Regex corrections + fix for merged rows (2afc6b8a) · Commits · Centre for Testing and Interoperability / Markdown specifications development / Specification tools

grid_table_filter.rb

+426 −414

Original line number	Diff line number	Diff line
		@@ -25,9 +25,10 @@ module Banzai
		# Anything, starting with \| blocks which are ignored by this filter
		# +---+---+---+---+

		^\s\+-.\+\s$ # First separator line
		(?:.\n)? # Any number of rows (non-greedy)
		\s\+-.\+\s$ # Last separator line
		^\s*\+(-+\+)+$\n # First separator line
		(?:^\s[\|+][^\n]$\n)*
		^\s*\+(-+\+)+$ # Last separator line

		)
		}mx

		@@ -63,7 +64,7 @@ module Banzai
		raise "Invalid table formatting"
		end
		end
		end
		end # end of class Cell

		class Row
		attr_accessor :cells
		@@ -79,7 +80,7 @@ module Banzai
		def []=(index, value)
		@cells[index] = value
		end
		end
		end # end of class Row

		class RowTracker
		attr_accessor :row_tracker
		@@ -95,13 +96,17 @@ module Banzai
		def []=(index, value)
		@row_tracker[index] = value
		end

		def maxValue
		@row_tracker.max
		end
		end # end of class RowTracker

		# Add these regex constants at the top of the file, after the require statement
		GRID_TABLE_SEPARATOR = /\s\+([-:=]+\+)+\s$/
		GRID_TABLE_HEADER_SEPARATOR = /.\+([=:]+\+)+.$/
		GRID_TABLE_BODY_SEPARATOR = /.\+([:-]+\+)+.$/
		GRID_TABLE_BODY_SEPARATOR_LINE = /[-:]+$/
		GRID_TABLE_SEPARATOR = /^\s\+([-:=]+\+)+\s$/
		GRID_TABLE_HEADER_SEPARATOR = /^\s\+([=:]+\+)+\s$/
		GRID_TABLE_BODY_SEPARATOR = /[^\n]\+([:-]+\+)+[^\n]$/
		GRID_TABLE_BODY_SEPARATOR_LINE = /^[-:]+$/

		def parse_pandoc_table_with_spans(pandoc_table)
		# Split the input into lines
		@@ -181,6 +186,7 @@ module Banzai
		row[column_index]
		end

		# Retrieve separator indices
		separator_indices = lines.each_index.select { \|i\| is_separator(lines[i]) }

		raise "No valid separators found in the provided Pandoc table." if separator_indices.empty?
		@@ -189,6 +195,7 @@ module Banzai
		delimiter_positions = []
		number_of_columns = separator_indices.map { \|i\| lines[i].count("+") - 1 }.max

		# Determine delimiter positions
		separator_index_max_columns = separator_indices.find { \|i\| lines[i].count("+") - 1 == number_of_columns }
		number_of_columns.times do \|j\|
		start_pos = j.zero? ? 0 : delimiter_positions[j - 1]
		@@ -203,6 +210,7 @@ module Banzai
		header_rows = []
		header_separator_index = nil

		# Determine header delimiter positions
		separator_indices.each do \|index\|
		if GRID_TABLE_HEADER_SEPARATOR.match?(lines[index])
		has_header = true
		@@ -226,18 +234,19 @@ module Banzai
		end
		end

		# Process table body
		# Process table body (including rows belonging to header as they are processed in the same way)
		data_rows = []
		(separator_indices.length - 1).times do \|row\|
		rows = []
		rows_tracker = nil
		in_data_row = false
		start, end_idx = separator_indices[row], separator_indices[row + 1]
		start, end_idx = separator_indices[row], separator_indices[row + 1] # Lines between separators including separator line start as it gives information about the number of columns of the row
		row_lines = lines[start...end_idx]

		next if row_lines.empty?

