gpt-plus-gpt/pkg/provider/email/outlook.go

package email

import (
	"bufio"
	"context"
	"encoding/json"
	"fmt"
	"io"
	"log"
	"net/http"
	"net/url"
	"os"
	"regexp"
	"strconv"
	"strings"
	"sync"
	"time"
)

const (
	msTokenURL     = "https://login.microsoftonline.com/common/oauth2/v2.0/token"
	outlookMailURL = "https://outlook.office.com/api/v2.0/me/mailFolders/inbox/messages"
)

// OutlookAccount holds credentials for a single Outlook email account.
type OutlookAccount struct {
	Email        string
	Password     string
	ClientID     string // Microsoft OAuth client_id (field 3 in file)
	RefreshToken string // Microsoft refresh token (field 4 in file)
}

// OutlookProvider implements EmailProvider using pre-existing Outlook accounts
// and Outlook REST API for email retrieval.
type OutlookProvider struct {
	accounts []OutlookAccount
	mu       sync.Mutex
	nextIdx  int
}

// NewOutlookProvider creates an OutlookProvider by loading accounts from a file.
// File format: email----password----client_id----refresh_token (one per line).
// pop3Server and pop3Port are ignored (kept for API compatibility).
func NewOutlookProvider(accountsFile, pop3Server string, pop3Port int) (*OutlookProvider, error) {
	accounts, err := loadOutlookAccounts(accountsFile)
	if err != nil {
		return nil, err
	}
	if len(accounts) == 0 {
		return nil, fmt.Errorf("no accounts found in %s", accountsFile)
	}

	log.Printf("[outlook] loaded %d accounts from %s (using REST API)", len(accounts), accountsFile)

	return &OutlookProvider{
		accounts: accounts,
	}, nil
}

func loadOutlookAccounts(path string) ([]OutlookAccount, error) {
	f, err := os.Open(path)
	if err != nil {
		return nil, fmt.Errorf("open accounts file: %w", err)
	}
	defer f.Close()

	var accounts []OutlookAccount
	scanner := bufio.NewScanner(f)
	scanner.Buffer(make([]byte, 0, 4096), 4096)
	lineNum := 0
	for scanner.Scan() {
		lineNum++
		line := strings.TrimSpace(scanner.Text())
		if line == "" || strings.HasPrefix(line, "#") {
			continue
		}
		parts := strings.Split(line, "----")
		if len(parts) < 2 {
			log.Printf("[outlook] skipping line %d: expected at least email----password", lineNum)
			continue
		}
		acct := OutlookAccount{
			Email:    strings.TrimSpace(parts[0]),
			Password: strings.TrimSpace(parts[1]),
		}
		if len(parts) >= 3 {
			acct.ClientID = strings.TrimSpace(parts[2])
		}
		if len(parts) >= 4 {
			acct.RefreshToken = strings.TrimSpace(parts[3])
		}
		accounts = append(accounts, acct)
	}
	return accounts, scanner.Err()
}

// CreateMailbox returns the next available Outlook account email.
func (o *OutlookProvider) CreateMailbox(ctx context.Context) (string, string, error) {
	o.mu.Lock()
	defer o.mu.Unlock()

	if o.nextIdx >= len(o.accounts) {
		return "", "", fmt.Errorf("all %d outlook accounts exhausted", len(o.accounts))
	}

	acct := o.accounts[o.nextIdx]
	mailboxID := strconv.Itoa(o.nextIdx)
	o.nextIdx++

	log.Printf("[outlook] using account #%s: %s", mailboxID, acct.Email)
	return acct.Email, mailboxID, nil
}

// WaitForVerificationCode polls the Outlook inbox via REST API for a 6-digit OTP.
func (o *OutlookProvider) WaitForVerificationCode(ctx context.Context, mailboxID string, timeout time.Duration, notBefore time.Time) (string, error) {
	idx, err := strconv.Atoi(mailboxID)
	if err != nil || idx < 0 || idx >= len(o.accounts) {
		return "", fmt.Errorf("invalid mailbox ID: %s", mailboxID)
	}
	acct := o.accounts[idx]

	if acct.RefreshToken == "" || acct.ClientID == "" {
		return "", fmt.Errorf("outlook account %s has no refresh_token/client_id for REST API", acct.Email)
	}

	deadline := time.Now().Add(timeout)
	ticker := time.NewTicker(5 * time.Second)
	defer ticker.Stop()

	for attempt := 1; ; attempt++ {
		log.Printf("[outlook] polling REST API for OTP (attempt %d, account=%s)", attempt, acct.Email)

		code, err := o.checkRESTForOTP(acct, notBefore)
		if err != nil {
			log.Printf("[outlook] REST API error (attempt %d): %v", attempt, err)
		} else if code != "" {
			log.Printf("[outlook] verification code found: %s", code)
			return code, nil
		}

		if time.Now().After(deadline) {
			return "", fmt.Errorf("timeout waiting for OTP after %s (%d attempts)", timeout, attempt)
		}

		select {
		case <-ctx.Done():
			return "", ctx.Err()
		case <-ticker.C:
		}
	}
}

