Imported from aspose-words/agentic-net-examples (extraction/AGENTS.md). Install upstream with npx skills add aspose-words/agentic-net-examples --skill extraction. Copyright stays with the author.

AGENTS.md - Extraction

Purpose

This folder is a live, curated example set for Extraction scenarios. Treat every .cs file as a standalone console application. The goal is correct, warning-free examples that use documented Aspose APIs and match the original task intent.

Non-negotiable conventions

Use native Aspose.Words extraction APIs directly.
Bootstrap local source documents, files, images, streams, or folders whenever the task implies an existing source.
Enumerate actual document nodes and validate their types before extracting content.
Use Aspose.Words.Tables for Table, Row, and Cell when structured extraction is required.
Use Newtonsoft.Json for JSON serialization tasks and Aspose.Drawing instead of System.Drawing when drawing-related types are needed.
Guard maybe-null values to avoid nullable-reference warnings such as CS8600, CS8602, and CS8604.

Recommended workflow selection

Text Range Extraction workflow: 11 examples
Targeted Node Extraction workflow: 9 examples
Table Structured Extraction workflow: 6 examples
Input Bootstrap workflow: 1 examples
Image Shape Extraction workflow: 3 examples

Use the simplest workflow that satisfies the task.

Validation priorities

The code must compile and run without manual input.
API usage must be supported by the configured package versions.
Exported outputs must actually be written by the example.
Validation scenarios must inspect only the behavior requested by the task.
Examples that depend on files, folders, streams, images, or data should bootstrap those inputs locally during the example run.

