## How do I install PDF4LLM for .NET? Install via the .NET CLI: ```bash theme={null} dotnet add package PDF4LLM ``` Or via the Visual Studio Package Manager Console: ```powershell theme={null} Install-Package PDF4LLM ``` Or by adding a `PackageReference` directly to your `.csproj`: ```xml theme={null} ``` `MuPDF.NET` is installed automatically as a dependency — you do not need to add it separately. ## Which .NET targets are supported? PDF4LLM targets .NET Standard 2.0, making it compatible with any framework that implements that standard: | Target framework | Supported | | -------------------- | --------- | | .NET 8.0 | ✓ | | .NET 7.0 | ✓ | | .NET 6.0 | ✓ | | .NET 5.0 | ✓ | | .NET Standard 2.0 | ✓ | | .NET Framework 4.8 | ✓ | | .NET Framework 4.7.2 | ✓ | | .NET Framework 4.6.1 | ✓ | ## How do I verify my installation? Add a `using` directive and call `ToMarkdown` on any PDF: ```csharp theme={null} using MuPDF.NET; using PDF4LLM; string markdown = PdfExtractor.ToMarkdown("document.pdf"); Console.WriteLine(markdown); ``` If this prints Markdown to the console, everything is wired up correctly. ## I'm seeing an "assembly with the same simple name" conflict. How do I fix it? Your installed version of `MuPDF.NET` already bundles PDF4LLM internally. Having both packages referenced simultaneously causes the conflict. Remove the explicit `PDF4LLM` package reference and rely on the bundled version: ```bash theme={null} dotnet remove package PDF4LLM ``` Or remove the line from your `.csproj` manually: ```xml theme={null} ``` The API is identical either way — `using PDF4LLM;` and `PdfExtractor.*` work regardless of which package supplies the assembly. > A future release of MuPDF.NET will stop bundling PDF4LLM, allowing both packages to coexist without conflict. ## How do I convert a PDF to Markdown? Open a `Document` and pass it to `PdfExtractor.ToMarkdown()`: ```csharp theme={null} using MuPDF.NET; using PDF4LLM; Document doc = new Document("my-document.pdf"); string markdown = PdfExtractor.ToMarkdown(doc); doc.Close(); Console.WriteLine(markdown); ``` To save the result to a file: ```csharp theme={null} File.WriteAllText("output.md", markdown, System.Text.Encoding.UTF8); ``` ## What output formats are supported? There are three extraction methods on `PdfExtractor`, all sharing a consistent interface: | Method | Returns | Best for | | -------------- | --------------- | ----------------------------------------------------------- | | `ToMarkdown()` | `string` | LLM ingestion and RAG pipelines | | `ToJson()` | `string` (JSON) | Custom pipelines needing bounding boxes and layout metadata | | `ToText()` | `string` | Search indexing and NLP preprocessing | ```csharp theme={null} string markdown = PdfExtractor.ToMarkdown(doc); string json = PdfExtractor.ToJson(doc); string text = PdfExtractor.ToText(doc); ``` ## How do I extract only specific pages? Pass a zero-based list of page indices to the `pages` parameter. This works across all three extraction methods: ```csharp theme={null} string markdown = PdfExtractor.ToMarkdown( doc, pages: new List { 0, 1, 2 } ); ``` Page numbers are zero-indexed — page 1 of the document is `0`, page 2 is `1`, and so on. ## What document formats are supported as input? Standard formats — PDF, XPS, EPUB, MOBI, and more — are supported out of the box. See the [Supported Formats guide](/dotnet/getting-started/supported-formats) for a full list of supported input and output formats. ## How do I analyse visual layout regions (columns, figures, sidebars)? Use `PdfExtractor.ParseDocument()`. It analyses the document and returns a typed `ParsedDocument` object containing a list of `ParsedPage` objects, each with its detected blocks, tables, images, and bounding boxes in reading order. ```csharp theme={null} ParsedDocument parsed = PdfExtractor.ParseDocument(doc); foreach (var page in parsed.Pages) { Console.WriteLine($"Page {page.Number}: {page.Blocks.Count} blocks"); } ``` ## How do I extract AcroForm field values from a filled PDF? Use `PdfExtractor.GetKeyValues()`. It returns a `List`, each with `Name`, `Value`, and `Page` properties: ```csharp theme={null} List fields = PdfExtractor.GetKeyValues(doc); foreach (var field in fields) { Console.WriteLine($"{field.Name} (page {field.Page}): {field.Value}"); } ``` > Note: `GetKeyValues()` does not accept a `pages` parameter — it always processes the full document. ## Does it handle scanned or image-based PDFs? Yes, via Tesseract OCR. Unlike the Python library, OCR is **not** triggered automatically — you must opt in with `useOcr: true`: ```csharp theme={null} string markdown = PdfExtractor.ToMarkdown(doc, useOcr: true); ``` The same flag works across all three extraction methods: ```csharp theme={null} string text = PdfExtractor.ToText(doc, useOcr: true); ParsedDocument parsed = PdfExtractor.ParseDocument(doc, useOcr: true); ``` OCR output goes through the same layout analysis as native text, so reading order, heading detection, and table detection all apply. ## How do I install Tesseract for OCR? Tesseract must be installed on the host system and available on the `PATH`. PDF4LLM does not bundle it. **Windows** — Download the installer from [UB Mannheim Tesseract builds](https://github.com/UB-Mannheim/tesseract/wiki) and add the install directory (e.g. `C:\Program