python-hwpx-automation 6.0.3__py3-none-any.whl

hwpx_automation/office/authoring/design/harvest.py

@@ -0,0 +1,305 @@
+# SPDX-License-Identifier: Apache-2.0
+"""Harvest a profile from a real Hancom-saved document (plan §2 Phase E, task 1).
+
+Given a genuine ``.hwpx`` it writes, under ``profiles/<id>/``:
+
+* ``template.hwpx`` — a **body-stripped skeleton**: the real document with its
+  section body reduced to a single ``secPr``-carrying paragraph. ``header.xml``
+  (every style/font/borderFill) and the page setup survive untouched, so the
+  committed artifact is small and carries no document content (no PII), yet is
+  genuinely Hancom-derived — never imagined XML.
+* ``fragments/{title,heading,body,info_table}.xml`` — real ``<hp:p>`` / ``<hp:tbl>``
+  elements harvested from the source, with their text **sanitised to placeholders**
+  (the style is kept, the content dropped).
+* ``profile.json`` — the manifest the loader reads.
+
+This is an authoring-time tool (run once per profile); the committed artifacts are
+what production uses.
+"""
+from __future__ import annotations
+
+import json
+from collections import Counter
+from copy import deepcopy
+from pathlib import Path
+
+from lxml import etree  # pyright: ignore[reportAttributeAccessIssue]
+
+from hwpx.document import HwpxDocument
+
+import hwpx_automation.office.authoring.design._support as S
+
+PROFILE_SCHEMA = "hwpx.design.profile.v1"
+
+_PLACEHOLDER = {
+    "title": "{{title}}",
+    "heading": "{{heading}}",
+    "body": "{{body}}",
+}
+
+
+def _font_height(doc, ref) -> int:
+    style = doc.char_property(ref)
+    if style is None:
+        return 0
+    try:
+        return int(style.attributes.get("height") or 0)
+    except ValueError:
+        return 0
+
+
+def _text_paragraphs(doc):
+    out = []
+    for index, para in enumerate(doc.sections[0].paragraphs):
+        if getattr(para, "tables", []):
+            continue
+        text = (para.text or "").strip()
+        if not text:
+            continue
+        runs = para.runs
+        cref = runs[0].char_pr_id_ref if runs else "0"
+        out.append(
+            {
+                "index": index,
+                "para": para.element.get("paraPrIDRef"),
+                "char": cref,
+                "h": _font_height(doc, cref),
+                "len": len(text),
+                "element": para.element,
+            }
+        )
+    return out
+
+
+def _pick_fragments(doc) -> dict[str, etree._Element]:
+    tps = _text_paragraphs(doc)
+    if not tps:
+        raise ValueError("no text paragraphs to harvest")
+    prose = [t for t in tps if t["len"] >= 8] or tps
+
+    title = max(tps, key=lambda t: t["h"])
+    combo = Counter((t["para"], t["char"]) for t in prose)
+    body_key = combo.most_common(1)[0][0]
+    body = next(t for t in prose if (t["para"], t["char"]) == body_key)
+    head_pool = [
+        t for t in tps
+        if (t["para"], t["char"]) != body_key
+        and Counter((x["para"], x["char"]) for x in tps)[(t["para"], t["char"])] >= 2
+        and t["h"] >= body["h"]
+    ]
+    heading = max(head_pool, key=lambda t: t["h"], default=title)
+
+    picks = {"title": title, "heading": heading, "body": body}
+    fragments = {role: deepcopy(p["element"]) for role, p in picks.items()}
+    for role, frag in fragments.items():
+        S.set_paragraph_text(frag, _PLACEHOLDER[role])
+
+    info = _pick_info_table(doc)
+    if info is not None:
+        fragments["info_table"] = info
+    return fragments
+
+
+def _references_binary(element: etree._Element) -> bool:
+    """True if *element* embeds an image / binary (so harvesting would dangle)."""
+
+    for node in element.iter():
+        if S.ln(node.tag).lower() in ("pic", "img", "image", "ole", "drawing"):
+            return True
+        for key in node.attrib:
+            if "binaryitemidref" in S.ln(key).lower() or "bindata" in S.ln(key).lower():
+                return True
+    return False
+
+
+def _is_simple_grid(table_element: etree._Element) -> bool:
+    """True if every cell is 1×1 (no row/col merges) — safe to row-clone."""
+
+    for tc in S.iter_local(table_element, "tc"):
+        span = next((c for c in tc if S.ln(c.tag) == "cellSpan"), None)
+        if span is not None:
+            if int(span.get("rowSpan", "1")) > 1 or int(span.get("colSpan", "1")) > 1:
+                return False
+    return True
+
+
+def _pick_info_table(doc) -> etree._Element | None:
+    best = None
+    for para in doc.sections[0].paragraphs:
+        for table in getattr(para, "tables", []):
+            rows, cols = table.row_count, table.column_count
+            if not (2 <= cols <= 4 and 2 <= rows <= 12):
+                continue
+            if _references_binary(table.element):
+                continue  # text-only key-value tables, never image grids
+            if not _is_simple_grid(table.element):
+                continue  # merge-free grids only — the composer row-clones them
+            score = rows * cols
+            if best is None or score < best[0]:
+                best = (score, table)
+    if best is None:
+        return None
+    # Harvest the *paragraph* that wraps the table — a table is not a valid
+    # section child on its own; it must ride inside an <hp:p>.
+    frag = deepcopy(best[1].paragraph.element)
+    # Sanitise every cell to a neutral placeholder; structure is what we keep.
+    for tc in S.iter_local(frag, "tc"):
+        sublist = _subList(tc)
