curl2info/parse_curl.go

package curl2info

import (
	"fmt"
	"io/ioutil"
	"log"
	"net/http"
	"net/http/cookiejar"
	"net/url"
	"os"
	"regexp"
	"strconv"
	"strings"

	"474420502.top/eson/requests"
)

// CURL 信息结构
type CURL struct {
	ParsedURL *url.URL
	Method    string
	Header    http.Header
	CookieJar http.CookieJar
	Cookies   []*http.Cookie
	Body      *requests.Body
	Auth      *requests.BasicAuth
	Timeout   int // second
	Insecure  bool

	CallBack string
}

// NewCURL new 一个 curl 出来
func NewCURL() *CURL {

	u := &CURL{}
	u.Insecure = false

	u.Header = make(http.Header)
	u.CookieJar, _ = cookiejar.New(nil)
	u.Body = requests.NewBody()
	u.Timeout = 30

	return u
}

func (curl *CURL) String() string {
	if curl != nil {
		return fmt.Sprintf("Method: %s\nParsedURL: %s\nHeader: %s\nCookie: %s",
			curl.Method, curl.ParsedURL.String(), curl.Header, curl.Cookies)
	}
	return ""
}

// CreateSession 创建Session
func (curl *CURL) CreateSession() *requests.Session {
	ses := requests.NewSession()
	ses.SetHeader(curl.Header)
	ses.SetCookies(curl.ParsedURL, curl.Cookies)
	ses.SetConfig(requests.CRequestTimeout, curl.Timeout)

	if curl.Auth != nil {
		ses.SetConfig(requests.CBasicAuth, curl.Auth)
	}

	if curl.Insecure {
		ses.SetConfig(requests.CInsecure, curl.Insecure)
	}

	return ses
}

// CreateWorkflow 根据Session 创建Workflow
func (curl *CURL) CreateWorkflow(ses *requests.Session) *requests.Workflow {
	var wf *requests.Workflow

	if ses == nil {
		ses = curl.CreateSession()
	}

	switch curl.Method {
	case "HEAD":
		wf = ses.Head(curl.ParsedURL.String())
	case "GET":
		wf = ses.Get(curl.ParsedURL.String())
	case "POST":
		wf = ses.Post(curl.ParsedURL.String())
	case "PUT":
		wf = ses.Put(curl.ParsedURL.String())
	case "PATCH":
		wf = ses.Patch(curl.ParsedURL.String())
	case "OPTIONS":
		wf = ses.Options(curl.ParsedURL.String())
	case "DELETE":
		wf = ses.Delete(curl.ParsedURL.String())
	}

	wf.SetBody(curl.Body)
	return wf
}

func init() {
	optionTrie = NewTrie()
	oelist := []*optionExecute{
		{"-H", 10, parseHeader, nil},
		{"-X", 10, parseOptX, nil},
		{"-A", 15, parseUserAgent, &extract{re: "^-A +(.+)", execute: extractData}},
		{"-I", 15, parseOptI, nil},
		{"-d", 10, parseBodyASCII, &extract{re: "^-d +(.+)", execute: extractData}},
		{"-u", 15, parseUser, &extract{re: "^-u +(.+)", execute: extractData}},
		{"-k", 15, parseInsecure, nil},
		// Body
		{"--data", 10, parseBodyASCII, &extract{re: "--data +(.+)", execute: extractData}},
		{"--data-urlencode", 10, parseBodyURLEncode, &extract{re: "--data-urlencode +(.+)", execute: extractData}},
		{"--data-binary", 10, parseBodyBinary, &extract{re: "--data-binary +(.+)", execute: extractData}},
		{"--data-ascii", 10, parseBodyASCII, &extract{re: "--data-ascii +(.+)", execute: extractData}},
		{"--data-raw", 10, parseBodyRaw, &extract{re: "--data-raw +(.+)", execute: extractData}},
		//"--"
		{"--header", 10, parseHeader, nil},
		{"--insecure", 15, parseInsecure, nil},
		{"--call", 10, parseCallBack, &extract{re: "--call +(.+)", execute: extractData}},
		{"--user-agent", 15, parseUserAgent, &extract{re: "--user-agent +(.+)", execute: extractData}},
		{"--user", 15, parseUser, &extract{re: "--user +(.+)", execute: extractData}},
		{"--connect-timeout", 15, parseTimeout, &extract{re: "--connect-timeout +(.+)", execute: extractData}},
	}

	for _, oe := range oelist {
		optionTrie.Insert(oe)
	}

	log.Println("support options:", optionTrie.AllWords())
}

// ParseRawCURL curl_bash 可以用trie改进 没空改
func ParseRawCURL(scurl string) (cURL *CURL, err error) {

