Superminaren/50ohm
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

refactor: avoid parsing document twice for figure numbers

Browse Source
This commit is contained in:
Tim Grelka
2026-06-24 16:18:39 +02:00
parent c15b3f709b
commit f6ceb4cf49
32 changed files with 263 additions and 349 deletions
Download Patch File Download Diff File Expand all files Collapse all files
renderer/document.py
+6 -3
View File
@@ -2,6 +2,9 @@ import mistletoe
class Document(mistletoe.Document):
# margin_count = 0
# margin_anchor_count = 0
references = {}
def __init__(self, *args, **kwargs):
# references must exist before super().__init__ tokenizes the lines,
# since ReferencedToken reads it from _root_node during tokenization.
self.references: dict[str, str] = {}
super().__init__(*args, **kwargs)
renderer/fifty_ohm_html_renderer.py
+37 -81
View File
@@ -8,16 +8,15 @@ from .dash import Dash
from .document import Document
from .formula import Formula
from .halfwidth_spaces import HalfwidthSpaces
from .image import Image
from .include import Include
from .index import Index
from .morse import Morse
from .nonbreaking_spaces import NonbreakingSpaces, NonbreakingSpacesDots
from .photo import Photo
from .picture import Picture
from .qso import Qso
from .question import Question
from .quote import Quote
from .references import References
from .reference import Reference
from .table import Table, TableBody, TableCell, TableHeader, TableRow
from .tag import Tag
from .underline import Underline
@@ -30,14 +29,6 @@ class FiftyOhmHtmlRenderer(HtmlRenderer):
margin_anchor_id = 0
margin_id = 0
edition: str = None
chapter: str = None
section: str = None
section_url: str = None
# TODO: Add type
references = {}
def __init__(
self,
*extras,
@@ -62,10 +53,9 @@ class FiftyOhmHtmlRenderer(HtmlRenderer):
HalfwidthSpaces,
NonbreakingSpaces,
NonbreakingSpacesDots,
References,
Reference,
Question,
Picture,
Photo,
Image,
Table,
TableHeader,
TableRow,
@@ -90,23 +80,21 @@ class FiftyOhmHtmlRenderer(HtmlRenderer):
self.section = section
# Set section URL if provided, otherwise use default
if section_url is not None:
self.section_url = section_url
self.section_url = section_url if section_url is not None else "section.html"
# Figure map: stores ref -> hierarchical number mapping
self.figure_map = {}
# Single unified counter for all figure types (pictures, photos, tables)
self.figure_counter = 0
self.references: dict[str, str] = {}
# Keep track of index ids already emitted in this document.
self.index_anchor_ids = set()
def get_reference_label(self, marker: str):
def _format_reference_label(self, marker: str):
if marker in self.references:
return f"{self.edition}-{self.chapter}.{self.section}.{self.references[marker]}", True
label = self.references[marker]
if self.edition and self.chapter and self.section:
label = f"{self.edition}-{self.chapter}.{self.section}.{label}"
return label
else:
return marker, False
return None
def render_dash(self, token):
return " – "
@@ -222,47 +210,11 @@ class FiftyOhmHtmlRenderer(HtmlRenderer):
lookup = {"": "", " ": " "}
return f"{lookup[token.first]}{token.second}{lookup[token.third]}"
def collect_figures(self, document):
"""First pass: collect all figures and assign hierarchical numbers."""
self.figure_counter = 0
self.figure_map = {}
self._collect_figures_recursive(document)
def _collect_figures_recursive(self, token):
"""Recursively traverse the document tree to collect figures."""
if hasattr(token, "children") and token.children is not None:
for child in token.children:
# Check if this is a Picture, Photo, or Table token
if isinstance(child, Picture | Photo | Table):
# Get the reference ID (ref for Picture/Photo, name for Table)
ref_id = getattr(child, "ref", None) or getattr(child, "name", None)
if ref_id: # Only number figures with a reference ID
self.figure_counter += 1
hierarchical_num = self._format_figure_number(self.figure_counter)
self.figure_map[ref_id] = hierarchical_num
child.number = hierarchical_num
else:
child.number = ""
# Recurse into children
self._collect_figures_recursive(child)
def _format_figure_number(self, counter):
"""Format the hierarchical figure number based on context."""
if self.edition and self.chapter and self.section:
return f"{self.edition}-{self.chapter}.{self.section}.{counter}"
else:
# Fallback for tests or when context is not provided
return str(counter)
def render_references(self, token):
def render_reference(self, token: Reference):
# Look up the figure number from the map
figure_num = self.figure_map.get(token.first, "?")
label = self._format_reference_label(token.marker) or "?"
