面向零基础用户 · 完整操作手册

DMS.app 用户手册

DMS 是一款运行在您自己电脑上的文档与照片管理系统。它把您硬盘上的真实文件，组织成一棵可以随意展开、搜索的文件夹树，全部通过浏览器操作，不上传到任何云端服务器。本手册假定您此前从未使用过 DMS，会从"如何安装"开始，逐步讲到每一项功能的具体点击步骤。

文档
用户手册 · 2026-10-23
适用对象
完全没用过 DMS 的新用户
支持平台
macOS · Windows
章节
01–17 + 常见问题
Mac 版下载 DMS-Mac.zip Windows 版下载 DMS-Windows.zip 查看全部历史版本
下载中文手册 PDF DMS_User_Manual-261023-CN.pdf 下载英文手册 PDF DMS_User_Manual-261023-EN.pdf
01

安装与首次启动

DMS 不需要注册账号，也不需要安装到系统目录——下载后就是一个可以直接双击运行的程序。您的所有文件，从第一步起，就只会保存在您自己指定的硬盘位置上。

1.1 下载

在本页顶部选择您电脑对应的版本下载：Mac 电脑选"Mac 版下载"，Windows 电脑选"Windows 版下载"。下载得到的是一个 .zip 压缩包。

  1. 下载完成后，在"下载"文件夹中找到该 .zip 文件，双击解压。
  2. 解压出的文件夹中会看到 DMS.app（Mac）或 DMS.exe（Windows）。建议把它拖到"应用程序"（Mac）或桌面上您习惯的位置，方便以后打开。

1.2 首次启动

  1. 双击 DMS.app 或 DMS.exe。
  2. 会先弹出一个很小的状态窗口，标题是 "QC Document Management System (QCDMS)"，里面显示当前状态和一个网址链接。不要关闭这个窗口——它就是 DMS 的后台服务，关掉它整个程序就会退出。
  3. 大约一秒后，系统会自动打开您的默认浏览器（推荐 Chrome），显示 DMS 的网页界面。如果没有自动打开，可以点击状态窗口里那行带下划线的网址。
提示 这个小状态窗口右下角有一个 "Quit DMS" 按钮，用来彻底退出程序；左下角的 "Remote Upload" 按钮会在第 13 章"手机与远程访问"中用到，现在可以先不管它。双击 DMS.exe 如果没反应、或 Windows 弹出安全警告，见下方 FAQ 中的"Windows 已保护你的电脑"一条。

1.3 设置存储位置

首次启动时，浏览器会显示"设置存储位置"页面，要求您指定一个文件夹，DMS 以后上传的所有 PDF、图片和索引数据都会实实在在地存放在这个文件夹里。

  1. 在"存储路径"输入框中，粘贴或手动输入一个文件夹路径。如果这个文件夹还不存在，DMS 会自动创建它。例如：
    Mac：/Users/yourname/Documents/DMS_Storage
    Windows：C:\Users\yourname\Documents\DMS_Storage
  2. 点击 "保存并继续"。
  3. 系统初始化后会自动进入主界面。
重要提醒 在 Mac 上，请不要把存储路径设置在 iCloud Drive 内部（例如已开启"桌面与文稿同步"的桌面或文稿文件夹）——iCloud 会把不常用的文件自动"云化"、从本地移走，可能导致 DMS 读不到文件。如果想要免费云备份，请改用 OneDrive 或 Dropbox 同步的文件夹（详见第 14 章）。
02

认识主界面

进入主界面后，花两分钟认清下面这几块区域，后面所有章节都会反复用到它们。

2.1 顶部工具栏

  • 中文 / EN 切换按钮 — 点击可随时切换整个应用界面的语言，不影响已保存的数据。
  • 项目 按钮 — 点开是一个下拉菜单，包含导出/导入备份、全局设置、密码保护等，第 14、15、16 章会详细介绍。
  • 当前存储路径按钮 — 显示正在使用的文件夹路径；点击可更改到另一个文件夹（即切换到另一个"项目"）。
  • 如果当前项目设置了项目密码，这里还会显示一个带锁图标的项目名称标签。

2.2 四个标签页

工具栏下方是四个标签页，点击切换：

树形默认视图。左边是文件夹树，右边是选中文件夹里的文档——本手册第 3–5 章主要在这里操作。
搜索关键词搜索，或切换到"AI 问答"整库提问——见第 6、8 章。
数据手册把多份文档合并生成一份带封面的 PDF 手册——见第 11 章。
人脸按人物浏览已标记的照片——见第 10 章。

2.3 树形视图的左右结构

在"树形"标签下：

  • 左侧：文件夹树，可展开/折叠，右键点击任意文件夹弹出操作菜单。树的正上方还有一排快捷按钮：新建文件夹、上传文档、重命名、删除、文本树（显示当前选中文件夹的结构）、趋势图。
  • 右侧：选中文件夹里的文档列表。
几乎所有输入框都能语音输入 文件夹/文档名称、序列号、登录信息、描述、搜索框、AI 提问框……几乎每一处文字输入框旁边都有一个麦克风图标，点一下对着话筒说话，识别结果就会自动追加到光标处，再点一下停止；可以在"中文"和"EN"之间切换识别语言，完全在浏览器里完成，不需要配置任何 AI Key（详见 4.5 节）。
03

创建与整理文件夹

文件夹树是 DMS 的骨架，建议先花几分钟按您实际的分类方式（比如按设备、按客户、按家庭成员）搭好大致框架，再往里面添加文档。

3.1 新建文件夹

  1. 在左侧树中，先点击选中一个"父"文件夹（新文件夹会建在它下面）。
  2. 右键点击该文件夹，在弹出菜单中选择 "新建子文件夹"；或者直接点击树上方工具栏里的 "新建文件夹" 按钮。
  3. 输入文件夹名称，按 Enter 确认。

3.2 重命名与删除

  1. 重命名：右键文件夹 →"重命名"，或双击文件夹名称，改好后按 Enter。磁盘上的实际文件夹会同步改名。
  2. 删除：右键文件夹 →"删除文件夹"，确认后该文件夹从树里消失，其中的文档不会立即从磁盘抹除，而是被移动到一个叫 Not Show in Tree 的隐藏文件夹里暂存。
想找回被删文件夹里的文档？ 在左侧树的底部找到 Not Show in Tree 节点展开，里面的文档可以重新"链接"或移动回任意文件夹（见 5.3 节）；只有在这个隐藏节点里再次点删除，文件才会真正从磁盘上永久清除、无法撤销。但文件夹被删除后就不在树里了，这个动作本身不会撤销——如果只是想临时挪走、以后可能还要用，建议改用拖拽把它移到别处，而不是删除。

3.3 拖拽移动、排序与合并

  1. 按住鼠标左键拖动一个文件夹。
  2. 拖到目标文件夹正上方（高亮显示）再松手 → 移动到该文件夹内部，成为其子文件夹。
  3. 拖到目标文件夹的上边缘或下边缘再松手 → 移动为该文件夹的前一个/后一个同级文件夹（仅调整顺序，不改变所属关系）。
自动合并 如果拖入的目标文件夹内已经存在一个同名子文件夹，DMS 会自动把两者合并，源文件夹里的文档和子文件夹会全部并入目标文件夹，不会出现重复的同名文件夹。

3.4 文件夹编号（SN）

可以给每个文件夹设置一个编号（如 A-001），方便与您原有的资产/物料编号体系对应；子文件夹和其中的文档会默认"继承"父文件夹的编号，除非它们自己单独设置了编号。

  1. 在右侧的"文件夹属性"/详情面板中找到"序列号（SN）"字段。
  2. 输入编号并保存。编号会显示在文档卡片上，方便快速识别归属。

3.5 从一份清单批量建立整棵文件夹树

如果您已经有一份文件夹结构的清单（比如物料清单 BOM、组织架构表），不需要一个个手动新建——可以用一份简单的 CSV（逗号分隔）文本一次性批量导入。

第 1 列文件夹名。必填，这一行要创建的文件夹叫什么。
第 2 列父文件夹名，即这个文件夹挂在哪个已出现过的文件夹下面；留空表示挂在根节点下。文件的第一行本身就是根文件夹，这一列必须留空。
第 3 列说明文字，可选，整列都可以省略。填写时会成为该文件夹的初始描述。
  1. 点击顶部工具栏的 "项目" → "从层级文件创建文件夹树"。
  2. 粘贴或上传您的 CSV 文本，例如：
    总装配
    泵壳组件,总装配
    转子组件,总装配
    螺栓,转子组件
  3. 点击"验证"检查解析结果。如果同一个文件夹名在文件里重复出现（例如不同总成下各有一个同名的"螺栓"），默认会报错；勾选"自动为重复文件夹名添加编号"后重新验证，DMS 会把这些重名文件夹依次编号为"螺栓 01""螺栓 02"……全部导入成功。
  4. 确认预览无误后点击"创建节点树"，所有文件夹立即出现在树里。
从物料清单软件导出？ 带有"Level 1"这类分级标题行、或第一行是"文件夹名, 父文件夹名, 说明"这样的列标题行，DMS 都会自动识别并跳过，不用手动删除。

3.6 在描述中直接粘贴截图

文件夹和文档的描述框都支持直接粘贴图片：截一张屏（或复制一张图片），把光标放进描述框，按 Ctrl+V（Mac：Cmd+V），截图会内嵌显示在光标所在位置，与文字混排。

  • 粘贴的截图会作为一份真实的图片文档保存在当前文件夹中（会出现在文档列表里，删除后进入“Not Show in Tree”，可以找回）；描述里只是引用它。
  • 在描述框里粘贴图片，不会再同时触发 4.3 的“粘贴截图为新文档”；粘贴网页上复制的带格式文字时，只保留纯文字。
  • AI 整理润色、检查错别字只处理文字部分，已插入的截图会保留；导出为 .txt / PDF 或在搜索结果中显示时，截图位置显示为“[截图]”。旧版本保存的纯文字描述会自动兼容。
04

添加文档

支持 PDF、JPG、PNG、HEIC、TIFF、GIF、WEBP、BMP 等图片格式，以及 TXT 等文本文件；原则上任何格式的文件都可以上传保存，只是并非所有格式都能在浏览器里直接预览。单个文件最大 4GB。

4.1 拖放上传（最快的方式）

  1. 在左侧树中点击选中目标文件夹。
  2. 直接把文件从桌面或访达/资源管理器窗口，拖到右侧的文档区域。
  3. 松开鼠标，文件自动上传，完成后出现在文档列表中。

4.2 通过按钮上传

  1. 选中目标文件夹后，点击树上方工具栏的 "上传文档" 按钮，或右键文件夹 → "上传文档"。
  2. 在弹出的文件选择对话框中选择文件（可按住 Cmd/Ctrl 多选），点击"打开"。

4.3 从剪贴板直接粘贴截图

不需要先截图保存到桌面再上传：截图后直接在 DMS 页面里按 Cmd+V（Mac）或 Ctrl+V（Windows）即可粘贴进当前文件夹。

复制的文字也可以直接粘贴：如果剪贴板里是一段文字而不是图片，DMS 会把它排版成一份 PDF，命名为 Pasted text 日期 时间.pdf，存入当前文件夹。

4.4 一次上传整个文件夹，三种归档方式任选

如果您电脑上已经有一整套按文件夹分类好的资料（比如一个项目的完整目录、或手机相册导出的整个相册文件夹），不用一个个拖文件，可以整个文件夹一次性上传。

  1. 选中目标文件夹后，点击工具栏的 "上传文件夹"，弹出"上传多个文件夹"对话框。
  2. 点击"+ 添加文件夹"选择电脑上的一个文件夹；可以重复这一步一次排进多个文件夹，逐个列在上方，点右侧的 × 可以移除。
  3. 三种归档方式二选一：
    • 按年/月文件夹上传（默认）：每个所选文件夹（含其子文件夹）里的所有文件都按拍摄/归档日期自动归入"年/月"两级子文件夹，不保留原来的子文件夹结构——最适合相册这类"只关心日期、不关心原来怎么分类"的资料。
    • 按来源文件夹结构上传：子文件夹原样在此处重建，不按日期生成年/月文件夹——适合已经整理好、想保留原目录结构的资料。
    • 在当前文件夹下重建所选文件夹：与上一种的区别是，所选文件夹本身也会变成当前文件夹下的一个新子文件夹（前两种只上传其内容，不建这一层），整棵结构原样落地，本地磁盘上也会同步建好对应目录。
  4. 如果需要，在"在文件名中添加文字（可选）"里填一段文字——有日期时插在日期后面，没有日期时加在文件名最前面；旁边也有麦克风按钮可以直接口述。留空则不加任何文字。
  5. 点击右下角"上传"完成。

4.5 描述、语音输入与 AI 整理润色

每一份文档和每张照片都可以添加一段自由文字的"描述"，这段文字会被保存进搜索索引，全文可搜索。

  • 语音输入：点击描述框旁边的语音输入按钮，对着电脑说话即可，识别出的文字会自动追加到描述中（Chrome / Edge 浏览器支持效果最佳），打字与口述可以混用；语音识别的语言可以在"中"和"EN"之间切换，中英文内容都能口述录入。
  • AI 整理润色（需先按第 8 章配置 AI）：口述或随手打的文字往往夹杂口语、错别字或病句，点击 "用 AI 整理润色"，可选择"整理结构""润色文字"或"精简概括"三种方式之一；AI 处理后的结果会先给您预览确认，满意后才会替换原文，不会自动覆盖。
  • 检查错别字：点击 "检查错别字"，AI 会专门检查描述里的错别字、标点和成语误用，逐条列出供您参考修改。
  • 保存方式：写好后点击 "保存并可搜索" 只把这段文字写入索引；如果想把这段描述单独存成一份可以像其他文档一样打开、下载、分享的 .txt 文件，点击 "另存为新文档"（已经存过一次后，这里会变成"更新描述文档"，方便持续修改同一份）。
用法举例：整理一份课程资料 比如管理某门课程的讲义、习题、往年考卷，可以给每份资料都写一段说明（口述+AI 整理润色即可，不用逐字手打），并都"另存为新文档"；这样每份说明本身也成了库里的独立文档，随时可单独打开、检索。如果还想要一份汇总说明材料用于分享或自行存档，可以用第 11 章的"数据手册"功能把这些说明文档一次性合并成一份 PDF——它不会移动或影响任何原始文档，每份说明依旧各自保留、可单独查阅（第 5.4 节的快速合并则会把参与合并的原文档移入回收文件夹，这种场景不建议使用）。

4.6 照片和视频自动按年月归档

把某个文件夹设为"照片根目录"（第 10 章）后，上传的照片会自动读取拍摄日期，在这个根目录下自动创建"2026"→"08"这样的年/月子文件夹并归入对应位置；如果照片带 GPS 定位信息，拍摄地点也会一并识别出来。iPhone 拍摄的 HEIC 格式照片无需手动转换即可直接预览。