	defer func() {
		if _err := recover(); _err != nil {
			cURL = nil
			err = _err.(error)
		}
	}()

	executor := newPQueueExecute()
	curl := NewCURL()

	if scurl[0] == '"' && scurl[len(scurl)-1] == '"' {
		scurl = strings.Trim(scurl, `"`)
	} else if scurl[0] == '\'' && scurl[len(scurl)-1] == '\'' {
		scurl = strings.Trim(scurl, `'`)
	}

	scurl = strings.TrimSpace(scurl)
	scurl = strings.TrimLeft(scurl, "curl")

	mathches := regexp.MustCompile(`--[^ ]+ +'[^']+'|--[^ ]+ +[^ ]+|-[A-Za-z] +'[^']+'|-[A-Za-z] +[^ ]+| '[^']+'|--[a-z]+ {0,}`).FindAllString(scurl, -1)
	for _, m := range mathches {
		m = strings.TrimSpace(m)
		switch v := m[0]; v {
		case '\'':
			purl, err := url.Parse(strings.Trim(m, "'"))
			if err != nil {
				panic(err)
			}
			curl.ParsedURL = purl
		case '-':
			exec := judgeOptions(curl, m)
			if exec != nil {
				executor.Push(exec)
			}
		}
	}

	for executor.Len() > 0 {
		exec := executor.Pop()
		exec.Execute()
	}

	if curl.Method == "" {
		curl.Method = "GET"
	}

	return curl, nil
}

func judgeOptions(u *CURL, soption string) *parseFunction {
	word := TrieStrWord(soption)
	if ioe := optionTrie.SearchMostPrefix(&word); ioe != nil {
		oe := ioe.(*optionExecute)
		return oe.BuildFunction(u, soption)
	}

	log.Println(soption, " no haved this option")
	return nil
}

func judgeAndParseOptions(u *CURL, soption string) *parseFunction {
	switch prefix := soption[0:2]; prefix {
	case "-H":
		return &parseFunction{ParamCURL: u, ParamData: soption, ExecuteFunction: parseHeader, Priority: 10}
	case "-X":
		return &parseFunction{ParamCURL: u, ParamData: soption, ExecuteFunction: parseOptX, Priority: 10}
	case "-A": // User-Agent 先后顺序的问题
		data := extractData("^-A +(.+)", soption)
		return &parseFunction{ParamCURL: u, ParamData: data, ExecuteFunction: parseUserAgent, Priority: 15}
	case "-I":
		return &parseFunction{ParamCURL: u, ParamData: soption, ExecuteFunction: parseOptI, Priority: 15}
	case "--":
		return parseLongOption(u, soption)
	case "-d":
		data := extractData("^-d +(.+)", soption)
		return &parseFunction{ParamCURL: u, ParamData: data, ExecuteFunction: parseBodyASCII, Priority: 10}
	case "-u":
		data := extractData("^-u +(.+)", soption)
		return &parseFunction{ParamCURL: u, ParamData: data, ExecuteFunction: parseUser, Priority: 15}
	case "-k": // -k, --insecure      Allow insecure server connections when using SSL
		return &parseFunction{ParamCURL: u, ParamData: soption, ExecuteFunction: parseInsecure, Priority: 15}
	}
	return nil
}

func parseLongOption(u *CURL, soption string) *parseFunction {
	// -d, --data <data>   HTTP POST data
	// --data-ascii <data> HTTP POST ASCII data
	// --data-binary <data> HTTP POST binary data
	// --data-raw <data> HTTP POST data, '@' allowed
	// --data-urlencode <data> HTTP POST data url encoded

	switch {
	case regexp.MustCompile("^--data |^--data-urlencode|^--data-binary|^--data-ascii|^--data-raw").MatchString(soption):

		datas := regexp.MustCompile("^--data-(binary) +(.+)|^--data-(ascii) +(.+)|^--data-(raw) +(.+)|^--data-(urlencode) +(.+)|^--(data) +(.+)").FindStringSubmatch(soption)
		dtype := datas[1]
		data := strings.Trim(datas[2], "'")

		switch dtype {
		case "binary":
			return &parseFunction{ParamCURL: u, ParamData: data, ExecuteFunction: parseBodyBinary, Priority: 10}
		case "ascii":
			return &parseFunction{ParamCURL: u, ParamData: data, ExecuteFunction: parseBodyASCII, Priority: 10}
		case "raw":
			return &parseFunction{ParamCURL: u, ParamData: data, ExecuteFunction: parseBodyRaw, Priority: 10}
		case "urlencode":
			return &parseFunction{ParamCURL: u, ParamData: data, ExecuteFunction: parseBodyURLEncode, Priority: 10}
		case "data":
			return &parseFunction{ParamCURL: u, ParamData: data, ExecuteFunction: parseBodyASCII, Priority: 10}
		}