File-to-task reference

load-a-docx-file-extract-content-between-two-paragraphs-and-save-the-result-as-a-new-docx.cs
- Task: Load a DOCX file, extract content between two paragraphs, and save the result as a new DOCX.
- Workflow: text-range-extraction
- Outputs: docx
- Selected engine: mcp
load-a-docm-file-extract-content-between-a-macro-enabled-field-and-a-paragraph-and-save-as.cs
- Task: Load a DOCM file, extract content between a macro-enabled field and a paragraph, and save as DOCX.
- Workflow: targeted-node-extraction
- Outputs: docx, doc
- Selected engine: mcp
identify-a-start-run-node-and-an-end-bookmark-node-then-extract-the-intervening-nodes-into.cs
- Task: Identify a start run node and an end bookmark node, then extract the intervening nodes into a document.
- Workflow: targeted-node-extraction
- Outputs: docx
- Selected engine: mcp
programmatically-determine-start-and-end-nodes-based-on-paragraph-styles-then-extract-the.cs
- Task: Programmatically determine start and end nodes based on paragraph styles, then extract the styled content segment.
- Workflow: text-range-extraction
- Outputs: docx
- Selected engine: mcp
extract-a-mixed-node-range-that-starts-with-a-table-cell-and-ends-with-a-paragraph-maintai.cs
- Task: Extract a mixed node range that starts with a table cell and ends with a paragraph, maintaining layout.
- Workflow: table-structured-extraction
- Outputs: docx
- Selected engine: mcp
extract-a-range-of-nodes-that-includes-tables-images-and-fields-preserving-original-hierar.cs
- Task: Extract a range of nodes that includes tables, images, and fields, preserving original hierarchy in the output.
- Workflow: table-structured-extraction
- Outputs: docx
- Selected engine: mcp
extract-content-between-a-run-node-and-the-next-bookmark-then-convert-the-extracted-segmen.cs
- Task: Extract content between a run node and the next bookmark, then convert the extracted segment to HTML format.
- Workflow: targeted-node-extraction
- Outputs: html
- Selected engine: mcp
extract-content-between-a-run-node-and-the-following-table-then-convert-the-extracted-port.cs
- Task: Extract content between a run node and the following table, then convert the extracted portion to XPS format.
- Workflow: table-structured-extraction
- Outputs: xps
- Selected engine: mcp
use-the-extraction-api-to-copy-content-between-two-headings-and-insert-it-into-a-template.cs
- Task: Use the extraction API to copy content between two headings and insert it into a template document.
- Workflow: text-range-extraction
- Outputs: docx
- Selected engine: mcp
use-documentbuilder-to-prepend-extracted-node-collection-to-the-beginning-of-a-new-documen.cs
- Task: Use DocumentBuilder to prepend extracted node collection to the beginning of a new document before saving.
- Workflow: text-range-extraction
- Outputs: docx
- Selected engine: mcp
use-documentbuilder-to-insert-extracted-node-collection-into-a-new-document-at-a-custom-bo.cs
- Task: Use DocumentBuilder to insert extracted node collection into a new document at a custom bookmark location.
- Workflow: targeted-node-extraction
- Outputs: docx
- Selected engine: llm
duplicate-extracted-content-between-a-table-and-a-field-node-within-the-original-document.cs
- Task: Duplicate extracted content between a table and a field node within the original document without altering formatting.
- Workflow: table-structured-extraction
- Outputs: docx
- Selected engine: llm
save-extracted-content-as-a-docx-file-while-preserving-embedded-fields-and-their-evaluatio.cs
- Task: Save extracted content as a DOCX file while preserving embedded fields and their evaluation results.
- Workflow: targeted-node-extraction
- Outputs: docx
- Selected engine: mcp
batch-process-multiple-word-files-extracting-content-between-specified-nodes-and-saving-ea.cs
- Task: Batch process multiple Word files, extracting content between specified nodes and saving each extraction as an individual PDF.
- Workflow: input-bootstrap
- Outputs: pdf
- Selected engine: mcp
batch-extract-images-from-shape-nodes-in-documents-and-generate-a-csv-manifest-listing-ima.cs
- Task: Batch extract images from shape nodes in documents and generate a CSV manifest listing image names and sources.
- Workflow: table-structured-extraction
- Outputs: csv
- Selected engine: mcp
extract-all-images-from-shape-nodes-across-a-document-collection-and-compile-them-into-a-s.cs
- Task: Extract all images from shape nodes across a document collection and compile them into a single ZIP archive.
- Workflow: image-shape-extraction
- Outputs: docx
- Selected engine: mcp
extract-a-document-segment-that-includes-nested-tables-and-ensure-nested-structures-are-re.cs
- Task: Extract a document segment that includes nested tables and ensure nested structures are retained in the new file.
- Workflow: table-structured-extraction
- Outputs: docx
- Selected engine: mcp
create-a-reusable-extraction-utility-that-accepts-node-identifiers-and-returns-a-document.cs
- Task: Create a reusable extraction utility that accepts node identifiers and returns a Document containing the selected content.
- Workflow: text-range-extraction
- Outputs: docx
- Selected engine: mcp
implement-error-handling-for-cases-where-the-start-node-appears-after-the-end-node-during.cs
- Task: Implement error handling for cases where the start node appears after the end node during extraction.
- Workflow: text-range-extraction
- Outputs: docx
- Selected engine: mcp
implement-a-custom-node-filter-to-exclude-comments-while-extracting-content-between-two-pa.cs
- Task: Implement a custom node filter to exclude comments while extracting content between two paragraphs.
- Workflow: targeted-node-extraction
- Outputs: docx
- Selected engine: mcp
extract-content-between-two-bookmark-nodes-and-replace-the-original-range-with-a-placehold.cs
- Task: Extract content between two bookmark nodes and replace the original range with a placeholder paragraph.
- Workflow: targeted-node-extraction
- Outputs: docx
- Selected engine: existing_repo
extract-content-between-a-paragraph-and-a-comment-node-then-log-the-extracted-text-to-a-mo.cs
- Task: Extract content between a paragraph and a comment node, then log the extracted text to a monitoring system.
- Workflow: targeted-node-extraction
- Outputs: docx
- Selected engine: mcp
automate-extraction-of-footnote-content-between-specified-nodes-and-export-each-footnote-a.cs
- Task: Automate extraction of footnote content between specified nodes and export each footnote as a separate text file.
- Workflow: targeted-node-extraction
- Outputs: txt
- Selected engine: mcp
create-a-command-line-tool-that-accepts-start-and-end-node-ids-and-outputs-the-extracted-s.cs
- Task: Create a command-line tool that accepts start and end node IDs and outputs the extracted segment as PDF.
- Workflow: text-range-extraction
- Outputs: pdf
- Selected engine: llm
create-a-unit-test-that-verifies-extraction-of-content-between-two-specific-paragraphs-ret.cs
- Task: Create a unit test that verifies extraction of content between two specific paragraphs retains original text styling.
- Workflow: text-range-extraction
- Outputs: docx
- Selected engine: existing_repo
develop-a-macro-that-calls-the-extraction-api-to-copy-selected-content-into-the-clipboard.cs
- Task: Develop a macro that calls the extraction API to copy selected content into the clipboard for pasting elsewhere.
- Workflow: text-range-extraction
- Outputs: docx
- Selected engine: llm
extract-a-range-that-starts-inside-a-shape-s-image-and-ends-at-a-field-preserving-both-ele.cs
- Task: Extract a range that starts inside a shape's image and ends at a field, preserving both elements.
- Workflow: image-shape-extraction
- Outputs: docx
- Selected engine: mcp
extract-content-between-two-nodes-in-a-document-then-encrypt-the-resulting-file-using-a-pa.cs
- Task: Extract content between two nodes in a document, then encrypt the resulting file using a password.
- Workflow: text-range-extraction
- Outputs: docx
- Selected engine: existing_repo
implement-parallel-processing-to-extract-node-ranges-from-multiple-documents-simultaneousl.cs
- Task: Implement parallel processing to extract node ranges from multiple documents simultaneously, improving performance.
- Workflow: text-range-extraction
- Outputs: docx
- Selected engine: mcp
extract-images-from-shape-nodes-and-embed-them-directly-into-a-new-docx-document.cs
- Task: Extract images from shape nodes and embed them directly into a new DOCX document.
- Workflow: image-shape-extraction
- Outputs: docx
- Selected engine: mcp