视频文件现在同样会按年月自动归档：MP4、MOV 等格式会直接读取视频文件内部记录的实际拍摄时间；其他视频格式没有这类信息，则改用文件本身的"修改日期"来归档。

4.7 ZIP 批量导入

如果您有大量文档需要上传，并且文件名里已经能看出该归到哪个文件夹，可以按命名规则打包成一个 ZIP，一次性自动分配，文件夹必须已经在树里存在（不会像 3.5 节的层级导入那样顺带新建文件夹）。

  1. 把每个文件按 文件夹名#说明.pdf 的格式命名（例如 房屋保险#2026年保单.pdf 对应名为"房屋保险"的文件夹，# 号后面的部分可以随意填写），打包成一个 ZIP。
  2. 点击 "项目" → "从 ZIP 批量导入文档"，选择该 ZIP 文件。
  3. 系统会显示每个文件将被分配到哪个文件夹、哪些因为名称对不上而未能匹配。
树里有同名文件夹怎么办？ 如果同一个文件夹名在树里不止一处（比如"转子组件"和"泵壳组件"下各有一个同名的"螺栓"），只用文件夹名会匹配到排在前面的那个。这时可以在 # 号前改用一段"/"分隔的路径来精确指定，比如 转子组件/螺栓#清单.pdf，就能准确落到"转子组件"下的那个"螺栓"，而不是另一个同名文件夹。

4.8 重复文件

上传时 DMS 会自动检测内容完全相同的文件；如果发现重复，会提示您选择：跳过、把它链接到当前文件夹（不占用额外磁盘空间），或者仍然作为新文件上传。

4.9 自动识别文档日期

PDF 等文档上传后，DMS 会自动扫描文档首页的文字，尝试识别出这份文档自己的日期（比如报告日期、签发日期），省得您每份都手动填写。识别时会优先寻找带明确标签的日期（如"签发日期""检验日期""Issue Date"等），找不到时才退而使用文档里第一个看起来像日期的文字。

务必人工核对 如果一份文档里同时出现多个日期（比如出生日期、签发日期、有效期），自动识别未必总能挑中您真正想要的那一个。自动识别出的日期会用黄色提示标出、并注明"根据文档内容中的'……'自动识别"，在您确认或修改之前会一直保持这个提示；点击文档卡片上的"日期"按钮即可核对并保存正确的日期。

4.10 批量转换为 PDF

需要把一批照片或已有的 PDF 统一转成规范命名的独立 PDF 文件时，不用逐份另存。

  1. 勾选要转换的图片和/或 PDF 文档，点击工具栏的 "转换为 PDF"。
  2. 弹出的对话框会列出已勾选的文档，确认后点击"将 N 份文档转换为 PDF"。
  3. 每份文档被分别转换为独立的 PDF（能确定日期时文件名以"YYYY-MM-DD-"开头），原始文件则移入 Not Show in Tree 回收文件夹（见 3.2 节，可以找回）。

4.11 磁盘文件名与结构快照

这两点不需要手动操作，但了解一下有助于日后用系统自带的文件管理器查看文件、或排查问题：

  • 硬盘上保存的实际文件名是"名字在前、编号在后"（例如 体检报告+DOC-20260920-xxx.pdf），在 Finder / 资源管理器里翻文件夹时一眼就能看出是什么文档，不用先点开才知道。
  • 每次修改项目的文件夹结构（增删文件夹、移动、批量操作等）之前，DMS 都会自动先保存一份当时的快照；万一某次改动把结构弄乱了，可以借助该快照恢复——这是一层后台自动运行的安全网，正常使用不会看到任何界面。

4.12 照片按拍摄地点自动分文件夹

带定位信息的照片，除了按年/月分文件夹（见 4.6），现在还可以在月文件夹下再按拍摄地点分一层：同一“半径”范围内的照片放进同一个地点文件夹，文件夹自动以地名命名（例如“Boston, Massachusetts”）。

  1. 上传照片时，会先弹出确认窗口，勾选“照片按拍摄地点分子文件夹（在月文件夹内）”。
  2. 输入“半径”，选择“英里”或“公里”。第一次默认 10 英里，之后会记住您上次用的值，不想改直接点“上传”；取消勾选则只按年/月归档。
  3. 没有定位信息的照片留在月文件夹里；手机上传也按同样规则分。地名是上传完成后在后台查询的，不会拖慢上传；查不到时文件夹先叫“Location 1”“Location 2”……

分得不满意？半径太大会把不同地方混在一起，太小又会把同一地方拆开。在照片所在的文件夹上右键，选择“按地点重新分组照片…”，换一个半径，窗口会先预览“会分成几个地点文件夹”，满意后点“重新分组”，可以反复调整。这个功能也能把以前上传、还没按地点分过的照片整理一遍；被手动改过名称的地点文件夹会被重新命名，其他文件夹和文档不受影响。

4.13 扫描条码（识别图片中的条码/二维码）

照片里拍到条码或二维码（产品标签、设备铭牌等）时，DMS 可以直接把编码内容识别出来，无需手动抄录。

  1. 打开该图片（或含条码图片的 PDF）的查看器。
  2. 点击顶部的“扫描条码”。
  3. 识别成功后，顶部会显示编码类型与内容（例如“CODE128：6901234567892”）。
  4. 结果自动存为该文档的“Barcode”参数，可在元数据区域查看或编辑；同一张照片有多个条码时，依次存为 Barcode、Barcode2、Barcode3……
说明 本功能完全离线运行（基于开源 zbar 库），不上传任何数据，也不需要 AI 密钥；但要求条码清晰、无遮挡，四周留有空白边框。手机截图分享的二维码（尤其是中间嵌有图标、或背景色直接顶到条码边缘的，如微信“扫码入群”图片）通常无法识别，这是已知局限，并非软件故障。

4.14 上传设备记录（AN^SN^PN 命名）

适合设备资料管理：一份记录按“装配号（AN）^序列号（SN）^零件名称（PN）^原文件名”命名，并直接存入名称等于该装配号的文件夹，之后单看文件名就知道它属于哪台设备。

  1. 点击树形标签工具栏中的“上传设备记录”，弹出可拖动、可调整大小的窗口。
  2. 在左侧选择 AN（装配号 / 文件夹）——可先在筛选框里输入文字缩小范围——再按需填写 SN、PN（均可留空）。
  3. 点击“选择文件…”选一个文件，窗口会显示“将保存为：”的新文件名，右侧可预览图片、PDF 或文本。
  4. 确认无误后点击“上传”。AN / SN / PN 会保留，可直接选下一个文件。

一次传很多份时，点击“按文件名上传”，选择多个文件名已写成 AN^SN^PN.扩展名 的文件（例如 A100^SN001^轴承.pdf），DMS 自动拆出字段、匹配对应文件夹，逐行显示状态（已上传 / 已存在跳过 / 出错原因），点击“上传 N 个文件”完成。

说明 用“^”作分隔符，是因为工单号和零件名称本身常带连字符。AN 必须与某个文件夹名称一致，否则该行提示找不到文件夹。

4.15 质检审核（QC Review）与审核报告

质检审核用于对一批检验报告逐份填写审核意见，并在文件夹层面汇总成一份可交付的 PDF。结构是：每个“检验项目”一个文件夹，里面放该项目的检验报告；每个文件各有一份审核表；整个文件夹另有一条“总体意见”。

  1. 上传记录：点击树形标签工具栏的“上传质检审核”，用法同 4.14，字段为检验项目（文件夹）、SN、设备描述、WO/PO（工单号 / 采购订单号）；也可“按文件名上传”，文件名格式 检验项目^SN^设备描述^WO/PO。每份文件都会自动建立审核表并预先填好这些字段。
  2. 填写单个审核：打开文档查看器，点击“质检审核”（尚未填写时显示“添加质检审核”）。窗口自动变宽，左半边显示文件，右半边是审核表，分隔线可拖动。表中包含检验项目、SN、设备描述、WO/PO、审核日期、审核人、观察结果、缓解措施，以及审核结果（Pass / Fail / Conditional / Pending）。修改会自动保存，这些字段也可以被搜索到。
  3. 总体意见与报告：选中检验项目文件夹，在其质检审核区域填写“总体意见”，再点击“质检审核报告”。在对话框里选择要附在报告后面的实际检验报告（全选 / 全不选 / 逐份勾选），DMS 生成一份 PDF：先是总体意见和各文件的审核汇总页，后面附上所选检验报告；Word、PowerPoint 会先转成 PDF（需要电脑上安装 LibreOffice）。报告存在该文件夹里，再次生成会覆盖原报告，不会越存越多。

4.16 AI 识别服务日期

化验单、体检报告里往往同时有出生日期和检查日期，上传时的自动日期识别分不清二者。“AI 识别服务日期”用 Gemini 重新读取已有文档，专门找“服务 / 检验 / 检查发生的日期”，并明确排除出生日期。

  • 文档列表工具栏：点击“AI 识别服务日期”，勾选若干 PDF / 图片后开始。
  • 文件夹多选模式：勾选若干文件夹后点击同一按钮，会处理这些文件夹及所有子文件夹内的每份 PDF / 图片。
  • 识别结束后会弹出与上传日期核对相同的确认对话框，您核对后才会写入日期，不会静默覆盖；没识别到服务日期的文档只在提示里统计数量。需要先配置 Gemini 密钥（第 8 章）。
05

查看、下载与整理文档

5.1 预览文档

  1. 在文档列表中点击一份文档，即可在页面内直接预览：PDF 支持翻页和缩放，还提供"文档内查找"，即使 OCR 文字有一点小错误也能模糊匹配到并跳转高亮；图片按原尺寸显示；文本文件直接显示内容。
  2. 如果想用电脑上安装的其他软件打开（比如用 Word 打开一份 .docx），点击"用原生应用打开"，DMS 会调用系统默认程序。

5.2 重命名与下载

  • 在文档卡片上点击"重命名"，可同时修改磁盘上的文件名和显示名称。
  • 勾选一份或多份文档后点击"下载"，可以把它们复制到您指定的文件夹（默认是系统的"下载"文件夹），完成后能直接打开该文件夹查看。

5.3 把同一份文档关联到多个文件夹

同一份文档可以出现在多个文件夹中，而不会在磁盘上多占一份空间，这称为"链接"。

  1. 打开目标文件夹，点击"链接现有文档"。
  2. 在弹出的窗口中搜索或浏览要链接的文档；也可以直接把结果拖拽到左侧的文件夹上，松手即完成关联，不用再逐个点击确认。
  3. 该文档随即出现在当前文件夹中，同时仍保留在原文件夹里；文档卡片上会显示"已链接至 N 个节点"。

“链接现有文档”窗口也升级了：拖动标题栏可以移动窗口，拖动两栏之间的分隔线可以调整宽度；左侧文件夹树有“全部展开 / 全部折叠”按钮，展开后直接勾选要链接的文档；右侧的“目标文件夹”既可以勾选，也可以把文档直接拖到目标文件夹上。窗口里还有“删除”按钮，会把选中的文档从磁盘永久删除，无法撤销，请谨慎使用。

关于删除 删除一个链接只是解除这个文件夹对该文档的引用，不会删除实际文件；只有当所有引用都被删除时，文件才会真正从磁盘上移除（并进入下方的回收区）。

5.4 合并为一份 PDF

可以把选中的多份扫描件或文件快速合并为一个 PDF，合并后的文件会归档到您指定的文件夹里。含有照片时，每张照片会先被压缩（长边不超过 900 像素、JPEG 85% 品质）再嵌入，肉眼几乎看不出画质差别，但整份文件体积能缩小到原来的百分之一左右——例如一份含 136 张照片的合同书，PDF 体积可以从 2.3 GB 降到约 18 MB。快速合并会跳过正式封面页，仅靠每份文档各自的书签导航；即便原始页面是横版，也会自动统一转正为竖版。

参与合并的原始文档会被移出当前文件夹 合并完成后，被合并的原始文档会从当前文件夹中移除，并移入"Not Show in Tree"回收文件夹（不会被立即彻底删除，可以去那里找回或永久清除）。如果希望合集之外还保留一份可单独查看的原始文档，建议改用第 11 章的"数据手册"功能——它会生成同样的合集，但不会动原始文档。

5.5 打开一份文档就会一直高亮

打开过的文档（无论是在内置预览里，还是用原生应用打开），会在所在文件夹的列表中持续高亮显示，方便您稍后一眼找回。

5.6 删除文档

点击文档卡片上的删除按钮，确认后：如果该文档被多个文件夹引用，只会解除当前文件夹的引用；如果这是最后一个引用，文件会被移入磁盘上的"Not Show in Tree"回收文件夹（不会立即彻底抹除，可以手动去该文件夹里找回或永久清除）。

5.7 切换视图、排序与批量重命名

  • 列表 / 图标视图：文档区域右上角可以在"列表视图"（详细信息一行一个）和"图标视图"（缩略图网格，类似 Windows 资源管理器）之间切换；图标视图下还能在"小图标"和"大图标"之间切换缩略图大小，方便快速浏览照片，选择会被记住。
  • 排序：文档区域上方的"排序"下拉菜单可按名称（A–Z）、文档日期（元数据）、到期日（元数据）或修改日期排列当前文件夹里的文档。
  • 上一份 / 下一份：预览窗口打开一份文档时，可以点击窗口两侧的箭头，或直接按键盘左右方向键，在当前文件夹的文档之间快速切换，不用每次都关掉预览重新点选。
  • 批量重命名：点击工具栏上的 "批量重命名"，DMS 会为当前文件夹里的每份文档生成一个新文件名——格式为"日期+地点前 10 个字符+原文件名"（例如 20260815旧金山-合同.pdf），日期和地点分别取自文档自身的日期与位置元数据；确认预览列表无误后一次性完成改名，磁盘上的实际文件名也会同步更新。
  • 批量用 AI 提取参数：想给一批文档统一填同一个关键参数，不用逐份打开手动录入——点击工具栏上的 "AI 填参数"，勾选文档后选一个已有或新建的参数名，DMS 会逐份读取每份文档的页面图像来识别对应数值，结果先列出来供确认，点"应用"才会真正写入（用法细节见 7.4 节，需先按第 8 章配置好 AI）。

5.8 到期日与"日程"颜色预警

可以给任意一份文档单独设置一个到期日（比如证件有效期、保修截止日、续费日期），到期临近时自动用颜色提醒，不用自己记日子。

  1. 在文档卡片上点击 "到期日" 按钮，选择日期后保存（留空并保存可以清除已设置的到期日）。
  2. 点击顶部工具栏的 "日程" 按钮，开启颜色预警模式：所有设置了到期日的文档、以及包含这些文档的文件夹，都会按紧急程度自动变色——1 天内到期或已逾期标红，1 周内标黄，2 周内标蓝；工具栏上也会同步显示这三种颜色对应的图例。
  3. "日程"模式开启的同时，会自动弹出一份 "临近到期文档" 清单，按紧急程度列出所有 2 周内到期的文档、它们所在的文件夹路径和具体到期日，点击其中任意一行即可直接打开对应文档；清单是一个独立的浮窗，关闭后仍可随时通过再次点击"日程"按钮重新打开。

