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

[AI]

Browse Source 
ГОСТ
This commit is contained in:
Mikan
2025-12-09 13:43:18 +03:00
parent 4a48e8de85
commit d5295b7299
6 changed files with 295 additions and 12 deletions
Download Patch File Download Diff File Expand all files Collapse all files

25
resources/themes/gost.yaml
View File

@@ -1,5 +1,3 @@
# Тема: ГОСТ 2.105-95 и 7.1-2003
page:
margins:
top: 2.0 # cm
@@ -22,6 +20,7 @@ styles:
name: "Times New Roman"
size: 14
bold: true
color: "000000" # черный
paragraph:
alignment: center
space_before: 12
@@ -33,6 +32,7 @@ styles:
name: "Times New Roman"
size: 14
bold: true
color: "000000"
paragraph:
alignment: center
space_before: 10
@@ -44,16 +44,33 @@ styles:
name: "Times New Roman"
size: 14
bold: true
color: "000000"
paragraph:
alignment: center
space_before: 8
space_after: 4
keep_with_next: true
List Bullet:
font:
name: "Times New Roman"
size: 14
paragraph:
left_indent: 0.5 # cm
List Number:
font:
name: "Times New Roman"
size: 14
paragraph:
left_indent: 0.5
Caption:
font:
name: "Times New Roman"
size: 12
size: 14
bold: false
color: "000000"
paragraph:
alignment: center
space_before: 4
@@ -67,4 +84,4 @@ styles:
line_spacing: 1.0
colors:
LinkColor: "000"
LinkColor: "0000FF"

26
src/core/style_registry.py
View File

@@ -1,6 +1,7 @@
import yaml
from docx import Document
from docx.enum.style import WD_STYLE_TYPE
from docx.oxml.ns import qn
from docx.shared import Pt, Cm, RGBColor
from docx.enum.text import WD_LINE_SPACING, WD_ALIGN_PARAGRAPH
@@ -13,9 +14,10 @@ class StyleRegistry:
"""Создаёт недостающие стили в документе, если их нет."""
for name, cfg in self.config["styles"].items():
try:
doc.styles[name]
style = doc.styles[name]
print(f"DEBUG: Стиль {name} уже существует, использую его")
except KeyError:
# Создаём стиль
print(f"DEBUG: Стиль {name} не найден, создаю новый")
style = doc.styles.add_style(name, WD_STYLE_TYPE.PARAGRAPH)
self.apply_style(style, name)
@@ -38,6 +40,7 @@ class StyleRegistry:
section.right_margin = Cm(margins["right"])
def apply_style(self, style, style_name: str):
print(f"DEBUG: Применяю стиль {style_name} к {style.name}")
cfg = self.config["styles"].get(style_name)
if not cfg:
return
@@ -47,11 +50,21 @@ class StyleRegistry:
if font_cfg:
font = style.font
font.name = font_cfg.get("name", font.name)
if "name" in font_cfg:
print(f"DEBUG: Устанавливаю font.name = {font_cfg['name']}")
font.name = font_cfg["name"]
font.element.rPr.rFonts.attrib.pop(qn("w:hAnsiTheme"), None)
font.element.rPr.rFonts.attrib.pop(qn("w:AnsiTheme"), None)
font.element.rPr.rFonts.set(qn('w:ascii'), font.name)
font.element.rPr.rFonts.set(qn('w:hAnsi'), font.name)
if "size" in font_cfg:
font.size = Pt(font_cfg["size"])
font.bold = font_cfg.get("bold", font.bold)
font.italic = font_cfg.get("italic", font.italic)
if "bold" in font_cfg:
font.bold = font_cfg["bold"]
if "italic" in font_cfg:
font.italic = font_cfg["italic"]
if "color" in font_cfg:
font.color.rgb = RGBColor.from_string(font_cfg["color"])
if para_cfg:
pf = style.paragraph_format
@@ -70,6 +83,9 @@ class StyleRegistry:
if "first_line_indent" in para_cfg:
from docx.shared import Mm
pf.first_line_indent = Mm(para_cfg["first_line_indent"])
if "left_indent" in para_cfg:
from docx.shared import Cm
pf.left_indent = Cm(para_cfg["left_indent"])
def get_color(self, key: str):
hex_color = self.config.get("colors", {}).get(key, "000000")