return (
f'<a href="{self.section_url}#ref_{token.first}"'
f" onclick=\"highlightRef('{token.first}');\">{figure_num}</a>"
f'<a href="{self.section_url}#ref_{token.marker}" onclick="highlightRef(\'{token.marker}\');">{label}</a>'
)
def render_question(self, token):
@@ -282,31 +234,34 @@ class FiftyOhmHtmlRenderer(HtmlRenderer):
return (
f'<figure class="picture" id="ref_{marker}" name="{marker}">\n'
f' <img src="pictures/{id}.svg" alt="{alt_text}">\n'
f" <figcaption>Abbildung {self.get_reference_label(marker)}: {text}</figcaption>\n"
f" <figcaption>Abbildung {label}: {text}</figcaption>\n"
"</figure>\n"
)
def render_picture(self, token):
alt_text = ""
if self.picture_handler is not None:
alt_text = self.picture_handler(token.id)
return self.render_picture_helper(token.id, token.marker, token.text, token.label, alt_text)
@staticmethod
def render_photo_helper(id, ref, text, number, alt_text):
def render_photo_helper(id, ref, text, label, alt_text):
return f"""
<figure class="photo" id="ref_{ref}" name="{ref}">
<img src="photos/{id}.png" alt="{alt_text}">
<figcaption>Abbildung {number}: {text}</figcaption>
<figcaption>Abbildung {label}: {text}</figcaption>
</figure>
"""
def render_photo(self, token):
alt_text = ""
if self.photo_handler is not None:
alt_text = self.photo_handler(token.id) or alt_text
return self.render_photo_helper(token.id, token.ref, token.text, token.number, alt_text)
def render_image(self, token: Image):
label = self._format_reference_label(token.marker) or "?"
# Only supported kinds are photo and picture
if token.kind == "photo":
alt_text = ""
if self.photo_handler is not None:
alt_text = self.photo_handler(token.id) or alt_text
return self.render_photo_helper(token.id, token.marker, token.text, label, alt_text)
if token.kind == "picture":
alt_text = ""
if self.picture_handler is not None:
alt_text = self.picture_handler(token.id)
return self.render_picture_helper(token.id, token.marker, token.text, label, alt_text)
return ""
def render_table(self, token: Table):
# Add id and name attributes if table has a name
@@ -317,10 +272,11 @@ class FiftyOhmHtmlRenderer(HtmlRenderer):
table = f'<table class="table table-hover"{table_attrs}>\n{self.render_inner(token)}'
if token.caption != "":
# Include hierarchical number in caption if available
# Include hierarchical number in caption if the table is named.
caption_text = token.caption
if hasattr(token, "number") and token.number:
caption_text = f"Tabelle {token.number}: {token.caption}"
label = self._format_reference_label(token.name)
if label:
caption_text = f"Tabelle {label}: {token.caption}"
table += f"<caption>{caption_text}</caption>\n"
table += "</table>"
renderer/fifty_ohm_latex_renderer.py
+8 -9
View File
@@ -3,10 +3,9 @@ from mistletoe.latex_renderer import LaTeXRenderer
from .comment import BlockComment
from .dash import Dash
from .halfwidth_spaces import HalfwidthSpaces
from .image import Image
from .index import Index
from .nonbreaking_spaces import NonbreakingSpaces, NonbreakingSpacesDots
from .photo import Photo
from .picture import Picture
from .question import Question
from .quote import Quote
from .table import Table, TableBody, TableCell, TableHeader, TableRow
@@ -26,8 +25,7 @@ class FiftyOhmLaTeXRenderer(LaTeXRenderer):
NonbreakingSpaces,
NonbreakingSpacesDots,
Question,
Picture,
Photo,
Image,
Table,
TableBody,
TableRow,
@@ -115,17 +113,18 @@ class FiftyOhmLaTeXRenderer(LaTeXRenderer):
\captionof{{figure}}{{{text}}}
\label{{{ref}}}"""
def render_picture(self, token):
return self.render_picture_helper(token.id, token.ref, token.text, token.number)
@staticmethod
def render_photo_helper(id, ref, text, number):
return rf"""\includegraphics[width=1.0\linewidth]{{photo/{id}}}
\captionof{{figure}}{{{text}}}
\label{{{ref}}}"""
def render_photo(self, token):
return self.render_photo_helper(token.id, token.ref, token.text, token.number)
def render_image(self, token: Image):
if token.kind == "photo":
return self.render_photo_helper(token.id, token.marker, token.text, token.label)
if token.kind == "picture":
return self.render_picture_helper(token.id, token.marker, token.text, token.label)
return ""
def render_table(self, token):
table = "\\begin{DARCtabular}" + f"{self.render_inner(token)}" + "\\end{DARCtabular}"
renderer/image.py
+40
View File
@@ -0,0 +1,40 @@
import re
from mistletoe.block_tokenizer import FileWrapper
from renderer.referenced_token import ReferencedToken
class Image(ReferencedToken):
"""
Block token for pictures and photos.