5.9 图标视图弹出成独立窗口

5.7 节的"图标视图"还可以点击"弹出"按钮，变成一个独立的浏览器窗口，方便拖到另一台显示器上单独摆着，一边看图一边操作树形视图的其他部分。弹出的窗口里，多选文档后除了 5.2 节的批量下载，还可以：

  • 批量删除：勾选多份文档后一次性删除（进入 Not Show in Tree 回收文件夹，见 3.2 节，可以找回），不用一份一份点。
  • 批量加前缀：给勾选的一批文件名统一加上一段前缀文字（比如在日期后面补一句说明），和 5.7 节按"日期+地点"生成新文件名的批量重命名是两种不同的改名方式，各有用途。

5.10 PDF 翻译、保存译文与原文｜译文对照

打开 PDF，点击查看器顶部的“中文翻译”，右半边会显示译文，左边是原文，两边同步滚动。右上角的下拉框可选“快速翻译（免费）”（默认，无需 AI 密钥）、Gemini 或 DeepSeek；后两者需要先配置密钥（见第 8 章）。

  • 保存为 PDF：把整份译文按页排成纯文字 PDF，存进原文所在的文件夹，文件名为“原文件名 + 语言名”。
  • 保存为 PDF（保持版式）：把译文放回原文所在的位置，保留表格、分栏等版式（要求原 PDF 有可选中的文字，可能需要一两分钟）。
  • 原文 | 译文：保存过译文之后，打开原文或译文任意一份，点击这个按钮即可左右对照，同步滚动。
  • 编辑译文：点击右栏的“编辑”可以逐页（版式译文则逐段）修改，支持语音输入；先选中文字再点调色板里的颜色，可以给所选文字标色或高亮，“清除”可去掉颜色，改完点“保存”。

5.11 在文档中查找文字、编辑元数据

查找文字：查看器顶部的“查找文字”可在当前文档内查找一段文字，采用模糊匹配（相似度约 70% 即算命中），所以不必与原文完全一致，OCR 有少量错字也能找到；命中后可直接跳转到对应页。没有可搜索文字的文档会提示先运行 OCR。

编辑元数据：点击“编辑元数据”（信息栏里为“元数据”）可修改文档的元数据。同一份文档如果被关联到多个文件夹，这里的修改对所有引用它的位置同时生效。

06

文字识别与搜索

6.1 提取文字（OCR）

对于扫描件或图片（没有文字层，无法直接选中复制的那种 PDF/图片），DMS 可以自动识别出其中的文字，识别后的文字既能被搜索到，也能用来自动填写关键参数。

  1. 打开一份文档的预览。
  2. 点击 "提取文字 (OCR)"。如果文档本身带有可选中的文字层，系统会直接读取；如果是纯图片或扫描件，则会调用 OCR 引擎识别。
  3. 识别完成后，文字会显示出来，可以手动修改纠错，再点击保存。
扫描件文字有误？ 如果 PDF 本身带了一层内容错误的文字（常见于某些扫描软件生成的文件），可以点击 "强制 OCR"，让 DMS 忽略原有文字层，直接对页面图像重新识别一遍。

6.2 关键词搜索

  1. 切换到顶部"搜索"标签页（或树形视图右上角的搜索框）。
  2. 输入编号、文件名、描述文字或元数据关键词，结果会实时过滤显示。

搜索范围包括文件名、OCR 识别出的全文，以及您为文档添加的所有元数据字段（见第 7 章）。系统还会自动标记出可能不可靠的乱码 OCR 结果，避免它们干扰搜索。

输入多个词（用空格分隔）时，搜索框旁的"多词匹配"决定它们怎么组合：全部包含 (AND)（默认，必须同时包含每一个词才算匹配）、任一包含 (OR)（包含其中任意一个词就算匹配，比如同时输入"发票 收据 账单"这类近义词时很好用）、整句匹配（不拆分成单独的词，把输入框里的全部内容当一句话原样精确匹配，适合确定就要找某一句原文的时候）。

6.3 跳转到命中页并红色高亮

搜索命中的是 PDF 时，查看器里会出现“跳转到第 N 页”按钮：点击后直接翻到命中所在的那一页，并用红色高亮标出匹配的文字，不用再逐页寻找。

6.4 使用筛选器

  • 文件类型：勾选图片、PDF、文本或文档，把结果限定为该类型。
  • 范围节点：启用“限定范围至所选节点”，只在当前文件夹及其子文件夹中搜索。
  • 元数据：输入参数名称和 / 或值，查找具有匹配元数据的文档（例如：压力 = 3000 psi）。
07

关键参数：从文档中提取结构化数据

从文档中提取特定参数——比如体检报告里的"血糖值"、材料检测报告里的"屈服强度"、成绩单上的"年级排名"——并用于后续分析和报告，是 DMS 的一项核心特色；"关键参数"就是实现它的方式：您给每份文档附加的自定义信息字段。填好之后，这些信息既可以在搜索里用到，也能变成第 9 章"趋势图表"里的数据点。对完全没概念的用户，可以把它简单理解成"给每份文档打一些自定义标签，并写上对应的值"。

对于大多数文档，仅靠 7.4 节的关键词匹配提取（或直接手动录入）就已经能达到很高的准确率；只有版式复杂、参数散落在图表或表格里的文档，才需要用到 8.2 节的 AI 辅助提取作为补充，进一步提升识别的适用性和稳定性。

7.1 管理全局字段列表

系统预设了一些常用字段名，您可以按需增删，设置好后，之后给任意文档添加字段时，输入框会自动弹出这些名称供选择，保持命名统一。

  1. 点击 "项目" → "全局关键参数"。
  2. 点击"添加"输入新字段名；点击字段旁边的删除图标可移除；可拖动调整顺序。
  3. 点击"保存"，全局参数列表立即生效。

7.2 给一份文档添加参数值

  1. 打开一份文档，找到"元数据"/关键参数区域，点击展开。
  2. 点击"添加字段"，从下拉列表选一个已有字段名（或直接输入一个只属于这份文档、不进入全局列表的新字段名）。
  3. 在字段右侧填入具体的值，点击保存，即时写入。

7.3 设计值与实测值

每个参数都可以同时保存"设计值"和"实测值"两组数据，各自包含数值、单位、说明三部分，形成一套完整、可直接使用的数据结构，适合需要核对设计要求与现场实测结果是否一致的场景（比如设备验收）。

这套结构不只是方便核对：它同样可以直接用于合规性检查、设备采购比价、结构安全性分析等实际工作场景；随着数据日积月累，也为将来做设备升级决策、趋势分析，甚至交给厂家或第三方 AI 做进一步分析，提前积累好一份干净、结构化的高质量数据。

7.4 自动从文档文字中提取

如果文档已经完成第 6 章的文字识别，就可以让 DMS 自动在文字里查找关键词并抓取对应数值，不用逐字手动录入。

  1. 在文字识别面板中点击 "提取关键参数"。
  2. 勾选需要提取的字段名。
  3. 点击 "开始提取"，系统会在全文中查找字段关键词及其后面的数值。
  4. 预览提取结果——每条结果会自动拆分为数值、单位、说明三部分，而不是一整段原始文字；同一字段在文档里出现多次时，会分别编号（如"压力_01""压力_02"）单独列出，不会遗漏。
  5. 确认无误后点击"应用到文档"。
表格类文档小技巧 勾选"仅同行提取"，DMS 只会匹配与字段名处在同一行的内容，更适合表格；不勾选则在全文范围内搜索。

以上是打开单份文档、逐份提取的用法。如果要给一整批文档统一填同一个参数，不用一份一份打开——在文档列表工具栏点 "AI 填参数"，勾选要处理的文档，选一个参数名后点"开始提取"，DMS 会逐份读取每份文档的页面图像来识别该参数的值（不要求页面上有可选中的文字层，扫描件、照片同样有效），结果列出来供确认，点"应用"才会写入（见 5.7 节；这个批量流程本身用的就是 8.2 节的 AI 图像识别能力，需要先按第 8 章配置好 AI）。

7.5 核对时的浮窗预览

人工录入或核对参数时，可以弹出一个可拖拽、可自由缩放的原文档预览小窗口，一边看着原文一边填写，不需要来回切换页面。

7.6 把所有元数据导出成表格

点击 "项目" → "导出元数据 (CSV)"，即可把库里每一份文档及其全部关键参数导出成一张电子表格，直接用 Excel 打开查阅或做进一步分析。

08

AI 辅助（可选功能）

这一整章的功能都是可选的：不配置也完全不影响其他所有功能的正常使用。它需要您自己申请一个免费或低价的 AI 服务密钥（API Key），并保持联网。DMS 现在支持两种 AI 服务：Google 的 Gemini 和 DeepSeek（国内用户访问更快更稳定），二选一或都配置、随时切换均可。

8.1 申请并配置密钥

  1. 点击 "项目" → "AI 辅助提取设置"，打开"AI 提供商设置"对话框。
  2. Gemini：点击"没有 API Key？前往 Google AI Studio 免费获取"，网址是 aistudio.google.com/apikey；按页面提示登录 Google 账号并创建密钥，复制后粘贴回 DMS 的输入框。
  3. DeepSeek：点击"没有 API Key？前往 DeepSeek 开放平台获取"，同样登录、创建、复制、粘贴。
  4. 保存后密钥会被加密保存在本地，不会以明文形式存储。
费用说明 Gemini 免费额度每次最多处理 3 页文档，对大多数体检报告、产品测试报告已经足够；需要处理几十甚至上百页的大型文档，可订阅 Gemini 付费版（约 10 美元/月）。Gemini 付费版还承诺不会把您上传的文件用于训练其模型，更适合涉及隐私或商业敏感信息的文档。

8.2 AI 辅助提取关键参数

对于版式复杂、普通文字匹配抓不准的文档（比如参数散落在图表或复杂表格里），可以让 AI 直接"看"文档页面的图像来提取数值，而不是只匹配文字。在第 7.4 节的提取面板中，勾选"使用 AI"再开始提取即可。

8.3 就整个项目提问（Ask AI）

  1. 切换到顶部"搜索"标签页，点击 "AI 问答"。
  2. 在输入框中用一句话提问，例如"设备的工作压力是多少？"
  3. 选择使用 Gemini 还是 DeepSeek，点击提问。AI 会自动跨文档查找答案，并标注具体来自哪一份文档，方便您核实原文。

也可以只把范围限定在某一个文件夹（勾选"仅限此节点及子节点"），而不是整个项目。

如果只想针对正在看的这一份文档提问，不用切到"搜索"标签页——直接在文档预览窗口顶部点 "提问 AI"，右侧会展开一个提问面板，AI 依据这份文档已提取的文字、描述和关键参数来回答；同一面板里也能对这份文档的描述做"用 AI 整理润色"（用法与 4.5 节完全一致）。

8.4 问答记录自动保存

每次向 AI 提问，问题和回答都会自动追加进一份与该文档关联的 PDF 记录里，随时可以在文档列表中打开查阅、也能被搜索到，不需要自己另外整理。

8.5 打包成 PDF，交给任意 AI 平台（ChatGPT、Gemini 等）

DMS 自带的"AI 问答"（8.3 节）适合快速查资料，但如果您想用一批照片讲一个故事，再让 AI 生成一段视频、一篇文章或其他 DMS 本身不提供的创作，可以改用"生成 PDF"：把照片、您对 AI 的要求，以及一段关于这些照片的故事，一起打包成一份 PDF，交给 ChatGPT、Gemini 或您习惯使用的任何 AI 平台去完成。

  1. 在"对文件夹提问 AI"对话框中（第 8.3 节同一入口），勾选要用到的照片。
  2. 在提问框里写下内容——既可以是一句给 AI 的具体要求（例如"根据这些照片生成一段视频，配轻音乐"），也可以是一段关于这些照片的故事或说明，两者也可以写在一起。
  3. 打字之外，也可以点击语音输入按钮直接口述要求或故事；说完后点击 "用 AI 整理润色"，把随口讲出的内容整理得更通顺（用法与 4.5 节的描述框完全一致）。
  4. 点击 "生成 PDF（供 ChatGPT 等）"，DMS 会把选中的照片压缩转换为体积很小但清晰度不打折的 PDF 页面，连同您写的要求和故事一起合并成一份 PDF 并下载下来。
  5. 把下载好的 PDF 上传到 ChatGPT、Gemini 或其他任意支持文件上传、具备图像理解能力的 AI 平台，照常提出您的请求即可。因为照片已经转换为体积很小的 PDF，即便选了很多张照片，上传和处理也会顺畅很多。
和第 8.3 节的区别 8.3 节的"直接问 AI"由 DMS 内置调用 Gemini/DeepSeek 直接给出文字回答；这里的"生成 PDF"则是把素材打包好，交给您自己选择的、能力更强或更适合创作类任务的外部 AI 平台去处理，DMS 本身并不生成视频或其他创作内容。
10

照片库与人脸识别

10.1 设置照片根目录

先指定一个文件夹作为照片库的根节点，之后所有自动整理的照片都会归入它的子文件夹中。

  1. 右键点击要作为根节点的文件夹（例如新建一个叫"照片库"的文件夹）。
  2. 选择 "设为照片根目录"。
  3. 该文件夹图标会变成相机图标，表示设置成功。若要取消，右键选择"取消照片根目录"。
注意 每个项目只能设置一个照片根节点。

10.2 上传照片

上传照片到照片根目录（或其子文件夹）时，DMS 会自动读取照片的拍摄日期（EXIF 信息），在根目录下创建对应的"年/月"子文件夹并归入；如果照片带 GPS 信息，拍摄地点也会一并识别，显示为文档元数据中的"Location"字段，并可一键在地图上查看拍摄地点。

10.3 标记人脸

  1. 选中一张或多张照片。
  2. 点击"标记人脸"，在照片中框选出人物，输入姓名（首次标记某人时会新建一个"人物"条目，此后重复使用同一个名字即可归入同一人）。

10.4 按人物浏览照片

  1. 切换到顶部 "人脸" 标签页。
  2. 点击任意一个已标记的人物，即可看到这个人在整个库中出现过的所有照片。
  3. 点击"自动扫描匹配"，可以在某个文件夹或整个库范围内查找这位（或所有已标记）人物可能出现的更多照片，匹配的置信度阈值可以调整。

