图片兼容完善可达性校验与请求日志状态展示

desktop-tauri/i18n/i18n-src/en/requests.json

@@ -19,6 +19,7 @@
   "↻ 重试 {n} 次": "↻ Retried {n} times",
   "✅ 压缩完成": "✅ Vacuum complete",
   "「仅看进行中」开启时，状态固定为进行中": "When \"In progress only\" is on, the status is pinned to In progress",
+  "上游拒图补救": "Recovery after upstream image rejection",
   "上游模型繁忙，请求已排进上游队列；网关正按上游建议的时长等待后重发（不是登录态或额度问题）": "Upstream model is busy; the request is queued upstream and the Gateway will resend it after the wait suggested by the upstream (not a login-session or quota problem)",
   "上游没有返回积分消耗": "The upstream did not report points consumption",
   "上游状态码: {status}": "Upstream status code: {status}",
@@ -29,9 +30,14 @@
   "仅看进行中": "In progress only",
   "令牌": "Tokens",
   "体积": "Size",
+  "供应商": "Provider",
+  "供应商归属规则": "Provider attribution rules",
+  "保存中…": "Saving…",
+  "保存归属规则": "Save attribution rule",
   "保留 {count} 条请求记录 · 将清理上游调试报文 {upstream} 条、下游正文 {downstream} 条 · 数据库占用 {db}": "Keep {count} request records · Clear {upstream} upstream debug payloads and {downstream} downstream bodies · Database size {db}",
   "保留 {count} 条请求记录；将清理上游调试报文 {upstream} 条、下游正文 {downstream} 条。用量统计与报表不变。": "Keeps {count} request records; clears {upstream} upstream debug payloads and {downstream} downstream bodies. Usage statistics and reports remain unchanged.",
   "保留统计行与报表，只抹掉请求 / 响应正文。清理后详情弹窗的「预览对话」将无原文可看，统计数字分毫不动。": "Keeps stats rows and reports, and only wipes the request / response bodies. After clearing, \"Conversation preview\" in the detail dialog will have no raw content to show and the stats numbers stay exactly the same.",
+  "保留草稿并读取最新设置": "Keep draft and reload latest settings",
   "保留请求记录、用量统计与报表，同时清理上游调试报文和下游请求 / 响应正文。清理后「预览对话」与「原始报文」不再保留原文。有筛选时只清理命中请求的报文；无筛选时清空两侧全部报文。此操作无法恢复。": "Keeps request records, usage statistics, and reports while clearing upstream debug payloads and downstream request / response bodies. Conversation preview and Raw payload will no longer contain the original content. With filters active, only payloads for matching requests are cleared; without filters, all payloads on both sides are cleared. This cannot be undone.",
   "全部删除": "Delete all",
   "全部提供商": "All providers",
@@ -44,8 +50,10 @@
   "其它": "Other",
   "内容按原样保存，请求头中的凭据字段已替换为 [redacted]": "Content is saved as-is; credential fields in the request headers have been replaced with [redacted]",
   "内部重试": "Internal retry",
+  "切换视觉模型": "Vision model reroute",
   "切换路径：": "Switch path:",
   "列表数据已刷新，请重试（关闭后重新打开这条详情）": "List data has been refreshed; please try again (close and reopen this detail)",
+  "删除归属规则": "Delete attribution rule",
   "删除日志记录本身。当前有筛选时只删命中的条目，按天聚合报表不受影响；无筛选时清空全部并重置报表。此方式无法恢复。": "Deletes the log records themselves. With filters active, only matching entries are deleted and daily aggregate reports are unaffected; with no filters, everything is cleared and reports are reset. This cannot be undone.",
   "删除请求记录，并同步清理对应的上游调试报文与下游正文。有筛选时只删除命中的记录，按天聚合报表不受影响；无筛选时清空全部并重置报表。此操作无法恢复。": "Deletes request records together with their upstream debug payloads and downstream bodies. With filters active, only matching records are deleted and daily reports are preserved; without filters, all data is cleared and reports are reset. This cannot be undone.",
   "加载更早的请求": "",
@@ -59,23 +67,29 @@
   "原始报文": "Raw payload",
   "原模型": "Original model",
   "发给上游的模型": "",
+  "取消": "Cancel",
   "另有 {missing} 次尝试未记录明细（只保留最早的 {kept} 条）": "{missing} more attempts have no details recorded (only the earliest {kept} are kept)",
   "另有 {n} 个词命中": "{n} more words hit",