+        for para in S.children_local(sublist if sublist is not None else tc, "p"):
+            S.set_paragraph_text(para, "{{cell}}")
+    S.strip_lineseg(frag)
+    return frag
+
+
+def _subList(tc: etree._Element):
+    for child in tc:
+        if S.ln(child.tag) == "subList":
+            return child
+    return None
+
+
+def _build_skeleton(source: Path, out_path: Path) -> dict:
+    doc = HwpxDocument.open(source)
+    section = doc.sections[0]
+    secpr = S.find_secpr(section.element)
+    if secpr is None:
+        doc.close()
+        raise ValueError("template has no secPr (page setup)")
+    secpr = deepcopy(secpr)
+
+    empty = etree.SubElement(section.element, S.hp(section.element, "p"))
+    run = etree.SubElement(empty, S.hp(section.element, "run"), {"charPrIDRef": "0"})
+    etree.SubElement(run, S.hp(section.element, "t")).text = ""
+    section.element.remove(empty)  # detach; replace_section_body re-inserts
+    S.replace_section_body(section.element, [empty])
+    S.move_secpr_into(empty, secpr)
+    section.mark_dirty()
+
+    stripped = _strip_binaries(doc)
+    _scrub_metadata(doc)
+
+    page = {}
+    try:
+        ps, pm = section.properties.page_size, section.properties.page_margins
+        page = {
+            "width": ps.width, "height": ps.height, "orientation": ps.orientation,
+            "margins": {"left": pm.left, "right": pm.right, "top": pm.top, "bottom": pm.bottom},
+        }
+    except Exception:  # pragma: no cover - diagnostics only
+        page = {}
+
+    report = doc.save_report(out_path)
+    style_count = len(doc.char_properties)
+    doc.close()
+    return {
+        "page": page,
+        "char_pr_count": style_count,
+        "open_safe": report.open_safety.ok,
+        "stripped_parts": stripped,
+    }
+
+
+def _strip_binaries(doc) -> list[str]:
+    """Drop embedded images + the source preview (size + PII) from the skeleton.
+
+    The body is already reduced to one empty paragraph, so nothing references
+    these parts; removing them (and their manifest items) keeps the committed
+    skeleton small and content-free while leaving every style in header.xml.
+    """
+
+    package = doc._package
+    removed: list[str] = []
+
+    for item in list(package._manifest_items()):
+        href = item.get("href", "") or ""
+        media = item.get("media-type", "") or ""
+        base = href.rsplit("/", 1)[-1]
+        if (
+            "BinData/" in href
+            or base.startswith("PrvImage")
+            or media.lower().startswith("image/")
+        ):
+            item_id = item.get("id")
+            if item_id:
+                package.remove_manifest_item(item_id)
+
+    for name in list(package.part_names()):
+        if name.startswith("BinData/") or name.startswith("Preview/PrvImage"):
+            try:
+                package.delete(name)
+                removed.append(name)
+            except Exception:  # pragma: no cover - defensive
+                pass
+
+    if package.has_part("Preview/PrvText.txt"):
+        package.set_part("Preview/PrvText.txt", b"")  # clear source text (PII)
+    return removed
+
+
+# Manifest (content.hpf) metadata that carries source PII — author, the person who
+# last saved, the source title, and timestamps. Cleared so generated docs inherit
+# nothing identifying.
+_SCRUB_META_NAMES = {
+    "creator", "lastsaveby", "subject", "description", "keyword",
+    "date", "CreatedDate", "ModifiedDate",
+}
+
+
+def _scrub_metadata(doc) -> None:
+    """Blank source-identifying OPF metadata in Contents/content.hpf (PII)."""
+
+    package = doc._package
+    try:
+        tree = package.manifest_tree()
+    except Exception:  # pragma: no cover - defensive
+        return
+    for element in tree.iter():
+        name = S.ln(element.tag)
+        if name == "title":
+            element.text = ""
+        elif name == "meta" and element.get("name") in _SCRUB_META_NAMES:
+            element.text = ""
+    package.set_part(package.main_content.full_path, tree)
+
+
+def harvest_profile(source: str | Path, profile_id: str, out_root: str | Path) -> dict:
+    """Harvest *source* into ``<out_root>/<profile_id>/`` and return the manifest."""
+
+    source = Path(source)
+    out_dir = Path(out_root) / profile_id
+    (out_dir / "fragments").mkdir(parents=True, exist_ok=True)
+
+    # Fragments first (reads the full source), then the skeleton (rewrites body).
+    doc = HwpxDocument.open(source)
+    fragments = _pick_fragments(doc)
+    doc.close()
+
+    roles = []
+    for role, frag in fragments.items():
+        path = out_dir / "fragments" / f"{role}.xml"
+        path.write_bytes(etree.tostring(frag, encoding="UTF-8"))
+        roles.append(role)
+
+    skel = _build_skeleton(source, out_dir / "template.hwpx")
+
+    manifest = {
+        "schemaVersion": PROFILE_SCHEMA,
+        "id": profile_id,
+        "template": "template.hwpx",
+        "fragments": {role: f"fragments/{role}.xml" for role in roles},
+        "page": skel["page"],
+        "char_pr_count": skel["char_pr_count"],
+        "style_coverage_threshold": 0.98,
+        "source_basename": source.name,
+    }
+    (out_dir / "profile.json").write_text(
+        json.dumps(manifest, ensure_ascii=False, indent=2), encoding="utf-8"
+    )
+    return {
+        **manifest,
+        "skeleton_open_safe": skel["open_safe"],
+        "stripped_parts": skel.get("stripped_parts", []),
+    }
+
+
+__all__ = ["harvest_profile", "PROFILE_SCHEMA"]

