DCIT/DCIT.Core/Services/OfficeDocConversionService.cs

using System;
using System.Collections.Generic;
using System.Diagnostics;
using System.IO;
using System.Linq;
using System.Runtime.InteropServices;

namespace DCIT.Core.Services
{
    public class OfficeDocConversionService : IDocConversionService, IDocumentFieldUpdateService
    {
        private const int WdFormatDocument97 = 0;
        private const int WdFormatDocumentDefault = 16;
        private static readonly string[] WpsProgIds =
        {
            "KWPS.Application",
            "kwps.Application",
            "WPS.Application",
            "wps.Application",
        };

        private readonly string _applicationBaseDirectory;

        public OfficeDocConversionService(string applicationBaseDirectory = null)
        {
            _applicationBaseDirectory = string.IsNullOrWhiteSpace(applicationBaseDirectory)
                ? AppContext.BaseDirectory
                : applicationBaseDirectory;
        }

        public void Convert(string sourcePath, string outputPath)
        {
            if (string.IsNullOrWhiteSpace(sourcePath))
            {
                throw new ArgumentNullException("sourcePath");
            }

            if (string.IsNullOrWhiteSpace(outputPath))
            {
                throw new ArgumentNullException("outputPath");
            }

            var sourceExtension = NormalizeExtension(sourcePath);
            var outputExtension = NormalizeExtension(outputPath);
            if (!IsSupportedConversion(sourceExtension, outputExtension))
            {
                throw new NotSupportedException("仅支持 .doc 与 .docx 之间互相转换。");
            }

            var outputDirectory = Path.GetDirectoryName(outputPath);
            if (!string.IsNullOrWhiteSpace(outputDirectory))
            {
                Directory.CreateDirectory(outputDirectory);
            }

            if (File.Exists(outputPath))
            {
                File.Delete(outputPath);
            }

            var errors = new List<string>();
            foreach (var progId in GetOfficeAutomationProgIds())
            {
                try
                {
                    ConvertWithOfficeAutomationBackend(progId, sourcePath, outputPath, outputExtension);
                    return;
                }
                catch (Exception ex)
                {
                    errors.Add(progId + ": " + ex.Message);
                }
            }

            var docToPath = ResolveDocToPath();
            if (!string.IsNullOrWhiteSpace(docToPath))
            {
                try
                {
                    ConvertWithDocToBackend(docToPath, sourcePath, outputPath, outputExtension);
                    return;
                }
                catch (Exception ex)
                {
                    errors.Add("DocTo: " + ex.Message);
                }
            }

            throw new InvalidOperationException(
                "未找到可用的 .doc 转换链路。请确认已安装可自动化的 Microsoft Word 或 WPS；如自动化不可用，可提供 docto.exe 作为兜底转换工具。"
                + Environment.NewLine
                + string.Join(Environment.NewLine, errors.Where(item => !string.IsNullOrWhiteSpace(item))));
        }

        public void UpdateDynamicContent(string documentPath)
        {
            if (string.IsNullOrWhiteSpace(documentPath))
            {
                throw new ArgumentNullException("documentPath");
            }

            if (!File.Exists(documentPath))
            {
                throw new FileNotFoundException("要更新动态内容的文档不存在。", documentPath);
            }

            var extension = NormalizeExtension(documentPath);
            if (!string.Equals(extension, ".docx", StringComparison.OrdinalIgnoreCase)
                && !string.Equals(extension, ".doc", StringComparison.OrdinalIgnoreCase))
            {
                throw new NotSupportedException("动态内容更新仅支持 .doc 或 .docx 文档。");
            }

            var errors = new List<string>();
            foreach (var progId in GetOfficeAutomationProgIds())
            {
                try
                {
                    UpdateDynamicContentWithOfficeAutomationBackend(progId, documentPath);
                    return;
                }
                catch (Exception ex)
                {
                    errors.Add(progId + ": " + ex.Message);
                }
            }

            throw new InvalidOperationException(
                "未找到可用的目录/域刷新链路。请确保已安装可自动化的 Microsoft Word 或 WPS。"
                + Environment.NewLine
                + string.Join(Environment.NewLine, errors.Where(item => !string.IsNullOrWhiteSpace(item))));
        }

        protected virtual IEnumerable<string> GetOfficeAutomationProgIds()
        {
            yield return "Word.Application";
            foreach (var progId in WpsProgIds)
            {
                yield return progId;
            }
        }

        protected virtual string ResolveDocToPath()
        {
            var candidates = new List<string>();
            var envPath = Environment.GetEnvironmentVariable("DCIT_DOCTO_PATH");
            if (!string.IsNullOrWhiteSpace(envPath))
            {
                candidates.Add(envPath);
            }