	case regexp.MustCompile("^--header").MatchString(soption):
		return &parseFunction{ParamCURL: u, ParamData: soption, ExecuteFunction: parseHeader, Priority: 10}
	case regexp.MustCompile("^--call").MatchString(soption):
		data := extractData("^--call +(.+)", soption)
		return &parseFunction{ParamCURL: u, ParamData: data, ExecuteFunction: parseCallBack, Priority: 10}
	case regexp.MustCompile("^--user-agent").MatchString(soption):
		data := extractData("^--user-agent +(.+)", soption)
		return &parseFunction{ParamCURL: u, ParamData: data, ExecuteFunction: parseUserAgent, Priority: 15}
	case regexp.MustCompile("^--user").MatchString(soption):
		data := extractData("^--user +(.+)", soption)
		return &parseFunction{ParamCURL: u, ParamData: data, ExecuteFunction: parseUser, Priority: 15}
	case regexp.MustCompile("^--insecure").MatchString(soption):
		return &parseFunction{ParamCURL: u, ParamData: soption, ExecuteFunction: parseInsecure, Priority: 15}
	case regexp.MustCompile("^--connect-timeout").MatchString(soption):
		data := extractData("^--connect-timeout +(.+)", soption)
		return &parseFunction{ParamCURL: u, ParamData: data, ExecuteFunction: parseTimeout, Priority: 15}
	}

	log.Println("can't parseOption", soption)
	return nil
}

func extractData(re, soption string) string {
	datas := regexp.MustCompile(re).FindStringSubmatch(soption)
	return strings.Trim(datas[1], "'")
}

func parseCallBack(u *CURL, value string) {
	u.CallBack = value
}

func parseTimeout(u *CURL, value string) {
	timeout, err := strconv.Atoi(value)
	if err != nil {
		panic(err)
	}
	u.Timeout = timeout
}

func parseInsecure(u *CURL, soption string) {
	u.Insecure = true
}

func parseUser(u *CURL, soption string) {
	auth := strings.Split(soption, ":")
	u.Auth = &requests.BasicAuth{User: auth[0], Password: auth[1]}
}

func parseUserAgent(u *CURL, value string) {
	u.Header.Add("User-Agent", value)
}

func parseOptI(u *CURL, soption string) {
	u.Method = "HEAD"
}

func parseOptX(u *CURL, soption string) {
	matches := regexp.MustCompile("-X +(.+)").FindStringSubmatch(soption)
	method := strings.Trim(matches[1], "'")
	u.Method = method
}

func parseBodyURLEncode(u *CURL, data string) {
	if u.Method != "" {
		u.Method = "POST"
	}

	u.Body.SetPrefix(requests.TypeURLENCODED)
	u.Body.SetIOBody(data)
}

func parseBodyRaw(u *CURL, data string) {
	if u.Method != "" {
		u.Method = "POST"
	}

	u.Body.SetPrefix(requests.TypeURLENCODED)
	u.Body.SetIOBody(data)
}

func parseBodyASCII(u *CURL, data string) {
	if u.Method != "" {
		u.Method = "POST"
	}

	u.Body.SetPrefix(requests.TypeURLENCODED)

	if data[0] != '@' {
		u.Body.SetIOBody(data)
	} else {
		f, err := os.Open(data[1:])
		if err != nil {
			panic(err)
		}
		defer f.Close()

		bdata, err := ioutil.ReadAll(f)
		if err != nil {
			panic(err)
		}
		u.Body.SetIOBody(bdata)
	}
}

// 处理@ 并且替/r/n符号
func parseBodyBinary(u *CURL, data string) {
	if u.Method != "" {
		u.Method = "POST"
	}

	u.Body.SetPrefix(requests.TypeURLENCODED)

	if data[0] != '@' {
		u.Body.SetIOBody(data)
	} else {
		f, err := os.Open(data[1:])
		if err != nil {
			panic(err)
		}
		defer f.Close()
		bdata, err := ioutil.ReadAll(f)
		if err != nil {
			panic(err)
		}
		bdata = regexp.MustCompile("\n|\r").ReplaceAll(bdata, []byte(""))
		u.Body.SetIOBody(bdata)
	}
}

func parseHeader(u *CURL, soption string) {
	matches := regexp.MustCompile(`'([^:]+): ([^']+)'`).FindAllStringSubmatch(soption, 1)[0]
	key := matches[1]
	value := matches[2]

	switch key {
	case "Cookie":
		u.Cookies = ReadRawCookies(value, "")
		u.CookieJar.SetCookies(u.ParsedURL, u.Cookies)
	case "Content-Type":
		u.Body.SetPrefix(value)
	default:
		u.Header.Add(key, value)
	}

}