Files\Tesseract-OCR`) to your `PATH`. **macOS** ```bash theme={null} brew install tesseract ``` **Linux (Debian / Ubuntu)** ```bash theme={null} sudo apt-get install tesseract-ocr ``` Verify Tesseract is reachable from the application's environment: ```bash theme={null} tesseract --version ``` If Tesseract is installed but not on the `PATH`, you will get a `TesseractNotFoundException` at runtime. ## How do I use OCR with a non-English language? Pass a Tesseract language code to `ocrLanguage`. The default is `"eng"`. Combine multiple languages with a `+`: ```csharp theme={null} // Single language string markdown = PdfExtractor.ToMarkdown(doc, useOcr: true, ocrLanguage: "fra"); // Multiple languages mixed on the same pages string markdown = PdfExtractor.ToMarkdown(doc, useOcr: true, ocrLanguage: "eng+deu"); ``` The corresponding Tesseract language packs must be installed on your system first. On Debian/Ubuntu: ```bash theme={null} sudo apt-get install tesseract-ocr-fra tesseract-ocr-deu ``` Common language codes: `eng` (English), `fra` (French), `deu` (German), `spa` (Spanish), `jpn` (Japanese), `chi_sim` (Simplified Chinese), `chi_tra` (Traditional Chinese). ## How do I handle documents that mix scanned and native text pages? Use a per-page native probe to identify which pages need OCR, then extract each set separately: ```csharp theme={null} var scannedPages = new List(); for (int i = 0; i < doc.PageCount; i++) { string native = PdfExtractor.ToText(doc, pages: new List { i }); if (native.Trim().Length < 50) scannedPages.Add(i); } string ocrMarkdown = scannedPages.Count > 0 ? PdfExtractor.ToMarkdown(doc, pages: scannedPages, useOcr: true) : string.Empty; var nativePages = Enumerable.Range(0, doc.PageCount).Except(scannedPages).ToList(); string nativeMarkdown = nativePages.Count > 0 ? PdfExtractor.ToMarkdown(doc, pages: nativePages) : string.Empty; ``` Adjust the character threshold (`< 50`) to suit your documents — pages with only a page number or short heading will score low. ## How do I run OCR in a Docker container? Add Tesseract to your `Dockerfile`: ```dockerfile theme={null} FROM mcr.microsoft.com/dotnet/runtime:8.0 RUN apt-get update && apt-get install -y \ tesseract-ocr \ tesseract-ocr-eng \ && rm -rf /var/lib/apt/lists/* WORKDIR /app COPY --from=build /app/publish . ENTRYPOINT ["dotnet", "MyApp.dll"] ``` Verify Tesseract is on the `PATH` inside the built image: ```bash theme={null} docker run --rm your-image tesseract --version ``` If you install language data to a custom location, set `TESSDATA_PREFIX` explicitly: ```dockerfile theme={null} ENV TESSDATA_PREFIX=/usr/share/tesseract-ocr/4.00/tessdata/ ``` ## Does it integrate with LlamaIndex? Yes. Use `PdfExtractor.LlamaMarkdownReader()` to get a `PDFMarkdownReader` instance, then call `LoadData()` to get one `LlamaDocument` per page with Markdown text and metadata: ```csharp theme={null} var reader = PdfExtractor.LlamaMarkdownReader(); var pages = reader.LoadData("product-manual.pdf"); foreach (var page in pages) { int pageNum = (int)page.ExtraInfo["page"]; string filePath = (string)page.ExtraInfo["file_path"]; Console.WriteLine($"Page {pageNum}: {page.Text.Length} chars"); } ``` ## How do I use PDF4LLM with Azure OpenAI? Install the Azure OpenAI SDK alongside PDF4LLM: ```bash theme={null} dotnet add package PDF4LLM dotnet add package Azure.AI.OpenAI ``` Extract Markdown and pass it to a chat completion for summarisation or Q\&A: ```csharp theme={null} using Azure; using Azure.AI.OpenAI; using MuPDF.NET; using PDF4LLM; AzureOpenAIClient client = new( new Uri(Environment.GetEnvironmentVariable("AZURE_OPENAI_ENDPOINT")!), new AzureKeyCredential(Environment.GetEnvironmentVariable("AZURE_OPENAI_API_KEY")!) ); Document doc = new Document("briefing.pdf"); string markdown = PdfExtractor.ToMarkdown(doc); doc.Close(); var chatClient = client.GetChatClient("gpt-4o"); var result = await chatClient.CompleteChatAsync( [ new SystemChatMessage("You are a precise document summariser."), new UserChatMessage($"Summarise this document in five bullet points:\n\n{markdown}") ]); Console.WriteLine(result.Value.Content[0].Text); ``` For RAG pipelines, embed per-page chunks using `LlamaMarkdownReader` and `text-embedding-3-small`, then retrieve by cosine similarity before passing context to a chat model. See the [Azure OpenAI integration guide](https://docs.pdf4llm.com/dotnet/integrations/azure) for full patterns including batched embedding, parallel OCR, multimodal image input, and Managed Identity auth. ## How do I use Managed Identity instead of an API key for Azure OpenAI? Replace `AzureKeyCredential` with `DefaultAzureCredential`: ```csharp theme={null} using Azure.Identity; AzureOpenAIClient client = new( new Uri(Environment.GetEnvironmentVariable("AZURE_OPENAI_ENDPOINT")!), new DefaultAzureCredential() ); ``` `DefaultAzureCredential` works in Azure App Service, Azure Functions, AKS, and other managed environments. For local development, run `az login` first, or use `AzureCliCredential` explicitly. Assign the `Cognitive Services OpenAI User` role to the managed identity in the Azure Portal or via the CLI: ```bash theme={null} az role assignment create \ --role "Cognitive Services OpenAI User" \ --assignee \ --scope /subscriptions/_{/resourceGroups//providers/Microsoft.CognitiveServices/accounts/
```}