From 4299b86171693b1493f3f6fd3e006b7db731a1e1 Mon Sep 17 00:00:00 2001
From: ankraft <an.kraft@gmail.com>
Date: Wed, 23 Oct 2024 22:53:21 +0200
Subject: [PATCH] Added table separator correction.
the length of the separators is used in pandoc to calculate column widths. this is not what we need, and could be wrong in some cases.
---
generateChangemarks/pandocFilter.py | 39 ++++++++++++++++++++++++++++-
1 file changed, 38 insertions(+), 1 deletion(-)
diff --git a/generateChangemarks/pandocFilter.py b/generateChangemarks/pandocFilter.py
index 23e6a25..2550dc3 100644
--- a/generateChangemarks/pandocFilter.py
+++ b/generateChangemarks/pandocFilter.py
@@ -8,7 +8,7 @@
#
import argparse, os, re, sys
-from rich import print
+from rich import print, inspect
from rich.progress import Progress, TextColumn, TimeElapsedColumn
def readMDFile(progress:Progress, document:str) -> list[str]:
@@ -39,6 +39,16 @@ def writeMDFile(progress:Progress, mdLines:list[str], document:str, outDirectory
progress.stop_task(_taskID)
+_inCodeBlock = False
+def checkInCodeBlock(line:str) -> bool:
+ global _inCodeBlock
+
+ # Check if we are in a code block
+ if line.strip().startswith('```'):
+ _inCodeBlock = not _inCodeBlock
+ return _inCodeBlock
+
+
def correctTOC(progress:Progress, mdLines:list[str], tocSection:str = 'Contents') -> list[str]:
""" Correct the TOC to be compatible with pandoc.
"""
@@ -176,6 +186,31 @@ def replaceLineBreaks(progress: Progress, mdLines: list[str]) -> list[str]:
return _lines
+def correctTableSeparators(progress: Progress, mdLines: list[str]) -> list[str]:
+ """ Replace all table header separators with very simple ones. This is
+ needed because pandoc calculates the column width based on the separators.
+ And that is not always wanted
+ """
+ _taskID = progress.add_task('[blue]Correcting table separators', total=0)
+ tableSeparatorRegex = re.compile('^(\s*)(\|[-\|]*\|)\s*$')
+
+ _lines: list[str] = []
+ for line in mdLines:
+ if checkInCodeBlock(line):
+ _lines.append(line)
+ continue
+
+ match = re.match(tableSeparatorRegex, line)
+ if match:
+ # Replace the separators with a simple one
+ _lines.append(f'{match.group(1)}{re.sub("-+", "-", match.group(2))}\n')
+ else:
+ _lines.append(line)
+
+ progress.stop_task(_taskID)
+ return _lines
+
+
def process(document:str, outDirectory:str) -> None:
with Progress(TextColumn('{task.description}'), TimeElapsedColumn()) as progress:
mdLines = readMDFile(progress, document)
@@ -184,8 +219,10 @@ def process(document:str, outDirectory:str) -> None:
mdLines = replaceFigureCaptions(progress, mdLines)
mdLines = replaceFiguresPathSvgToPng(progress, mdLines)
mdLines = replaceLineBreaks(progress, mdLines)
+ mdLines = correctTableSeparators(progress, mdLines)
writeMDFile(progress, mdLines, document, outDirectory)
+
def main(args=None):
# Parse command line arguments
parser = argparse.ArgumentParser(formatter_class=argparse.ArgumentDefaultsHelpFormatter)
--
GitLab