            candidates.Add(Path.Combine(_applicationBaseDirectory, "docto.exe"));
            candidates.Add(Path.Combine(_applicationBaseDirectory, "tools", "DocTo", "docto.exe"));

            foreach (var candidate in candidates)
            {
                if (!string.IsNullOrWhiteSpace(candidate) && File.Exists(candidate))
                {
                    return candidate;
                }
            }

            var pathEnv = Environment.GetEnvironmentVariable("PATH") ?? string.Empty;
            foreach (var segment in pathEnv.Split(new[] { ';' }, StringSplitOptions.RemoveEmptyEntries))
            {
                try
                {
                    var candidate = Path.Combine(segment.Trim(), "docto.exe");
                    if (File.Exists(candidate))
                    {
                        return candidate;
                    }
                }
                catch
                {
                }
            }

            return null;
        }

        protected virtual void ConvertWithDocToBackend(string docToPath, string sourcePath, string outputPath, string outputExtension)
        {
            var format = string.Equals(outputExtension, ".docx", StringComparison.OrdinalIgnoreCase)
                ? "wdFormatDocumentDefault"
                : "wdFormatDocument97";

            var startInfo = new ProcessStartInfo
            {
                FileName = docToPath,
                Arguments = string.Format("-F \"{0}\" -O \"{1}\" -T {2}", sourcePath, outputPath, format),
                UseShellExecute = false,
                RedirectStandardOutput = true,
                RedirectStandardError = true,
                CreateNoWindow = true,
                WorkingDirectory = Path.GetDirectoryName(docToPath) ?? _applicationBaseDirectory,
            };

            using (var process = Process.Start(startInfo))
            {
                if (process == null)
                {
                    throw new InvalidOperationException("无法启动 docto.exe。");
                }

                var standardOutput = process.StandardOutput.ReadToEnd();
                var standardError = process.StandardError.ReadToEnd();
                process.WaitForExit(120000);

                if (process.ExitCode != 0 || !File.Exists(outputPath))
                {
                    throw new InvalidOperationException(
                        "docto.exe 转换失败。"
                        + Environment.NewLine
                        + (string.IsNullOrWhiteSpace(standardOutput) ? string.Empty : standardOutput + Environment.NewLine)
                        + standardError);
                }
            }
        }

        protected virtual void ConvertWithOfficeAutomationBackend(string progId, string sourcePath, string outputPath, string outputExtension)
        {
            ConvertWithWordAutomation(progId, sourcePath, outputPath, outputExtension);
        }

        protected virtual void UpdateDynamicContentWithOfficeAutomationBackend(string progId, string documentPath)
        {
            UpdateDynamicContentWithWordAutomation(progId, documentPath);
        }

        private static bool IsSupportedConversion(string sourceExtension, string outputExtension)
        {
            if (string.Equals(sourceExtension, outputExtension, StringComparison.OrdinalIgnoreCase))
            {
                return false;
            }

            return (string.Equals(sourceExtension, ".doc", StringComparison.OrdinalIgnoreCase)
                    || string.Equals(sourceExtension, ".docx", StringComparison.OrdinalIgnoreCase))
                && (string.Equals(outputExtension, ".doc", StringComparison.OrdinalIgnoreCase)
                    || string.Equals(outputExtension, ".docx", StringComparison.OrdinalIgnoreCase));
        }

        private static string NormalizeExtension(string path)
        {
            return Path.GetExtension(path) ?? string.Empty;
        }

        private static void ConvertWithWordAutomation(string progId, string sourcePath, string outputPath, string outputExtension)
        {
            var appType = Type.GetTypeFromProgID(progId, false);
            if (appType == null)
            {
                throw new InvalidOperationException("未注册 COM ProgID: " + progId);
            }

            object application = null;
            object documents = null;
            object document = null;

            try
            {
                application = Activator.CreateInstance(appType);
                if (application == null)
                {
                    throw new InvalidOperationException("无法创建 COM 实例: " + progId);
                }

                TrySetProperty(application, "Visible", false);
                TrySetProperty(application, "DisplayAlerts", 0);
                documents = GetProperty(application, "Documents");
                if (documents == null)
                {
                    throw new InvalidOperationException("无法获取 Documents 集合。");
                }

                document = InvokeMethod(documents, "Open", sourcePath, false, true);
                if (document == null)
                {
                    throw new InvalidOperationException("无法打开文档。");
                }

                var fileFormat = string.Equals(outputExtension, ".docx", StringComparison.OrdinalIgnoreCase)
                    ? WdFormatDocumentDefault
                    : WdFormatDocument97;

                try
                {
                    InvokeMethod(document, "SaveAs2", outputPath, fileFormat);
                }
                catch (MissingMethodException)
                {
                    InvokeMethod(document, "SaveAs", outputPath, fileFormat);
                }