		row_lines.each do \|line\|
		# First line (normally a separator) of each block
		if is_separator(line) && !in_data_row
		in_data_row = true
		parts = line.strip.delete_prefix("+").split("+")
		@@ -258,9 +267,10 @@ module Banzai
		end
		i += 1
		end

		# Lines in a block
		elsif in_data_row
		if GRID_TABLE_BODY_SEPARATOR.match?(line)
		# Regular data row or partial separator
		if GRID_TABLE_BODY_SEPARATOR.match?(line) # Partial separator
		cells_content = line.strip.delete_prefix("\|").delete_prefix("+")
		.delete_suffix("\|").delete_suffix("+").split(/[\\|\+]/)

		@@ -283,20 +293,21 @@ module Banzai

		if cells_content.length <= number_of_columns
		column_index = 0
		maxRowTracker = rows_tracker.maxValue
		cells_content.each_with_index do \|content, i\|
		if GRID_TABLE_BODY_SEPARATOR_LINE.match?(content)
		rows_tracker[column_index] += 1
		if GRID_TABLE_BODY_SEPARATOR_LINE.match?(content) # Separator - split row
		rows_tracker[column_index] = maxRowTracker + 1
		rows[rows_tracker[column_index]][column_index].list_flag = false

		column_forward = 0
		(column_index...delimiter_positions.length).each do \|del_index\|
		if rows[rows_tracker[column_index]][column_index].position >= delimiter_positions[del_index]
		column_forward += 1
		rows_tracker[column_index + column_forward - 1] += 1 if column_forward > 1
		#rows_tracker[column_index + column_forward - 1] += 1 if column_forward > 1
		end
		end
		column_index += column_forward
		else
		else # Regular cell in Partial separator line
		rows[rows_tracker[column_index]][column_index] =
		handling_content(rows[rows_tracker[column_index]][column_index], content)
		rows[rows_tracker[column_index]][column_index].rowspan += 1
		@@ -317,8 +328,8 @@ module Banzai
		else
		raise "More cells than columns found"
		end
		else
		cells_content = line.strip.delete_prefix("\|").split(/\s\\|\s/)
		else # Data row
		cells_content = line.strip.delete_prefix("\|").delete_suffix("\|").split(/\\|/)
		column_index = 0

		if cells_content.length < number_of_columns
		@@ -419,7 +430,7 @@ module Banzai
		end

		[header_rows, data_rows]
		end
		end # end of parse_pandoc_table_with_spans

		def generate_html_table_with_spans(pandoc_table)
		begin
		@@ -427,7 +438,7 @@ module Banzai
		rescue StandardError => e
		logger = Logger.new(STDOUT)
		logger.error("Grid table could not be generated: #{e.message}")
		return "HTML TABLE COULD NOT BE GENERATED FROM MARKDOWN GRID TABLE. CHECK LOGS"
		return "HTML TABLE COULD NOT BE GENERATED FROM MARKDOWN GRID TABLE. CHECK LOGS FILE"
		else
		html = "<table>\n"
		has_header = false
		@@ -497,15 +508,16 @@ module Banzai
		html += "</table>"
		html
		end
		end
		end # end of def generate_html_table_with_spans

		def call
		return @text if MarkdownFilter.glfm_markdown?(context)
		return @text unless MarkdownFilter.glfm_markdown?(context)

		regex = Gitlab::UntrustedRegexp.new(MARKDOWN_GRID_TABLE_BLOCK_REGEX, multiline: true)
		regex = MARKDOWN_GRID_TABLE_BLOCK_REGEX
		return @text unless regex.match?(@text)

		regex.replace_gsub(@text) do \|match\|
		@text.gsub(regex) do
		match = Regexp.last_match
		# Extract the grid table content from the match
		grid_table = match[:code]
		if grid_table
		@@ -516,7 +528,7 @@ module Banzai
		match.to_s
		end
		end
		end
		end # end of def call
		end # end of class GridTableFilter
		end # end of module Filter
		end # end of module Banzai