RIGHT", "left", "right"})
 
+# LaTeX commands emitted from EqEdit operator tokens that latex2mathml otherwise
+# classifies as identifiers. The render boundary uses this metadata only while
+# producing MathML; the public LaTeX remains the canonical command above.
+EQEDIT_MATHML_OPERATOR_COMMANDS = frozenset({OPERATORS["triangle"]})
+
 
 __all__ = [
     "ACCENTS",
     "BIG_OPERATORS",
     "DELIMITERS",
+    "EQEDIT_MATHML_OPERATOR_COMMANDS",
     "FUNCTIONS",
     "GREEK",
     "MATRIX_ENVIRONMENTS",

hwpx/oxml/_paragraph_text_edit.py

@@ -0,0 +1,42 @@
+# SPDX-License-Identifier: Apache-2.0
+"""Small helpers for a style-preserving paragraph text edit."""
+
+from __future__ import annotations
+
+import xml.etree.ElementTree as ET
+
+from ..errors import HwpxValueError
+from ._document_primitives import _is_tab_control_element
+from .namespaces import tag_local_name
+
+
+def plain_text_nodes_for_edit(runs: list[ET.Element]) -> list[ET.Element]:
+    nodes: list[ET.Element] = []
+    for run in runs:
+        for child in run:
+            if tag_local_name(child.tag) == "t":
+                if len(child):
+                    raise HwpxValueError(
+                        "mixed text markup cannot be edited safely",
+                        code="paragraph-mixed-text-unsupported",
+                    )
+                nodes.append(child)
+            elif tag_local_name(child.tag) == "tab" or _is_tab_control_element(child):
+                raise HwpxValueError(
+                    "tab controls require an explicit run target",
+                    code="paragraph-tab-target-required",
+                )
+    return nodes
+
+
+def edit_node_candidates(
+    nodes: list[ET.Element], prefix: int, end: int
+) -> list[tuple[ET.Element, int]]:
+    offset = 0
+    candidates: list[tuple[ET.Element, int]] = []
+    for node in nodes:
+        length = len(node.text or "")
+        if offset <= prefix and end <= offset + length:
+            candidates.append((node, offset))
+        offset += length
+    return candidates

hwpx/oxml/paragraph.py

@@ -40,6 +40,7 @@ from .note_authoring import (
     _paragraph_endnotes,
     _paragraph_footnotes,
 )
+from ._paragraph_text_edit import edit_node_candidates, plain_text_nodes_for_edit
 from .objects import (
     HwpxOxmlInlineObject,
     _create_picture_element,
@@ -336,6 +337,42 @@ class HwpxOxmlParagraph:
         _clear_paragraph_layout_cache(self.element)
         self.section.mark_dirty()
 
+    def set_text_preserving_runs(self, value: str) -> None:
+        """Replace one unambiguous text span without changing run boundaries.
+
+        The edit must fit inside one plain ``hp:t`` node. This covers a
+        correction within a styled run while retaining every other run and
+        control; cross-style replacements require an explicit run target.
+        """
+        from ..errors import HwpxValueError
+
+        nodes = plain_text_nodes_for_edit(self._run_elements())
+        old = "".join(node.text or "" for node in nodes)
+        if old == value:
+            return
+        if not nodes:
+            raise HwpxValueError("paragraph has no plain text node", code="paragraph-text-missing")
+        prefix = 0
+        while prefix < min(len(old), len(value)) and old[prefix] == value[prefix]:
+            prefix += 1
+        suffix = 0
+        while (suffix < len(old) - prefix and suffix < len(value) - prefix
+               and old[len(old) - suffix - 1] == value[len(value) - suffix - 1]):
+            suffix += 1
+        end = len(old) - suffix
+        replacement = value[prefix:len(value) - suffix if suffix else len(value)]
+        candidates = edit_node_candidates(nodes, prefix, end)
+        if len(candidates) != 1:
+            raise HwpxValueError(
+                "text edit crosses run boundaries or has ambiguous insertion style",
+                code="paragraph-text-style-ambiguous",
+            )
+        node, offset = candidates[0]
+        current = node.text or ""
+        node.text = _sanitize_text(current[:prefix - offset] + replacement + current[end - offset:])
+        _clear_paragraph_layout_cache(self.element)
+        self.section.mark_dirty()
+
     def clear_text(self) -> None:
         """Remove all text content while preserving styles and non-text elements.
 

