build_index_from_sources

Build a complete search index from multiple source files and directories. This is the primary method for index construction. It handles file discovery, text extraction, chunking, embedding generation, and storage.

Parameters

sources

list[Path]Required

List of Path objects pointing to files and/or directories to index.

output_file

strRequired

Output path for the .swsearch file (SQLite backend) or collection name (pgvector).

file_types

list[str]Required

File extensions to include when scanning directories (e.g., ["md", "txt", "py"]).

exclude_patterns

Optional[list[str]]

Glob patterns for files to exclude (e.g., ["**/node_modules/**"]).

languages

Optional[list[str]]

List of language codes to support. Defaults to ["en"].

overwrite

boolDefaults to false

For the pgvector backend, drop and recreate the collection if it already exists.

Returns

None

Example

1 from pathlib import Path
2 from signalwire.search import IndexBuilder
3 
4 builder = IndexBuilder(chunking_strategy="markdown", verbose=True)
5 builder.build_index_from_sources(
6     sources=[Path("./docs"), Path("./examples")],
7     output_file="knowledge.swsearch",
8     file_types=["md", "txt", "py"],
9     exclude_patterns=["**/test_*"],
10     tags=["documentation"],
11 )