如果需要删除某个人物，可以一键删除该人物及其全部标记，不会影响照片本身。首次使用人脸识别功能前，系统会引导您安装底层识别引擎，并实时显示安装进度。

10.5 查找相似照片，清理重复内容

连续拍摄、反复保存或多次导出，常常会在同一批照片里留下好几张几乎一样的重复照片。DMS 可以自动挑出这些"近似重复"的照片，并推荐保留其中最好的一张。

  1. 打开一个装着照片的文件夹，点击工具栏上 "多选" 旁边的 "查找相似照片" 按钮，进入照片选择模式。
  2. 勾选想要比较的几张照片（只能勾选图片文件），设置右侧的 "相似度阈值"（默认为 75%，可自行调整，数值越高判定越严格）。
  3. 点击 "检测相似照片"，DMS 会在这批照片之间两两比较，把相似度达到阈值的照片自动归为一组，并标出每组的最低相似度。
  4. 每组照片中，DMS 会用一枚 ★ 星标 推荐其中分辨率最高、文件体积最大（即压缩最少、质量通常最好）的一张作为"建议保留"的版本，供您参考——最终留哪张仍由您自己判断。
  5. 确认无误后点击 "标记以便审查"，DMS 会把每一组照片链接（而非移动）到根目录下新建的 "疑似重复照片" 文件夹中对应的"组N"子文件夹里，原照片仍留在原来的文件夹中，方便您稍后统一查看、比对、删除。
缩略图大小 在结果窗口和正常的"图标视图"里，都可以在"小图标"和"大图标"之间切换，方便仔细比对照片细节；该设置会被记住，下次打开自动沿用（见 5.7 节）。
在"疑似重复照片"文件夹里删除，才是真正的清理 这个文件夹里的照片只是"链接"过来的，原照片仍保留在其原始文件夹中——按 5.6 节的一般规则，在别处删除通常只会解除当前文件夹的引用。但如果直接在"疑似重复照片 / 组N"里点删除，DMS 会把这份照片从它所有的位置一并移除（包括原始文件夹），并移入"Not Show in Tree"回收文件夹，这样才能真正达到清理重复照片、释放空间的目的。
11

生成数据手册（PDF 或 Word）

"数据手册"功能可以把文件夹树里的文档合并成一份带封面、目录和分节书签的正式手册，非常适合作为交付资料或存档打印，有手动选择生成 PDF、按文件夹自动生成 Word 两种模式。

11.1 手动选择文档，生成 PDF

  1. 切换到顶部 "数据手册" 标签页，选择"手动选择（PDF）"。
  2. 在文档选择器中勾选要合并的文档，可以跨多个文件夹选择。
  3. 填写手册的标题和副标题（如需要，可以在自动生成的封面之前再插入一张自定义封面）。
  4. 点击 "生成并下载"，系统会把所有选中文档按文件夹顺序、序列号自动排列成章节结构，合并成一份 PDF。
  • 只包含图片的文件夹会自动转换为可打印的 PDF 页面。
  • 横版的原始页面会自动旋转为竖版，保证整本手册方向统一。
  • 含大量照片时，图片会在嵌入前智能重新压缩，页面观感几乎无损，但整份文件体积可以缩小到原来的百分之一左右。
提示 已经生成过一次的选择列表和标题会自动记住，下次打开"数据手册"标签页时可以直接在上次的基础上继续操作。

11.2 按文件夹自动生成 Word 数据手册

与逐份手动勾选不同，这个模式只需选中一个"顶层"文件夹，DMS 会按文件夹结构自动生成章节，得到一份可以在 Word 里继续编辑排版的 .docx 文件——适合提交前还要调整格式、或对方明确要求 Word 而不是 PDF 的场合。

  1. 切换到顶部 "数据手册" 标签页，选择"按文件夹自动生成（Word）"。
  2. 勾选一个文件夹——它的每个子文件夹（可多级嵌套）都会成为一个章节，章节内文档按文件夹里原有顺序排列，每份文档后面附上它的描述文字。
  3. 点击生成按钮，DMS 在服务器端生成 .docx 文件，浏览器自动下载。
和 PDF 模式的区别 两种模式生成的都是全新文件，不会改动文件夹里原有的文档；Word 模式里的表格、标题、正文都是可编辑的原生 Word 内容，不是把 PDF 页面当图片贴进去，方便生成后再手动增删小段文字或调整格式。
12

分享文档

12.1 通过邮件发送

需要先配置一次 SMTP 邮箱设置（Mac 和 Windows 都适用）：

  1. 点击 "项目" → "邮件发送设置"。
  2. 点击顶部对应邮箱服务商的快捷按钮（Gmail / Yahoo / Outlook / iCloud / QQ 邮箱 / 网易163），自动填入服务器地址和端口。
  3. 填写用户名（您的邮箱地址）和密码。Gmail 用户请使用"应用专用密码"，不是账户登录密码——需要单独在 Google 账号设置里生成。
  4. 点击"发送测试邮件"验证设置，成功后您的邮箱会收到一封测试邮件；确认无误后点击"保存设置"。

设置好之后，发送文档时：

  1. 在文档卡片上点击邮件图标。
  2. 填写收件人地址（系统会记住上次填写过的地址）、主题（默认是文档名称）和正文（可选）。
  3. 点击"发送邮件"，系统通过 SMTP 直接发送，片刻后收件人即可收到。
未配置 SMTP 时 在 Mac 上会自动改用"信息"或 Mail App 帮您打开一封草稿；在 Windows 上则必须先完成上面的 SMTP 配置才能发送邮件。

12.2 发送到手机

点击文档卡片上的手机图标，输入对方手机号码，点击发送。系统会生成一个有效期 1 小时的下载链接，通过 Mac 的"信息"App（iMessage/短信）或 Windows 的"你的手机"应用发送过去，对方无需安装任何 DMS 相关的 App，点开链接即可下载。

13

手机与远程访问

这一章介绍反过来的方向——用手机往电脑里传照片/文件，或者用手机直接浏览电脑上的整棵文件夹树。这两项功能都是从启动时弹出的那个小状态窗口（第 1.2 节提到的 "QCDMS" 窗口）里操作的，而不是在浏览器网页里。

13.1 从手机上传照片或文件到某个文件夹

  1. 在电脑上找到那个小状态窗口（如果被最小化了，可以在任务栏/程序坞里找回），点击 "Remote Upload" 按钮。
  2. 在弹出的对话框中，可选填入手机号码（含国家代码，如 +8613800138000），并从下拉列表中选择希望文件传到哪个文件夹。
  3. 点击 "Generate Link"，系统会启动一条安全的公网隧道并生成一个二维码和链接（此过程通常需要几秒到十几秒）。
  4. 用手机扫描二维码，或点击"Send via iMessage / SMS / WeChat"直接把链接发到手机上，也可以点"Copy URL"手动复制发送。
  5. 手机上打开链接后，即可拍照或从相册中选图直接上传，文件会自动出现在您第 2 步选择的那个文件夹里。
链接有效期 这个远程上传链接有效期为 7 天，过期后需要重新点击一次"Generate Link"生成新的。

13.2 在手机上浏览电脑里的文件

用手机打开第 13.1 节生成的同一条链接后，页面上会有一个"浏览"入口，点进去即可像在电脑上一样浏览整棵文件夹树、打开文档查看，而不仅仅是上传新文件。

13.3 同一 Wi-Fi 下的局域网访问

如果手机和电脑连接的是同一个 Wi-Fi 网络，也可以不经过公网隧道，直接在手机浏览器里访问电脑的局域网地址（形如 http://192.168.x.x:端口号/mobile）打开同样的上传页面。如果不确定电脑的局域网地址，最简单的方式仍是使用上面 13.1 节的 "Remote Upload"，它在任何网络环境下都能用。

14

备份、导出与导入

DMS 里的一切都是普通文件，不会被"锁死"在专有格式里。点击顶部 "项目" 按钮即可看到全部备份/导出选项。

14.1 三种导出方式

导出索引备份（.dms）只保存文件夹树结构和元数据，不含实体文件，文件名以 METADATA- 开头，体积很小，适合快速留存结构。
导出完整备份（含文件）把文件夹树、元数据和全部实体文档、照片一起打包，文件名以 COMP- 开头，是最完整的一份备份。
导出选定文件夹（含文件）先勾选部分文件夹，只导出这些文件夹树及其中的文件，文件名以 PART- 开头，适合只想分享/备份某一部分内容的场景。

另外还有 "导出元数据 (CSV)"——把所有文档的元数据导出成表格，方便在 Excel 中查阅或记账。

自动保留最近 10 份 导出的 .dms 备份文件会保存在存储文件夹内。项目目录中最多保留 10 个 .dms 备份文件，更早的会自动移入"Not Shown in The Tree"文件夹，不会无限占用空间。
已设项目密码时 如果当前项目设置了项目密码（见第 15.2 节），以上每一种导出都会自动用这个密码加密，不需要另外再单独设置导出密码。

14.2 导入一份 .dms 备份

  1. 点击 "项目" → "导入项目（.dms）"，或者直接把 .dms 文件拖到上传区域。
  2. 选择一个目标文件夹作为解压位置。
  3. 确认后系统开始解压并导入，完成后自动切换到这个新导入的项目。

导入只会恢复文件夹树结构，不会删除磁盘上任何已有的文件。

14.3 新建一个全新的空白项目

点击 "项目" → "新建空白项目"，输入一个新的空文件夹路径即可，当前正在使用的项目不受影响，随时可以通过顶部的存储路径按钮切回来。

14.4 免费云备份

只要把存储路径设置在一个已经开启同步的 OneDrive 或 Dropbox 文件夹里，DMS 的所有数据就会随文件夹自动同步备份到云端，不需要额外操作。应用内提供分步的 OneDrive 配置指南（在"项目"菜单的"文档下载"区域可以下载）。

再次提醒 Mac 上请不要把存储路径放进 iCloud Drive（详见第 1.3 节的说明）。
15

安全与密码保护

DMS 提供两层完全独立的密码保护，可以只用一层，也可以两层都用。

15.1 应用登录密码（保护整台电脑上的 DMS）

这是一个可选的登录密码，保护本机上的整个 DMS 应用，不管当前打开的是哪个项目——适合公用电脑或希望别人完全无法打开 DMS 的场景。

  1. 点击 "项目" → "设置应用登录密码"。
  2. 输入新密码（两次确认），点击保存，立即生效。之后每次打开浏览器访问 DMS 都需要先输入这个密码。
  3. 如需修改或移除，回到同一入口（此时会显示为"修改或移除应用登录密码"），输入当前密码验证身份后操作；移除时把新密码留空即可。
  4. 已设置密码时，菜单里会出现"退出登录"选项，点击后当前浏览器会话结束，下次访问需要重新输入密码。
存储方式 密码以加密形式（SHA256 加随机盐、10 万次迭代）存储在本机的配置文件中，不会以明文保存。

15.2 项目密码（只保护某一个具体项目）

这是与上面完全独立的第二重密码，只属于某一个具体的项目——密码信息就存储在这个项目的文件夹内部，会随着文件夹一起被复制、同步或交给别人。打开这个项目、或者在别的电脑上导入它导出的 .dms 备份时，都需要这个密码。

  1. 点击 "项目" → "设置项目密码"。
  2. 可以给这个项目起一个便于识别的名称（会显示在锁定界面和文件选择器中），设置密码后保存。
  3. 已设置密码时，工具栏会出现带锁图标的项目名称标签；菜单里出现"锁定当前项目"选项，点击后需要重新输入项目密码才能继续查看（不影响上面的应用登录状态）。

15.3 按文件夹保存网站登录信息

可以在某个文件夹下保存一组或多组相关的网站地址、用户名和密码，点击"添加登录信息"即可继续添加，只有点击"显示"时才会解密显示；同一个文件夹下可以保存任意多组，并提供"复制用户名"的快捷按钮。

常见用法是把某个在线账号的登录信息，和与它相关的文件存在同一个文件夹里，做到"资料"与"入口"合一：比如在"体检报告"文件夹里顺便存好医院/医生网上个人账户的登录信息，以后要复查报告或预约，不用再到处翻找账号密码；同理也适用于设备厂家的管理后台、保险公司官网、银行或基金账户等场景。

15.4 一次导出所有登录信息

点击文档工具栏上的“导出登录信息”，DMS 会生成一份 Excel 表格并直接用默认的表格程序打开。每一行是一组登录信息，列依次为：文件夹、网站、用户名、密码、文件夹描述。

注意 密码列是明文。这份表格是临时文件，请核对完毕后妥善保管或删除，不要随意转发。显示“(cannot decrypt…)”表示该密码是在另一台电脑上保存的，当前电脑无法解密。
16

其他设置

  • 下载用户手册 / 设置指南：点击"项目"菜单最下方的"文档下载"区域，可以下载本手册的 .docx 版本、OneDrive 配置指南（中/英文）、Windows 人脸识别安装指南等。
  • 切换存储路径：点击顶部工具栏的存储路径按钮，可以随时切换到电脑上另一个已有的 DMS 项目文件夹（第 17 章还介绍了启动时的项目选择界面、同时多开，以及把另一个项目合并进来）。
  • 语言切换：整个应用界面可随时在顶部一键切换中文/英文，不影响已保存的数据。
  • 文档数量不再受限：单个项目可保存的文档数量没有上限。

16.1 键盘快捷键

  • 确认文件夹重命名 / 新文件夹名称：Enter；取消：Escape
  • 开始重命名文件夹：双击文件夹名称；打开文件夹右键菜单：右键单击文件夹
  • 关闭文档查看器：Escape 或点击“关闭”
  • 粘贴剪贴板图片 / 文字到当前文件夹：Ctrl+V（Windows）/ Cmd+V（Mac）
  • 复制 / 全选 / 撤销（文本框内）：Ctrl+C / Ctrl+A / Ctrl+Z
  • 查看器内切换上一份 / 下一份文档：← / →
17

切换、合并与多开项目

DMS 里的"项目"就是一个存储路径——一整套独立的文件夹树、文档和设置。现在可以很方便地在多个项目之间切换、同时开着好几个、甚至把两个项目合并成一个。

17.1 启动时选择项目

双击 DMS.app / DMS.exe 打开浏览器后，如果电脑上已经用过 DMS，不会再像第一次那样直接要求设置存储位置，而是先出现一个"选择项目"界面：

  • 继续上次的项目：直接打开上次正常退出前所在的那个项目。
  • 选择别的项目文件夹：浏览到电脑上另一个已有的 DMS 项目文件夹。
  • 新建一个全新的空项目：从零开始，不会带入任何示例数据。

只要是通过窗口右下角"Quit DMS"按钮、Cmd+Q、关闭状态窗口等正常方式退出的，DMS 都会记住当时打开的是哪个项目；如果同时开着好几个项目窗口，"上次"记录的是最后一个被正常关闭的那个。

17.2 同时打开多个项目

可以同时开启多个 DMS 项目窗口/实例，各自独立运行、各自登录，互不干扰，不需要每次都先关掉一个再打开另一个。

  1. 点击顶部"项目"菜单里"打开另一个项目"一类的入口，或者直接再运行一次 DMS.app / DMS.exe，在 17.1 节的启动界面里选另一个项目文件夹。
  2. DMS 会为新项目单独开一个窗口和一套服务，用不同的本机端口，与已经打开的窗口完全独立。
  3. 如果对同一个项目文件夹重复发起"打开"，DMS 会直接切换到已经在服务这个项目的窗口，不会重复启动。

17.3 合并另一个项目进来

适合两台电脑（或两个人）各自整理了一段时间的资料，现在要拼成一份的场景——把另一个 DMS 项目的文件夹和文档，整体并入当前正在用的这个项目。

  1. 点击顶部"项目"菜单里的"合并项目"。
  2. 选择要并入的另一个项目的存储文件夹。
  3. 确认后开始合并：同名文件夹会自动合并到一起，内容完全相同的文档会自动跳过、不会重复导入，其余文档和文件夹正常并入当前项目的树里。
合并前建议先备份 合并是把另一个项目"拼"进当前项目，操作前建议先按第 14 章导出一份当前项目的完整备份，以防合并结果不是预期的样子。

