concordance

name	concordance
description	Create a LibreOffice Writer concordance file (.sdi, semicolon-delimited) for use with the AutoMark feature to automatically generate a book index. Use when given text, a chapter, or a list of terms that need to be indexed.
metadata	{"author":"cascade","version":"1.0","date":"2026-04-17"}

Concordance File Skill

A concordance file is a semicolon-delimited text file used by LibreOffice Writer's Insert → Table of Contents and Index → AutoMark feature to automatically find and mark index entries throughout a document.

File Format

Each line is one index entry with six semicolon-separated fields:

Search term;Alternative entry;1st key;2nd key;Case sensitive;Word only

Field Reference

Field	Required	Description
Search term	Yes	Exact text to find in the document
Alternative entry	No	Text shown in the index (uses search term if empty)
1st key	No	Primary category / top-level index heading
2nd key	No	Secondary category for nested sub-entries
Case sensitive	No	1 = match case, 0 or empty = ignore case
Word only	No	1 = whole word only, 0 or empty = match substrings

Example Lines

NetCDF;NetCDF;File Formats;;0;1
HDF5;HDF5;File Formats;;0;1
dispatch table;dispatch table;Architecture;;0;1
nc_open;;API functions;;0;1

How to Generate a Concordance CSV

Given source text (a chapter, section, or term list), follow these steps:

Step 1 — Identify Index-Worthy Terms

Extract terms that are:

• Proper nouns: product names, library names, standards (e.g., NetCDF, HDF5, Zarr)
• Technical terms: functions, data structures, algorithms, protocols
• Concepts: architecture patterns, design principles, key ideas
• Acronyms: with and without expansion (index both forms)
• People and organizations: authors, standards bodies

Avoid indexing:

• Common words (the, and, is, for)
• Terms appearing only once and not significant
• Overly generic terms that would produce hundreds of entries

Step 2 — Assign Categories (1st key)

Group related terms under a shared 1st key. Typical categories for technical books:

• API functions — function/method names like nc_open, nc_close
• Data structures — structs, classes, types
• File formats — NetCDF, HDF5, CSV, GeoTIFF
• Concepts — chunking, compression, dispatch
• Tools — command-line tools, utilities
• Standards — CF Conventions, OGC, ISO

For non-technical books, categories reflect subject matter:

• People, Places, Events, Terms, Organizations

Step 3 — Decide Case and Word Matching

• Use Word only=1 for most technical terms to avoid false matches (e.g., "var" matching "variable")
• Use Case sensitive=1 for proper nouns, acronyms, and function names
• Use Case sensitive=0 for concepts that may appear mid-sentence in lowercase

Step 4 — Write the CSV

Output as plain text, one entry per line, no header row, semicolons as delimiters:

Search term;Alternative entry;1st key;2nd key;Case sensitive;Word only

Save with .sdi extension in UTF-8 or ANSI encoding.

Complete Example

Given text about NetCDF file I/O, a concordance might look like:

NetCDF;NetCDF;File formats;;0;1
HDF5;HDF5;File formats;;0;1
Zarr;Zarr;File formats;;0;1
nc_open;nc_open();API functions;;1;1
nc_close;nc_close();API functions;;1;1
nc_get_vara;nc_get_vara();API functions;;1;1
dispatch table;dispatch table;Architecture;;0;1
NC_Dispatch;NC_Dispatch;Data structures;;1;1
chunking;chunking;Concepts;;0;1
compression;compression;Concepts;;0;1
OPeNDAP;OPeNDAP;Protocols;;1;1
DAP2;DAP2;Protocols;;1;1
DAP4;DAP4;Protocols;;1;1

Output Format Rules

1. No header row in the output file — LibreOffice expects data on every line
2. Semicolons as delimiters (not commas)
3. No quotes around fields unless the field contains a semicolon
4. One entry per line
5. Sort alphabetically by search term for readability (optional but helpful)
6. UTF-8 encoding for documents with non-ASCII characters

Tips

• A term can appear multiple times under different 1st keys if it belongs to multiple categories
• The Alternative entry can be used to normalize capitalization in the index (e.g., search hdf5, display HDF5)
• For acronyms, add two entries: one for the acronym, one for the expanded form, both pointing to the same 1st key
• Keep 2nd key sparse — only use it when a category genuinely has sub-groupings (e.g., 1st key API functions, 2nd key read vs write)