XCSDD/src/lib/parser.ts

import type { ParsedDoc, DocMetadata, DocSection, DocTable, DocCodeBlock } from './types'

const METADATA_REGEX = /^\*\*([^\*]+)\*\*:\s*(.+)$/
const TABLE_REGEX = /^\|(.+)\|$/
const CODE_BLOCK_REGEX = /^```(\w*)$/
const HEADING_REGEX = /^(#{1,6})\s+(.+)$/
const HR_REGEX = /^---+$/
const REFERENCE_REGEX = /@see\s+([^\s]+)/g

export function parseMarkdown(content: string): ParsedDoc {
  if (!content || typeof content !== 'string') {
    return { title: '', metadata: {}, sections: [], references: [] };
  }
  
  const lines = content.split('\n')
  const result: ParsedDoc = {
    title: '',
    metadata: {},
    sections: [],
    references: [],
  }

  const references: string[] = []
  let metadata: DocMetadata = {}
  let sections: DocSection[] = []
  let currentSection: DocSection | null = null
  let currentContent: Array<{ type: 'text' | 'table' | 'code'; data: string | DocTable | DocCodeBlock }> = []
  let inCodeBlock = false
  let codeBlockLanguage = ''
  let codeBlockLines: string[] = []
  let tableBuffer: string[] = []
  let inTable = false

  try {
    for (let i = 0; i < lines.length; i++) {
      const line = lines[i]
      const trimmedLine = line.trim()

      try {
        const refMatch = trimmedLine.match(REFERENCE_REGEX)
        if (refMatch) {
          let match
          const refRegex = /@see\s+([^\s]+)/g
          while ((match = refRegex.exec(trimmedLine)) !== null) {
            references.push(match[1])
          }
        }

        const codeMatch = line.match(CODE_BLOCK_REGEX)
        if (codeMatch) {
          if (currentSection) {
            currentSection.content = [...currentContent]
            sections.push(currentSection)
            currentSection = null
          }
          inCodeBlock = true
          codeBlockLanguage = codeMatch[1] || ''
          currentContent = []
          continue
        }

        if (inCodeBlock) {
          if (trimmedLine === '```' || trimmedLine.startsWith('```')) {
            currentContent.push({
              type: 'code',
              data: { language: codeBlockLanguage, code: codeBlockLines.join('\n') },
            } as { type: 'code'; data: DocCodeBlock })
            codeBlockLines = []
            codeBlockLanguage = ''
            inCodeBlock = false
            
            if (!currentSection) {
              currentSection = { title: '', level: 2, content: [] }
            }
          } else {
            codeBlockLines.push(line)
          }
          continue
        }

        if (HR_REGEX.test(trimmedLine)) {
          if (inTable && tableBuffer.length > 0) {
            const table = parseTable(tableBuffer)
            if (table) {
              currentContent.push({ type: 'table', data: table })
            }
            tableBuffer = []
            inTable = false
          }
          continue
        }

        if (!result.title && line.startsWith('# ')) {
          result.title = line.slice(2).trim()
          continue
        }

        const headingMatch = line.match(HEADING_REGEX)
        if (headingMatch) {
          if (currentSection) {
            currentSection.content = [...currentContent]
            sections.push(currentSection)
          }
          const level = headingMatch[1].length
          const title = headingMatch[2].trim()
          currentSection = { title, level, content: [] }
          currentContent = []
          inTable = false
          tableBuffer = []
          continue
        }

        const metaMatch = trimmedLine.match(METADATA_REGEX)
        if (metaMatch) {
          const key = metaMatch[1].toLowerCase()
          const value = metaMatch[2].trim()
          if (key === 'namespace') metadata.namespace = value
          else if (key === 'description') metadata.description = value
          else if (key === 'type') metadata.type = value
          else if (key === 'inherits') metadata.inherits = value
          else if (key === 'package') metadata.package = value
          continue
        }

        const tableMatch = line.match(TABLE_REGEX)
        if (tableMatch) {
          tableBuffer.push(trimmedLine)
          inTable = true
          continue
        } else if (inTable && tableBuffer.length > 0) {
          const table = parseTable(tableBuffer)
          if (table) {
            currentContent.push({ type: 'table', data: table })
          }
          tableBuffer = []
          inTable = false
        }

        if (trimmedLine) {
          currentContent.push({ type: 'text', data: trimmedLine })
        }
      } catch (lineError) {
        console.warn('Error parsing line', i, ':', lineError)
      }
    }

