mikan/md-to-docx
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

table + list

Browse Source
This commit is contained in:
Mikan
2025-12-08 17:09:16 +03:00
parent 848eb462b5
commit 1dbd05e57e
2 changed files with 129 additions and 38 deletions
Download Patch File Download Diff File Expand all files Collapse all files

56
src/handlers/builtin.py
View File

@@ -23,17 +23,26 @@ def _render_inline_tokens(p, tokens, doc, images_dir, style_reg):
elif token["type"] == "strong":
run = p.add_run()
run.bold = True
run.text = token["children"][0]["raw"] if token["children"] else ""
# Обрабатываем children внутри этого же run
for child in token["children"]:
if child["type"] == "text":
run.text += child["raw"]
elif token["type"] == "emphasis":
run = p.add_run()
run.italic = True
run.text = token["children"][0]["raw"] if token["children"] else ""
for child in token["children"]:
if child["type"] == "text":
run.text += child["raw"]
elif token["type"] == "link":
run = p.add_run()
run.text = token["children"][0]["raw"] if token["children"] else ""
run.font.color.rgb = style_reg.get_color("LinkColor")
elif token["type"] == "softbreak" or token["type"] == "linebreak":
p.add_run().add_break()
elif token["type"] == "html":
if "<br" in token["raw"]:
p.add_run().add_break()
elif token["type"] == "image":
# Если image внутри параграфа — вызываем обработчик
handle_image(token, doc, images_dir, style_reg)
@register_handler("paragraph")
@@ -48,37 +57,58 @@ def handle_image(node: dict, doc, images_dir: Path, style_reg):
alt = node.get("alt", "")
img_path = images_dir / src
if img_path.exists():
logger.debug(f"Обработка изображения: {img_path}, exists: {img_path.exists()}")
p = doc.add_paragraph()
run = p.add_run()
logger.debug(f"Добавлен параграф: {len(doc.paragraphs)} до add_picture")
try:
run.add_picture(str(img_path), width=Inches(5))
logger.debug("add_picture успешно выполнен")
p.alignment = WD_ALIGN_PARAGRAPH.CENTER
except Exception as e:
logger.error(f"Ошибка при добавлении изображения {img_path}: {e}")
doc.add_paragraph(f"[Изображение не загружено: {src}]")
return
logger.debug(f"После add_picture: {len(doc.paragraphs)} параграфов")
if alt:
try:
cap_p = doc.add_paragraph(alt, style="Caption")
logger.debug(f"Добавлена подпись: {alt}, всего параграфов: {len(doc.paragraphs)}")
except KeyError:
cap_p = doc.add_paragraph(alt)
style_reg.apply_style(cap_p.style, "Caption")
logger.debug(f"Добавлена подпись без стиля: {alt}")
else:
doc.add_paragraph(f"[Изображение не найдено: {src}]")
logger.error(f"[Изображение не найдено: {src}]")
@register_handler("list")
def handle_list(node: dict, doc, images_dir: Path, style_reg):
ordered = node.get("ordered", False)
for item in node["children"]:
p = doc.add_paragraph(style="List Number" if ordered else "List Bullet")
for child in item["children"]:
if child["type"] == "text":
p.add_run(child["raw"])
_render_inline_tokens(p, item["children"], doc, images_dir, style_reg)
@register_handler("table")
def handle_table(node: dict, doc, images_dir: Path, style_reg):
rows_data = node["children"] # [header_row, row1, row2...]
if not rows_data:
return
header_row = rows_data[0]["children"]
body_rows = rows_data[1:]
num_cols = len(header_row)
table = doc.add_table(rows=1, cols=num_cols)
table.style = "Table Grid" # можно настроить в YAML
# Заполняем заголовок
hdr_cells = table.rows[0].cells
for i, cell_data in enumerate(header_row):
if i < num_cols:
cell = hdr_cells[i]
_render_inline_tokens(cell.paragraphs[0], cell_data["children"], doc, images_dir, style_reg)
# Заполняем тело
for row_data in body_rows:
cells = table.add_row().cells
for i, cell_data in enumerate(row_data["children"]):
if i < num_cols:
cell = cells[i]
_render_inline_tokens(cell.paragraphs[0], cell_data["children"], doc, images_dir, style_reg)

