rs-chunks 0.1.0

Fast, high-fidelity document chunking for RAG — a pure-Rust engine covering 36 file formats (Office, OpenDocument, PDF, email, ebooks, notebooks, and more).
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310

//! Source-agnostic dispatch: route a file to the right engine by extension,
//! mirroring the Python `get_chunks()` routing (including the delimited/
//! spreadsheet special-casing) so behaviour matches the reference package.

use std::path::Path;

use crate::chunk::Chunk;
use crate::error::{ChunkError, Result};
use crate::formats;

/// Chunk a document supplied as raw bytes. `filename` is used only for extension
/// detection (routing) — never persisted under that name.
///
/// Routes to each format's no-filesystem `chunk_from_bytes` where available (these
/// work on wasm32). Formats not yet refactored fall back to a temp file on native
/// targets; on wasm32 they return [`ChunkError::Unsupported`] until refactored.
pub fn get_chunks_from_bytes(
    data: &[u8],
    filename: &str,
    mode: &str,
    window_size: usize,
    overlap: usize,
    sentences_per_chunk: usize,
    paragraphs_per_page: usize,
) -> Result<Vec<Chunk>> {
    let ext = ext_of(filename);
    match ext.as_str() {
        "csv" | "tsv" => {
            let csv_mode = if mode == "default" { "row" } else { mode };
            let rows_per_chunk = if csv_mode == "page_aware" { paragraphs_per_page } else { csv_rows_per_chunk(sentences_per_chunk) };
            let delimiter = if ext == "tsv" { Some(b'\t') } else { None };
            formats::csv::chunk_from_bytes(data, csv_mode, rows_per_chunk, window_size, overlap, true, delimiter, "utf-8", true)
        }
        "xlsx" | "xls" | "xlsm" | "xlsb" | "ods" | "xltx" | "xltm" => {
            let xmode = if mode == "default" { "row" } else { mode };
            let rows_per_chunk = if sentences_per_chunk == 3 { 1 } else { sentences_per_chunk };
            formats::xlsx::chunk_from_bytes(data, &ext, xmode, rows_per_chunk, window_size, overlap, true, Vec::new(), true, 2000)
        }
        "md" => formats::md::chunk_from_bytes(data, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        "txt" => formats::txt::chunk_from_bytes(data, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        "html" | "htm" => formats::html::chunk_from_bytes(data, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        "docx" | "docm" | "dotx" | "dotm" => formats::docx::chunk_from_bytes(data, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        "pptx" | "potx" | "potm" | "ppsx" | "ppsm" => formats::pptx::chunk_from_bytes(data, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        "json" | "jsonl" | "ndjson" => formats::json::chunk_from_bytes(data, filename, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        "eml" | "mbox" => formats::eml::chunk_from_bytes(data, filename, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        "odt" | "odp" => formats::odf::chunk_from_bytes(data, filename, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        "ipynb" => formats::ipynb::chunk_from_bytes(data, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        "rtf" => formats::rtf::chunk_from_bytes(data, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        "epub" => formats::epub::chunk_from_bytes(data, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        "msg" => formats::msg::chunk_from_bytes(data, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        "doc" => formats::doc::chunk_from_bytes(data, filename, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        "ppt" => formats::ppt::chunk_from_bytes(data, filename, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        _ => bytes_fallback_chunks(data, &ext, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
    }
}