// checkRESTForOTP uses Outlook REST API to fetch recent emails and extract OTP.
func (o *OutlookProvider) checkRESTForOTP(acct OutlookAccount, notBefore time.Time) (string, error) {
	// Exchange refresh token for access token (no scope param — token already has outlook.office.com scopes)
	accessToken, err := exchangeMSToken(acct.ClientID, acct.RefreshToken)
	if err != nil {
		return "", fmt.Errorf("token exchange: %w", err)
	}

	// Fetch recent emails without $filter (Outlook v2.0 often rejects filter on
	// consumer mailboxes with InefficientFilter). We filter client-side instead.
	reqURL := fmt.Sprintf("%s?$top=10&$orderby=ReceivedDateTime+desc&$select=Subject,BodyPreview,ReceivedDateTime,From",
		outlookMailURL)

	req, err := http.NewRequest("GET", reqURL, nil)
	if err != nil {
		return "", fmt.Errorf("build request: %w", err)
	}
	req.Header.Set("Authorization", "Bearer "+accessToken)
	req.Header.Set("Content-Type", "application/json")

	client := &http.Client{Timeout: 15 * time.Second}
	resp, err := client.Do(req)
	if err != nil {
		return "", fmt.Errorf("REST API request: %w", err)
	}
	defer resp.Body.Close()

	body, _ := io.ReadAll(resp.Body)
	if resp.StatusCode != 200 {
		return "", fmt.Errorf("REST API returned %d: %s", resp.StatusCode, truncateStr(string(body), 200))
	}

	var result struct {
		Value []struct {
			Subject          string `json:"Subject"`
			BodyPreview      string `json:"BodyPreview"`
			ReceivedDateTime string `json:"ReceivedDateTime"`
			From             struct {
				EmailAddress struct {
					Address string `json:"Address"`
				} `json:"EmailAddress"`
			} `json:"From"`
		} `json:"value"`
	}
	if err := json.Unmarshal(body, &result); err != nil {
		return "", fmt.Errorf("parse REST API response: %w", err)
	}

	log.Printf("[outlook] REST API returned %d emails", len(result.Value))

	otpRegexp := regexp.MustCompile(`\b(\d{6})\b`)

	for _, email := range result.Value {
		subject := email.Subject
		preview := email.BodyPreview
		sender := strings.ToLower(email.From.EmailAddress.Address)

		log.Printf("[outlook] email: from=%s subject=%s", sender, subject)

		// Client-side sender filter: must be from OpenAI
		if !strings.Contains(sender, "openai") {
			continue
		}

		// Client-side time filter: skip emails before notBefore
		if !notBefore.IsZero() && email.ReceivedDateTime != "" {
			if t, err := time.Parse("2006-01-02T15:04:05Z", email.ReceivedDateTime); err == nil {
				if t.Before(notBefore) {
					log.Printf("[outlook] skipping old email (received %s, notBefore %s)", email.ReceivedDateTime, notBefore.UTC().Format(time.RFC3339))
					continue
				}
			}
		}

		// Try subject first ("Your ChatGPT code is 884584")
		if m := otpRegexp.FindStringSubmatch(subject); len(m) >= 2 {
			return m[1], nil
		}

		// Try body preview
		if m := otpRegexp.FindStringSubmatch(preview); len(m) >= 2 {
			return m[1], nil
		}
	}

	return "", nil
}

// exchangeMSToken exchanges a Microsoft refresh token for an access token.
// Note: do NOT pass scope parameter — the token already has outlook.office.com scopes.
func exchangeMSToken(clientID, refreshToken string) (string, error) {
	data := url.Values{
		"client_id":     {clientID},
		"grant_type":    {"refresh_token"},
		"refresh_token": {refreshToken},
	}

	client := &http.Client{Timeout: 15 * time.Second}
	resp, err := client.PostForm(msTokenURL, data)
	if err != nil {
		return "", fmt.Errorf("token request: %w", err)
	}
	defer resp.Body.Close()

	body, _ := io.ReadAll(resp.Body)
	if resp.StatusCode != 200 {
		return "", fmt.Errorf("token exchange failed (%d): %s", resp.StatusCode, truncateStr(string(body), 200))
	}