hwpx_automation/office/authoring/design/plan.py

@@ -0,0 +1,69 @@
+# SPDX-License-Identifier: Apache-2.0
+"""``DocumentPlan`` — the model-facing description of a new document (Appendix B).
+
+The LLM emits a plan (``hwpx.document_plan.v1``), never XML. Phase E lowers it
+natively onto a verified Hancom-saved template + harvested fragments
+(:mod:`hwpx.design.composer`).
+"""
+from __future__ import annotations
+
+from dataclasses import dataclass, field
+from typing import Any, Literal
+
+DOCUMENT_PLAN_SCHEMA = "hwpx.document_plan.v1"
+
+BlockType = Literal["paragraph", "table"]
+# Paragraph roles map to harvested fragments; table → the info-table fragment.
+BlockRole = Literal["title", "heading", "subheading", "body", "info"]
+
+
+@dataclass(slots=True)
+class Block:
+    """One block of a :class:`DocumentPlan`."""
+
+    type: BlockType
+    role: BlockRole = "body"
+    text: str = ""
+    columns: list[str] = field(default_factory=list)
+    rows: list[list[str]] = field(default_factory=list)
+
+    @classmethod
+    def from_dict(cls, data: dict[str, Any]) -> "Block":
+        return cls(
+            type=data.get("type", "paragraph"),
+            role=data.get("role", "body"),
+            text=str(data.get("text", "")),
+            columns=[str(c) for c in data.get("columns", [])],
+            rows=[[str(c) for c in row] for row in data.get("rows", [])],
+        )
+
+
+@dataclass(slots=True)
+class DocumentPlan:
+    """A new-document plan (Appendix B ``hwpx.document_plan.v1``)."""
+
+    profile: str
+    title: str = ""
+    blocks: list[Block] = field(default_factory=list)
+
+    @classmethod
+    def from_dict(cls, data: dict[str, Any]) -> "DocumentPlan":
+        if not data.get("profile"):
+            raise ValueError("DocumentPlan requires a 'profile'")
+        return cls(
+            profile=str(data["profile"]),
+            title=str(data.get("title", "")),
+            blocks=[Block.from_dict(b) for b in data.get("blocks", [])],
+        )
+
+    def iter_blocks(self) -> list[Block]:
+        """Blocks with the title promoted to a leading title block."""
+
+        blocks: list[Block] = []
+        if self.title:
+            blocks.append(Block(type="paragraph", role="title", text=self.title))
+        blocks.extend(self.blocks)
+        return blocks
+
+
+__all__ = ["DocumentPlan", "Block", "BlockType", "BlockRole", "DOCUMENT_PLAN_SCHEMA"]