   "只显示正在转发中的请求（状态列按阶段显示：连接中 / 等待响应 / 响应中 / 重试中，用时列显示已用时）": "Only shows requests currently being forwarded (the status column shows the phase: Connecting / Awaiting response / Streaming / Retrying; the duration column shows elapsed time)",
   "只看失败": "Failed only",
   "只看成功": "Success only",
   "只看进行中": "In progress only",
   "命中了敏感词": "Sensitive word hit",
+  "命中依据：{header} · {pattern}": "Matched on: {header} · {pattern}",
   "命中的敏感词": "Sensitive words hit",
+  "命中规则：{provider} → {client}": "Matched rule: {provider} → {client}",
   "响应中": "Streaming",
   "响应体": "Response body",
   "响应头": "Response headers",
   "回收已删除数据占用的磁盘空间（checkpoint + VACUUM，后台执行）": "Reclaims the disk space used by deleted data (checkpoint + VACUUM, runs in the background)",
+  "图片兼容未应用": "Image compatibility not applied",
   "图片处理用时：{n} 秒": "Image processing time: {n} seconds",
+  "图片能力判据": "Image capability check",
   "在设置 → 通用里开启「调试模式」后，新发生的请求才会保存报文": "Raw payloads are only saved for requests made after \"Debug mode\" is enabled in Settings → General",
   "在路由页查看": "View on the Routing page",
   "在途 {n}": "",
   "处理中": "Processing",
+  "失败": "Failed",
   "失败请求不记录用量": "Usage is not recorded for failed requests",
   "失败（{status}）：{error}": "Failed ({status}): {error}",
   "失败：{error}": "Failed: {error}",
@@ -95,10 +109,13 @@
   "已删除 {count} 条": "{count} logs deleted",
   "已删除 {count} 条请求记录，并清理上游调试报文 {upstream} 条、下游正文 {downstream} 条。": "Deleted {count} request records and cleared {upstream} upstream debug payloads and {downstream} downstream bodies.",
   "已受理终止：上游连接已断开，列表稍后刷新为「请求已被手动终止」": "Abort accepted: the upstream connection is closed; the list will refresh to \"Request manually aborted\" shortly",
+  "已完成": "Completed",
   "已抹掉 {count} 条报文原文": "Raw payloads wiped for {count} logs",
   "已清理上游调试报文 {upstream} 条、下游正文 {downstream} 条；请求记录与报表已保留。": "Cleared {upstream} upstream debug payloads and {downstream} downstream bodies; request records and reports were preserved.",
   "已用时（请求仍在转发中）": "Elapsed time (request is still being forwarded)",
   "开启调试模式后仅记录新请求，历史对话无法补录；关闭不会删除已有记录。": "Debug mode records only new requests; past conversations cannot be recovered. Turning it off does not delete existing records.",
+  "归属在选路后确定：命中供应商规则用规则客户端，未命中用默认归属客户端，都没设置保留识别结果": "",
+  "归属客户端": "Client",
   "当前筛选下没有记录": "",
   "当前筛选范围": "the current filter results",
   "当前网关未提供上游 / 下游清理统计，请更新网关后再清理。": "This gateway does not provide upstream / downstream cleanup counts. Update the gateway before clearing data.",
@@ -119,6 +136,8 @@
   "提供商 / 账号": "Provider / Account",
   "提供商: {provider}": "Provider: {provider}",
   "提示": "Notice",
+  "搜索供应商": "Search providers",
+  "搜索客户端": "Search clients",
   "敏": "SW",
   "敏感词": "Sensitive words",
   "数据库占用包含主文件与 WAL 文件；清理后可点击「压缩数据库」回收空闲空间。": "Database size includes the main file and WAL file. After clearing data, use Vacuum database to reclaim unused space.",
@@ -129,21 +148,29 @@
   "时间": "Time",
   "暂无请求日志": "",
   "暂无请求日志，网关还没有转发过请求": "No request logs yet; the Gateway hasn't forwarded any requests",
+  "最终归属（记账时的客户端）": "Final attribution (the client recorded for accounting)",
   "最终承载：{picked}": "Finally served by: {picked}",
   "有重试": "",
+  "未应用": "Not applied",
+  "未应用原因": "Reason not applied",
+  "未改写，保留识别结果": "Not rewritten; the identified client is kept",
+  "未知状态": "Unknown status",
   "本次请求消耗 {credit} 积分": "Points consumed by this request: {credit}",
   "本轮尝试失败，正在退避等待或切换到下一个账号": "This attempt failed; backing off or switching to the next account",
   "来源客户端": "",
   "来源客户端：{client}": "Source client: {client}",
   "查看该请求的上游原始报文": "View the raw upstream payload for this request",
   "模型": "Model",
+  "模型图片能力未知，先发送原图": "Model image capability is unknown; the original image was sent first",