常见问题

Gemini 提示“API 额度已用完”（429）怎么办？
免费版 Gemini 有每日（或每分钟）额度上限，提示里会写明大约多久后恢复。可以稍后重试、为该 Google 项目开通计费、换另一个 API 密钥，或改用 DeepSeek；翻译可以改选“快速翻译（免费）”。
忘记了应用登录密码怎么办？
可以手动编辑本机的配置文件重置密码。文件位于 Mac 的 ~/.pms_dms_config.json 或 Windows 的 C:\Users\yourname\.pms_dms_config.json。用文本编辑器打开，把其中 password_hash 和 password_salt 两个字段的值都改成空字符串 ""，保存后重启 DMS，密码保护即被移除。（这是应用登录密码，与第 15.2 节的"项目密码"是两回事，项目密码保存在项目文件夹内，同理可用文本编辑器打开项目文件夹里的相应配置文件清空重置。）
误删了文件夹或文档，能恢复吗？
删除文档时，只要它还被其他文件夹引用，就只是解除引用，文件本身还在；只有最后一个引用被删除，文件才会移入磁盘上的"Not Show in Tree"文件夹（见 3.2 节），可以手动去那里找回。删除整个文件夹也是同样的机制：文件夹本身从树里消失，但里面的文档同样先进入"Not Show in Tree"暂存，而不是立刻从磁盘上抹掉——只有在那个隐藏文件夹里再次点删除，才是真正、不可撤销的永久清除。不过文件夹从树上消失这件事本身无法一键撤销，重要结构变动前仍建议先按第 14 章导出一份备份。如果存储路径本身开启了云盘同步（如 OneDrive），也可以尝试通过云盘的版本历史找回。
为什么发邮件提示"SMTP 错误"？
常见原因：Gmail 未使用"应用专用密码"（用的是账户登录密码）；服务商的 SMTP 服务未开启（QQ、163 邮箱需要先在网页版里手动开启）；公司网络防火墙屏蔽了 587 或 465 端口，需联系网络管理员。
文档可以在多台电脑上共享使用吗？
可以。把存储路径设置到一个共享网络位置即可，比如家用/办公室的 NAS 网络存储盒子（映射为网络磁盘，例如 Z:\photo），或云同步文件夹（如 OneDrive、Dropbox；Mac 上请避开 iCloud Drive，见第 1.3 节）——多台电脑上的 DMS 都指向同一路径即可共享数据，但一般同一时间只有一台电脑能联接读写，也请不要在两台电脑上同时编辑同一份文档，以免产生冲突，建议按文件夹分工使用。如果人不在家、需要从外部访问家里的 NAS，得依靠 NAS 自带的 VPN 功能，或额外用 Tailscale、WireGuard 之类的工具打通网络——DMS 本身不提供联网穿透能力，只负责读写您指定路径下的文件。如果需要真正意义上多人同时读写（而不是错开时间共享），技术上可行，但目前不作为通用产品功能提供，有此需求可以联系开发者另行商谈。
Windows 上双击 DMS.exe 提示"Windows 已保护你的电脑"（SmartScreen 警告）怎么办？
这是因为 DMS.exe 是新生成的可执行文件，还没有向微软付费购买"数字签名认证"，Windows 默认会拦截未知来源的新程序以保护用户，这不代表程序有问题。两种处理方式：

方法一（推荐，最简单）：在警告窗口中点击"更多信息"链接，窗口底部会出现一个"仍要运行"按钮，点击它即可正常启动。

方法二：如果窗口已经关闭或无法点击，可以右键点击 DMS.exe → "属性" → 在"常规"选项卡最下方勾选"解除锁定"复选框 → 点击"应用" → "确定"。这样设置后以后也不会再触发 SmartScreen 警告。
支持哪些文件格式？
支持 PDF、JPG、JPEG、PNG、HEIC、TIFF、GIF、WEBP、BMP（图片格式），以及 TXT、CSV 等文本文件。原则上任何格式的文件都可以上传并保存，但页面内直接预览功能仅支持以上列出的格式，其他格式可以下载后用对应软件打开。
DMS.app 和 DMS.exe 有什么区别？只是分别对应 Mac 和 Windows 吗？
是的，两者功能完全相同，分别是打包给 Mac 和 Windows 用户的可执行程序，双击即可运行，不需要额外安装 Python 或任何依赖库——唯一的例外是人脸识别功能（详见第 10.4 节）：其底层识别引擎体积较大，没有打包进安装包，需要另外单独安装，应用内会引导您完成整个过程。
DMS.app — 用户手册 面向零基础新用户编写，根据当前版本整理 · 2026-10-23
For Complete Beginners — Full Step-by-Step Guide

DMS.app User Manual

DMS is a document and photo management system that runs on your own computer. It organizes the real files on your hard drive into a folder tree you can browse and search, all through your web browser — nothing is uploaded to any cloud server. This manual assumes you have never used DMS before: it starts with "how do I install it" and walks through the exact clicks for every feature.

Document
User Manual — rev. 2026-10-23
Audience
First-time DMS users
Platforms
macOS · Windows
Chapters
01–17 + FAQ
Download for Mac DMS-Mac.zip Download for Windows DMS-Windows.zip See all past releases
Download Manual PDF (Chinese) DMS_User_Manual-261023-CN.pdf Download Manual PDF (English) DMS_User_Manual-261023-EN.pdf
01

Install & First Launch

DMS needs no account and no system-level install — download it and you have a program you can double-click straight away. From the very first step, every file you add lives only on the hard drive location you choose yourself.

1.1 Download

Use the button at the top of this page for your computer: "Download for Mac" or "Download for Windows". You'll get a .zip archive.

  1. Once downloaded, find the .zip in your Downloads folder and double-click to unzip it.
  2. Inside you'll find DMS.app (Mac) or DMS.exe (Windows). Drag it into Applications (Mac) or wherever you'll remember on your Desktop, so it's easy to find later.

1.2 First launch

  1. Double-click DMS.app or DMS.exe.
  2. A small status window titled "QC Document Management System (QCDMS)" appears first, showing status and a URL link. Don't close this window — it's DMS's background server; closing it quits the whole program.
  3. After about a second, your default browser (Chrome recommended) opens automatically to the DMS web interface. If it doesn't, click the underlined URL in the status window.
Tip That small status window has a "Quit DMS" button (bottom-right) to fully exit the program, and a "Remote Upload" button (bottom-left) used in Chapter 13, "Mobile & Remote Access" — you can ignore it for now. If double-clicking DMS.exe does nothing, or Windows shows a security warning, see the "Windows protected your PC" entry in the FAQ below.

1.3 Set your storage location

On first launch, the browser shows a "Set storage location" page asking you to pick one folder — every PDF, image, and index DMS ever saves will physically live inside it.

  1. In the "Storage path" field, paste or type a folder path. If it doesn't exist yet, DMS creates it automatically. For example:
    Mac: /Users/yourname/Documents/DMS_Storage
    Windows: C:\Users\yourname\Documents\DMS_Storage
  2. Click "Save and continue".
  3. The app initializes and takes you straight to the main screen.
Important On Mac, do not pick a folder inside iCloud Drive (e.g. a Desktop or Documents folder with "Desktop & Documents sync" turned on) — iCloud can silently move infrequently-used files off your local disk, which can make DMS unable to read them. For free cloud backup, use a OneDrive- or Dropbox-synced folder instead (see Chapter 14).
02

Tour of the Main Screen

Spend two minutes getting familiar with these areas — every later chapter refers back to them.

2.1 The top toolbar

  • A 中文 / EN toggle — switches the whole interface's language instantly; your saved data is unaffected.
  • The Project button — opens a dropdown with export/import backups, global settings, and password protection, covered in Chapters 14, 15, and 16.
  • The current storage path button — shows the folder in use; click it to switch to a different folder (i.e. a different "project").
  • If the current project has a project password set, a padlock-badged project name shows here too.

2.2 Four tabs

Below the toolbar are four tabs:

TreeThe default view. Folders on the left, documents in the selected folder on the right — Chapters 3–5 mostly happen here.
SearchKeyword search, or switch to "Ask AI" to question the whole library — see Chapters 6 and 8.
Data bookMerge multiple documents into one cover-paged PDF — see Chapter 11.
FacesBrowse tagged photos by person — see Chapter 10.

2.3 Layout of the Tree tab

  • Left: the folder tree, expandable/collapsible, right-click any folder for its menu. Just above it sits a row of quick buttons: New folder, Upload document, Rename, Delete, Text tree (shows the structure of the currently selected folder), Plot.
  • Right: the list of documents in the currently selected folder.
Voice input works in almost every text field Folder/document names, serial numbers, login credentials, descriptions, the search box, the Ask AI box — nearly every text field has a small microphone icon next to it. Click it and speak; the recognized text is appended right where your cursor is, and clicking again stops it. You can switch the recognition language between Chinese and English, all done in the browser with no AI key needed (see Section 4.5).
03

Building Your Folder Tree

The folder tree is DMS's backbone. It's worth spending a few minutes sketching out the rough structure that matches how you actually think about your files (by equipment, by client, by family member) before adding documents.

3.1 Create a folder

  1. In the left tree, click to select a "parent" folder (the new folder will be created inside it).
  2. Right-click that folder and choose "New subfolder", or click the "New folder" button in the toolbar above the tree.
  3. Type a name and press Enter.

3.2 Rename & delete

  1. Rename: right-click a folder → "Rename", or double-click its name, edit, and press Enter. The real folder on disk is renamed to match.
  2. Delete: right-click a folder → "Delete folder" and confirm — the folder disappears from the tree, and its documents aren't erased from disk right away; they're moved into a hidden folder called Not Show in Tree.
Want a document back after deleting its folder? Scroll to the bottom of the left-hand tree and expand Not Show in Tree — documents there can be re-linked or moved back into any folder (Section 5.3). Only deleting a document again from inside that hidden folder permanently erases it from disk, with no undo. The folder disappearing from the tree itself isn't reversible in one click though — if you might need it again later, dragging it elsewhere is safer than deleting it.

3.3 Drag to move, reorder, or merge

  1. Press and hold the left mouse button on a folder to drag it.
  2. Drop it directly on top of another folder (it highlights) → moves it inside that folder, becoming a subfolder of it.
  3. Drop it on the top or bottom edge of another folder → moves it to become that folder's previous/next sibling (reorders only, doesn't change parents).
Automatic merging If the target folder already has a subfolder with the exact same name, DMS automatically merges the two — every document and subfolder from the source is folded into the target, so you never end up with duplicate same-named folders.

3.4 Folder serial number (SN)

Each folder can carry a serial number (e.g. A-001) to match your existing asset or part numbering. Subfolders and their documents default to "inheriting" their parent's SN unless they set their own.

  1. In the folder details panel on the right, find the "Serial number (SN)" field.
  2. Type the number and save. It appears on document cards so you can identify what they belong to at a glance.

3.5 Bulk-build a whole tree from a list

If you already have a written-out folder structure (e.g. a bill of materials, an org chart), you don't have to create each folder by hand — import it all at once from a simple comma-separated (CSV) text file.

Column 1Folder name. Required — the folder this row creates.
Column 2Parent folder name — which already-listed folder this nests under; leave blank for the root. The file's very first row is the root folder itself, so its own parent column must be blank.
Column 3Description, optional — can be left out of the file entirely. When present, it becomes that folder's initial description.
  1. Click "Project" → "Create folder tree from a hierarchy file" in the top toolbar.
  2. Paste or upload your CSV text, e.g.:
    Top Assembly
    Pump Housing,Top Assembly
    Rotor Assembly,Top Assembly
    Bolt,Rotor Assembly
  3. Click "Validate" to check it. If the same folder name repeats in the file (e.g. an identical "Bolt" sub-part under two different assemblies), it's flagged as an error by default; check "Auto-number duplicate folder names" and re-validate, and DMS numbers them "Bolt 01," "Bolt 02," and so on so the whole file imports successfully.
  4. Once the preview looks right, click "Create"; every folder appears in the tree immediately.
Exporting from a BOM tool? Level-heading rows (lines that just say something like "Level 1") are skipped automatically, and a leading column-title row ("folder name, parent folder name, description") is detected and ignored too — no need to delete either by hand.

3.6 Pasting screenshots straight into a description

Folder and document Description boxes accept pasted images: take a screenshot (or copy an image), click into a Description box and press Ctrl+V (Cmd+V on Mac). The screenshot appears inline at the cursor, mixed with your text.

  • The pasted screenshot is saved as a real image document in the current folder (it shows in the document list, and if deleted goes to “Not Show in Tree” where it can be recovered); the description just references it.
  • Pasting into a Description does not also trigger 4.3's “paste a screenshot as a new document”; pasting formatted text copied from a web page keeps plain text only.
  • AI polish and the typo checker only handle the text portion — inserted screenshots are kept. In .txt / PDF exports and search results a screenshot shows as “[screenshot]”. Plain-text descriptions saved by older versions keep working.
04

Getting Documents In

Supported image formats include PDF, JPG, PNG, HEIC, TIFF, GIF, WEBP, and BMP, plus text files like TXT. In principle any file type can be uploaded and stored, though not every format can be previewed directly in the browser. Maximum size per file is 4GB.

4.1 Drag and drop (the fastest way)

  1. Click to select the target folder in the left tree.
  2. Drag files straight from the Desktop or a Finder/Explorer window onto the document area on the right.
  3. Release — the file uploads automatically and appears in the document list.

4.2 Upload via the button

  1. With a folder selected, click "Upload document" in the toolbar above the tree, or right-click the folder → "Upload document".
  2. In the file picker, select files (hold Cmd/Ctrl to multi-select) and click "Open".

4.3 Paste a screenshot straight from the clipboard

No need to save a screenshot to the Desktop first — take the screenshot, then press Cmd+V (Mac) or Ctrl+V (Windows) anywhere on the DMS page to drop it straight into the current folder.

It works for copied text too: if the clipboard holds a passage of text instead of an image, pasting typesets it into a PDF named Pasted text <date> <time>.pdf in the current folder.

4.4 Upload a whole folder, three filing modes to choose from

If you already have a whole set of material organized into folders on your computer (e.g. a project's full directory, or a photo album folder exported from your phone), you don't have to drag files one at a time — upload an entire folder in one go.

  1. With the destination folder selected, click "Upload Folder" in the toolbar — this opens the "Upload Multiple Folders" dialog.
  2. Click "+ Add folder" and pick a folder on your computer; repeat to queue up several at once, each listed above with an × to remove it.
  3. Choose one of three filing modes:
    • Sort into year/month folders (default): every file in each picked folder (including subfolders) is auto-filed by capture/archive date into "year/month" subfolders, and the original subfolder layout isn't kept — best for a photo album where you only care about the date, not the original organization.
    • Upload as the sources: subfolders are recreated here exactly as they are, with no date-based year/month folders — for material that's already organized the way you want to keep it.
    • Recreate the picked folder under the current folder: unlike the other two, the picked folder itself also becomes a new subfolder here (not just its contents); the whole structure lands as-is, and matching directories are created on the local drive too.
  4. Optionally type something into "Add text to filenames (optional)" — it's inserted right after the date when there is one, or at the very front of the filename when there isn't; the mic button next to it accepts dictation too. Leave it blank to add nothing.
  5. Click "Upload" in the bottom-right corner to finish.