hwpx/oxml/table.py

@@ -464,6 +464,30 @@ class HwpxOxmlTable:
             ")"
         )
 
+    @property
+    def height(self) -> int:
+        """Declared table height in HWP units."""
+        size = self.element.find(f"{_HP}sz")
+        return int(size.get("height", "0")) if size is not None else 0
+
+    @property
+    def treat_as_char(self) -> bool:
+        """Whether Hancom lays out this table as one inline character."""
+        position = self.element.find(f"{_HP}pos")
+        return position is not None and position.get("treatAsChar", "1") == "1"
+
+    def set_treat_as_char(self, value: bool) -> None:
+        """Choose inline layout or flowing table layout across pages."""
+        position = self.element.find(f"{_HP}pos")
+        if position is None:
+            from ..errors import HwpxValueError
+
+            raise HwpxValueError("table has no hp:pos", code="table-position-missing")
+        desired = "1" if value else "0"
+        if position.get("treatAsChar") != desired:
+            position.set("treatAsChar", desired)
+            self.mark_dirty()
+
     # --- caption (hp:caption) -----------------------------------------
 
     @property

{python_hwpx-6.4.0.dist-info → python_hwpx-6.5.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: python-hwpx
-Version: 6.4.0
+Version: 6.5.0
 Summary: 한글 없이 HWPX 문서를 열고, 편집하고, 생성하고, 검증하는 Python 문서 라이브러리
 Author: python-hwpx Maintainers
 License-Expression: Apache-2.0