/// Convert bytes to Markdown; see [`get_chunks_from_bytes`] for the routing note.
pub fn get_markdown_from_bytes(data: &[u8], filename: &str) -> Result<String> {
    let ext = ext_of(filename);
    match ext.as_str() {
        "csv" => formats::csv::to_markdown_from_bytes(data, None, "utf-8"),
        "tsv" => formats::csv::to_markdown_from_bytes(data, Some(b'\t'), "utf-8"),
        "xlsx" | "xls" | "xlsm" | "xlsb" | "ods" | "xltx" | "xltm" => formats::xlsx::to_markdown_from_bytes(data, &ext),
        "md" => formats::md::to_markdown_from_bytes(data),
        "txt" => formats::txt::to_markdown_from_bytes(data),
        "html" | "htm" => formats::html::to_markdown_from_bytes(data),
        "docx" | "docm" | "dotx" | "dotm" => formats::docx::to_markdown_from_bytes(data),
        "pptx" | "potx" | "potm" | "ppsx" | "ppsm" => formats::pptx::to_markdown_from_bytes(data),
        "json" | "jsonl" | "ndjson" => formats::json::to_markdown_from_bytes(data, filename),
        "eml" | "mbox" => formats::eml::to_markdown_from_bytes(data, filename),
        "odt" | "odp" => formats::odf::to_markdown_from_bytes(data, filename),
        "ipynb" => formats::ipynb::to_markdown_from_bytes(data),
        "rtf" => formats::rtf::to_markdown_from_bytes(data),
        "epub" => formats::epub::to_markdown_from_bytes(data),
        "msg" => formats::msg::to_markdown_from_bytes(data),
        "doc" => formats::doc::to_markdown_from_bytes(data),
        "ppt" => formats::ppt::to_markdown_from_bytes(data),
        _ => bytes_fallback_markdown(data, &ext),
    }
}

/// Chunk bytes and also return extracted image bytes (`list_images=True`).
/// Formats without embedded-image support return an empty image list and the
/// same chunks as [`get_chunks_from_bytes`].
#[allow(clippy::too_many_arguments)]
pub fn get_chunks_with_images_from_bytes(
    data: &[u8],
    filename: &str,
    mode: &str,
    window_size: usize,
    overlap: usize,
    sentences_per_chunk: usize,
    paragraphs_per_page: usize,
) -> Result<(Vec<Chunk>, Vec<(String, Vec<u8>)>)> {
    let ext = ext_of(filename);
    match ext.as_str() {
        "xlsx" | "xls" | "xlsm" | "xlsb" | "ods" | "xltx" | "xltm" => {
            let xmode = if mode == "default" { "row" } else { mode };
            let rows_per_chunk = if sentences_per_chunk == 3 { 1 } else { sentences_per_chunk };
            formats::xlsx::chunk_with_images_from_bytes(data, &ext, xmode, rows_per_chunk, window_size, overlap, true, Vec::new(), true, 2000)
        }
        "html" | "htm" => formats::html::chunk_with_images_from_bytes(data, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        "docx" | "docm" | "dotx" | "dotm" => formats::docx::chunk_with_images_from_bytes(data, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        "pptx" | "potx" | "potm" | "ppsx" | "ppsm" => formats::pptx::chunk_with_images_from_bytes(data, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        "eml" | "mbox" => formats::eml::chunk_with_images_from_bytes(data, filename, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        "odt" | "odp" => formats::odf::chunk_with_images_from_bytes(data, filename, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        "ipynb" => formats::ipynb::chunk_with_images_from_bytes(data, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        "epub" => formats::epub::chunk_with_images_from_bytes(data, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        "doc" => formats::doc::chunk_with_images_from_bytes(data, filename, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        "ppt" => formats::ppt::chunk_with_images_from_bytes(data, filename, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        // No embedded-image support: chunks only, empty image list.
        _ => Ok((get_chunks_from_bytes(data, filename, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page)?, Vec::new())),
    }
}

/// Convert bytes to Markdown and return extracted image bytes (`list_images=True`).
pub fn get_markdown_with_images_from_bytes(data: &[u8], filename: &str) -> Result<(String, Vec<(String, Vec<u8>)>)> {
    let ext = ext_of(filename);
    match ext.as_str() {
        "xlsx" | "xls" | "xlsm" | "xlsb" | "ods" | "xltx" | "xltm" => formats::xlsx::to_markdown_with_images_from_bytes(data, &ext),
        "html" | "htm" => formats::html::to_markdown_with_images_from_bytes(data),
        "docx" | "docm" | "dotx" | "dotm" => formats::docx::to_markdown_with_images_from_bytes(data),
        "pptx" | "potx" | "potm" | "ppsx" | "ppsm" => formats::pptx::to_markdown_with_images_from_bytes(data),
        "eml" | "mbox" => formats::eml::to_markdown_with_images_from_bytes(data, filename),
        "odt" | "odp" => formats::odf::to_markdown_with_images_from_bytes(data, filename),
        "ipynb" => formats::ipynb::to_markdown_with_images_from_bytes(data),
        "epub" => formats::epub::to_markdown_with_images_from_bytes(data),
        "doc" => formats::doc::to_markdown_with_images_from_bytes(data),
        "ppt" => formats::ppt::to_markdown_with_images_from_bytes(data),
        _ => Ok((get_markdown_from_bytes(data, filename)?, Vec::new())),
    }
}