hwpx_automation/office/authoring/design/profile.py

@@ -0,0 +1,88 @@
+# SPDX-License-Identifier: Apache-2.0
+"""Profile loader + registry (plan §2 Phase E, task 1).
+
+A :class:`Profile` bundles a verified Hancom-saved skeleton ``template.hwpx`` with
+the harvested ``<hp:p>``/``<hp:tbl>`` fragments keyed by role. The default registry
+loads the profiles committed under ``hwpx/design/profiles/``.
+"""
+from __future__ import annotations
+
+import json
+from dataclasses import dataclass, field
+from functools import lru_cache
+from pathlib import Path
+from typing import Any
+
+from lxml import etree  # pyright: ignore[reportAttributeAccessIssue]
+
+PROFILES_DIRNAME = "profiles"
+
+
+@dataclass(slots=True)
+class Profile:
+    """A verified template skeleton + its harvested fragments."""
+
+    id: str
+    root: Path
+    manifest: dict[str, Any]
+    template_bytes: bytes
+    _fragments: dict[str, bytes] = field(default_factory=dict)
+
+    @property
+    def style_coverage_threshold(self) -> float:
+        return float(self.manifest.get("style_coverage_threshold", 0.98))
+
+    @property
+    def roles(self) -> list[str]:
+        return list(self._fragments)
+
+    def has_role(self, role: str) -> bool:
+        return role in self._fragments
+
+    def fragment(self, role: str) -> etree._Element:
+        """Return a fresh (deep, detached) clone of the *role* fragment element."""
+
+        if role not in self._fragments:
+            raise KeyError(f"profile {self.id!r} has no fragment for role {role!r}")
+        return etree.fromstring(self._fragments[role])
+
+    @classmethod
+    def load(cls, root: str | Path) -> "Profile":
+        root = Path(root)
+        manifest = json.loads((root / "profile.json").read_text(encoding="utf-8"))
+        template_bytes = (root / manifest["template"]).read_bytes()
+        fragments: dict[str, bytes] = {}
+        for role, rel in manifest.get("fragments", {}).items():
+            fragments[role] = (root / rel).read_bytes()
+        return cls(
+            id=manifest["id"],
+            root=root,
+            manifest=manifest,
+            template_bytes=template_bytes,
+            _fragments=fragments,
+        )
+
+
+@lru_cache(maxsize=1)
+def _registry_root() -> Path:
+    return Path(__file__).resolve().parent / PROFILES_DIRNAME
+
+
+def available_profiles() -> list[str]:
+    root = _registry_root()
+    return sorted(p.name for p in root.iterdir() if (p / "profile.json").exists())
+
+
+@lru_cache(maxsize=None)
+def load_profile(profile_id: str) -> Profile:
+    """Load a committed profile by id (cached)."""
+
+    root = _registry_root() / profile_id
+    if not (root / "profile.json").exists():
+        raise KeyError(
+            f"unknown profile {profile_id!r}; available: {available_profiles()}"
+        )
+    return Profile.load(root)
+
+
+__all__ = ["Profile", "load_profile", "available_profiles", "PROFILES_DIRNAME"]

hwpx_automation/office/authoring/design/profiles/application_form/fragments/body.xml

@@ -0,0 +1 @@
+<hp:p xmlns:hp="http://www.hancom.co.kr/hwpml/2011/paragraph" id="2147483648" paraPrIDRef="107" styleIDRef="0" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="142"><hp:t>{{body}}</hp:t></hp:run><hp:run charPrIDRef="32"><hp:t></hp:t></hp:run><hp:run charPrIDRef="143"><hp:t></hp:t></hp:run><hp:run charPrIDRef="160"><hp:t></hp:t></hp:run><hp:run charPrIDRef="174"><hp:t></hp:t></hp:run><hp:run charPrIDRef="163"><hp:t></hp:t></hp:run><hp:run charPrIDRef="145"><hp:t></hp:t></hp:run><hp:run charPrIDRef="146"><hp:t></hp:t></hp:run></hp:p>

hwpx_automation/office/authoring/design/profiles/application_form/fragments/heading.xml

@@ -0,0 +1 @@
+<hp:p xmlns:hp="http://www.hancom.co.kr/hwpml/2011/paragraph" id="0" paraPrIDRef="1" styleIDRef="0" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="141"><hp:t>{{heading}}</hp:t></hp:run></hp:p>

hwpx_automation/office/authoring/design/profiles/application_form/fragments/info_table.xml