    // 结束未闭合的代码块
    if (inCodeBlock && codeBlockLines.length > 0) {
      currentContent.push({
        type: 'code',
        data: { language: codeBlockLanguage, code: codeBlockLines.join('\n') },
      } as { type: 'code'; data: DocCodeBlock })
    }

    // 结束未闭合的表格
    if (inTable && tableBuffer.length > 0) {
      const table = parseTable(tableBuffer)
      if (table) {
        currentContent.push({ type: 'table', data: table })
      }
    }

    if (currentSection) {
      currentSection.content = [...currentContent]
      sections.push(currentSection)
    }

    result.metadata = metadata
    result.sections = sections
    result.references = references
  } catch (error) {
    console.error('Markdown parse error:', error)
    return {
      title: '',
      metadata: {},
      sections: [],
      references: [],
      parseError: true,
      rawContent: content
    }
  }

  return result
}

function parseTable(tableLines: string[]): DocTable | null {
  try {
    if (tableLines.length < 1) return null

    const headers = tableLines[0].split('|').filter(h => h.trim()).map(h => h.trim())
    const rows: string[][] = []

    for (let i = 1; i < tableLines.length; i++) {
      const line = tableLines[i]
      const cleaned = line.replace(/\|/g, '').trim()
      const isSeparator = /^[-\s]+$/.test(cleaned)
      if (isSeparator) continue
      
      const cells = line.split('|').filter(c => c.trim()).map(c => c.trim())
      if (cells.length > 0) {
        rows.push(cells)
      }
    }

    return { headers, rows }
  } catch {
    return null
  }
}

export function getDisplayName(filename: string): string {
  const name = filename.replace(/\.md$/, '')
  return name
}

export function buildFileTree(files: string[], basePath: string): import('./types').DocFile[] {
  const tree: Map<string, import('./types').DocFile> = new Map()

  for (const file of files) {
    const relativePath = file.split('/').filter(Boolean).join('/')
    const parts = relativePath.split('/')

    let currentPath = ''
    for (let i = 0; i < parts.length; i++) {
      const part = parts[i]
      const isLast = i === parts.length - 1
      currentPath = currentPath ? `${currentPath}/${part}` : part

      if (!tree.has(currentPath)) {
        tree.set(currentPath, {
          name: part,
          path: `${basePath}/${currentPath}`,
          relativePath: currentPath,
          isDir: !isLast,
          children: isLast ? undefined : [],
        })
      }

      const parentPath = currentPath.split('/').slice(0, -1).join('/')
      if (parentPath && tree.has(parentPath)) {
        const parent = tree.get(parentPath)!
        if (parent.children) {
          const existing = parent.children.find(c => c.name === part)
          if (!existing) {
            parent.children.push(tree.get(currentPath)!)
          }
        }
      }
    }
  }

  const rootFiles: import('./types').DocFile[] = []
  for (const [path, file] of tree) {
    const parentPath = path.split('/').slice(0, -1).join('/')
    if (!parentPath) {
      rootFiles.push(file)
    }
  }

  const sortFiles = (files: import('./types').DocFile[]): import('./types').DocFile[] => {
    return files.sort((a, b) => {
      if (a.isDir !== b.isDir) return a.isDir ? -1 : 1
      return a.name.localeCompare(b.name)
    }).map(f => {
      if (f.children) {
        f.children = sortFiles(f.children)
      }
      return f
    })
  }

  return sortFiles(rootFiles)
}