#[cfg(not(target_arch = "wasm32"))]
#[allow(clippy::too_many_arguments)]
fn bytes_fallback_chunks(
    data: &[u8],
    ext: &str,
    mode: &str,
    window_size: usize,
    overlap: usize,
    sentences_per_chunk: usize,
    paragraphs_per_page: usize,
) -> Result<Vec<Chunk>> {
    let tmp = TempFile::new(data, ext)?;
    get_chunks(tmp.path(), mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page)
}

#[cfg(target_arch = "wasm32")]
#[allow(clippy::too_many_arguments)]
fn bytes_fallback_chunks(_: &[u8], ext: &str, _: &str, _: usize, _: usize, _: usize, _: usize) -> Result<Vec<Chunk>> {
    Err(ChunkError::Unsupported(format!("bytes chunking for '.{ext}' is not yet available on wasm32")))
}

#[cfg(not(target_arch = "wasm32"))]
fn bytes_fallback_markdown(data: &[u8], ext: &str) -> Result<String> {
    let tmp = TempFile::new(data, ext)?;
    get_markdown(tmp.path())
}

#[cfg(target_arch = "wasm32")]
fn bytes_fallback_markdown(_: &[u8], ext: &str) -> Result<String> {
    Err(ChunkError::Unsupported(format!("bytes markdown for '.{ext}' is not yet available on wasm32")))
}

/// Minimal RAII temp file (no external deps). Removed on drop.
#[cfg(not(target_arch = "wasm32"))]
struct TempFile {
    path: String,
}

#[cfg(not(target_arch = "wasm32"))]
impl TempFile {
    fn new(data: &[u8], ext: &str) -> Result<Self> {
        use std::io::Write;
        let mut dir = std::env::temp_dir();
        let unique = format!(
            "chunks_rs_{}_{}.{ext}",
            std::process::id(),
            std::time::SystemTime::now()
                .duration_since(std::time::UNIX_EPOCH)
                .map(|d| d.as_nanos())
                .unwrap_or(0)
        );
        dir.push(unique);
        let path = dir.to_string_lossy().to_string();
        let mut f = std::fs::File::create(&path).map_err(ChunkError::Io)?;
        f.write_all(data).map_err(ChunkError::Io)?;
        Ok(TempFile { path })
    }
    fn path(&self) -> &str {
        &self.path
    }
}

#[cfg(not(target_arch = "wasm32"))]
impl Drop for TempFile {
    fn drop(&mut self) {
        let _ = std::fs::remove_file(&self.path);
    }
}

fn ext_of(file_path: &str) -> String {
    Path::new(file_path)
        .extension()
        .and_then(|e| e.to_str())
        .map(|e| e.to_ascii_lowercase())
        .unwrap_or_default()
}

fn csv_rows_per_chunk(sentences_per_chunk: usize) -> usize {
    sentences_per_chunk.max(1)
}