@@ -0,0 +1 @@
+<hp:p xmlns:hp="http://www.hancom.co.kr/hwpml/2011/paragraph" id="0" paraPrIDRef="93" styleIDRef="0" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="161"><hp:t>  </hp:t><hp:tbl id="2052787869" zOrder="32" numberingType="TABLE" textWrap="TOP_AND_BOTTOM" textFlow="BOTH_SIDES" lock="0" dropcapstyle="None" pageBreak="CELL" repeatHeader="1" rowCnt="3" colCnt="2" cellSpacing="0" borderFillIDRef="4" noAdjust="0"><hp:sz width="46946" widthRelTo="ABSOLUTE" height="19347" heightRelTo="ABSOLUTE" protect="0"/><hp:pos treatAsChar="1" affectLSpacing="0" flowWithText="1" allowOverlap="0" holdAnchorAndSO="0" vertRelTo="PARA" horzRelTo="PARA" vertAlign="TOP" horzAlign="LEFT" vertOffset="0" horzOffset="0"/><hp:outMargin left="0" right="0" top="0" bottom="0"/><hp:inMargin left="510" right="510" top="141" bottom="141"/><hp:tr><hp:tc name="" header="0" hasMargin="0" protect="0" editable="0" dirty="0" borderFillIDRef="238"><hp:subList id="" textDirection="HORIZONTAL" lineWrap="BREAK" vertAlign="CENTER" linkListIDRef="0" linkListNextIDRef="0" textWidth="0" textHeight="0" hasTextRef="0" hasNumRef="0"><hp:p id="2147483648" paraPrIDRef="66" styleIDRef="0" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="79"><hp:t>{{cell}}</hp:t></hp:run></hp:p></hp:subList><hp:cellAddr colAddr="0" rowAddr="0"/><hp:cellSpan colSpan="1" rowSpan="1"/><hp:cellSz width="5253" height="1639"/><hp:cellMargin left="510" right="510" top="141" bottom="141"/></hp:tc><hp:tc name="" header="0" hasMargin="0" protect="0" editable="0" dirty="0" borderFillIDRef="238"><hp:subList id="" textDirection="HORIZONTAL" lineWrap="BREAK" vertAlign="CENTER" linkListIDRef="0" linkListNextIDRef="0" textWidth="0" textHeight="0" hasTextRef="0" hasNumRef="0"><hp:p id="2147483648" paraPrIDRef="6" styleIDRef="21" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="79"><hp:t>{{cell}}</hp:t></hp:run></hp:p></hp:subList><hp:cellAddr colAddr="1" rowAddr="0"/><hp:cellSpan colSpan="1" rowSpan="1"/><hp:cellSz width="41693" height="1639"/><hp:cellMargin left="510" right="510" top="141" bottom="141"/></hp:tc></hp:tr><hp:tr><hp:tc name="" header="0" hasMargin="0" protect="0" editable="0" dirty="0" borderFillIDRef="258"><hp:subList id="" textDirection="HORIZONTAL" lineWrap="BREAK" vertAlign="CENTER" linkListIDRef="0" linkListNextIDRef="0" textWidth="0" textHeight="0" hasTextRef="0" hasNumRef="0"><hp:p id="2147483648" paraPrIDRef="66" styleIDRef="0" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="69"><hp:t>{{cell}}</hp:t></hp:run></hp:p></hp:subList><hp:cellAddr colAddr="0" rowAddr="1"/><hp:cellSpan colSpan="1" rowSpan="1"/><hp:cellSz width="5253" height="8854"/><hp:cellMargin left="510" right="510" top="141" bottom="141"/></hp:tc><hp:tc name="" header="0" hasMargin="0" protect="0" editable="0" dirty="0" borderFillIDRef="258"><hp:subList id="" textDirection="HORIZONTAL" lineWrap="BREAK" vertAlign="CENTER" linkListIDRef="0" linkListNextIDRef="0" textWidth="0" textHeight="0" hasTextRef="0" hasNumRef="0"><hp:p id="2147483648" paraPrIDRef="103" styleIDRef="21" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="80"><hp:t>{{cell}}</hp:t></hp:run></hp:p><hp:p id="0" paraPrIDRef="103" styleIDRef="21" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="80"><hp:t>{{cell}}</hp:t></hp:run></hp:p><hp:p id="0" paraPrIDRef="103" styleIDRef="21" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="80"><hp:t>{{cell}}</hp:t></hp:run></hp:p><hp:p id="0" paraPrIDRef="103" styleIDRef="21" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="80"><hp:t>{{cell}}</hp:t></hp:run></hp:p><hp:p id="0" paraPrIDRef="103" styleIDRef="21" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="80"><hp:t>{{cell}}</hp:t></hp:run></hp:p></hp:subList><hp:cellAddr colAddr="1" rowAddr="1"/><hp:cellSpan colSpan="1" rowSpan="1"/><hp:cellSz width="41693" height="8854"/><hp:cellMargin left="510" right="510" top="141" bottom="141"/></hp:tc></hp:tr><hp:tr><hp:tc name="" header="0" hasMargin="0" protect="0" editable="0" dirty="0" borderFillIDRef="259"><hp:subList id="" textDirection="HORIZONTAL" lineWrap="BREAK" vertAlign="CENTER" linkListIDRef="0" linkListNextIDRef="0" textWidth="0" textHeight="0" hasTextRef="0" hasNumRef="0"><hp:p id="0" paraPrIDRef="66" styleIDRef="0" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="69"><hp:t>{{cell}}</hp:t></hp:run></hp:p></hp:subList><hp:cellAddr colAddr="0" rowAddr="2"/><hp:cellSpan colSpan="1" rowSpan="1"/><hp:cellSz width="5253" height="8854"/><hp:cellMargin left="510" right="510" top="141" bottom="141"/></hp:tc><hp:tc name="" header="0" hasMargin="0" protect="0" editable="0" dirty="0" borderFillIDRef="259"><hp:subList id="" textDirection="HORIZONTAL" lineWrap="BREAK" vertAlign="CENTER" linkListIDRef="0" linkListNextIDRef="0" textWidth="0" textHeight="0" hasTextRef="0" hasNumRef="0"><hp:p id="0" paraPrIDRef="103" styleIDRef="21" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="80"><hp:t>{{cell}}</hp:t></hp:run></hp:p><hp:p id="0" paraPrIDRef="103" styleIDRef="21" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="80"><hp:t>{{cell}}</hp:t></hp:run></hp:p><hp:p id="0" paraPrIDRef="103" styleIDRef="21" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="80"><hp:t>{{cell}}</hp:t></hp:run></hp:p><hp:p id="0" paraPrIDRef="103" styleIDRef="21" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="80"><hp:t>{{cell}}</hp:t></hp:run></hp:p></hp:subList><hp:cellAddr colAddr="1" rowAddr="2"/><hp:cellSpan colSpan="1" rowSpan="1"/><hp:cellSz width="41693" height="8854"/><hp:cellMargin left="510" right="510" top="141" bottom="141"/></hp:tc></hp:tr></hp:tbl><hp:t/></hp:run></hp:p>

