fix: wrong annex links (d7dcd70c) · Commits · CIM - Context Information Management / NGSI-LD API

md_to_docx_converter/md_to_html_2.lua

+35 −8

Original line number	Diff line number	Diff line
		@@ -51,6 +51,10 @@ end
		function AnnexLink(text, number)
		text = text:gsub("%-", "‑") --always use non-breaking hyphens for links
		first_character = number:sub(1, 1)
		-- if first character is a letter and it is lowercase, make it uppercase and replace the number
		if first_character:match("%a") and first_character == first_character:lower() then
		number = first_character:upper() .. number:sub(2)
		end
		return pandoc.Link(text, "annex-" .. first_character:lower() .. ".html#" .. number)
		end

		@@ -58,19 +62,19 @@ end
		function Substitute(el, word)
		local newContent = pandoc.List({})
		local pattern = "(%w%.?%d%.?%d%.?%d%.?%d%.?%d%.?%d+%-?%d*)"
		-- if word == "annex" then
		-- pattern = "(%u)"
		-- end
		local annexPattern = "(%a%.?%d%.?%d%.?%d%.?%d%.?%d%.?%d)"

		local i = 1
		while el.content[i] do
		local elem = el.content[i]
		-- logfile:write("DEBUG: Paragrafo con contenuto -> " .. pandoc.utils.stringify(elem) .. "\n")
		-- logfile:flush()
		if elem.t == "Str" and elem.text:lower():find(word) and not elem.text:find("+++") then --check the next strings to see if we need to link it
		local wordStartIndex = elem.text:lower():find(word)
		local precedentChar = elem.text:sub(wordStartIndex - 1, wordStartIndex - 1)
		local startIndex, endIndex, number = elem.text:gsub("‑", "-"):find(pattern) --check if number is in the same Str elem
		local textWithoutWord = elem.text:sub(wordStartIndex + #word)
		local startIndex, endIndex, number = textWithoutWord:gsub("‑", "-"):find(pattern) --check if number is in the same Str elem
		if not number then
		startIndex, endIndex, number = textWithoutWord:gsub("‑", "-"):find(annexPattern) --check if number is in the same Str elem
		end
		if not precedentChar:match("~") then
		if number then --create link
		first_character = number:sub(1, 1)
		@@ -84,6 +88,9 @@ function Substitute(el, word)
		local succ = el.content[i + 2] --next string if it exists should be number
		if succ and succ.t == "Str" then
		number = succ.text:gsub("‑", "-"):match(pattern)
		if not number then
		number = succ.text:gsub("‑", "-"):match(annexPattern)
		end
		if succ.t == "Str" and number then --we continue searching
		local succ_succ = el.content
		[i + 4] -- next next string if it exists and is of could link to other documents
		@@ -100,13 +107,23 @@ function Substitute(el, word)
		elseif word == "table" then
		newContent:insert(TableLink(elem.text .. " " .. number, number))
		elseif word == "annex" then
		local lastNumberChar = number:sub(-1)
		if lastNumberChar == "." or lastNumberChar == "," then
		number = number:sub(1, -2) --remove the last char if it is a dot or comma
		end
		newContent:insert(AnnexLink(elem.text .. " " .. number, number))
		if lastNumberChar == "." or lastNumberChar == "," then
		newContent:insert(pandoc.Str(lastNumberChar)) --reinsert the dot or comma if it was there
		end
		else
		if debug then print("Unkown behavior for " .. word .. " or link does not exists") end
		end
		if debug then print(succ.text, number) end
		local text, substitutions = succ.text:gsub("‑", "-") --if we substituted we need to account for this, because nbh is not acii anf thus is more then one byte
		local startIndex, endIndex = text:find(pattern)
		if not startIndex then
		startIndex, endIndex = text:find(annexPattern)
		end
		if endIndex + substitutions * 2 < #succ.text then -- if something remains (like a comma or a bracket), append it too
		newContent:insert(pandoc.Str(succ.text:sub(endIndex + substitutions * 2 + 1, -1)))
		end
		@@ -132,7 +149,8 @@ end

		function MultipleClauses(el)
		local newContent = pandoc.List({})
		local pattern = "(%w%.?%d%.?%d%.?%d%.?%d%.?%d%.?%d+%-?%d*)"
		local pattern = "(%w%.?%d%.?%d%.?%d%.?%d%.?%d%.?%d+)"
		local annexPattern = "(%a%.?%d%.?%d%.?%d%.?%d%.?%d%.?%d)"
		local clausesFound = false --this is true when we found the word "clauses" and remain true until we habe clauses numbers following

		local i = 1
		@@ -160,6 +178,10 @@ function MultipleClauses(el)
		if clausesFound then
		--check if we found a clause number
		local startIndex, endIndex, number = elem.text:find(pattern)
		if number == nil then
		--check if it is an annex clause number
		startIndex, endIndex, number = elem.text:find(annexPattern)
		end

		if elem.text:match("^,$") or elem.text:match("and") then
		--this is just a comma or the word 'and' separating the clauses
		@@ -167,7 +189,12 @@ function MultipleClauses(el)
		elseif number then
		-- we did in fact find a clause number
		--substitute with linkS
		-- if first letter is a letter it is an annex
		if elem.text:sub(1, 1):match("%a") then
		newContent:insert(AnnexLink(elem.text:sub(startIndex, endIndex), number))
		else
		newContent:insert(ClauseLink(elem.text:sub(startIndex, endIndex), number))
		end
		if endIndex < #elem.text then
		--add the rest of the string too
		newContent:insert(pandoc.Str(elem.text:sub(endIndex + 1, -1)))

md_to_docx_converter/md_to_html_3.lua

+7 −0

Original line number	Diff line number	Diff line
		@@ -107,14 +107,21 @@ function CustomTagsToLinks(el, prefix)
		local clause_number = filename_numbers_mapping[filename]
		-- check that number is not a letter to differentiate between clauses and annexes
		if type(clause_number) ~= "number" then
		-- replace the first char with a uppercase letter
		clause_number = clause_number:sub(1,1):upper() .. clause_number:sub(2)
		html_filename = "annex-" .. clause_number .. ".html"
		if id:find("root.") then
		id = id:gsub("root%.", clause_number .. ".")
		elseif id:match("root$") then
		id = id:gsub("root", clause_number)
		end
		prefix = prefix:gsub("Clause", "Annex"):gsub("clause", "annex")
		else
		html_filename = "clause-" .. math.floor(clause_number) .. ".html"
		if id:find("root.") then
		id = id:gsub("root%.", math.floor(clause_number) .. ".")
		elseif id:match("root$") then
		id = id:gsub("root", math.floor(clause_number))
		end
		end
		end