111
tests/test_handlers.py
View File

@@ -1,7 +1,7 @@
import pytest
from pathlib import Path
from docx import Document
from src.handlers.builtin import handle_heading, handle_paragraph, handle_image, handle_list
from src.handlers.builtin import handle_heading, handle_paragraph, handle_image, handle_list, handle_table
from src.core.style_registry import StyleRegistry
# Тесты для обработчиков
@@ -40,29 +40,6 @@ def test_handle_image():
assert len(doc.paragraphs) == 2
assert doc.paragraphs[1].text == "Логотип" # подпись
def test_debug_handle_image():
from PIL import Image
doc = Document()
style_reg = StyleRegistry(Path("resources/style_config.yaml"))
style_reg.ensure_styles_in_doc(doc)
# Проверим, есть ли стиль "Caption" в документе
print("Доступные стили:", [s.name for s in doc.styles if s.type.name == 'PARAGRAPH'])
assert "Caption" in [s.name for s in doc.styles if s.type.name == 'PARAGRAPH']
node = {"type": "image", "src": "logo.png", "alt": "Логотип"}
img_path = Path("tests/fixtures/images/logo.png")
img_path.parent.mkdir(exist_ok=True)
img = Image.new("RGB", (100, 100), color="red")
img.save(img_path)
handle_image(node, doc, Path("tests/fixtures"), style_reg)
print("Количество параграфов:", len(doc.paragraphs))
for i, p in enumerate(doc.paragraphs):
print(f"Параграф {i}: '{p.text}', стиль: {p.style.name if p.style else 'None'}")
def test_handle_paragraph_with_inline():
doc = Document()
style_reg = StyleRegistry(Path("resources/style_config.yaml"))
@@ -87,4 +64,88 @@ def test_handle_paragraph_with_inline():
assert runs[2].text == " и "
assert runs[3].text == "курсив"
assert runs[3].italic is True
assert runs[4].text == "."
assert runs[4].text == "."
def test_handle_list():
doc = Document()
style_reg = StyleRegistry(Path("resources/style_config.yaml"))
node = {
"type": "list",
"ordered": False,
"children": [
{"type": "list_item", "children": [{"type": "text", "raw": "Элемент 1"}]},
{"type": "list_item", "children": [{"type": "text", "raw": "Элемент 2"}]},
]
}
handle_list(node, doc, Path("."), style_reg)
assert len(doc.paragraphs) == 2
assert doc.paragraphs[0].style.name == "List Bullet"
assert doc.paragraphs[0].text == "Элемент 1"
assert doc.paragraphs[1].text == "Элемент 2"
def test_handle_table():
doc = Document()
style_reg = StyleRegistry(Path("resources/style_config.yaml"))
node = {
"type": "table",
"children": [
{ # header
"type": "table_row",
"children": [
{"type": "table_cell", "children": [{"type": "text", "raw": "Заголовок 1"}]},
{"type": "table_cell", "children": [{"type": "text", "raw": "Заголовок 2"}]},
]
},
{ # row 1
"type": "table_row",
"children": [
{"type": "table_cell", "children": [{"type": "text", "raw": "Ячейка 1"}]},
{"type": "table_cell", "children": [{"type": "text", "raw": "Ячейка 2"}]},
]
}
]
}
handle_table(node, doc, Path("."), style_reg)
assert len(doc.tables) == 1
table = doc.tables[0]
assert len(table.rows) == 2
assert table.cell(0, 0).text == "Заголовок 1"
assert table.cell(1, 1).text == "Ячейка 2"
def test_handle_table_with_br():
doc = Document()
style_reg = StyleRegistry(Path("resources/style_config.yaml"))
node = {
"type": "table",
"children": [
{ # header
"type": "table_row",
"children": [
{"type": "table_cell", "children": [{"type": "text", "raw": "Заголовок 1"}]},
{"type": "table_cell", "children": [{"type": "text", "raw": "Заголовок 2"}]},
]
},
{ # row 1
"type": "table_row",
"children": [
{"type": "table_cell", "children": [
{"type": "text", "raw": "Первая строка"},
{"type": "softbreak"}, # или linebreak/html
{"type": "text", "raw": "Вторая строка"}
]},
{"type": "table_cell", "children": [{"type": "text", "raw": "Ячейка 2"}]},
]
}
]
}
handle_table(node, doc, Path("."), style_reg)
assert len(doc.tables) == 1
table = doc.tables[0]
cell_text = table.cell(1, 0).text # "Первая строка\nВторая строка"
assert "Первая строка" in cell_text
assert "Вторая строка" in cell_text