hwpx_automation/office/authoring/design/profiles/application_form/fragments/title.xml

@@ -0,0 +1 @@
+<hp:p xmlns:hp="http://www.hancom.co.kr/hwpml/2011/paragraph" id="2147483648" paraPrIDRef="100" styleIDRef="0" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="141"><hp:t>{{title}}</hp:t></hp:run></hp:p>

hwpx_automation/office/authoring/design/profiles/application_form/profile.json

@@ -0,0 +1,25 @@
+{
+  "schemaVersion": "hwpx.design.profile.v1",
+  "id": "application_form",
+  "template": "template.hwpx",
+  "fragments": {
+    "title": "fragments/title.xml",
+    "heading": "fragments/heading.xml",
+    "body": "fragments/body.xml",
+    "info_table": "fragments/info_table.xml"
+  },
+  "page": {
+    "width": 59528,
+    "height": 84188,
+    "orientation": "WIDELY",
+    "margins": {
+      "left": 5669,
+      "right": 5669,
+      "top": 4251,
+      "bottom": 4251
+    }
+  },
+  "char_pr_count": 249,
+  "style_coverage_threshold": 0.98,
+  "source_basename": "2026%EB%85%84+AI+%EC%A4%91%EC%A0%90%ED%95%99%EA%B5%90+%EC%8B%A0%EC%B2%AD%EC%84%9C+%EB%B0%8F+%EC%9A%B4%EC%98%81%EA%B3%84%ED%9A%8D%EC%84%9C%28%EC%9C%A0%ED%98%951_%EB%A7%A4%EC%9B%90%EA%B3%A0%29_%EC%97%90%EB%93%80%EB%84%B7+%ED%83%91%EC%9E%AC%EC%9A%A9.hwpx"
+}

hwpx_automation/office/authoring/design/profiles/home_notice/fragments/body.xml

@@ -0,0 +1 @@
+<hp:p xmlns:hp="http://www.hancom.co.kr/hwpml/2011/paragraph" id="2147483648" paraPrIDRef="1" styleIDRef="0" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="26"><hp:t>{{body}}</hp:t></hp:run></hp:p>

hwpx_automation/office/authoring/design/profiles/home_notice/fragments/heading.xml

@@ -0,0 +1 @@
+<hp:p xmlns:hp="http://www.hancom.co.kr/hwpml/2011/paragraph" id="2147483648" paraPrIDRef="1" styleIDRef="0" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="25"><hp:t>{{heading}}</hp:t></hp:run></hp:p>

hwpx_automation/office/authoring/design/profiles/home_notice/fragments/title.xml

@@ -0,0 +1 @@
+<hp:p xmlns:hp="http://www.hancom.co.kr/hwpml/2011/paragraph" id="2147483648" paraPrIDRef="25" styleIDRef="0" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="17"><hp:t>{{title}}</hp:t></hp:run></hp:p>

hwpx_automation/office/authoring/design/profiles/home_notice/profile.json

@@ -0,0 +1,24 @@
+{
+  "schemaVersion": "hwpx.design.profile.v1",
+  "id": "home_notice",
+  "template": "template.hwpx",
+  "fragments": {
+    "title": "fragments/title.xml",
+    "heading": "fragments/heading.xml",
+    "body": "fragments/body.xml"
+  },
+  "page": {
+    "width": 59528,
+    "height": 84186,
+    "orientation": "WIDELY",
+    "margins": {
+      "left": 4251,
+      "right": 4251,
+      "top": 1417,
+      "bottom": 0
+    }
+  },
+  "char_pr_count": 54,
+  "style_coverage_threshold": 0.98,
+  "source_basename": "ganghwa_records.hwpx"
+}

hwpx_automation/office/authoring/design/profiles/official_notice/fragments/body.xml

