notex.nvim/lua/notex/parser/init.lua

-- Parser coordination module
local M = {}

local yaml_parser = require('notex.parser.yaml')
local markdown_parser = require('notex.parser.markdown')
local utils = require('notex.utils')

-- Parse complete markdown document
function M.parse_document(file_path)
  local result = {
    file_path = file_path,
    success = false,
    yaml_data = {},
    markdown_analysis = {},
    properties = {},
    errors = {},
    warnings = {}
  }

  -- Validate file
  if not utils.file_exists(file_path) then
    table.insert(result.errors, "File does not exist: " .. file_path)
    return result
  end

  if not utils.is_utf8(file_path) then
    table.insert(result.errors, "File is not valid UTF-8: " .. file_path)
    return result
  end

  -- Read file content
  local content, err = utils.read_file(file_path)
  if not content then
    table.insert(result.errors, "Cannot read file: " .. err)
    return result
  end

  -- Parse YAML header
  local yaml_data, yaml_err = yaml_parser.parse_markdown_file(file_path)
  if yaml_data then
    result.yaml_data = yaml_data

    -- Process properties
    result.properties = yaml_parser.process_properties(yaml_data)
  elseif yaml_err then
    table.insert(result.errors, "YAML parsing error: " .. yaml_err)
  else
    table.insert(result.warnings, "No YAML header found")
  end

  -- Analyze markdown content
  local markdown_analysis = markdown_parser.analyze_structure(content)
  result.markdown_analysis = markdown_analysis

  result.success = #result.errors == 0

  return result
end

-- Parse multiple documents
function M.parse_documents(file_paths)
  local results = {
    total = #file_paths,
    successful = 0,
    failed = 0,
    documents = {},
    errors = {}
  }

  for _, file_path in ipairs(file_paths) do
    local doc_result = M.parse_document(file_path)
    table.insert(results.documents, doc_result)

    if doc_result.success then
      results.successful = results.successful + 1
    else
      results.failed = results.failed + 1
      for _, error in ipairs(doc_result.errors) do
        table.insert(results.errors, {
          file_path = file_path,
          error = error
        })
      end
    end
  end

  return results
end

-- Extract document summary
function M.get_document_summary(file_path)
  local parse_result = M.parse_document(file_path)

  if not parse_result.success then
    return nil, "Failed to parse document: " .. table.concat(parse_result.errors, ", ")
  end

  local summary = {
    file_path = file_path,
    title = parse_result.yaml_data.title or parse_result.markdown_analysis.headings[1] and parse_result.markdown_analysis.headings[1].title or "Untitled",
    status = parse_result.yaml_data.status or "unknown",
    priority = parse_result.yaml_data.priority or 0,
    tags = parse_result.yaml_data.tags or {},
    word_count = parse_result.markdown_analysis.word_count,
    created_at = parse_result.yaml_data.created_at,
    updated_at = parse_result.yaml_data.updated_at,
    summary = parse_result.markdown_analysis.summary,
    properties_count = #parse_result.properties,
    has_headings = #parse_result.markdown_analysis.headings > 0,
    has_links = #parse_result.markdown_analysis.links > 0,
    has_code = #parse_result.markdown_analysis.code_blocks > 0,
    reading_time = parse_result.markdown_analysis.reading_time_minutes
  }

  return summary
end

-- Validate document against schema
function M.validate_document_schema(file_path, schema_requirements)
  local parse_result = M.parse_document(file_path)

  if not parse_result.success then
    return false, parse_result.errors
  end

  local validation_errors = {}

  -- Check required properties
  if schema_requirements.required then
    for _, required_prop in ipairs(schema_requirements.required) do
      local found = false
      for _, prop in ipairs(parse_result.properties) do
        if prop.key == required_prop then
          found = true
          break
        end
      end

      if not found then
        table.insert(validation_errors, string.format("Missing required property: %s", required_prop))
      end
    end
  end

  -- Check property types
  if schema_requirements.property_types then
    for _, prop in ipairs(parse_result.properties) do
      local expected_type = schema_requirements.property_types[prop.key]
      if expected_type and prop.value_type ~= expected_type then
        table.insert(validation_errors, string.format("Property '%s' should be %s, got %s", prop.key, expected_type, prop.value_type))
      end
    end
  end