Common failure patterns and preferred agent fixes

Invalid node insertion in destination document
- Symptom: Runtime failures such as 'Cannot insert a node of this type at this location'.
- Preferred fix: Rebuild a valid destination document structure explicitly and insert cloned inline or block nodes only into supported parents.
Table namespace assumptions
- Symptom: Compile errors because Table, Row, or Cell were used without Aspose.Words.Tables.
- Preferred fix: Import Aspose.Words.Tables or use fully qualified table node type names.
Weak bookmark boundary logic
- Symptom: Bookmark-bounded extraction or replacement fails because the wrong nodes are used or placeholder insertion is invalid.
- Preferred fix: Use real BookmarkStart and BookmarkEnd boundaries and insert a valid placeholder Paragraph into a supported block container.
Footnote-specific API issues
- Symptom: Footnote export or enumeration fails because footnote APIs or namespaces are used incorrectly.
- Preferred fix: Use Aspose.Words.Notes.Footnote and Aspose.Words.Notes.FootnoteType explicitly and enumerate actual footnote nodes.
Font or drawing ambiguity
- Symptom: Compile errors due to System.Drawing usage or ambiguous Font references.
- Preferred fix: Use Aspose.Drawing only and prefer explicit Aspose.Drawing type names when ambiguity is possible.

Build and run contract

Target framework: net8.0
Package: Aspose.Words 26.8.0
Package: Newtonsoft.Json

Command reference

Create a temporary console project

dotnet new console -n ExampleProject --framework net8.0
cd ExampleProject

Add required packages

dotnet add package Aspose.Words --version 26.8.0
dotnet add package Newtonsoft.Json

Copy a category example into the temp project

Copy-Item ..\extraction\<example-file>.cs .\Program.cs

Build and run

dotnet build --configuration Release --verbosity minimal
dotnet run --configuration Release --no-build

Category update guidance

Preserve file-to-task traceability. Any future update should keep the original task text associated with the file in metadata.
When replacing a file, prefer the latest verified publish-ready version rather than a merely compiling draft.
Keep examples standalone, bootstrap missing inputs locally, and prefer documented Aspose APIs over speculative shortcuts.

extraction

About