+  "模型已声明支持图片": "Model declares image support",
   "模型映射：{from} → {to}": "",
   "模型测试": "Model test",
   "模型管理页操作列那颗「测试」发起的请求：与真实请求同一条转发链路，但不计入报表统计": "Requests started by the \"Test\" button in the action column of the Model Management page: the same forwarding path as real requests, but not counted in reports",
   "正在加载请求日志…": "Loading request logs…",
   "正在统计…": "Calculating…",
   "正文超过单条上限，可能已被截断": "Body exceeds the per-record limit and may have been truncated",
   "没有符合筛选条件的请求": "No requests match the current filters",
+  "添加自定义头匹配规则…": "Add a custom header rule…",
   "清理": "Clear",
   "清理中…": "Clearing…",
   "清理失败：{message}": "Clear failed: {message}",
@@ -168,28 +195,37 @@
   "结果": "Result",
   "缓存读取: - / 命中率: -": "Cache reads: - / Hit rate: -",
   "缓存读取: {cache} / 命中率: {rate}": "Cache reads: {cache} / Hit rate: {rate}",
+  "编辑这条头匹配规则…": "Edit this header rule…",
   "耗时": "Duration",
   "脱敏命中：{words}": "",
+  "自定义 · {name}": "Custom · {name}",
   "被跳过：": "Skipped: ",
+  "设置识别规则": "Set identification rules",
   "识别到 User-Agent，但不属于已知的客户端类型：{ua}": "User-Agent present but not a known client type: {ua}",
+  "识别结果（请求头识别出的真实客户端）": "Identification result (the real client recognized from request headers)",
   "识图用量按实际视觉账号单独记录，不混入原模型的回答用量。": "Vision usage is recorded separately against the actual vision account, not mixed with the original model's answer usage.",
+  "该供应商已有归属规则，保存将覆盖它": "This provider already has an attribution rule; saving overwrites it",
   "详情": "Details",
   "详情内容": "Detail content",
   "请求体": "Request body",
   "请求图片": "Request image",
   "请求在任何账号接手之前就失败了": "The request failed before any account picked it up",
   "请求头": "Request headers",
+  "请求头识别为 {identified}，已改写归属为 {client}": "Identified as {identified} from request headers; attribution rewritten to {client}",
   "请求已受理，正在选路、取凭证、建立上游连接": "Request accepted; selecting a route, fetching credentials, and establishing the upstream connection",
   "请求日志列表": "Request Logs",
   "请求日志时间范围": "Request Logs time range",
   "请求日志表": "Request Logs table",
   "请求正在转发中，用时列显示的是已用时": "The request is being forwarded; the duration column shows elapsed time",
   "请求筛选": "Request filters",
   "请求详情": "Request details",
+  "请选择供应商": "Select a provider",
+  "请选择归属客户端": "Select a client",
   "读取请求日志失败，详见控制台": "Failed to load request logs; see the console for details",
   "路由回放": "Routing replay",
   "转发到上游的模型：{model}": "Model forwarded to the upstream: {model}",
   "轮次": "Round",
+  "辅助识图": "Assisted image reading",
   "辅助识图调用": "Assisted image reading call",
   "输入 Token": "",
   "输入 {input} · 输出 {output} · 总计 {total} · 缓存读 {cache}": "Input {input} · Output {output} · Total {total} · Cache reads {cache}",
@@ -201,10 +237,13 @@
   "这条记录命中了敏感词，但没有留下命中明细": "This record hit a sensitive word, but no hit details were kept",
   "这条记录的尝试明细未采集（该字段上线前的旧数据，或请求在转发前就失败）": "Attempt details were not collected for this record (old data from before this field existed, or the request failed before forwarding)",
   "这条请求没有保存这一段报文": "This request did not save this payload section",
+  "这条请求没有记录供应商（旧数据或未走到转发），不能从这里改归属": "No provider recorded for this request (legacy data or it never reached forwarding); attribution cannot be changed here",
   "进入「{phase}」阶段已持续 {time}": "In \"{phase}\" phase for {time}",
   "进行中": "In progress",
   "进行中…": "In progress…",
   "连接中": "Connecting",
+  "选择供应商": "Select provider",
+  "选择归属客户端": "Select client",
   "速度": "Speed",
   "重试 {n}": "Retry {n}",
   "重试 {n} 次": "",
@@ -220,6 +259,7 @@
   "首响：上游首帧到达的耗时": "TTFB: time to first byte from the upstream",
   "首响：上游首帧到达的耗时（请求仍在转发中）": "TTFB: time to first byte from the upstream (request is still being forwarded)",