4.5 Descriptions, voice input & AI polish

Every document and photo can carry a free-text "description" — it's saved into the search index and fully searchable.

  • Voice input: click the voice-input button next to the description box and just speak — the transcribed text is appended automatically (works best in Chrome/Edge), and you can freely mix typing and dictation; the recognition language can be switched between "中" (Chinese) and "EN" (English), so either language can be dictated.
  • AI organize & polish (needs AI configured, Chapter 8): dictated or quickly-typed text often comes out rough — click "Organize & polish" and pick one of "Organize", "Polish wording", or "Make concise". The AI's result is shown for you to review first; it only replaces your original text once you accept it.
  • Check typos: click "Check typos" to have AI specifically flag typos, punctuation issues, and misused idioms in the description, listed one by one for you to fix.
  • Saving: click "Save and make searchable" to store the text in the index only; to also save this description as its own standalone .txt document you can open, download, or share like any other file, click "Save as new document" (once saved, this becomes "Update description file" so you can keep revising the same one).
Example: organizing course material Managing handouts, worksheets, and past exams for a class? Write a description for each item (dictate it and let AI organize/polish it — no need to type it all out by hand) and save each one as a new document; every description then becomes its own document in the library, always individually searchable and openable. If you also want one combined write-up to share or archive, use Chapter 11's "Data book" feature to merge those description documents into a single PDF — it never moves or touches the originals, so each description still stays individually available (Section 5.4's quick merge, by contrast, moves the merged-away originals into the recycle folder, so it isn't the right tool for this).

4.6 Photos and videos auto-filed by year and month

Once a folder is set as the "photo root" (Chapter 10), uploaded photos automatically read their capture date and get filed into "2026" → "08"-style year/month subfolders under it. If a photo carries GPS data, the shooting location is picked up too. HEIC photos from an iPhone preview correctly with no manual conversion.

Videos are now filed the same way: MP4/MOV-family files use the recording timestamp embedded in the video itself, while other video formats fall back to the file's own modified date.

4.7 Bulk import from a ZIP

If you have a large batch of documents and can tell from the filename which folder each belongs in, name them with a convention and zip them up for DMS to sort all at once. The target folders must already exist in the tree (unlike Section 3.5's hierarchy import, this doesn't create new folders).

  1. Name each file like FolderName#description.pdf (e.g. Home Insurance#2026 Policy.pdf for a folder named "Home Insurance" — everything after the # can be anything you like), then zip them up.
  2. Click "Project" → "Bulk import documents from ZIP" and pick that ZIP file.
  3. DMS shows which files match which folder, and which couldn't be matched.
Same folder name in more than one place? If the same folder name occurs more than once in the tree (say, "Bolt" under both "Rotor Assembly" and "Housing Assembly"), matching by bare name lands on whichever comes first. Use a "/"-separated path before the # instead, e.g. Rotor Assembly/Bolt#List.pdf, to land specifically on the "Bolt" folder under "Rotor Assembly" rather than the identically-named one elsewhere.

4.8 Duplicate files

On upload, DMS automatically checks for files with identical content already in the library. If it finds a match, it asks whether to skip, link the existing file into this folder instead (no extra disk space used), or upload it anyway as a new file.

4.9 Auto-detected document date

Once a PDF or similar document is uploaded, DMS automatically scans the text on its first page for the document's own date (a report date, an issue date), so you don't have to type one in by hand for every file. It looks first for a date that follows a recognized label (like "Issue Date", "签发日期", "Inspection Date"), and only falls back to the first date-shaped text anywhere in the document if no such label is found.

Always double-check it If a document contains more than one date — a birth date, an issue date, an expiry date — the auto-detected one isn't guaranteed to be the one you actually want. An auto-detected date is flagged in amber, noting it was "auto-detected from '...' in the document's own text," and keeps that flag until you confirm or correct it — click "Date" on the document's card to check and save the right one.

4.10 Batch convert to PDF

Need to convert a batch of photos or existing PDFs into individually named PDF files? No need to save them one at a time.

  1. Check the images and/or PDFs to convert, then click "Convert to PDF" in the toolbar.
  2. The dialog lists your checked documents; confirm and click "Convert N documents to PDF".
  3. Each one is converted into its own PDF (filenames start with "YYYY-MM-DD-" when a date can be determined), and the originals move into the Not Show in Tree recycle folder (Section 3.2, recoverable).

4.11 On-disk filenames & structural snapshots

Neither of these needs any action from you, but they're worth knowing about when browsing files in Finder/Explorer or troubleshooting:

  • Filenames on disk are name-first, id-second (e.g. Inspection Report+DOC-20260920-xxx.pdf), so browsing the storage folder directly tells you what a file is without opening it first.
  • Before any change to a project's folder structure (adding/removing folders, moving, batch operations), DMS automatically saves a snapshot of the structure at that moment; if a change ever leaves things in a bad state, that snapshot can be used to roll back — a background safety net you won't see any UI for during normal use.

4.12 Grouping photos into location folders

Photos that carry location data can now be split one level further than year/month (see 4.6): inside each month folder, photos taken within the same radius of each other go into one place folder, named automatically after the place (for example “Boston, Massachusetts”).

  1. When you upload photos, a confirmation window appears first. Tick “Group photos into location folders (inside each month)”.
  2. Enter a Radius and choose miles or km. The first time it defaults to 10 miles; afterwards DMS remembers your last value, so just click “Upload” if you don't want to change it. Untick the box to file by year/month only.
  3. Photos without location data stay in the month folder, and phone uploads follow the same rule. Place names are looked up in the background after the upload, so uploads aren't slowed down; if a name can't be found the folder is temporarily called “Location 1”, “Location 2”, …

Not happy with the grouping? A radius that is too large lumps different places together; too small splits one place apart. Right-click the folder holding the photos and choose “Regroup photos by location…”, pick a new radius, and the window previews how many place folders it will make; click “Regroup” when you're happy. You can repeat as often as you like. It also tidies photos you uploaded earlier that were never grouped by place; place folders you renamed by hand get renamed again, while other folders and documents are untouched.

4.13 Barcode scanning

If a photo shows a barcode or QR code (a product label, an equipment nameplate…), DMS can read the code straight from the image — no retyping.

  1. Open the viewer for that image (or a PDF containing one).
  2. Click “Scan Barcode” at the top.
  3. On success a banner shows the code type and value (e.g. “CODE128: 6901234567892”).
  4. The result is saved automatically as the document's “Barcode” parameter, viewable and editable in the metadata section; several barcodes on one photo are saved as Barcode, Barcode2, Barcode3, …
Good to know This runs entirely offline (open-source zbar library), sends nothing anywhere and needs no AI key — but the code must be sharp, unobstructed and have a blank margin around it. QR codes shared as phone screenshots (especially with a logo in the middle, or a background running right up to the edge, like a WeChat “join group” image) usually can't be read; that's a known limitation, not a bug.

4.14 Upload Eq Record (AN^SN^PN naming)

For equipment record-keeping: a record is renamed to Assembly Number (AN)^Serial Number (SN)^Part Name (PN)^original file name and filed straight into the folder whose name equals that AN, so the file name alone tells you which unit it belongs to.

  1. Click “Upload Eq Record” in the Tree tab toolbar. A window opens that you can drag and resize.
  2. On the left pick the AN (assembly number / folder) — type in the filter box first to narrow the list — then optionally fill in SN and PN (both may be blank).
  3. Click “Choose file…”. The window shows “Will be saved as:” with the new name and previews images, PDFs and text on the right.
  4. Click “Upload” when it looks right. AN / SN / PN stay filled in so you can pick the next file.

To upload many at once, click “Upload by file name” and pick several files already named AN^SN^PN.extension (e.g. A100^SN001^Bearing.pdf). DMS splits out the fields, matches each AN to its folder and shows a status per row (Uploaded / Already exists, skipped / the reason for an error); click “Upload N file(s)” to finish.

Good to know “^” is the separator because work-order numbers and part names often contain hyphens. The AN must match a folder name, otherwise that row reports no folder was found.

4.15 QC Review and the QC Review Report

QC Review lets you fill in a review for each of a batch of inspection reports and roll everything up into a deliverable PDF at folder level. The structure: one folder per “Inspection” holding its report files; each file has its own review form; the folder as a whole has one General Comment.

  1. Upload the records: click “Upload QC Review” in the Tree tab toolbar. It works like 4.14 but the fields are Inspection (folder), SN, Equipment Description and WO/PO (Work Order / Purchase Order); “Upload by file name” accepts Inspection^SN^Equipment Description^WO/PO. Each file automatically gets a review form pre-filled with those fields.
  2. Fill in one file's review: open the viewer and click “QC Review” (“Add QC Review” until one exists). The window widens: the file on the left, the form on the right, with a draggable divider. The form has Inspection, SN, Equipment Description, WO/PO, Date Reviewed, Reviewed By, Observation, Mitigation and Review Result (Pass / Fail / Conditional / Pending). Changes save automatically, and the fields are searchable.
  3. General Comment and report: select the inspection folder, fill in the General Comment in its QC Review section, then click “QC Review Report”. Choose which actual inspection reports to attach (Select all / None / tick one by one) and DMS builds a PDF: the General Comment and per-file review summary first, then the chosen reports. Word and PowerPoint files are converted to PDF first (needs LibreOffice installed). The report is filed in that folder and regenerating overwrites it instead of piling up copies.

4.16 AI: Detect service date

Lab reports and medical exams often show both a date of birth and an exam date, and the upload-time date detector can't tell them apart. “AI: Detect service date” re-reads existing documents with Gemini, looking specifically for the date the service / inspection / exam took place and explicitly excluding the date of birth.

  • Document list toolbar: click “AI: Detect service date”, tick several PDFs / images, and start.
  • Folder multi-select mode: tick some folders and click the same button to process every PDF / image in them and all their subfolders.
  • When it finishes, the same confirmation dialog used for upload-time dates opens — nothing is written until you review it. Documents with no service date found are only counted in the message. Needs a Gemini key (chapter 8).
05

Viewing, Downloading & Organizing Documents

5.1 Previewing a document

  1. Click any document in the list to preview it right on the page: PDFs support page navigation and zoom, plus a "find in document" search that tolerates small OCR typos and jumps straight to the highlighted match; images show at full size; text files display their content directly.
  2. To open it in another app installed on your computer instead (e.g. Word for a .docx), click "Open in native app" and DMS hands it to your system's default program.

5.2 Rename & download

  • Click "Rename" on a document's card to change both the file on disk and its display name at once.
  • Select one or more documents and click "Download" to copy them to a folder of your choice (the system Downloads folder by default), with the option to open that folder right away.

5.3 Linking one document into several folders

The same document can appear in multiple folders without using extra disk space — this is called "linking".

  1. Open the target folder and click "Link existing document".
  2. Search or browse for the document in the dialog that opens; you can also drag a result straight onto a folder on the left to link it in one motion, with no per-item confirm click needed.
  3. The document now appears in the current folder while still remaining in its original one; its card shows "Linked to N node(s)".

The Link Existing Document window has been upgraded too: drag its title bar to move it and drag the divider between the two columns to resize them; the folder tree on the left has Expand all / Collapse all buttons, so you can open folders and tick the documents to link; on the right, tick the Target folders or simply drag a document onto one. A Delete button permanently deletes the selected documents from disk — it cannot be undone, so use it with care.

About deleting a link Removing a link only detaches that one folder's reference to the document — it doesn't delete the actual file. Only once every reference is removed does the file actually move off disk (into the recycle area below).

5.4 Combine into one PDF

Merge several selected scans or files into a single PDF, filed back into a folder you choose. Any photos among them are compressed first — longest edge capped at 900px, JPEG quality 85% — before being embedded, with no visible quality loss to the eye but the whole file shrinking to roughly a hundredth of its original size: a contract bundling 136 photos, for example, can go from 2.3GB down to about 18MB as a PDF. Quick merges skip the formal cover page — per-document bookmarks alone handle navigation — and any landscape page is automatically rotated to portrait so the whole file stays consistently oriented.

The originals are removed from the current folder Once merged, the original documents are removed from the current folder and moved into the "Not Show in Tree" recycle folder (not immediately erased — you can find or permanently clear them there). If you want the originals to also stay individually viewable elsewhere, use Chapter 11's "Data book" feature instead — it produces the same kind of combined PDF without touching the originals.

5.5 Opening a document keeps it highlighted

Any document you've opened (in the built-in preview or a native app) stays highlighted in its folder's list, so you can spot it again at a glance later.

5.6 Deleting a document

Click the delete button on a document's card and confirm. If the document is linked into other folders, only the current folder's reference is removed; if this was its last reference, the file is moved into the "Not Show in Tree" recycle folder on disk (not immediately erased — you can find or permanently clear it there manually).

5.7 View mode, sorting & batch rename

  • List / icon view: toggle between "List view" (one detailed row per document) and "Icon view" (a thumbnail grid, like Windows Explorer) at the top right of the document area; in icon view, a further "Small"/"Large" toggle controls thumbnail size — handy for browsing photos quickly, and your choice is remembered.
  • Sorting: the "Sort by" dropdown above the document list orders the current folder's documents by name (A–Z), document date (metadata), due date (metadata), or date modified.
  • Previous / next: with a document open in the preview window, click the arrows on either side of it, or just press the left/right arrow keys, to step through the current folder's documents without closing and reopening the preview each time.
  • Batch rename: click "Batch rename" in the toolbar to generate a new filename for every document in the current folder — in the form "date + first 10 characters of location + original name" (e.g. 20260815SanFrancisco-contract.pdf), using each document's own date and location metadata. Review the preview list, confirm, and every file is renamed at once — including the actual file on disk.
  • Batch AI-fill a parameter: to fill in the same key parameter across a batch of documents without opening each one, click "AI-fill parameter" in the toolbar, check the documents, and pick an existing or new parameter name. DMS reads each document's page images to find the value, lists the results for you to review, and only writes them in once you click "Apply" (full flow in Section 7.4; needs AI configured per Chapter 8).

5.8 Due dates & "Schedule" color warnings

Set a due date on any individual document — a permit's expiry, a warranty deadline, a renewal date — and DMS warns you with color as it approaches, so you don't have to track the date yourself.

  1. Click "Due date" on a document's card, pick a date, and save (leave it blank and save to clear an existing due date).
  2. Click "Schedule" in the top toolbar to turn on color warnings: every document with a due date, and every folder containing one, is automatically color-coded by urgency — red for due within 1 day or overdue, yellow for within 1 week, blue for within 2 weeks — with a matching color legend shown right in the toolbar.
  3. Turning on Schedule mode also opens a "Documents due soon" list, showing every document due within 2 weeks ranked by urgency, along with its folder path and exact due date — click any row to open that document directly. It's a standalone popup you can close and reopen anytime by clicking "Schedule" again.