	var tokenResp struct {
		AccessToken string `json:"access_token"`
		Error       string `json:"error"`
		ErrorDesc   string `json:"error_description"`
	}
	if err := json.Unmarshal(body, &tokenResp); err != nil {
		return "", fmt.Errorf("parse token response: %w", err)
	}
	if tokenResp.AccessToken == "" {
		return "", fmt.Errorf("no access_token: error=%s desc=%s", tokenResp.Error, truncateStr(tokenResp.ErrorDesc, 100))
	}

	return tokenResp.AccessToken, nil
}

func truncateStr(s string, maxLen int) string {
	if len(s) <= maxLen {
		return s
	}
	return s[:maxLen-3] + "..."
}

// WaitForTeamAccountID polls for a Team workspace creation email and extracts account_id.
func (o *OutlookProvider) WaitForTeamAccountID(ctx context.Context, mailboxID string, timeout time.Duration, notBefore time.Time) (string, error) {
	idx, err := strconv.Atoi(mailboxID)
	if err != nil || idx < 0 || idx >= len(o.accounts) {
		return "", fmt.Errorf("invalid mailbox ID: %s", mailboxID)
	}
	acct := o.accounts[idx]

	if acct.RefreshToken == "" || acct.ClientID == "" {
		return "", fmt.Errorf("outlook account %s has no refresh_token/client_id", acct.Email)
	}

	deadline := time.Now().Add(timeout)
	ticker := time.NewTicker(5 * time.Second)
	defer ticker.Stop()

	teamURLRegexp := regexp.MustCompile(`account_id=([a-f0-9-]+)`)

	for attempt := 1; ; attempt++ {
		accessToken, tokenErr := exchangeMSToken(acct.ClientID, acct.RefreshToken)
		if tokenErr != nil {
			log.Printf("[outlook] token exchange error (attempt %d): %v", attempt, tokenErr)
			goto wait
		}

		{
			// Fetch recent emails without $filter (consumer mailboxes reject filters).
			// We filter client-side by sender and time.
			reqURL := fmt.Sprintf("%s?$top=10&$orderby=ReceivedDateTime+desc&$select=Subject,Body,ReceivedDateTime,From",
				outlookMailURL)

			req, _ := http.NewRequest("GET", reqURL, nil)
			req.Header.Set("Authorization", "Bearer "+accessToken)
			req.Header.Set("Content-Type", "application/json")
			req.Header.Set("Prefer", "outlook.body-content-type=\"text\"")

			httpClient := &http.Client{Timeout: 15 * time.Second}
			resp, reqErr := httpClient.Do(req)
			if reqErr != nil {
				log.Printf("[outlook] REST API error (attempt %d): %v", attempt, reqErr)
				goto wait
			}

			body, _ := io.ReadAll(resp.Body)
			resp.Body.Close()

			if resp.StatusCode != 200 {
				log.Printf("[outlook] REST API returned %d (attempt %d)", resp.StatusCode, attempt)
				goto wait
			}

			var result struct {
				Value []struct {
					Subject          string `json:"Subject"`
					ReceivedDateTime string `json:"ReceivedDateTime"`
					Body             struct {
						Content string `json:"Content"`
					} `json:"Body"`
					From struct {
						EmailAddress struct {
							Address string `json:"Address"`
						} `json:"EmailAddress"`
					} `json:"From"`
				} `json:"value"`
			}
			json.Unmarshal(body, &result)

			for _, email := range result.Value {
				sender := strings.ToLower(email.From.EmailAddress.Address)
				// Client-side sender filter
				if !strings.Contains(sender, "openai") {
					continue
				}
				// Client-side time filter
				if !notBefore.IsZero() && email.ReceivedDateTime != "" {
					if t, parseErr := time.Parse("2006-01-02T15:04:05Z", email.ReceivedDateTime); parseErr == nil {
						if t.Before(notBefore) {
							continue
						}
					}
				}
				subject := strings.ToLower(email.Subject)
				if strings.Contains(subject, "team") || strings.Contains(subject, "workspace") {
					if m := teamURLRegexp.FindStringSubmatch(email.Body.Content); len(m) >= 2 {
						log.Printf("[outlook] found team account_id: %s", m[1])
						return m[1], nil
					}
				}
			}
		}

	wait:
		if time.Now().After(deadline) {
			return "", fmt.Errorf("timeout waiting for team email after %s (%d attempts)", timeout, attempt)
		}

		select {
		case <-ctx.Done():
			return "", ctx.Err()
		case <-ticker.C:
		}
	}
}