                if (!File.Exists(outputPath))
                {
                    throw new InvalidOperationException("Office 自动化未生成目标文件。");
                }
            }
            finally
            {
                SafeCloseDocument(document);
                SafeQuitApplication(application);
                ReleaseComObject(document);
                ReleaseComObject(documents);
                ReleaseComObject(application);
            }
        }

        private static void UpdateDynamicContentWithWordAutomation(string progId, string documentPath)
        {
            var appType = Type.GetTypeFromProgID(progId, false);
            if (appType == null)
            {
                throw new InvalidOperationException("未注册 COM ProgID: " + progId);
            }

            object application = null;
            object documents = null;
            object document = null;

            try
            {
                application = Activator.CreateInstance(appType);
                if (application == null)
                {
                    throw new InvalidOperationException("无法创建 COM 实例: " + progId);
                }

                TrySetProperty(application, "Visible", false);
                TrySetProperty(application, "DisplayAlerts", 0);
                documents = GetProperty(application, "Documents");
                if (documents == null)
                {
                    throw new InvalidOperationException("无法获取 Documents 集合。");
                }

                document = InvokeMethod(documents, "Open", documentPath, false, false);
                if (document == null)
                {
                    throw new InvalidOperationException("无法打开文档。");
                }

                TryInvokeMethod(document, "Activate");
                TryInvokeMethod(document, "Repaginate");
                TryUpdateCollectionItems(document, "TablesOfContents", "Update", "UpdatePageNumbers");
                TryUpdateCollectionItems(document, "TablesOfFigures", "Update");
                TryUpdateCollectionItems(document, "TablesOfAuthorities", "Update");
                TryUpdateCollection(document, "Fields", "Update");
                TryUpdateCollection(document, "Indexes", "Update");
                TryInvokeMethod(document, "Repaginate");
                InvokeMethod(document, "Save");
            }
            finally
            {
                SafeCloseDocument(document);
                SafeQuitApplication(application);
                ReleaseComObject(document);
                ReleaseComObject(documents);
                ReleaseComObject(application);
            }
        }

        private static object GetProperty(object instance, string propertyName)
        {
            return instance.GetType().InvokeMember(
                propertyName,
                System.Reflection.BindingFlags.GetProperty,
                null,
                instance,
                null);
        }

        private static void TrySetProperty(object instance, string propertyName, object value)
        {
            try
            {
                instance.GetType().InvokeMember(
                    propertyName,
                    System.Reflection.BindingFlags.SetProperty,
                    null,
                    instance,
                    new[] { value });
            }
            catch
            {
            }
        }

        private static object InvokeMethod(object instance, string methodName, params object[] arguments)
        {
            return instance.GetType().InvokeMember(
                methodName,
                System.Reflection.BindingFlags.InvokeMethod,
                null,
                instance,
                arguments);
        }

        private static bool TryInvokeMethod(object instance, string methodName, params object[] arguments)
        {
            try
            {
                InvokeMethod(instance, methodName, arguments);
                return true;
            }
            catch
            {
                return false;
            }
        }

        private static void TryUpdateCollection(object instance, string propertyName, string updateMethodName)
        {
            object collection = null;
            try
            {
                collection = GetProperty(instance, propertyName);
                if (collection == null)
                {
                    return;
                }

                TryInvokeMethod(collection, updateMethodName);
            }
            finally
            {
                ReleaseComObject(collection);
            }
        }

        private static void TryUpdateCollectionItems(object instance, string propertyName, params string[] methodNames)
        {
            object collection = null;
            try
            {
                collection = GetProperty(instance, propertyName);
                if (collection == null)
                {
                    return;
                }

                var countObject = GetProperty(collection, "Count");
                var count = System.Convert.ToInt32(countObject);
                for (var index = 1; index <= count; index++)
                {
                    object item = null;
                    try
                    {
                        item = InvokeMethod(collection, "Item", index);
                        foreach (var methodName in methodNames.Where(name => !string.IsNullOrWhiteSpace(name)))
                        {
                            TryInvokeMethod(item, methodName);
                        }
                    }
                    finally
                    {
                        ReleaseComObject(item);
                    }
                }
            }
            catch
            {
            }
            finally
            {
                ReleaseComObject(collection);
            }
        }

        private static void SafeCloseDocument(object document)
        {
            if (document == null)
            {
                return;
            }

            try
            {
                InvokeMethod(document, "Close", false);
            }
            catch
            {
            }
        }

        private static void SafeQuitApplication(object application)
        {
            if (application == null)
            {
                return;
            }

            try
            {
                InvokeMethod(application, "Quit", false);
            }
            catch
            {
            }
        }

        private static void ReleaseComObject(object instance)
        {
            try
            {
                if (instance != null && Marshal.IsComObject(instance))
                {
                    Marshal.FinalReleaseComObject(instance);
                }
            }
            catch
            {
            }
        }
    }
}