sub-store/internal/proxy/format.go

package proxy

import (
	"encoding/json"
	"regexp"
	"strings"

	"github.com/peterqiu0516/sub-store/internal/model"
	"github.com/peterqiu0516/sub-store/internal/util"
	"gopkg.in/yaml.v3"
)

var (
	// reURIScheme matches a URI scheme like "vless://", "ss://", etc.
	reURIScheme = regexp.MustCompile(`(?im)^[a-z][a-z0-9+.-]*://`)
	// reYamlKeys matches YAML top-level keys: proxies, proxy-groups, rules
	reYamlKeys = regexp.MustCompile(`(?m)^\s*(proxies|proxy-groups|rules)\s*:`)
	// reJSONStart matches the start of a JSON array or object
	reJSONStart = regexp.MustCompile(`^\s*[\[{]`)
	// reQxConfig matches QX-style config lines: "shadowsocks = ...", "vmess = ...", etc.
	reQxConfig = regexp.MustCompile(`(?im)^\s*(shadowsocks|vmess|vless|trojan|http|socks5|anytls)\s*=`)
	// reNamedConfig matches Surge/Loon named config lines
	reNamedConfig = regexp.MustCompile(`(?im)^\s*[^=\n]{1,80}\s*=\s*(ss|shadowsocks|ssr|vmess|vless|trojan|http|https|socks5|socks5-tls|hysteria2|hysteria|anytls|tuic|tuic-v5)\s*,`)
	// reCommentOrSection matches lines to skip: comments (#, ;) and section headers ([...])
	reCommentOrSection = regexp.MustCompile(`^\s*(#|;|\[[^\]]+\])`)
)

// ParseProxies is the main entry point for parsing subscription content.
// It trims the input, detects JSON/YAML/URI-line format, and dispatches accordingly.
func ParseProxies(raw string) []model.ProxyNode {
	text := strings.TrimSpace(raw)
	if text == "" {
		return nil
	}
	if reJSONStart.MatchString(text) {
		return ParseJsonProxies(text)
	}
	if reYamlKeys.MatchString(text) {
		return ParseYamlProxies(text)
	}
	return ParseProxyLines(text)
}

// DecodeMaybeBase64 tries structured detection first; if the input doesn't
// look like a structured subscription, attempts base64 decode.
func DecodeMaybeBase64(raw string) string {
	text := strings.TrimSpace(raw)
	if LooksLikeStructuredSubscription(text) {
		return raw
	}
	// Try base64 decode (strip all whitespace first)
	cleaned := regexp.MustCompile(`\s+`).ReplaceAllString(text, "")
	decoded, err := util.DecodeBase64Auto(cleaned)
	if err != nil {
		return raw
	}
	if LooksLikeStructuredSubscription(strings.TrimSpace(decoded)) {
		return decoded
	}
	return raw
}

// LooksLikeStructuredSubscription checks if the text looks like structured
// subscription content (URI scheme, YAML keys, JSON, or client config lines).
func LooksLikeStructuredSubscription(text string) bool {
	return reURIScheme.MatchString(text) ||
		reYamlKeys.MatchString(text) ||
		reJSONStart.MatchString(text) ||
		reQxConfig.MatchString(text) ||
		reNamedConfig.MatchString(text)
}

// ParseJsonProxies parses a JSON array of proxies or {"proxies": [...]}.
func ParseJsonProxies(raw string) []model.ProxyNode {
	var payload any
	if err := json.Unmarshal([]byte(raw), &payload); err != nil {
		return nil
	}

	var rawList []any
	switch v := payload.(type) {
	case []any:
		rawList = v
	case map[string]any:
		if arr, ok := v["proxies"].([]any); ok {
			rawList = arr
		}
	default:
		return nil
	}

	var result []model.ProxyNode
	for _, item := range rawList {
		if m, ok := item.(map[string]any); ok {
			normalized := NormalizeProxy(m)
			if IsProxyNode(normalized) {
				result = append(result, normalized)
			}
		}
	}
	return result
}

// ParseYamlProxies parses a YAML document with a "proxies:" key.
func ParseYamlProxies(raw string) []model.ProxyNode {
	var payload map[string]any
	if err := yaml.Unmarshal([]byte(raw), &payload); err != nil {
		return nil
	}

	rawList, ok := payload["proxies"].([]any)
	if !ok {
		return nil
	}

	var result []model.ProxyNode
	for _, item := range rawList {
		if m, ok := item.(map[string]any); ok {
			normalized := NormalizeProxy(m)
			if IsProxyNode(normalized) {
				result = append(result, normalized)
			}
		}
	}
	return result
}

// ParseProxyLines splits raw text into lines, skips comments/blanks/section headers,
// and parses each line as a URI or client config line.
//
// Literal "\n" / "\r\n" sequences (backslash + n, as produced by some frontends
// stuffing multi-line content into a single string) are normalized to real
// newlines before splitting so each link is parsed as its own line.
func ParseProxyLines(raw string) []model.ProxyNode {
	if strings.Contains(raw, `\n`) {
		raw = strings.ReplaceAll(raw, `\r\n`, "\n")
		raw = strings.ReplaceAll(raw, `\n`, "\n")
	}
	lines := strings.Split(raw, "\n")
	var result []model.ProxyNode
	index := 0
	for _, line := range lines {
		line = strings.TrimSpace(line)
		if line == "" {
			continue
		}
		if strings.HasPrefix(line, "#") || strings.HasPrefix(line, ";") {
			continue
		}
		if reCommentOrSection.MatchString(line) {
			continue
		}

		var node model.ProxyNode
		if node = ParseProxyUri(line, index); node == nil {
			node = ParseClientProxyLine(line, index)
		}
		if node != nil {
			result = append(result, node)
			index++
		}
	}
	return result
}