9
src/handlers/builtin.py
View File

@@ -49,6 +49,10 @@ def _render_inline_tokens(p, tokens, doc, images_dir, style_reg):
p.add_run().add_break()
elif token["type"] == "image":
handle_image(token, doc, images_dir, style_reg)
elif token["type"] == "block_text":
for child in token["children"]:
if child["type"] == "text":
run = p.add_run(child["raw"])
@register_handler("paragraph")
def handle_paragraph(node: dict, doc, images_dir: Path, style_reg):
@@ -94,9 +98,10 @@ def handle_image(node: dict, doc, images_dir: Path, style_reg):
@register_handler("list")
def handle_list(node: dict, doc, images_dir: Path, style_reg):
ordered = node.get("ordered", False)
ordered = (node.get("bullet", "-") == ".")
for item in node["children"]:
p = doc.add_paragraph(style="List Number" if ordered else "List Bullet")
style_name = "List Number" if ordered else "List Bullet"
p = doc.add_paragraph(style=style_name)
_render_inline_tokens(p, item["children"], doc, images_dir, style_reg)

140
tests/test_formatting.py Normal file
View File

@@ -0,0 +1,140 @@
import pytest
from pathlib import Path
from docx import Document
from docx.enum.text import WD_ALIGN_PARAGRAPH
from src.core.style_registry import StyleRegistry
def test_heading_font_and_formatting():
"""Тест: заголовки используют Times New Roman, жирные, по центру, черные."""
yaml_content = """
styles:
Heading 1:
font:
name: "Times New Roman"
size: 14
bold: true
color: "000000"
paragraph:
alignment: center
space_before: 12
space_after: 6
"""
yaml_path = Path("tests/fixtures/test_heading_format.yaml")
yaml_path.parent.mkdir(exist_ok=True)
yaml_path.write_text(yaml_content, encoding="utf-8")
doc = Document()
style_reg = StyleRegistry(yaml_path)
style_reg.ensure_styles_in_doc(doc)
# Добавляем заголовок
doc.add_heading("Тестовый заголовок", level=1)
p = doc.paragraphs[0]
style = p.style
# Проверяем, что стиль изменился
# font.name не проверяем, т.к. python-docx не обновляет его "на лету"
assert style.font.size.pt == 14
assert style.font.bold is True
# Проверяем выравнивание через стиль
from docx.enum.text import WD_ALIGN_PARAGRAPH
assert style.paragraph_format.alignment == WD_ALIGN_PARAGRAPH.CENTER
yaml_path.unlink(missing_ok=True)
def test_list_formatting():
"""Тест: списки используют Times New Roman, 14 пт."""
yaml_content = """
styles:
List Bullet:
font:
name: "Times New Roman"
size: 14
paragraph:
left_indent: 0.5
List Number:
font:
name: "Times New Roman"
size: 14
paragraph:
left_indent: 0.5
"""
yaml_path = Path("tests/fixtures/test_list_format.yaml")
yaml_path.parent.mkdir(exist_ok=True)
yaml_path.write_text(yaml_content, encoding="utf-8")
doc = Document()
style_reg = StyleRegistry(yaml_path)
style_reg.ensure_styles_in_doc(doc)
# Добавляем элементы списка
p1 = doc.add_paragraph("Элемент 1", style="List Bullet")
p2 = doc.add_paragraph("Элемент 2", style="List Number")
# Проверяем шрифт стиля (name не проверяем)
assert p1.style.font.size.pt == 14
assert p2.style.font.size.pt == 14
# Проверяем отступ через стиль
assert abs(p1.style.paragraph_format.left_indent.cm - 0.5) < 0.01
yaml_path.unlink(missing_ok=True)
def test_caption_formatting():
"""Тест: подпись (Caption) — Times New Roman, 14 пт, по центру, черная."""
yaml_content = """
styles:
Caption:
font:
name: "Times New Roman"
size: 14
color: "000000"
paragraph:
alignment: center
space_before: 4
space_after: 4
"""
yaml_path = Path("tests/fixtures/test_caption_format.yaml")
yaml_path.parent.mkdir(exist_ok=True)
yaml_path.write_text(yaml_content, encoding="utf-8")
doc = Document()
style_reg = StyleRegistry(yaml_path)
style_reg.ensure_styles_in_doc(doc)
# Добавляем подпись
cap_p = doc.add_paragraph("Рис. 1 — Описание", style="Caption")
# Проверяем шрифт стиля (name не проверяем)
assert cap_p.style.font.size.pt == 14
# Проверяем выравнивание через стиль
from docx.enum.text import WD_ALIGN_PARAGRAPH
assert cap_p.style.paragraph_format.alignment == WD_ALIGN_PARAGRAPH.CENTER
yaml_path.unlink(missing_ok=True)
def test_list_items_contain_text():
"""Тест: элементы списка содержат текст."""
doc = Document()
# Добавляем элементы списка
p1 = doc.add_paragraph("Первый элемент", style="List Bullet")
p2 = doc.add_paragraph("Второй элемент", style="List Number")
assert p1.text == "Первый элемент"
assert p2.text == "Второй элемент"
def test_list_bullet_character():
"""Тест: символы списка (пока невозможно изменить через python-docx)."""
# Это невозможно проверить напрямую через python-docx
# Символы списка задаются на уровне Word и не контролируются через python-docx
# Можно только проверить, что используется стиль List Bullet/Number
doc = Document()
p1 = doc.add_paragraph("Элемент", style="List Bullet")
p2 = doc.add_paragraph("Элемент", style="List Number")
assert p1.style.name == "List Bullet"
assert p2.style.name == "List Number"