Examples:
[picture:727:amplitude_periode_halbwellen:Positive und negative Halbwellen einer Sinusschwingung]
[photo:123:n_wasserfalldiagramm_trx_collage:Ansichten verschiedener Transceiver]
"""
pattern = re.compile(r"^\[(?P<kind>picture|photo):(?P<id>\d+):(?P<marker>[^:\]]+):(?P<text>[^:\]]+)\]$")
@classmethod
def start(cls, line: str):
return cls.pattern.match(line.strip()) is not None
@classmethod
def check_interrupts_paragraph(cls, lines: FileWrapper):
line = lines.peek()
if line is not None:
return cls.start(line)
else:
return False
@classmethod
def read(cls, lines: FileWrapper):
match = cls.pattern.match(next(lines).strip())
return match.group("kind"), match.group("id"), match.group("marker"), match.group("text")
def __init__(self, match):
kind, id, marker, text = match
super().__init__(marker)
self.kind = kind
self.id = id
self.text = text
renderer/photo.py
-19
View File
@@ -1,19 +0,0 @@
import re
from mistletoe.span_token import SpanToken
class Photo(SpanToken):
"""
Span token for photos.
Example: [photo:123:n_wasserfalldiagramm_trx_collage:Ansichten verschiedener Transceiver]
"""
parse_inner = False
pattern = re.compile(r"\[photo:(\d+):([^:\]]+):([^:\]]+)\]")
def __init__(self, match_object):
self.id = match_object.group(1)
self.ref = match_object.group(2)
self.text = match_object.group(3)
self.number = "TODO"
renderer/picture.py
-33
View File
@@ -1,33 +0,0 @@
import re
from mistletoe.block_tokenizer import FileWrapper
from renderer.referenced_token import ReferencedToken
class Picture(ReferencedToken):
"""
Span token for pictures.
Example: [picture:727:amplitude_periode_halbwellen:Positive und negative Halbwellen einer Sinusschwingung]
"""
pattern = re.compile(r"^\[picture:(?P<id>\d+):(?P<marker>[^:\]]+):(?P<text>[^:\]]+)\]$")
@classmethod
def start(cls, line: str):
return cls.pattern.match(line.strip()) is not None
@classmethod
def check_interrupts_paragraph(cls, lines: FileWrapper):
return cls.pattern.match(lines.peek())
@classmethod
def read(cls, lines: FileWrapper):
match = cls.pattern.match(next(lines))
return match.group("id"), match.group("marker"), match.group("text")
def __init__(self, match):
id, marker, text = match
super().__init__(marker)
self.id = id
self.text = text
renderer/references.py → renderer/reference.py
+2 -2
View File
@@ -3,7 +3,7 @@ import re
from mistletoe.span_token import SpanToken
class References(SpanToken):
class Reference(SpanToken):
"""
Resolve references for html
Identifies "[ref:$]"
@@ -12,4 +12,4 @@ class References(SpanToken):
pattern = re.compile(r"\[ref\:(\S+)\]")
def __init__(self, match_object):
self.first = match_object.group(1)
self.marker: str = match_object.group(1)
renderer/referenced_token.py
+11 -10
View File
@@ -3,15 +3,16 @@ from mistletoe.block_token import BlockToken
class ReferencedToken(BlockToken):
line_number: int = -1
marker: str = None
label: str = None
def __init__(self, marker: str):
self.marker = marker
if marker not in mistletoe.token._root_node.references:
self.label = len(mistletoe.token._root_node.references) + 1
mistletoe.token._root_node.references[marker] = self.label
if marker != "":
self.marker = marker
references = mistletoe.token._root_node.references
if marker not in references:
label = str(len(references) + 1)
references[marker] = label
else:
label = references[marker]
self.label = label
else:
self.label = mistletoe._root_node.references[marker]
self.marker = None
self.label = None
renderer/table.py
+8 -2
View File
@@ -4,6 +4,8 @@ from enum import Enum
from mistletoe import span_token
from mistletoe.block_token import BlockToken
from renderer.referenced_token import ReferencedToken
class CellAlignment(Enum):
LEFT = "l"
@@ -76,7 +78,7 @@ class TableBody(BlockToken):
self.children, self.header = rows, header
class Table(BlockToken):
class Table(ReferencedToken):
name: str
caption: str
header: list[TableHeader]
@@ -115,4 +117,8 @@ class Table(BlockToken):
return children, name, caption
def __init__(self, match):
self.children, self.name, self.caption = match
children, name, caption = match
super().__init__(name)
self.children = children
self.name = name
self.caption = caption