   "首帧已到，上游内容正在下发": "First byte received; upstream content is streaming in",
+  "默认归属客户端": "Default attribution client",
   "（请求仍在转发中）": "(request is still being forwarded)",
   "（账号：{account}）": "(Account: {account})",
   "，{n}毫秒后重试": ", retry in {n}ms",

desktop-tauri/i18n/i18n-src/en/routing.json

@@ -1,7 +1,4 @@
 {
-  "不按余额排序": "Do not sort by balance",
-  "调整到期时间、可用余额及其他消耗规则的优先级": "Adjust the priority of expiry, available balance and other consumption rules",
-  "优先级：{rules}；前一项相同才比较后一项。": "Priority: {rules}. Later rules apply only when earlier rules tie.",
   "{account} · 账号设置": "{account} · Account settings",
   "{account} 参与本路由": "Include {account} in this route",
   "{account} 尝试失败后切换到此账号": "Switched to this account after {account} failed",
@@ -50,6 +47,7 @@
   "下一条请求": "Next request",
   "下一条请求将走": "The next request will go to",
   "不参与": "Not included",
+  "不按余额排序": "Do not sort by balance",
   "不按到期排序": "Do not sort by expiry",
   "与账号页共用全局消耗规则": "Shares global consumption rules with Accounts",
   "与账号页共用：{rules}。积分到期等规则直接影响网关实际选路。": "Shared with Accounts: {rules}. Rules such as credit expiry directly affect gateway routing.",
@@ -76,6 +74,7 @@
   "以下 {n} 个账号原本被全局路由排除。确认后会将它们设为禁用，所有路由都不能再选用；原有独立路由的参与名单保留。": "These {n} accounts were excluded from the global route. Confirming disables them for all routes. Existing participation lists in independent routes are preserved.",
   "优先使用队首账号，不可用时顺延。": "Prefer the first account in the queue and fall back when it is unavailable.",
   "优先尝试": "Try first",
+  "优先级：{rules}；前一项相同才比较后一项。": "Priority: {rules}. Later rules apply only when earlier rules tie.",
   "会话": "Sessions",
   "会话最长保持六小时无往来；可以随时在会话粘滞中解除。": "Bindings last for up to six hours of inactivity and can be released at any time under Session bindings.",
   "会话来源": "Session source",
@@ -279,6 +278,7 @@
   "完整详情": "Full details",
   "完整请求详情": "Full request details",
   "完整过程": "Full process",
+  "实时状态暂未更新，当前显示上次读取的结果：{error}": "Live status could not be refreshed. Showing the last retrieved results: {error}",
   "实际使用": "",
   "实际尝试失败": "Actual attempt failed",
   "实际模型": "Actual model",
@@ -582,7 +582,6 @@
   "没有需要恢复的提供商": "No providers need restoring",
   "沿用工具服务的视觉模型": "Use the tool service vision model",
   "流动：有请求在途": "Moving: requests in flight",
-  "实时状态暂未更新，当前显示上次读取的结果：{error}": "Live status could not be refreshed. Showing the last retrieved results: {error}",
   "消耗规则": "Consumption rules",
   "消耗规则已保存，但路由列表刷新失败：{error}": "Consumption rules were saved, but the routing list could not be refreshed: {error}",
   "添加到全局路由，新请求将按这里的顺序选择账号。": "Add to the global route. New requests select accounts in the order configured here.",
@@ -723,6 +722,7 @@
   "读图结果交回原模型继续回答": "",
   "谁参与及先后顺序": "Participation and priority",
   "调整全局消耗规则": "Edit global consumption rules",
+  "调整到期时间、可用余额及其他消耗规则的优先级": "Adjust the priority of expiry, available balance and other consumption rules",
   "调整后续请求的配置": "Configure subsequent requests",
   "调整模型顺序": "Reorder models",
   "调整积分到期远近及其他消耗规则": "Adjust credit expiry order and other consumption rules",

desktop-tauri/i18n/i18n-src/en/settings.json

@@ -252,6 +252,7 @@
   "允许浏览器里的页面跨来源调用网关（/v1/*）": "Allow pages in a browser to call the gateway cross-origin (/v1/*)",
   "允许的域名": "Allowed hostnames",
   "先发送原图，明确拒图后补救": "Send original images first; adapt only after an explicit rejection",
+  "沿用全局设置": "Use global setting",
   "先在当前路由范围内找可用的视觉模型，这次请求由视觉模型直接回答，不回原模型；{fallback}": "",
   "先把图片交给视觉模型读一遍，再让原模型结合读到的内容回答。": "The vision model reads the image first, then the original model answers using what was read.",
   "全局配置": "Global Configuration",