/// Chunk any supported document by path. `mode` is passed through to the engine
/// ("default" selects each format's natural strategy); the delimited formats map
/// it onto their row/window/page strategies exactly like the Python entry point.
pub fn get_chunks(
    file_path: &str,
    mode: &str,
    window_size: usize,
    overlap: usize,
    sentences_per_chunk: usize,
    paragraphs_per_page: usize,
) -> Result<Vec<Chunk>> {
    let ext = ext_of(file_path);
    match ext.as_str() {
        // ── Delimited text ──────────────────────────────────────────────
        "csv" | "tsv" => {
            let csv_mode = if mode == "default" { "row" } else { mode };
            let rows_per_chunk = if csv_mode == "page_aware" {
                paragraphs_per_page
            } else {
                csv_rows_per_chunk(sentences_per_chunk)
            };
            let delimiter = if ext == "tsv" { Some(b'\t') } else { None };
            formats::csv::chunk(
                file_path,
                csv_mode,
                rows_per_chunk,
                window_size,
                overlap,
                true,
                delimiter,
                "utf-8",
                true,
            )
        }
        // ── Spreadsheets (calamine) ─────────────────────────────────────
        "xlsx" | "xls" | "xlsm" | "xlsb" | "ods" | "xltx" | "xltm" => {
            let xmode = if mode == "default" { "row" } else { mode };
            let rows_per_chunk = if sentences_per_chunk == 3 { 1 } else { sentences_per_chunk };
            formats::xlsx::chunk(
                file_path, xmode, rows_per_chunk, window_size, overlap, true, Vec::new(), true, 2000,
            )
        }
        "doc" => formats::doc::chunk(file_path, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        // ── Word OOXML ──────────────────────────────────────────────────
        "docx" | "docm" | "dotx" | "dotm" => {
            formats::docx::chunk(file_path, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page)
        }
        // ── Prose / markdown-pipeline formats ───────────────────────────
        "ppt" => formats::ppt::chunk(file_path, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        "pptx" | "potx" | "potm" | "ppsx" | "ppsm" => {
            formats::pptx::chunk(file_path, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page)
        }
        "md" => formats::md::chunk(file_path, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        "txt" => formats::txt::chunk(file_path, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        "html" | "htm" => formats::html::chunk(file_path, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        "json" | "jsonl" | "ndjson" => formats::json::chunk(file_path, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        "eml" | "mbox" => formats::eml::chunk(file_path, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        "odt" | "odp" => formats::odf::chunk(file_path, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        "msg" => formats::msg::chunk(file_path, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        "ipynb" => formats::ipynb::chunk(file_path, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        "rtf" => formats::rtf::chunk(file_path, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        #[cfg(feature = "pdf-native")]
        "pdf" => formats::pdf::chunk(file_path, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        "epub" => formats::epub::chunk(file_path, mode, window_size, overlap, sentences_per_chunk, paragraphs_per_page),
        other => Err(ChunkError::Unsupported(format!("Unsupported file type '.{other}'"))),
    }
}

/// Convert a supported document to Markdown by path.
pub fn get_markdown(file_path: &str) -> Result<String> {
    let ext = ext_of(file_path);
    match ext.as_str() {
        "csv" | "tsv" => {
            let delimiter = if ext == "tsv" { Some(b'\t') } else { None };
            formats::csv::to_markdown(file_path, delimiter, "utf-8")
        }
        "xlsx" | "xls" | "xlsm" | "xlsb" | "ods" | "xltx" | "xltm" => formats::xlsx::to_markdown(file_path),
        "doc" => formats::doc::to_markdown(file_path),
        "docx" | "docm" | "dotx" | "dotm" => formats::docx::to_markdown(file_path),
        "ppt" => formats::ppt::to_markdown(file_path),
        "pptx" | "potx" | "potm" | "ppsx" | "ppsm" => formats::pptx::to_markdown(file_path),
        "md" => formats::md::to_markdown(file_path),
        "txt" => formats::txt::to_markdown(file_path),
        "html" | "htm" => formats::html::to_markdown(file_path),
        "json" | "jsonl" | "ndjson" => formats::json::to_markdown(file_path),
        "eml" | "mbox" => formats::eml::to_markdown(file_path),
        "odt" | "odp" => formats::odf::to_markdown(file_path),
        "msg" => formats::msg::to_markdown(file_path),
        "ipynb" => formats::ipynb::to_markdown(file_path),
        "rtf" => formats::rtf::to_markdown(file_path),
        #[cfg(feature = "pdf-native")]
        "pdf" => formats::pdf::to_markdown(file_path),
        "epub" => formats::epub::to_markdown(file_path),
        other => Err(ChunkError::Unsupported(format!("get_markdown does not support '.{other}'"))),
    }
}