uucore: add parse_glob module

src/uucore/Cargo.toml

@@ -23,6 +23,7 @@ clap = "3.2"
 dns-lookup = { version="1.0.5", optional=true }
 dunce = "1.0.0"
 wild = "2.0"
+glob = "0.3.0"
 # * optional
 itertools = { version="0.10.0", optional=true }
 thiserror = { version="1.0", optional=true }

src/uucore/src/lib/lib.rs

@@ -29,6 +29,7 @@ pub use crate::mods::ranges;
 pub use crate::mods::version_cmp;
 
 // * string parsing modules
+pub use crate::parser::parse_glob;
 pub use crate::parser::parse_size;
 pub use crate::parser::parse_time;
 

src/uucore/src/lib/parser.rs

@@ -1,2 +1,3 @@
+pub mod parse_glob;
 pub mod parse_size;
 pub mod parse_time;

src/uucore/src/lib/parser/parse_glob.rs

@@ -0,0 +1,109 @@
+//! Parsing a glob Pattern from a string.
+//!
+//! Use the [`from_str`] function to parse a [`Pattern`] from a string.
+
+// cSpell:words fnmatch
+
+use glob::{Pattern, PatternError};
+
+fn fix_negation(glob: &str) -> String {
+    let mut chars = glob.chars().collect::<Vec<_>>();
+
+    let mut i = 0;
+    while i < chars.len() {
+        if chars[i] == '[' && i + 4 <= glob.len() && chars[i + 1] == '^' {
+            match chars[i + 3..].iter().position(|x| *x == ']') {
+                None => (),
+                Some(j) => {
+                    chars[i + 1] = '!';
+                    i += j + 4;
+                    continue;
+                }
+            }
+        }
+
+        i += 1;
+    }
+
+    chars.into_iter().collect::<String>()
+}
+
+/// Parse a glob Pattern from a string.
+///
+/// This function amends the input string to replace any caret or circumflex
+/// character (^) used to negate a set of characters with an exclamation mark
+/// (!), which adapts rust's glob matching to function the way the GNU utils'
+/// fnmatch does.
+///
+/// # Examples
+///
+/// ```rust
+/// use std::time::Duration;
+/// use uucore::parse_glob::from_str;
+/// assert!(!from_str("[^abc]").unwrap().matches("a"));
+/// assert!(from_str("[^abc]").unwrap().matches("x"));
+/// ```
+pub fn from_str(glob: &str) -> Result<Pattern, PatternError> {
+    Pattern::new(&fix_negation(glob))
+}
+
+#[cfg(test)]
+mod tests {
+
+    use super::*;
+
+    #[test]
+    fn test_from_str() {
+        assert_eq!(from_str("[^abc]").unwrap(), Pattern::new("[!abc]").unwrap());
+    }
+
+    #[test]
+    fn test_fix_negation() {
+        // Happy/Simple case
+        assert_eq!(fix_negation("[^abc]"), "[!abc]");
+
+        // Should fix negations in a long regex
+        assert_eq!(fix_negation("foo[abc]  bar[^def]"), "foo[abc]  bar[!def]");
+
+        // Should fix multiple negations in a regex
+        assert_eq!(fix_negation("foo[^abc]bar[^def]"), "foo[!abc]bar[!def]");
+
+        // Should fix negation of the single character ]
+        assert_eq!(fix_negation("[^]]"), "[!]]");
+
+        // Should fix negation of the single character ^
+        assert_eq!(fix_negation("[^^]"), "[!^]");
+
+        // Should fix negation of the space character
+        assert_eq!(fix_negation("[^ ]"), "[! ]");
+
+        // Complicated patterns
+        assert_eq!(fix_negation("[^][]"), "[!][]");
+        assert_eq!(fix_negation("[^[]]"), "[![]]");
+
+        // More complex patterns that should be replaced
+        assert_eq!(fix_negation("[[]] [^a]"), "[[]] [!a]");
+        assert_eq!(fix_negation("[[] [^a]"), "[[] [!a]");
+        assert_eq!(fix_negation("[]] [^a]"), "[]] [!a]");
+    }
+
+    #[test]
+    fn test_fix_negation_should_not_amend() {
+        assert_eq!(fix_negation("abc"), "abc");
+
+        // Regex specifically matches either [ or ^
+        assert_eq!(fix_negation("[[^]"), "[[^]");
+
+        // Regex that specifically matches either space or ^
+        assert_eq!(fix_negation("[ ^]"), "[ ^]");
+
+        // Regex that specifically matches either [, space or ^
+        assert_eq!(fix_negation("[[ ^]"), "[[ ^]");
+        assert_eq!(fix_negation("[ [^]"), "[ [^]");
+
+        // Invalid globs (according to rust's glob implementation) will remain unamended
+        assert_eq!(fix_negation("[^]"), "[^]");
+        assert_eq!(fix_negation("[^"), "[^");
+        assert_eq!(fix_negation("[][^]"), "[][^]");
+    }
+}