@@ -0,0 +1 @@
+<hp:p xmlns:hp="http://www.hancom.co.kr/hwpml/2011/paragraph" id="2147483648" paraPrIDRef="33" styleIDRef="0" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="18"><hp:t>{{body}}</hp:t></hp:run></hp:p>

hwpx_automation/office/authoring/design/profiles/official_notice/fragments/heading.xml

@@ -0,0 +1 @@
+<hp:p xmlns:hp="http://www.hancom.co.kr/hwpml/2011/paragraph" id="0" paraPrIDRef="25" styleIDRef="0" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="13"><hp:t>{{heading}}</hp:t></hp:run></hp:p>

hwpx_automation/office/authoring/design/profiles/official_notice/fragments/info_table.xml

@@ -0,0 +1 @@
+<hp:p xmlns:hp="http://www.hancom.co.kr/hwpml/2011/paragraph" id="0" paraPrIDRef="28" styleIDRef="0" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="15"><hp:t>  </hp:t></hp:run><hp:run charPrIDRef="14"><hp:tbl id="1465524190" zOrder="10" numberingType="TABLE" textWrap="TOP_AND_BOTTOM" textFlow="BOTH_SIDES" lock="0" dropcapstyle="None" pageBreak="CELL" repeatHeader="1" rowCnt="2" colCnt="3" cellSpacing="0" borderFillIDRef="12" noAdjust="0"><hp:sz width="46490" widthRelTo="ABSOLUTE" height="7274" heightRelTo="ABSOLUTE" protect="0"/><hp:pos treatAsChar="1" affectLSpacing="0" flowWithText="1" allowOverlap="0" holdAnchorAndSO="0" vertRelTo="PARA" horzRelTo="PARA" vertAlign="TOP" horzAlign="LEFT" vertOffset="0" horzOffset="0"/><hp:outMargin left="141" right="141" top="141" bottom="141"/><hp:inMargin left="510" right="510" top="141" bottom="141"/><hp:tr><hp:tc name="" header="0" hasMargin="0" protect="0" editable="0" dirty="0" borderFillIDRef="15"><hp:subList id="" textDirection="HORIZONTAL" lineWrap="BREAK" vertAlign="CENTER" linkListIDRef="0" linkListNextIDRef="0" textWidth="0" textHeight="0" hasTextRef="0" hasNumRef="0"><hp:p id="0" paraPrIDRef="22" styleIDRef="0" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="42"><hp:t>{{cell}}</hp:t></hp:run></hp:p></hp:subList><hp:cellAddr colAddr="0" rowAddr="0"/><hp:cellSpan colSpan="1" rowSpan="1"/><hp:cellSz width="13610" height="2414"/><hp:cellMargin left="510" right="510" top="141" bottom="141"/></hp:tc><hp:tc name="" header="0" hasMargin="0" protect="0" editable="0" dirty="0" borderFillIDRef="15"><hp:subList id="" textDirection="HORIZONTAL" lineWrap="BREAK" vertAlign="CENTER" linkListIDRef="0" linkListNextIDRef="0" textWidth="0" textHeight="0" hasTextRef="0" hasNumRef="0"><hp:p id="0" paraPrIDRef="22" styleIDRef="0" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="42"><hp:t>{{cell}}</hp:t></hp:run></hp:p></hp:subList><hp:cellAddr colAddr="1" rowAddr="0"/><hp:cellSpan colSpan="1" rowSpan="1"/><hp:cellSz width="13610" height="2414"/><hp:cellMargin left="510" right="510" top="141" bottom="141"/></hp:tc><hp:tc name="" header="0" hasMargin="0" protect="0" editable="0" dirty="0" borderFillIDRef="15"><hp:subList id="" textDirection="HORIZONTAL" lineWrap="BREAK" vertAlign="CENTER" linkListIDRef="0" linkListNextIDRef="0" textWidth="0" textHeight="0" hasTextRef="0" hasNumRef="0"><hp:p id="0" paraPrIDRef="22" styleIDRef="0" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="52"><hp:t>{{cell}}</hp:t></hp:run></hp:p></hp:subList><hp:cellAddr colAddr="2" rowAddr="0"/><hp:cellSpan colSpan="1" rowSpan="1"/><hp:cellSz width="19270" height="2414"/><hp:cellMargin left="510" right="510" top="141" bottom="141"/></hp:tc></hp:tr><hp:tr><hp:tc name="" header="0" hasMargin="0" protect="0" editable="0" dirty="0" borderFillIDRef="12"><hp:subList id="" textDirection="HORIZONTAL" lineWrap="BREAK" vertAlign="CENTER" linkListIDRef="0" linkListNextIDRef="0" textWidth="0" textHeight="0" hasTextRef="0" hasNumRef="0"><hp:p id="0" paraPrIDRef="58" styleIDRef="0" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="42"><hp:t>{{cell}}</hp:t></hp:run></hp:p><hp:p id="0" paraPrIDRef="58" styleIDRef="0" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="42"><hp:t>{{cell}}</hp:t></hp:run></hp:p></hp:subList><hp:cellAddr colAddr="0" rowAddr="1"/><hp:cellSpan colSpan="1" rowSpan="1"/><hp:cellSz width="13610" height="4860"/><hp:cellMargin left="510" right="510" top="141" bottom="141"/></hp:tc><hp:tc name="" header="0" hasMargin="0" protect="0" editable="0" dirty="0" borderFillIDRef="12"><hp:subList id="" textDirection="HORIZONTAL" lineWrap="BREAK" vertAlign="CENTER" linkListIDRef="0" linkListNextIDRef="0" textWidth="0" textHeight="0" hasTextRef="0" hasNumRef="0"><hp:p id="0" paraPrIDRef="58" styleIDRef="0" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="42"><hp:t>{{cell}}</hp:t></hp:run></hp:p><hp:p id="0" paraPrIDRef="58" styleIDRef="0" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="42"><hp:t>{{cell}}</hp:t></hp:run></hp:p></hp:subList><hp:cellAddr colAddr="1" rowAddr="1"/><hp:cellSpan colSpan="1" rowSpan="1"/><hp:cellSz width="13610" height="4860"/><hp:cellMargin left="510" right="510" top="141" bottom="141"/></hp:tc><hp:tc name="" header="0" hasMargin="0" protect="0" editable="0" dirty="0" borderFillIDRef="12"><hp:subList id="" textDirection="HORIZONTAL" lineWrap="BREAK" vertAlign="CENTER" linkListIDRef="0" linkListNextIDRef="0" textWidth="0" textHeight="0" hasTextRef="0" hasNumRef="0"><hp:p id="0" paraPrIDRef="58" styleIDRef="0" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="52"><hp:t>{{cell}}</hp:t></hp:run></hp:p><hp:p id="0" paraPrIDRef="58" styleIDRef="0" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="52"><hp:t>{{cell}}</hp:t></hp:run></hp:p></hp:subList><hp:cellAddr colAddr="2" rowAddr="1"/><hp:cellSpan colSpan="1" rowSpan="1"/><hp:cellSz width="19270" height="4860"/><hp:cellMargin left="510" right="510" top="141" bottom="141"/></hp:tc></hp:tr></hp:tbl><hp:t/></hp:run></hp:p>

