notex.nvim/lua/notex/parser/markdown.lua

-- Markdown content parsing module
local M = {}

local utils = require('notex.utils')

-- Extract content between lines
function M.extract_content_between(content, start_pattern, end_pattern)
  local start_pos = content:find(start_pattern)
  if not start_pos then
    return nil
  end

  local end_pos = content:find(end_pattern, start_pos)
  if not end_pos then
    return content:sub(start_pos)
  end

  return content:sub(start_pos, end_pos)
end

-- Remove YAML header from content
function M.remove_yaml_header(content)
  return content:gsub("^%s*%-%-%-\n.-\n%-%-%-\n", "", 1)
end

-- Extract markdown body (content after YAML header)
function M.get_body(content)
  return M.remove_yaml_header(content)
end

-- Count words in markdown content
function M.count_words(content)
  local body = M.get_body(content)
  if not body then
    return 0
  end

  -- Remove markdown syntax for accurate word count
  local clean = body
    :gsub("#+ ", "") -- Headers
    :gsub("%*%*(.-)%*%*", "%1") -- Bold
    :gsub("%*(.-)%*", "%1") -- Italic
    :gsub("`(.-)`", "%1") -- Inline code
    :gsub("```.-```", "") -- Code blocks
    :gsub("%[.-%]%(.-%)", "") -- Links
    :gsub("%!%[.-%]%(.-%)", "") -- Images
    :gsub("%W+", " ") -- Replace non-word chars with spaces
    :gsub("%s+", " ") -- Collapse multiple spaces

  local words = {}
  for word in clean:gmatch("%S+") do
    if #word > 0 then
      table.insert(words, word)
    end
  end

  return #words
end

-- Count characters in markdown content
function M.count_characters(content)
  local body = M.get_body(content)
  return body and #body or 0
end

-- Extract headings from markdown
function M.extract_headings(content)
  local headings = {}
  local body = M.get_body(content)

  for level, title in body:gmatch("^(#+)%s+(.+)$") do
    table.insert(headings, {
      level = #level,
      title = title:trim(),
      raw = level .. " " .. title
    })
  end

  return headings
end

-- Extract links from markdown
function M.extract_links(content)
  local links = {}

  for text, url in content:gmatch("%[([^%]]*)%]%(([^)]+)%)") do
    table.insert(links, {
      text = text,
      url = url,
      raw = "[" .. text .. "](" .. url .. ")"
    })
  end

  return links
end

-- Extract code blocks from markdown
function M.extract_code_blocks(content)
  local code_blocks = {}

  for lang, code in content:gmatch("```(%w*)\n(.-)\n```") do
    table.insert(code_blocks, {
      language = lang ~= "" and lang or "text",
      code = code,
      lines = select(2, code:gsub("\n", "")) + 1
    })
  end

  return code_blocks
end

-- Get content summary (first paragraph)
function M.get_summary(content, max_length)
  local body = M.get_body(content)
  if not body then
    return ""
  end

  -- Remove code blocks to avoid including them in summary
  local clean_body = body:gsub("```.-```", "")

  -- Extract first paragraph
  local first_paragraph = clean_body:match("\n\n([^%[].-)\n\n") or
                           clean_body:match("^([^%[].-)\n\n") or
                           clean_body:match("^([^%[].-)")

  if not first_paragraph then
    return ""
  end

  -- Clean up markdown formatting
  local summary = first_paragraph
    :gsub("#+ ", "")
    :gsub("%*%*(.-)%*%*", "%1")
    :gsub("%*(.-)%*", "%1")
    :gsub("`(.-)`", "%1")
    :gsub("%[.-%]%(.-%)", "")

  summary = summary:gsub("%s+", " "):trim()

  if #summary > max_length then
    summary = summary:sub(1, max_length - 3) .. "..."
  end

  return summary
end

-- Analyze markdown structure
function M.analyze_structure(content)
  local body = M.get_body(content)

  return {
    word_count = M.count_words(content),
    character_count = M.count_characters(content),
    headings = M.extract_headings(content),
    links = M.extract_links(content),
    code_blocks = M.extract_code_blocks(content),
    summary = M.get_summary(content, 200),
    line_count = select(2, body:gsub("\n", "")) + 1,
    has_toc = body:find("^%s*%[TOC%]") ~= nil,
    reading_time_minutes = math.ceil(M.count_words(content) / 200) -- Assuming 200 WPM
  }
end

-- Validate markdown format
function M.validate_markdown(content)
  local errors = {}

  if not content or content == "" then
    table.insert(errors, "Empty content")
    return errors
  end

  -- Check for balanced markdown syntax
  local function check_balance(content, open, close)
    local count_open = select(2, content:gsub(open, ""))
    local count_close = select(2, content:gsub(close, ""))
    return count_open == count_close
  end

  -- Check balanced headers
  local headers = content:match("#+")
  if headers and not check_balance(content, "```", "```") then
    table.insert(errors, "Unbalanced code blocks")
  end

  -- Check for malformed links
  for link in content:gmatch("%[.-%]%(.-%)") do
    if not link:match("%[.-%]%(([^)]+)%)") or
       link:match("%[.-%]%(%s*%)") then
      table.insert(errors, "Malformed link: " .. link)
    end
  end

  return errors
end

-- Convert markdown to plain text
function M.to_plain_text(content)
  local body = M.get_body(content)
  if not body then
    return ""
  end

  local plain = body
    :gsub("^#%s+", "\n") -- Headers to newlines
    :gsub("\n#%s+", "\n") -- Headers to newlines
    :gsub("%*%*(.-)%*%*", "%1") -- Bold
    :gsub("%*(.-)%*", "%1") -- Italic
    :gsub("`(.-)`", "%1") -- Inline code
    :gsub("```%w*\n(.-)\n```", "%1") -- Code blocks
    :gsub("%[([^%]]*)%]%(([^)]+)%)", "%1") -- Links to text
    :gsub("%!%[([^%]]*)%]%(([^)]+)%)", "[Image: %1]") -- Images
    :gsub("\n%s*[-*+]%s+", "\n• ") -- List items
    :gsub("\n%s*%d+%.%s+", "\n• ") -- Numbered lists
    :gsub("\n%s*%[%s*%]%s+", "\n• ") -- Checkbox lists
    :gsub("\n%s*%[%s*x%s*%]%s+", "\n✓ ") -- Checked items
    :gsub("\n\n+", "\n\n") -- Multiple newlines
    :trim()

  return plain
end

return M