|
| 1 | +// |
| 2 | +// package_index.rs |
| 3 | +// |
| 4 | +// Copyright (C) 2025 by Posit Software, PBC |
| 5 | +// |
| 6 | +// |
| 7 | + |
| 8 | +use std::path::Path; |
| 9 | + |
| 10 | +pub struct Index { |
| 11 | + pub names: Vec<String>, |
| 12 | +} |
| 13 | + |
| 14 | +impl Index { |
| 15 | + pub fn load_from_folder(path: &Path) -> anyhow::Result<Self> { |
| 16 | + if !path.is_dir() { |
| 17 | + return Err(anyhow::anyhow!( |
| 18 | + "Can't load index as '{path}' is not a folder", |
| 19 | + path = path.to_string_lossy() |
| 20 | + )); |
| 21 | + } |
| 22 | + |
| 23 | + let index_path = path.join("INDEX"); |
| 24 | + if !index_path.is_file() { |
| 25 | + return Err(anyhow::anyhow!( |
| 26 | + "Can't load index as '{path}' does not contain an INDEX file", |
| 27 | + path = path.to_string_lossy() |
| 28 | + )); |
| 29 | + } |
| 30 | + |
| 31 | + let contents = std::fs::read_to_string(&index_path)?; |
| 32 | + Ok(Index::parse(&contents)) |
| 33 | + } |
| 34 | + |
| 35 | + /// Parses a package index text, extracting valid R symbol names from the first column. |
| 36 | + /// Only names starting at the beginning of a line and consisting of letters, digits, dots, or underscores are included. |
| 37 | + pub fn parse(input: &str) -> Self { |
| 38 | + let valid_name = regex::Regex::new(r"^[A-Za-z.][A-Za-z0-9._]*$").unwrap(); |
| 39 | + let mut names = Vec::new(); |
| 40 | + |
| 41 | + for line in input.lines() { |
| 42 | + // Only consider lines that start at column 0 (no leading whitespace) |
| 43 | + if line.starts_with(char::is_whitespace) { |
| 44 | + continue; |
| 45 | + } |
| 46 | + if let Some(first_col) = line.split_whitespace().next() { |
| 47 | + if valid_name.is_match(first_col) { |
| 48 | + names.push(first_col.to_string()); |
| 49 | + } |
| 50 | + } |
| 51 | + } |
| 52 | + |
| 53 | + Index { names } |
| 54 | + } |
| 55 | +} |
| 56 | + |
| 57 | +#[cfg(test)] |
| 58 | +mod tests { |
| 59 | + use std::fs; |
| 60 | + |
| 61 | + use tempfile::tempdir; |
| 62 | + |
| 63 | + use super::*; |
| 64 | + |
| 65 | + #[test] |
| 66 | + fn test_index_parses_simple_names() { |
| 67 | + let input = "\ |
| 68 | +foo Description of foo |
| 69 | +bar Description of bar |
| 70 | +baz Description of baz |
| 71 | +"; |
| 72 | + let idx = Index::parse(input); |
| 73 | + assert_eq!(idx.names, vec!["foo", "bar", "baz"]); |
| 74 | + } |
| 75 | + |
| 76 | + #[test] |
| 77 | + fn test_index_ignores_continuation_lines() { |
| 78 | + let input = "\ |
| 79 | +foo Description of foo |
| 80 | + Continuation of description |
| 81 | +bar Description of bar |
| 82 | +"; |
| 83 | + let idx = Index::parse(input); |
| 84 | + assert_eq!(idx.names, vec!["foo", "bar"]); |
| 85 | + } |
| 86 | + |
| 87 | + #[test] |
| 88 | + fn test_index_parses_names_with_dots_and_underscores() { |
| 89 | + let input = "\ |
| 90 | +foo.bar Description |
| 91 | +foo_bar Description |
| 92 | +.foo Description |
| 93 | +"; |
| 94 | + let idx = Index::parse(input); |
| 95 | + assert_eq!(idx.names, vec!["foo.bar", "foo_bar", ".foo"]); |
| 96 | + } |
| 97 | + |
| 98 | + #[test] |
| 99 | + fn test_index_skips_names_with_dashes() { |
| 100 | + let input = "\ |
| 101 | +foo-bar Description |
| 102 | +baz Description |
| 103 | +foo Description |
| 104 | +"; |
| 105 | + let idx = Index::parse(input); |
| 106 | + assert_eq!(idx.names, vec!["baz", "foo"]); |
| 107 | + } |
| 108 | + |
| 109 | + #[test] |
| 110 | + fn test_index_skips_lines_without_valid_names() { |
| 111 | + let input = "\ |
| 112 | +123foo Not a valid name |
| 113 | +_bar Not a valid name |
| 114 | +.foo Valid name |
| 115 | +"; |
| 116 | + let idx = Index::parse(input); |
| 117 | + assert_eq!(idx.names, vec![".foo"]); |
| 118 | + } |
| 119 | + |
| 120 | + #[test] |
| 121 | + fn test_index_parses_realistic_package_index() { |
| 122 | + let input = "\ |
| 123 | +.prt.methTit Print and Summary Method Utilities for Mixed |
| 124 | + Effects |
| 125 | +Arabidopsis Arabidopsis clipping/fertilization data |
| 126 | +Dyestuff Yield of dyestuff by batch |
| 127 | +GHrule Univariate Gauss-Hermite quadrature rule |
| 128 | +NelderMead-class Class '\"NelderMead\"' of Nelder-Mead optimizers |
| 129 | + and its Generator |
| 130 | +Nelder_Mead Nelder-Mead Optimization of Parameters, |
| 131 | + Possibly (Box) Constrained |
| 132 | +"; |
| 133 | + let idx = Index::parse(input); |
| 134 | + assert_eq!(idx.names, vec![ |
| 135 | + ".prt.methTit", |
| 136 | + "Arabidopsis", |
| 137 | + "Dyestuff", |
| 138 | + "GHrule", |
| 139 | + "Nelder_Mead" |
| 140 | + ]); |
| 141 | + } |
| 142 | + |
| 143 | + #[test] |
| 144 | + fn load_from_folder_returns_errors() { |
| 145 | + // From a file |
| 146 | + let file = tempfile::NamedTempFile::new().unwrap(); |
| 147 | + let result = Index::load_from_folder(file.path()); |
| 148 | + assert!(result.is_err()); |
| 149 | + |
| 150 | + // From a dir without an INDEX file |
| 151 | + let dir = tempdir().unwrap(); |
| 152 | + let result = Index::load_from_folder(dir.path()); |
| 153 | + assert!(result.is_err()); |
| 154 | + } |
| 155 | + |
| 156 | + #[test] |
| 157 | + fn load_from_folder_reads_and_parses_index() { |
| 158 | + let dir = tempdir().unwrap(); |
| 159 | + let index_path = dir.path().join("INDEX"); |
| 160 | + let content = "\ |
| 161 | +foo Description of foo |
| 162 | +bar Description of bar |
| 163 | +"; |
| 164 | + fs::write(&index_path, content).unwrap(); |
| 165 | + |
| 166 | + let idx = Index::load_from_folder(dir.path()).unwrap(); |
| 167 | + assert_eq!(idx.names, vec!["foo", "bar"]); |
| 168 | + } |
| 169 | +} |
0 commit comments