dcg/pkg/cmd/cmd_parser.go

package cmd

import (
	"database/sql"
	"git.noahlan.cn/northlan/ntools-go/stringn/ac"
	"strings"
)

type (
	Match struct {
		Prefix  string // 匹配前缀
		Content []rune // 内容
	}
	CMD struct {
		IsCMD   bool    // 是否CMD
		Matches []Match // 匹配项
	}
	Parser struct {
		ac       ac.AhoCorasick // ac自动机
		distinct bool           // 命令是否去重(总开关)

		patterns []Pattern // 匹配规则
	}
	Pattern struct {
		Prefix        string       // 前缀
		Alias         []string     // 前缀别名
		ContentMaxLen int          // 内容最大长度,以rune数组长度计算
		Distinct      sql.NullBool // Distinct 命令是否去重 单个权重大于总开关
		isAlias       bool         // alias
		realPattern   *Pattern     // 作为Alias时指向的实际Pattern
	}
)

func NewCMDParser(patterns ...Pattern) *Parser {
	p := &Parser{}

	keyPrefixArr := p.initPattern(patterns)

	builder := ac.NewAhoCorasickBuilder(ac.Opts{
		AsciiCaseInsensitive: true,
		MatchOnlyWholeWords:  false,
		MatchKind:            ac.LeftMostLongestMatch,
		DFA:                  true,
	})

	p.ac = builder.Build(keyPrefixArr)
	return p
}

func (p *Parser) initPattern(patterns []Pattern) []string {
	p.patterns = make([]Pattern, 0, len(patterns))

	result := make([]string, 0, len(patterns))
	for _, pattern := range patterns {
		p.patterns = append(p.patterns, pattern)
		result = append(result, pattern.Prefix)
		for _, alias := range pattern.Alias {
			p.patterns = append(p.patterns, Pattern{
				Prefix:  alias,
				isAlias: true,
				realPattern: &Pattern{
					Prefix:        pattern.Prefix,
					Alias:         pattern.Alias,
					ContentMaxLen: pattern.ContentMaxLen,
					Distinct:      pattern.Distinct,
				},
			})
			result = append(result, alias)
		}
	}
	return result
}

func (p *Parser) ParseTest(content string) []ac.Match {
	return p.ac.FindAll(content)
}

// SetDistinct 设置命令去重
func (p *Parser) SetDistinct(distinct bool) {
	p.distinct = distinct
}

// Parse 从弹幕内容解析命令
func (p *Parser) Parse(msg string) CMD {
	// 移除多余空格
	tmpContent := strings.TrimSpace(msg)
	resp := CMD{
		Matches: make([]Match, 0),
	}

	iter := p.ac.Iter(tmpContent)
	for match := iter.Next(); match != nil; {
		resp.IsCMD = true
		pattern := p.patterns[match.Pattern()]
		if pattern.isAlias {
			pattern = *pattern.realPattern
		}
		m := Match{
			Prefix: pattern.Prefix,
		}
		tmpNext := iter.Next()

		// 避免同类型指令重复
		if pattern.Distinct.Valid && pattern.Distinct.Bool {
			for _, m := range resp.Matches {
				if m.Prefix == pattern.Prefix {
					goto end // continue out for
				}
			}
		}
		if p.distinct {
			if !pattern.Distinct.Valid || pattern.Distinct.Bool {
				for _, m := range resp.Matches {
					if m.Prefix == pattern.Prefix {
						goto end // continue out for
					}
				}
			}
		}

		if tmpNext != nil {
			//nextPattern := p.patterns[tmpNext.Pattern()]
			content := []rune(tmpContent[match.End():tmpNext.Start()])
			if len(content) > pattern.ContentMaxLen {
				resp.IsCMD = false
				break
			}
			m.Content = content
		} else {
			content := []rune(tmpContent[match.End():])
			if len(content) > pattern.ContentMaxLen {
				resp.IsCMD = false
				break
			}
			m.Content = content
		}
		resp.Matches = append(resp.Matches, m)
	end:
		match = tmpNext
	}
	return resp
}