  -- Check property patterns
  if schema_requirements.patterns then
    for _, prop in ipairs(parse_result.properties) do
      local pattern = schema_requirements.patterns[prop.key]
      if pattern and not prop.value:match(pattern) then
        table.insert(validation_errors, string.format("Property '%s' does not match required pattern", prop.key))
      end
    end
  end

  if #validation_errors > 0 then
    return false, validation_errors
  end

  return true, parse_result
end

-- Extract document relationships
function M.extract_relationships(file_paths)
  local relationships = {
    links = {},
    references = {},
    backlinks = {}
  }

  -- Parse all documents
  local parse_results = M.parse_documents(file_paths)

  -- Build document lookup
  local docs_by_path = {}
  for _, doc_result in ipairs(parse_results.documents) do
    if doc_result.success then
      docs_by_path[doc_result.file_path] = doc_result
    end
  end

  -- Extract links and references
  for _, doc_result in ipairs(parse_results.documents) do
    if doc_result.success then
      local source_doc = doc_result.file_path

      -- Extract markdown links
      for _, link in ipairs(doc_result.markdown_analysis.links) do
        table.insert(relationships.links, {
          source = source_doc,
          target = link.url,
          text = link.text,
          type = "markdown_link"
        })
      end

      -- Extract property references (if any)
      for _, prop in ipairs(doc_result.properties) do
        if prop.key:match("ref") or prop.key:match("reference") then
          table.insert(relationships.references, {
            source = source_doc,
            target = prop.value,
            property = prop.key,
            type = "property_reference"
          })
        end
      end
    end
  end

  -- Build backlinks
  for _, link in ipairs(relationships.links) do
    for target_path, target_doc in pairs(docs_by_path) do
      if link.target == target_path or link.target:match(target_path) then
        table.insert(relationships.backlinks, {
          source = target_path,
          target = link.source,
          text = link.text,
          type = "backlink"
        })
      end
    end
  end

  return relationships
end

-- Generate document statistics
function M.generate_statistics(file_paths)
  local stats = {
    total_documents = #file_paths,
    total_words = 0,
    total_properties = 0,
    property_distribution = {},
    status_distribution = {},
    tag_distribution = {},
    file_size_distribution = {},
    average_word_count = 0,
    documents_with_headings = 0,
    documents_with_links = 0,
    documents_with_code = 0
  }

  -- Parse all documents
  local parse_results = M.parse_documents(file_paths)

  for _, doc_result in ipairs(parse_results.documents) do
    if doc_result.success then
      -- Word count
      stats.total_words = stats.total_words + doc_result.markdown_analysis.word_count

      -- Properties
      stats.total_properties = stats.total_properties + #doc_result.properties

      -- Property distribution
      for _, prop in ipairs(doc_result.properties) do
        if not stats.property_distribution[prop.key] then
          stats.property_distribution[prop.key] = 0
        end
        stats.property_distribution[prop.key] = stats.property_distribution[prop.key] + 1
      end

      -- Status distribution
      local status = doc_result.yaml_data.status or "unknown"
      if not stats.status_distribution[status] then
        stats.status_distribution[status] = 0
      end
      stats.status_distribution[status] = stats.status_distribution[status] + 1

      -- Tag distribution
      local tags = doc_result.yaml_data.tags or {}
      for _, tag in ipairs(tags) do
        if not stats.tag_distribution[tag] then
          stats.tag_distribution[tag] = 0
        end
        stats.tag_distribution[tag] = stats.tag_distribution[tag] + 1
      end

      -- Feature flags
      if doc_result.markdown_analysis.has_headings then
        stats.documents_with_headings = stats.documents_with_headings + 1
      end

      if doc_result.markdown_analysis.has_links then
        stats.documents_with_links = stats.documents_with_links + 1
      end

      if doc_result.markdown_analysis.has_code then
        stats.documents_with_code = stats.documents_with_code + 1
      end
    end
  end

  -- Calculate averages
  if stats.total_documents > 0 then
    stats.average_word_count = math.floor(stats.total_words / stats.total_documents)
  end

  return stats
end

return M