{python_hwpx-6.4.0.dist-info → python_hwpx-6.5.0.dist-info}/RECORD

@@ -43,11 +43,11 @@ hwpx/data/contract_docs/mutation-semantics.md,sha256=3GjuRgPd7W58y9N1aj7xK10U4DZ
 hwpx/data/contract_docs/recipes-traversal.md,sha256=tj9CsabtI4EmiFmUdNz4MwBQm4apS12s6uZ5v9vek0M,4419
 hwpx/data/contract_docs/support-matrix.md,sha256=vODB2JsgEGFRFHkeEQovwy6McxBSEimjlkhIG4N-93Y,65733
 hwpx/equation/__init__.py,sha256=0wks0_pS_tBokCVMeWUJGE6tbW35Ptc0I5Z_36FY-Jo,1328
-hwpx/equation/authoring.py,sha256=s8wJWNEEFJ2oS4xbs4-Hyv7Dlml_mSQrljdy0hbzpjU,16786
+hwpx/equation/authoring.py,sha256=xuLRxkIPPJmiDfSVX-ngcQYYKZRqSuHV1PBVvBvo6vY,17117
 hwpx/equation/eqedit.py,sha256=FjRNwc_JVC01yprqLFk5N6Nm5qCow2K2x-u3Dw7Wmio,11314
-hwpx/equation/mathml.py,sha256=mbmAojF-4e9cMEQQvgHx5wvp_dG3Mb_sPjqH7SpOaxQ,2150
-hwpx/equation/render.py,sha256=g698YmVWsfG4V-QNROXmSK0za-npJ4Z7xIe6Y5DZ1g0,2874
-hwpx/equation/tokens.py,sha256=RyFyqj3rMllFB5k4d2XO2L8CkiIySb573evJ-fwujhY,7097
+hwpx/equation/mathml.py,sha256=Wbuw9hikBJoxzUdKjMDasvBe8suGGjNUXBTeNrp-GSM,4424
+hwpx/equation/render.py,sha256=P475CF6xhEYveDgw2aiEHhg9TW-3ZTs3MlpyzU5PPvY,2888
+hwpx/equation/tokens.py,sha256=76T57UgeE-08NVPHjN0FD3wvxvIdPksbxgjegvUgf5w,7729
 hwpx/form_fit/__init__.py,sha256=l1HcFge4u_wL2vogB-VxSMrb-8h7zUT_IgTWMtnH9CM,1779
 hwpx/form_fit/apply.py,sha256=X2q7K-mLku4AXxm4-xP80QGzhBlpzpqA_Ade8aPXzNY,3450
 hwpx/form_fit/engine.py,sha256=2W5ePGnpaLDoVPH8RxKdRfVx3r3usy26RBWj0jU1fpE,21553
@@ -74,6 +74,7 @@ hwpx/opc/xml_utils.py,sha256=r4wc0irlf4tOXlOAwXoqReJOY3-EIqd7Uan56XR6dBs,4474
 hwpx/oxml/__init__.py,sha256=Ogl__NFyPORPxboCHqrsn2IrRdcjaQBzmnYm4pPWXpA,6731
 hwpx/oxml/_document_impl.py,sha256=41cxktJqXHER37fjA5HIFR8kw7dHMoISTxEv53PRBYo,5367
 hwpx/oxml/_document_primitives.py,sha256=IcGRT7B4gblNgJ5XDivV_DME-5cU5p88hRSAM1L8_Hk,56887
+hwpx/oxml/_paragraph_text_edit.py,sha256=8IHjDPRh1yXtCYSbL0LER9Z4_wVnXybpCh3qQ8jwmPo,1453
 hwpx/oxml/body.py,sha256=RRopsZ7rvHDMIYNoEwHPty7DXyaZaFRDqJQSI6gVIls,52196
 hwpx/oxml/canonical_defaults.py,sha256=WHAK7u_W-PDv3P3N1-1-aWAoiUgBh4x-UybFL4NqW-w,3791
 hwpx/oxml/color.py,sha256=K2lbU_JdqRxFkQelGD6BbBkH1IspJSQyGlKHNz1Nmf4,2393
@@ -96,7 +97,7 @@ hwpx/oxml/note_authoring.py,sha256=xOoYeX640XkwdftUQeHsdojGHHbiLbobZPi3aahh5ek,7
 hwpx/oxml/numbering.py,sha256=9a0ARGW1DkKsTF7mEEXM4FA5loQmzIBuqE4APA8UFIQ,669
 hwpx/oxml/numbering_kinds.py,sha256=dikaExXr8GZ0gNL4KM5xxvGQOZDCkCdcA5qKDLsstsg,4034
 hwpx/oxml/objects.py,sha256=kL4xy0tBmq8E598sJ67bY2xk5Zwg09apelPqQGnD44w,55570
-hwpx/oxml/paragraph.py,sha256=sdoUFlQ7tsZe0mdyTlpzcj4Peh2mkjtrfL1GuqYFFDw,60979
+hwpx/oxml/paragraph.py,sha256=2TN9XDwBUNA4qFBh7yTPM9M43zpnA-x_grsevh_B7fE,62747