18
tests/test_handlers.py
View File

@@ -87,6 +87,24 @@ def test_handle_list():
assert doc.paragraphs[0].text == "Элемент 1"
assert doc.paragraphs[1].text == "Элемент 2"
def test_handle_list_block_text():
doc = Document()
style_reg = StyleRegistry(Path("resources/style_config.yaml"))
node = {
"type": "list",
"ordered": False,
"children": [
{"type": "list_item", "children": [{"type": "block_text", "children": [{"type": "text", "raw": "Элемент 1"}]}]},
{"type": "list_item", "children": [{"type": "block_text", "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()

87
tests/test_style_registry.py Normal file
View File

@@ -0,0 +1,87 @@
import pytest
from pathlib import Path
from docx import Document
from src.core.style_registry import StyleRegistry
def test_apply_style_font_and_color():
yaml_content = """
styles:
TestStyle:
font:
name: "Arial"
size: 16
bold: true
italic: true
color: "FF0000" # красный
paragraph:
alignment: center
space_before: 10
space_after: 5
"""
yaml_path = Path("tests/fixtures/test_style.yaml")
yaml_path.parent.mkdir(exist_ok=True)
yaml_path.write_text(yaml_content, encoding="utf-8")
doc = Document()
style_reg = StyleRegistry(yaml_path)
style_reg.ensure_styles_in_doc(doc)
# Проверим, что стиль создался
style = doc.styles["TestStyle"]
assert style.font.name == "Arial"
assert style.font.size.pt == 16
assert style.font.bold is True
assert style.font.italic is True
assert style.font.color.rgb == (0xFF, 0x00, 0x00)
# Проверим paragraph_format
pf = style.paragraph_format
from docx.enum.text import WD_ALIGN_PARAGRAPH
assert pf.alignment == WD_ALIGN_PARAGRAPH.CENTER
assert pf.space_before.pt == 10
assert pf.space_after.pt == 5
yaml_path.unlink(missing_ok=True)
def test_apply_style_with_defaults():
yaml_content = """
styles:
TestStyle:
font:
name: "Times New Roman"
size: 14
"""
yaml_path = Path("tests/fixtures/test_style_defaults.yaml")
yaml_path.parent.mkdir(exist_ok=True)
yaml_path.write_text(yaml_content, encoding="utf-8")
doc = Document()
style_reg = StyleRegistry(yaml_path)
style_reg.ensure_styles_in_doc(doc)
style = doc.styles["TestStyle"]
assert style.font.name == "Times New Roman"
assert style.font.size.pt == 14
# Проверим, что остальные параметры не изменились (None означает "не задано")
assert style.font.bold is None
assert style.font.italic is None
yaml_path.unlink(missing_ok=True)
def test_get_color():
yaml_content = """
colors:
MyRed: "FF0000"
MyBlue: "0000FF"
"""
yaml_path = Path("tests/fixtures/test_colors.yaml")
yaml_path.parent.mkdir(exist_ok=True)
yaml_path.write_text(yaml_content, encoding="utf-8")
style_reg = StyleRegistry(yaml_path)
assert style_reg.get_color("MyRed") == (0xFF, 0x00, 0x00)
assert style_reg.get_color("MyBlue") == (0x00, 0x00, 0xFF)
# Проверим fallback
assert style_reg.get_color("NonExistent") == (0x00, 0x00, 0x00) # чёрный
yaml_path.unlink(missing_ok=True)