5.9 Pop out the icon view into its own window

Section 5.7's "Icon view" can also be popped out into a standalone browser window — click "Pop out" — handy for dragging to a second monitor while you keep working in the tree elsewhere. Inside the pop-out window, multi-selecting documents adds two more batch actions on top of Section 5.2's batch download:

  • Batch delete: check several documents and delete them all in one go (into the Not Show in Tree recycle folder, Section 3.2, recoverable), instead of one at a time.
  • Batch add a prefix: add a common piece of text as a prefix to every checked file's name (e.g. a note after the date) — a different renaming tool from Section 5.7's date+location batch rename, useful for different situations.

5.10 PDF translation, saving it, and Original | Translation

Open a PDF and click “Translate” at the top of the viewer: the right half shows the translation beside the original on the left, scrolling in sync. The dropdown at the top right offers “Fast (free)” (the default, no AI key needed), Gemini or DeepSeek; the latter two need a key configured first (chapter 8).

  • Save as PDF: translates the whole document and saves a plain-text PDF in the same folder, named “original name + language”.
  • Save with layout: puts each translation back where the original text was, keeping tables and columns (needs a PDF with selectable text; may take a minute or two).
  • Original | Translation: once a translation is saved, open either file and click this button for a side-by-side view that scrolls in sync.
  • Edit the translation: click “Edit” in the right pane to correct it page by page (paragraph by paragraph for layout translations); voice input is supported. Select text and click a colour in the palette to colour or highlight it, “Clear” removes the colour, then click “Save”.

5.11 Finding text inside a document, and editing metadata

Search text: the viewer's “Search text” finds a passage inside the current document using fuzzy matching (about 70% similarity counts as a hit), so it needn't match the original exactly and still works with a few OCR errors; a match can jump straight to its page. Documents with no searchable text prompt you to run OCR first.

Edit metadata: click “Edit metadata” (“Metadata” in the info panel) to change a document's metadata. If the same document is linked into several folders, the change applies everywhere it is referenced.

06

OCR & Search

6.1 Extracting text (OCR)

For a scan or an image with no selectable text layer, DMS can recognize the text inside it. Once recognized, that text becomes searchable and can also auto-fill key parameters.

  1. Open a document's preview.
  2. Click "Extract text (OCR)". If the document already has a selectable text layer, DMS reads it directly; for a pure image or scan, it runs the OCR engine.
  3. Once recognized, the text appears and can be hand-corrected before saving.
Scan has a bad text layer? If a PDF already has a text layer but it's wrong (common with certain scanning software), click "Force OCR" to ignore the existing layer and re-recognize the page image from scratch.

6.2 Keyword search

  1. Switch to the top "Search" tab (or use the search box in the top-right of the Tree view).
  2. Type a number, filename, description, or metadata keyword — results filter live as you type.

Search covers filenames, the full text from OCR, and every metadata field you've added to a document (Chapter 7). DMS also flags likely-unreliable garbled OCR output so it doesn't pollute your search results.

When you type more than one word, the "Multi-word match" control next to the search box decides how they combine: All words (AND) (the default — a document must contain every word you typed to match), Any word (OR) (matches if it contains at least one, handy when you're not sure which of several related terms a document uses, like "invoice receipt bill"), or Whole sentence (skips splitting into words and matches the entire box's contents verbatim as one exact phrase — useful when you know exactly which line you're looking for).

6.3 Jump to the matching page, highlighted in red

When a search hit is inside a PDF, the viewer shows a “Jump to page N” button. Click it to go straight to the page of the match, with the matching text highlighted in red — no more hunting page by page.

6.4 Using filters

  • File type: tick Image, PDF, Text or Documents to restrict results to that type.
  • Scope: enable “Scope to selected node” to search only within the selected folder and its sub-folders.
  • Metadata: enter a parameter name and/or value to find documents with matching metadata (e.g. Pressure = 3000 psi).
07

Key Parameters — Extracting Structured Data

Pulling specific parameters out of a document — a "blood glucose level" from a physical exam report, a "yield strength" from a material test report, a "class rank" from a report card — and using them for later analysis and reporting is one of DMS's core features. "Key parameters" are how it's done: custom information fields you attach to a document. Once filled in, this data is searchable, and can also become the data points behind Chapter 9's trend charts. If none of this is familiar yet, think of it simply as tagging each document with your own custom labels and values.

For most documents, keyword-match extraction (Section 7.4) — or just typing values in by hand — already gets very high accuracy. AI-assisted extraction (Section 8.2) is a supplement, only really needed for documents with a complicated layout or values scattered across charts and tables, to further improve how well extraction handles those trickier cases.

7.1 Managing the global field list

DMS ships with a few common field names already set up; you can add or remove your own. Once set, any field you start typing on a document will suggest these names, keeping naming consistent.

  1. Click "Project" → "Global key parameters".
  2. Click "Add" to type a new field name; click the ✕ next to a field to remove it; drag to reorder.
  3. Click "Save" — the global list takes effect immediately.

7.2 Adding a value to one document

  1. Open a document and expand its metadata / key-parameters area.
  2. Click "Add field" and pick an existing name from the dropdown (or type a new one that belongs only to this document, without joining the global list).
  3. Type the value next to it and save — it writes immediately.

7.3 Design value vs. actual value

Each parameter can hold both a "design" and an "actual" reading at once, each with its own value, unit, and description — a complete, ready-to-use data structure, useful whenever you need to check a design requirement against a real-world measurement (e.g. equipment acceptance testing).

That structure is good for more than just checking one reading against another: the same data applies directly to compliance checks, comparing equipment purchase options, and structural safety analysis. And as it accumulates over time, you're also building up a clean, structured, high-quality dataset ready for equipment-upgrade decisions, trend analysis, or handing off to a manufacturer's or third-party AI for further analysis down the road.

7.4 Extracting automatically from a document's text

Once a document has been through Chapter 6's text recognition, DMS can search that text for your field names and pull out the numbers that follow, instead of you typing each one by hand.

  1. In the text-recognition panel, click "Extract key parameters".
  2. Check the field names you want extracted.
  3. Click "Start extraction" — DMS searches the full text for each field's keyword and the value that follows it.
  4. Review the results — each match is automatically split into value, unit, and description rather than one raw blob of text; a field that appears more than once in the document is captured as separate, numbered results ("Pressure_01", "Pressure_02", ...) so nothing is missed.
  5. Click "Apply to document" once it looks right.
Tip for tables Check "Same-line only" so DMS only matches content on the same line as the field name — ideal for tabular documents; leave it unchecked to search the whole text instead.

The above is for one document at a time. To fill in the same parameter across a whole batch of documents at once, you don't need to open each one — in the document list toolbar, click "AI-fill parameter", check the documents to process, pick a parameter name, and click "Start extraction." DMS reads each document's page images to find the value (no selectable text layer required — scans and photos work too), lists the results for review, and only writes them in once you click "Apply" (see Section 5.7; this batch flow uses the same AI image-reading ability as Section 8.2, so it needs AI configured per Chapter 8).

7.5 The preview flyout while checking values

While entering or double-checking a parameter by hand, you can pop open a draggable, freely-resizable preview of the source document, so you can look at the original text while you type — no more tabbing back and forth.

7.6 Exporting all metadata as a spreadsheet

Click "Project" → "Export Metadata (CSV)" to export every document's full set of key parameters as a spreadsheet, ready to open directly in Excel for review or further analysis.

08

AI Assistance (Optional)

Everything in this chapter is optional — leaving it unconfigured doesn't affect anything else in the app. It requires an API key of your own from an AI service (free or low-cost) and an internet connection. DMS now supports two AI providers: Google's Gemini and DeepSeek (faster and steadier for users in mainland China). Set up one or both, and switch between them anytime.

8.1 Getting and configuring a key

  1. Click "Project" → "AI-assisted extraction settings" to open the "AI provider settings" dialog.
  2. Gemini: click "Don't have an API Key? Get one for free at Google AI Studio" — aistudio.google.com/apikey. Sign in with a Google account, create a key, then copy and paste it into DMS.
  3. DeepSeek: click "Don't have an API Key? Get one at the DeepSeek platform" and repeat the same steps.
  4. Once saved, the key is encrypted at rest — never stored as plain text.
On cost Gemini's free tier processes up to 3 pages of a document at a time — plenty for most medical reports or product test reports. For documents running dozens or even hundreds of pages, a Gemini paid subscription (about $10/month) lifts that limit; the paid tier also comes with Google's commitment not to use your uploaded files to train its models, which matters for anything privacy- or business-sensitive.

8.2 AI-assisted key parameter extraction

For documents with a complicated layout that plain text matching can't handle (values scattered across charts or complex tables), let AI read the actual page image instead of just matching text. In the extraction panel from Section 7.4, check "Use AI" before starting extraction.

8.3 Asking AI across the whole project

  1. Switch to the top "Search" tab and click "Ask AI".
  2. Type a plain-language question, e.g. "What is the equipment's working pressure?"
  3. Pick Gemini or DeepSeek and ask — AI searches across every document and cites exactly which one each answer came from, so you can verify the source.

You can also scope the question to just one folder (check "This node and its subfolders only") instead of the entire project.

To ask about just the document you're currently looking at, you don't need to switch to the Search tab — click "Ask AI" right at the top of the document preview window. A panel expands on the right, and AI answers using that document's own extracted text, description, and key parameters; the same panel also has "Clean up with AI" for that document's description (works exactly like Section 4.5).

8.4 Q&A history saved automatically

Every question you ask and every answer AI gives is automatically appended to a running PDF log tied to that document — always available to reopen from the document list, and fully searchable, with nothing extra to organize yourself.

8.5 Package as a PDF for any AI platform (ChatGPT, Gemini, etc.)

The built-in "Ask AI" (Section 8.3) is great for quick lookups, but if you want to turn a set of photos and a story into a video, an article, or some other creative output DMS doesn't produce itself, use "Build PDF" instead: it bundles your photos, your request to the AI, and a story about those photos into one PDF, for you to hand to ChatGPT, Gemini, or whichever AI platform you prefer.

  1. In the "Ask AI about this folder" dialog (same entry point as Section 8.3), check the photos you want to use.
  2. Write in the question box — it's a single free-text field, so use it however you like: a specific instruction to the AI (e.g. "make a video from these photos, with soft background music"), a story or description of the photos, or both together.
  3. Besides typing, you can also click the voice-input button and dictate your request or story directly; once you're done, click "Organize & polish" to clean up what you said into something more readable (works exactly like the description box in Section 4.5).
  4. Click "Build PDF (for ChatGPT etc.)". DMS compresses the selected photos into a small but sharp PDF and bundles your written request and story in alongside them, then downloads the result.
  5. Upload that PDF to ChatGPT, Gemini, or any other AI platform that accepts file uploads and can read images, and make your request there as usual. Since the photos are already a lightweight PDF, uploading and processing stays smooth even with a large batch selected.
How this differs from Section 8.3 "Ask AI now" in 8.3 calls Gemini/DeepSeek directly, inside DMS, for a text answer. "Build PDF" here instead hands your material to an external AI platform of your choice — better suited to creative tasks DMS itself doesn't generate, like video or long-form writing.
10

Photos & Face Recognition

10.1 Setting a photo root folder

First designate one folder as your photo library's root — every automatically-organized photo from then on lands somewhere under it.

  1. Right-click the folder you want as the root (e.g. create a new folder called "Photos").
  2. Choose "Set as photo root folder".
  3. Its icon changes to a camera icon, confirming it's set. To undo this, right-click and choose "Unset photo root folder".
Note Only one photo root is allowed per project.

10.2 Uploading photos

Uploading a photo into the photo root (or one of its subfolders) makes DMS automatically read its capture date (EXIF data) and create the matching "year/month" subfolders under the root to file it into. If the photo carries GPS data, the shooting location is picked up too, shown as a "Location" metadata field with a one-click link to view it on a map.

10.3 Tagging a face

  1. Select one or more photos.
  2. Click "Tag faces", draw a box around a person in the photo, and type their name (the first time you tag someone, a new "person" entry is created; reuse the same name afterward to group photos under that same person).

10.4 Browsing photos by person

  1. Switch to the top "Faces" tab.
  2. Click any tagged person to see every photo of them across the whole library.
  3. Click "Auto-scan for matches" to search a folder or the entire library for more photos of that person (or everyone), with an adjustable match-confidence threshold.

To remove someone, delete the person and every one of their tags in a single action — the photos themselves are untouched. The first time you use face recognition, DMS guides you through installing the underlying engine, with live progress shown.

10.5 Finding near-duplicate photos

Burst shots, repeated saves, or re-exports often leave several nearly-identical copies of the same photo scattered through a batch. DMS can automatically find these "near-duplicates" and recommend which one to keep.

  1. Open a folder containing photos and click "Find near-duplicates" next to the "Select" button in the toolbar to enter photo-selection mode.
  2. Check the photos you want to compare (only image files can be selected), and set the similarity threshold on the right (defaults to 75% — higher means a stricter match).
  3. Click "Check selected". DMS compares every pair of the checked photos and groups the ones that meet the threshold together, showing each group's lowest pairwise similarity.
  4. Within each group, DMS marks the photo with the highest resolution and largest file size (usually the least-compressed, best-quality copy) with a ★ badge as the suggested keeper — the final call is still yours.
  5. When you're happy with the groups, click "Mark for review". DMS links (not moves) each group's photos into a new "Group N" subfolder under a "Suspected duplicates" folder at the root, leaving the originals exactly where they were, so you can review, compare, and delete at your own pace.
Thumbnail size Both the results dialog and the normal grid view offer a "Small"/"Large" thumbnail toggle for closer comparison — your choice is remembered for next time (see Section 5.7).
Deleting inside "Suspected duplicates" is what actually cleans up Photos in this review folder are only linked there — the originals still live in their own folders, and by the general rule in Section 5.6, deleting elsewhere normally just removes that one reference. But deleting a photo directly from inside "Suspected duplicates / Group N" removes it from every location it's linked to, including the original folder, and moves it into the "Not Show in Tree" recycle folder — which is what actually frees up space and cleans up the duplicate.
11

Generating a Databook (PDF or Word)

The "Data book" feature merges documents into one formal book, complete with a cover page, table of contents, and per-section bookmarks — ideal as a handoff package or for archival printing. There are two modes: manually pick documents for a PDF, or auto-generate an editable Word file straight from a folder.

11.1 Hand-pick documents, generate a PDF

  1. Switch to the top "Data book" tab and choose "Manual selection (PDF)".
  2. Check the documents to merge in the document picker — you can select across multiple folders.
  3. Fill in a title and subtitle for the databook (optionally insert your own cover page ahead of the auto-generated one).
  4. Click "Generate and download" — DMS arranges everything into a section structure following folder order and serial numbers, then merges it into one PDF.
  • Image-only folders convert automatically into print-ready PDF pages.
  • A landscape source page rotates automatically to portrait, so the whole book stays consistently oriented.
  • With many photos involved, images are intelligently re-compressed before embedding — near-invisible on the page, but the whole file can shrink to roughly one-hundredth of its original size.
Tip Your selection and title are remembered automatically after you generate once, so reopening the "Data book" tab next time picks up right where you left off.