hwpx_automation/office/authoring/design/profiles/official_notice/fragments/title.xml

@@ -0,0 +1 @@
+<hp:p xmlns:hp="http://www.hancom.co.kr/hwpml/2011/paragraph" id="0" paraPrIDRef="25" styleIDRef="0" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="13"><hp:t>{{title}}</hp:t></hp:run></hp:p>

hwpx_automation/office/authoring/design/profiles/official_notice/profile.json

@@ -0,0 +1,25 @@
+{
+  "schemaVersion": "hwpx.design.profile.v1",
+  "id": "official_notice",
+  "template": "template.hwpx",
+  "fragments": {
+    "title": "fragments/title.xml",
+    "heading": "fragments/heading.xml",
+    "body": "fragments/body.xml",
+    "info_table": "fragments/info_table.xml"
+  },
+  "page": {
+    "width": 59528,
+    "height": 84188,
+    "orientation": "WIDELY",
+    "margins": {
+      "left": 5669,
+      "right": 5669,
+      "top": 4251,
+      "bottom": 2834
+    }
+  },
+  "char_pr_count": 65,
+  "style_coverage_threshold": 0.98,
+  "source_basename": "synthetic_official_notice_source.hwpx"
+}

hwpx_automation/office/authoring/design/profiles/report/fragments/body.xml

@@ -0,0 +1 @@
+<hp:p xmlns:hp="http://www.hancom.co.kr/hwpml/2011/paragraph" id="0" paraPrIDRef="0" styleIDRef="0" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="52"><hp:t>{{body}}</hp:t></hp:run></hp:p>

hwpx_automation/office/authoring/design/profiles/report/fragments/heading.xml

@@ -0,0 +1 @@
+<hp:p xmlns:hp="http://www.hancom.co.kr/hwpml/2011/paragraph" id="0" paraPrIDRef="22" styleIDRef="0" pageBreak="0" columnBreak="0" merged="0"><hp:run charPrIDRef="26"><hp:t>{{heading}}</hp:t></hp:run><hp:run charPrIDRef="26"/></hp:p>