oak_core/source/
streaming.rs

1use crate::source::{Source, TextChunk};
2use core::range::Range;
3use std::borrow::Cow;
4use triomphe::Arc;
5use url::Url;
6
7const CHUNK_SIZE: usize = 4096;
8
9/// A read-only, chunked source implementation for efficient handling of streamed or large files.
10#[derive(Clone, Debug)]
11pub struct ChunkedSource {
12    url: Option<Url>,
13    chunks: Arc<[Arc<str>]>,
14    starts: Arc<[usize]>,
15    len: usize,
16}
17
18/// A mutable buffer for chunked source code, supporting efficient appending of text.
19#[derive(Clone, Debug, Default)]
20pub struct ChunkedBuffer {
21    url: Option<Url>,
22    chunks: Vec<Arc<str>>,
23    starts: Vec<usize>,
24    len: usize,
25}
26
27impl ChunkedBuffer {
28    /// Creates a new empty `ChunkedBuffer`.
29    pub fn new() -> Self {
30        Self::default()
31    }
32
33    /// Creates a new `ChunkedBuffer` with the specified URL.
34    pub fn new_with_url(url: impl Into<Option<Url>>) -> Self {
35        Self { url: url.into(), ..Self::default() }
36    }
37
38    /// Appends the specified string to the end of the buffer.
39    pub fn push_str(&mut self, text: &str) {
40        if text.is_empty() {
41            return;
42        }
43        for chunk in chunkify(text) {
44            self.starts.push(self.len);
45            self.len += chunk.len();
46            self.chunks.push(chunk);
47        }
48    }
49
50    /// Returns a read-only snapshot of the current buffer.
51    pub fn snapshot(&self) -> ChunkedSource {
52        ChunkedSource { url: self.url.clone(), chunks: Arc::<[Arc<str>]>::from(self.chunks.clone()), starts: Arc::<[usize]>::from(self.starts.clone()), len: self.len }
53    }
54}
55
56impl Source for ChunkedBuffer {
57    fn length(&self) -> usize {
58        self.len
59    }
60
61    fn chunk_at(&self, offset: usize) -> TextChunk<'_> {
62        if offset >= self.len {
63            return TextChunk { start: self.len, text: "" };
64        }
65        let idx = match self.starts.binary_search(&offset) {
66            Ok(i) => i,
67            Err(0) => 0,
68            Err(i) => i - 1,
69        };
70        let start = self.starts[idx];
71        TextChunk { start, text: self.chunks[idx].as_ref() }
72    }
73
74    fn get_text_in(&self, range: Range<usize>) -> Cow<'_, str> {
75        text_in_chunks(&self.chunks, &self.starts, self.len, range)
76    }
77
78    fn get_url(&self) -> Option<&Url> {
79        self.url.as_ref()
80    }
81}
82
83impl Source for ChunkedSource {
84    fn length(&self) -> usize {
85        self.len
86    }
87
88    fn chunk_at(&self, offset: usize) -> TextChunk<'_> {
89        if offset >= self.len {
90            return TextChunk { start: self.len, text: "" };
91        }
92        let idx = match self.starts.binary_search(&offset) {
93            Ok(i) => i,
94            Err(0) => 0,
95            Err(i) => i - 1,
96        };
97        let start = self.starts[idx];
98        TextChunk { start, text: self.chunks[idx].as_ref() }
99    }
100
101    fn get_text_in(&self, range: Range<usize>) -> Cow<'_, str> {
102        text_in_chunks(&self.chunks, &self.starts, self.len, range)
103    }
104
105    fn get_url(&self) -> Option<&Url> {
106        self.url.as_ref()
107    }
108}
109
110fn chunkify(text: &str) -> Vec<Arc<str>> {
111    if text.is_empty() {
112        return vec![];
113    }
114    let mut out = Vec::new();
115    let mut start = 0usize;
116    while start < text.len() {
117        let mut end = (start + CHUNK_SIZE).min(text.len());
118        while end > start && !text.is_char_boundary(end) {
119            end -= 1;
120        }
121        if end == start {
122            end = text.len();
123        }
124        let part = &text[start..end];
125        out.push(Arc::<str>::from(part.to_string()));
126        start = end;
127    }
128    out
129}
130
131fn text_in_chunks<'a>(chunks: &'a [Arc<str>], starts: &'a [usize], len: usize, range: Range<usize>) -> Cow<'a, str> {
132    if range.start >= range.end || range.start >= len {
133        return Cow::Borrowed("");
134    }
135    let start = range.start;
136    let end = range.end.min(len);
137
138    let start_idx = match starts.binary_search(&start) {
139        Ok(i) => i,
140        Err(0) => 0,
141        Err(i) => i - 1,
142    };
143    let end_idx = match starts.binary_search(&end) {
144        Ok(i) => i,
145        Err(0) => 0,
146        Err(i) => i - 1,
147    };
148
149    if start_idx == end_idx {
150        let base = starts[start_idx];
151        let rel_start = start - base;
152        let rel_end = end - base;
153        let s = chunks[start_idx].as_ref();
154        return s.get(rel_start..rel_end).map(Cow::Borrowed).unwrap_or(Cow::Borrowed(""));
155    }
156
157    let mut buf = String::new();
158    for (i, c) in chunks.iter().enumerate().skip(start_idx).take(end_idx - start_idx + 1) {
159        let base = starts[i];
160        let cs = c.as_ref();
161        let seg_start = if i == start_idx { start.saturating_sub(base) } else { 0 };
162        let seg_end = if i == end_idx { end.saturating_sub(base) } else { cs.len() };
163        if let Some(seg) = cs.get(seg_start..seg_end) {
164            buf.push_str(seg);
165        }
166    }
167    Cow::Owned(buf)
168}
oak_core/source/streaming.rs

oak_core/source/
streaming.rs