11.2 Auto-generate a Word databook from a folder

Instead of hand-picking documents, this mode only needs one "top" folder selected — DMS builds the chapter structure from the folder tree itself, producing an editable .docx you can keep formatting in Word. Good for when you need to tweak formatting before handing it off, or someone specifically asked for Word instead of PDF.

  1. Switch to the top "Data book" tab and choose "By folder, auto (Word)".
  2. Check one folder — each of its subfolders (nesting is fine) becomes a chapter, documents within a chapter keep their existing folder order, and each document's description is appended after it.
  3. Click the generate button; DMS builds the .docx on the server and your browser downloads it.
How this differs from PDF mode Both modes produce a brand-new file without touching anything in your folders; the Word mode's tables, headings, and body text are genuine editable Word content, not PDF pages pasted in as images, so it's easy to add or tweak small pieces of text afterward.
12

Sharing Documents

12.1 Sending by email

Set up an SMTP mail account once first (works the same on Mac and Windows):

  1. Click "Project" → "Email sending settings".
  2. Click the quick-select button for your provider (Gmail / Yahoo / Outlook / iCloud / QQ Mail / 163) to auto-fill the server address and port.
  3. Fill in your username (your email address) and password. Gmail users must use an "App Password", not their normal account password — generate one separately in Google Account settings.
  4. Click "Send test email" to verify — a test message should arrive in your inbox. Once confirmed, click "Save settings".

Once set up, to send a document:

  1. Click the email icon on a document's card.
  2. Fill in the recipient's address (DMS remembers the last one you used), a subject (defaults to the document's name), and an optional message body.
  3. Click "Send email" — it goes out directly over SMTP and typically arrives within moments.
Without SMTP configured On Mac, DMS falls back to opening a draft in the Messages or Mail app for you; on Windows, SMTP must be configured first before you can send an email at all.

12.2 Send to phone

Click the phone icon on a document's card, enter the recipient's phone number, and click send. DMS generates a download link valid for one hour and sends it via the Mac Messages app (iMessage/SMS) or the Windows "Phone Link" app — the recipient needs no DMS-related app at all, just tap the link to download.

13

Mobile & Remote Access

This chapter covers the reverse direction — sending photos or files from your phone into DMS, or browsing your computer's whole folder tree straight from your phone. Both are driven from the small status window that appears at launch (the "QCDMS" window from Section 1.2), not from the browser page itself.

13.1 Uploading photos or files from your phone into a folder

  1. On your computer, find that small status window (check your taskbar/Dock if it's minimized) and click "Remote Upload".
  2. In the dialog, optionally enter a phone number (with country code, e.g. +16505551234), and pick which folder incoming files should land in from the dropdown.
  3. Click "Generate Link" — DMS starts a secure public tunnel and produces a QR code and URL (this usually takes a few seconds to around ten).
  4. Scan the QR code with your phone, or click "Send via iMessage / SMS / WeChat" to text the link directly — or "Copy URL" to send it manually.
  5. Opening the link on the phone lets you take a photo or pick one from the camera roll and upload it straight away — it lands in the folder you chose in step 2.
Link expiry This remote upload link is valid for 7 days. After it expires, click "Generate Link" again to make a new one.

13.2 Browsing your computer's files from your phone

Opening the same link from Section 13.1 on your phone shows a "Browse" option on the page — tap it to browse the whole folder tree and open documents just as you would on the computer, not just upload new ones.

13.3 Local access on the same Wi-Fi

If your phone and computer share the same Wi-Fi network, you can skip the public tunnel entirely and open your computer's local network address in the phone's browser (something like http://192.168.x.x:port/mobile) to reach the same upload page. If you're not sure of your computer's local address, the "Remote Upload" flow from 13.1 is the simplest option since it works on any network.

14

Backup, Export & Import

Everything in DMS is ordinary files — nothing is locked into a proprietary format. Click the "Project" button at the top to see every backup/export option.

14.1 Three ways to export

Export index backup (.dms)Saves just the folder tree structure and metadata — no actual files. Filenames start with METADATA-; small and fast, good for preserving structure quickly.
Export full backup (with files)Bundles the folder tree, metadata, and every actual document and photo together. Filenames start with COMP- — the most complete backup possible.
Export selected folders (with files)Check specific folders first, and export only their tree and files. Filenames start with PART- — good for sharing or backing up just part of the library.

There's also "Export Metadata (CSV)" — exports every document's metadata as a spreadsheet for review or bookkeeping in Excel.

Only the last 10 are kept Exported .dms backup files are saved inside the storage folder. The project folder keeps at most 10 .dms backups; older ones are automatically moved into the "Not Shown in The Tree" folder, so they never pile up indefinitely.
With a project password set If the current project has a project password (Section 15.2), every one of the above exports is automatically encrypted with it — there's no separate export password to set.

14.2 Importing a .dms backup

  1. Click "Project" → "Import project (.dms)", or drag a .dms file straight onto the upload area.
  2. Choose a destination folder to extract it into.
  3. Confirm — DMS extracts and imports, then automatically switches to the newly-imported project.

Import only restores the folder tree structure; it never deletes any existing files already on disk.

14.3 Starting a brand-new blank project

Click "Project" → "Create new blank project" and enter a new, empty folder path. The project you're currently using is unaffected, and you can switch back anytime using the storage path button at the top.

14.4 Free cloud backup

Point the storage path at a OneDrive or Dropbox folder that's already syncing, and every bit of DMS's data backs up to the cloud automatically along with that folder — no extra steps. A step-by-step OneDrive setup guide is available in-app (under "Downloads" in the Project menu).

One more reminder On Mac, don't put the storage path inside iCloud Drive (see Section 1.3).
15

Security & Passwords

DMS offers two completely independent layers of password protection — use just one, or both.

15.1 App login password (protects DMS on this whole machine)

An optional login password protecting the entire DMS application on this machine, regardless of which project is currently open — good for a shared computer, or whenever you want DMS unreachable by anyone without the password.

  1. Click "Project" → "Set app login password".
  2. Type a new password (twice to confirm) and save — it takes effect immediately. From then on, opening DMS in a browser requires this password first.
  3. To change or remove it, return to the same menu item (now shown as "Change or remove app login password"), verify your current password, and proceed — leave the new password blank to remove it entirely.
  4. Once a password is set, a "Log out" option appears in the menu — clicking it ends the current browser session, requiring the password again next time.
How it's stored The password is stored encrypted (SHA256 with a random salt, 100,000 iterations) in a local config file — never in plain text.

15.2 Project password (protects just one specific project)

A completely separate, second password that belongs to one specific project — it's stored inside that project's own folder, so it travels along whenever the folder is copied, synced, or handed off. It's required to open that project directly, or to import its exported .dms file anywhere else.

  1. Click "Project" → "Set project password".
  2. Optionally give the project a friendly name (shown on its lock screen and in the file picker), set a password, and save.
  3. Once set, a padlock-badged project name appears in the toolbar; a "Lock this project" option appears in the menu — clicking it requires the project password again to keep viewing (without affecting the app login above).

15.3 Storing website logins per folder

Save one or more sets of related website address, username, and password against a specific folder — click "Add another login" to keep adding more — decrypted only when you click "reveal". Any folder can hold any number of these logins, with a "copy username" quick-action button included.

A common use: keep an online account's login right alongside the files it relates to, so the record and the access point live in one place. For example, store your hospital or doctor's patient-portal login in the same folder as your "Physical Exam Reports" — next time you need to check a result or book an appointment, you're not hunting for the password separately. The same pattern works well for an equipment manufacturer's admin portal, an insurance company's website, or a bank or investment account.

15.4 Exporting every login at once

Click “Export logins” in the document toolbar and DMS builds an Excel spreadsheet and opens it in your default spreadsheet program. Each row is one login; the columns are Folder, Website, User name, Password and Folder description.

Careful The Password column is in plain text. The file is temporary — keep it safe or delete it when you're done, and don't forward it. A password shown as “(cannot decrypt…)” was saved on another computer and can't be decrypted here.
16

Other Settings

  • Download the user manual / setup guides: the "Downloads" section at the bottom of the "Project" menu offers a .docx copy of this manual, OneDrive setup guides (Chinese/English), and a Windows face-recognition install guide.
  • Switch storage location: click the storage path button in the top toolbar anytime to switch to another DMS project folder already on your computer (Chapter 17 covers the startup project picker, running several at once, and merging another project in).
  • Language switch: the whole interface can be toggled between Chinese and English anytime from the top of the screen, with no effect on your saved data.
  • No document count limit: a project can hold an unlimited number of documents.

16.1 Keyboard shortcuts

  • Confirm folder rename / new folder name: Enter; cancel: Escape
  • Start renaming a folder: double-click its name; open the folder context menu: right-click the folder
  • Close the Document Viewer: Escape, or click Close
  • Paste a clipboard image / text into the current folder: Ctrl+V (Windows) / Cmd+V (Mac)
  • Copy / select all / undo (in text fields): Ctrl+C / Ctrl+A / Ctrl+Z
  • Previous / next document in the viewer: ← / →
17

Switching, Combining & Multiple Projects

A "project" in DMS is just a storage path — one whole independent folder tree, its documents, and its settings. It's now easy to switch between projects, keep several open at once, or even combine two into one.

17.1 Choosing a project at startup

Double-click DMS.app / DMS.exe and, if DMS has already been used on this computer before, you won't be dropped straight into setting a storage location like the very first time — instead a "Choose Project" screen appears first:

  • Continue previous project: reopen whichever project was open the last time DMS quit normally.
  • Choose another project folder: browse to a different DMS project folder already on your computer.
  • Create a brand new empty project: start from scratch, with no sample data carried over.

As long as you quit normally (the "Quit DMS" button, Cmd+Q, closing the status window, and so on), DMS remembers which project was open. If several project windows were open at once, "previous" means whichever one was closed normally most recently.

17.2 Running multiple projects at once

Run several DMS project windows/instances side by side, each fully independent and logged in separately, with no need to close one before opening another.

  1. Click the "open another project" entry in the top "Project" menu, or simply launch DMS.app / DMS.exe again and pick a different project folder from Section 17.1's startup screen.
  2. DMS starts a separate window and its own service for the new project, on a different local port, fully independent of any window already open.
  3. Opening the same project folder again is detected, and DMS switches to the existing window for it instead of starting a duplicate.

17.3 Combining another project in

Good for when two computers (or two people) have each been organizing material separately for a while and now need to merge into one — folds another DMS project's folders and documents wholesale into the one you're currently using.

  1. Click "Combine Project" in the top "Project" menu.
  2. Pick the storage folder of the other project to fold in.
  3. Confirm to start: same-named folders merge together automatically, documents that are byte-for-byte identical are skipped rather than duplicated, and everything else merges normally into the current project's tree.
Back up first Combining folds another project into this one — export a full backup of the current project (Chapter 14) beforehand in case the result isn't quite what you expected.

Frequently Asked Questions

Gemini says “API quota exceeded” (429) — what now?
The free Gemini tier has daily (or per-minute) limits, and the message says roughly when it resets. Retry later, enable billing on the Google project, use another API key, or switch to DeepSeek; for translation you can pick “Fast (free)”.
I forgot the app login password — what do I do?
You can reset it by editing the local config file directly. It's located at ~/.pms_dms_config.json on Mac, or C:\Users\yourname\.pms_dms_config.json on Windows. Open it in a plain text editor and set both the password_hash and password_salt fields to an empty string "", save, and restart DMS — password protection is removed. (This is the app login password, distinct from Section 15.2's "project password", which lives inside the project's own folder — the same edit-and-clear approach works on its config file there.)
I accidentally deleted a folder or document — can I get it back?
Deleting a document only removes one reference as long as it's linked elsewhere — the file itself stays. Only once the last reference is deleted does the file move into the "Not Show in Tree" folder on disk (Section 3.2), where you can find it manually. Deleting an entire folder works the same way underneath: the folder itself disappears from the tree, but its documents also land in "Not Show in Tree" first rather than being erased from disk immediately — only deleting them again from inside that hidden folder is the real, unrecoverable step. The folder disappearing from the tree isn't something you can undo with one click, though, so it's still worth exporting a backup (Chapter 14) before any big structural change. If your storage path is inside a cloud-synced folder (like OneDrive), its version history may also help recover a deleted file.
Why does sending email show an "SMTP error"?
Common causes: Gmail without an "App Password" (using the regular account password instead); a provider's SMTP access isn't turned on (QQ Mail and 163 both require enabling it manually in their web interface first); or a company firewall blocking ports 587/465 — contact your network admin in that case.
Can documents be shared across multiple computers?
Yes. Point the storage path at a shared network location — a home or office NAS box mapped as a network drive (e.g. Z:\photo), or a cloud-synced folder (OneDrive, Dropbox — avoid iCloud Drive on Mac, see Section 1.3) — and every computer's DMS pointed at that same path shares the data. Typically only one computer can be connected at a time either way, and you should still avoid editing the same document from two computers at once to prevent conflicts — dividing work by folder is the safest approach. If you're away from home and need to reach your NAS remotely, that requires the NAS's own built-in VPN, or a separate tool like Tailscale or WireGuard to bridge the connection — DMS itself doesn't provide any network tunneling, it just reads and writes files at whatever path you give it. Genuinely simultaneous multi-user read/write (rather than time-sharing one path) is technically possible, but isn't offered as a standard product feature right now — reach out to the developer directly if you need that.
Windows says "Windows protected your PC" (SmartScreen) when I double-click DMS.exe — what now?
This happens because DMS.exe is a newly-built executable that hasn't been paid-certified with a digital signature from Microsoft — Windows blocks unrecognized new programs by default to protect users; it doesn't mean anything is wrong with it. Two ways around it:

Option 1 (recommended, simplest): on the warning screen, click the "More info" link — a "Run anyway" button appears at the bottom; click it to launch normally.

Option 2: if that window is already closed or unresponsive, right-click DMS.exe → "Properties" → check the "Unblock" checkbox near the bottom of the "General" tab → click "Apply" → "OK". This permanently clears the warning for that file going forward.
What file formats are supported?
PDF, JPG, JPEG, PNG, HEIC, TIFF, GIF, WEBP, and BMP for images, plus text formats like TXT and CSV. In principle any file type can be uploaded and stored, but in-browser preview is limited to the formats listed above — anything else can still be downloaded and opened with the matching application.
What's the difference between DMS.app and DMS.exe — are they just the Mac and Windows versions?
Exactly — they're functionally identical, just packaged executables for Mac and Windows respectively. Double-click to run; no separate Python install or dependencies needed — except for face recognition (see Section 10.4): its underlying engine is too large to bundle into the installer, so it's installed separately, with DMS guiding you through the whole process in-app.
DMS.app — User Manual Written for first-time users, compiled against the current release · rev. 2026-10-23