 hwpx/oxml/parser.py,sha256=pIfyNdW3WdFkCcE8JFY7hn1fobRbMngzfhVZICvWeVs,2937
 hwpx/oxml/run.py,sha256=khz9hf2_YpQjYSNR5Qq8XKJ4EmGfiJu9LtfCfLgblnU,16375
 hwpx/oxml/schema.py,sha256=ElR3_IIhhPPZEqJtNKMNCHA-VFVLdhL454W_4spuvlc,1284
@@ -106,7 +107,7 @@ hwpx/oxml/section_layout.py,sha256=TyzddL31pdKSFWYfxVKdIjs3vo8UiG0lERa5-X6bKHI,5
 hwpx/oxml/section_story.py,sha256=OqCXk78gPL0ILQJqDBnhmjtq_tUru0j23IeXXpoWApA,25896
 hwpx/oxml/settings.py,sha256=0cI0zODpxwUgcleVR0J6I4nGT3FGSW8CI0qxwZHXaAk,5523
 hwpx/oxml/simple_parts.py,sha256=XpeVmBWOjv1UVN6sGbskYqwyCIDsNSgzsnnG4QG80Us,3810
-hwpx/oxml/table.py,sha256=DD9NdBZJ2OAJU64D7uXsF4VGizao3ZCOEMLq8ZSaBdI,56582
+hwpx/oxml/table.py,sha256=Pgy3vW2sS2SqLEcnrQVaV8uHmmWivE5C86yX-fNrSlA,57561
 hwpx/oxml/utils.py,sha256=2L9au3SD_MFkFt2DXx3udXBT5caYuYO623lhGhgldVo,2972
 hwpx/oxml/version_part.py,sha256=kFXKQUKgClrM-LuVvBEaSD4RePV1kk1Kt8ltgttefS8,2627
 hwpx/plan/__init__.py,sha256=WvlBce8ic63hf8MsA6x-K_1BVrkNu6gaxKVIM7Axxks,1471
@@ -153,10 +154,10 @@ hwpx/tools/toc_fidelity.py,sha256=rvoKH8QJ65WNVrxS0d-i2AtODvBy-4O8NRudfQ244v4,19
 hwpx/tools/validator.py,sha256=U856izL9NcJZOiKDYoCpwOSaFNQ2Un8Jn4pzL20A96Q,7100
 hwpx/tools/_schemas/header.xsd,sha256=mJXuFMuHGT1JnFFaluUpYUglwjMCNlfbFCRVM26eHXE,664
 hwpx/tools/_schemas/section.xsd,sha256=MgvavVHG05RDfUnVPxVU10H4FQOja5ON04_m9Uk_m7E,522
-python_hwpx-6.4.0.dist-info/licenses/LICENSE,sha256=_ubz4wv-BkkT3l3gu-QuH7JGeVjuRYGZoZK95eNsCHU,9688
-python_hwpx-6.4.0.dist-info/licenses/NOTICE,sha256=he3PBLXnIZK0sNsozc-SAQ2jvWjV61goylWUZ8HPi3s,3837
-python_hwpx-6.4.0.dist-info/METADATA,sha256=i_FNveRXdfDhRbj7G_l5pvAJ5KmRVDR6qKVMzhHwMnI,14995
-python_hwpx-6.4.0.dist-info/WHEEL,sha256=YVMoNqKzERt-wjUZwJ33xBGAwnFl-4cqbYkTtWa4itE,91
-python_hwpx-6.4.0.dist-info/entry_points.txt,sha256=JUKRxbly9UaeHV7YzOea23y8IiqSTcrhUlooP3fS_Zc,405
-python_hwpx-6.4.0.dist-info/top_level.txt,sha256=R1iToqDh80Nf2oQhRjTN0rbN2X6kyDUizIocZjkhuxc,5
-python_hwpx-6.4.0.dist-info/RECORD,,
+python_hwpx-6.5.0.dist-info/licenses/LICENSE,sha256=_ubz4wv-BkkT3l3gu-QuH7JGeVjuRYGZoZK95eNsCHU,9688
+python_hwpx-6.5.0.dist-info/licenses/NOTICE,sha256=he3PBLXnIZK0sNsozc-SAQ2jvWjV61goylWUZ8HPi3s,3837
+python_hwpx-6.5.0.dist-info/METADATA,sha256=MLawxpU4hei8fmJR8XZXiU2Mc-RBjnI3dHxuywHr7tE,14995
+python_hwpx-6.5.0.dist-info/WHEEL,sha256=YVMoNqKzERt-wjUZwJ33xBGAwnFl-4cqbYkTtWa4itE,91
+python_hwpx-6.5.0.dist-info/entry_points.txt,sha256=JUKRxbly9UaeHV7YzOea23y8IiqSTcrhUlooP3fS_Zc,405
+python_hwpx-6.5.0.dist-info/top_level.txt,sha256=R1iToqDh80Nf2oQhRjTN0rbN2X6kyDUizIocZjkhuxc,5
+python_hwpx-6.5.0.dist-info/RECORD,,