Add --greedy-delimiter

riquito · May 23, 2022 · 9e3d690 · 9e3d690
1 parent 19f0654
commit 9e3d690
Show file tree

Hide file tree

Showing 5 changed files with 114 additions and 20 deletions.
diff --git a/README.md b/README.md
@@ -26,6 +26,7 @@ USAGE:
     tuc [FLAGS] [OPTIONS]
 
 FLAGS:
+    -g, --greedy-delimiter        Split fields using a greedy delimiter
     -p, --compress-delimiter      Collapse any sequence of delimiters
     -s, --only-delimited          Do not print lines not containing delimiters
     -V, --version                 Prints version information
@@ -73,10 +74,22 @@ Notes:
 
 ## Examples
 
+```sh
+# Cut and rearrange fields
+❯ echo "foo bar baz" | tuc -d ' ' -f 3,2,1
+bazbarfoo
+```
+
+```sh
+# Keep ranges
+❯ echo "foo bar baz" | tuc -d ' ' -f 2:
+bar baz
+```
+
 ```sh
 # Cut using a greedy delimiter
-❯ echo "foo    bar   baz" | tuc -d ' ' -f 2:
-bar   baz
+❯ echo "foo    bar" | tuc -g -d ' ' -f 1,2
+foobar
 ```
 
 ```sh

diff --git a/src/bin/tuc.rs b/src/bin/tuc.rs
@@ -17,6 +17,7 @@ USAGE:
     tuc [FLAGS] [OPTIONS]
 
 FLAGS:
+    -g, --greedy-delimiter        Split fields using a greedy delimiter
     -p, --compress-delimiter      Collapse any sequence of delimiters
     -s, --only-delimited          Do not print lines not containing delimiters
     -V, --version                 Prints version information
@@ -101,6 +102,7 @@ fn parse_args() -> Result<Opt, pico_args::Error> {
     let args = Opt {
         complement: pargs.contains(["-m", "--complement"]),
         only_delimited: pargs.contains(["-s", "--only-delimited"]),
+        greedy_delimiter: pargs.contains(["-g", "--greedy-delimiter"]),
         compress_delimiter: pargs.contains(["-p", "--compress-delimiter"]),
         version: pargs.contains(["-V", "--version"]),
         join: pargs.contains(["-j", "--join"]),

diff --git a/src/cut_str.rs b/src/cut_str.rs
@@ -19,23 +19,31 @@ fn complement_std_range(parts_length: usize, r: &Range<usize>) -> Vec<Range<usiz
     }
 }
 
-/*
- * Split a string into parts and build a vector of ranges that match those parts.
- *
- * `buffer` - empty vector that will be filled with ranges
- * `line` - the string to split
- * `delimiter` - what to search to split the string
- */
-fn build_ranges_vec(buffer: &mut Vec<Range<usize>>, line: &str, delimiter: &str) {
+// Split a string into parts and build a vector of ranges that match those parts.
+//
+// `buffer` - vector that will be filled with ranges
+// `line` - the string to split
+// `delimiter` - what to search to split the string
+// `greedy` - whether to consider consecutive delimiters as one or not
+fn build_ranges_vec(buffer: &mut Vec<Range<usize>>, line: &str, delimiter: &str, greedy: bool) {
+    buffer.clear();
+
+    if line.is_empty() {
+        return;
+    }
+
     let delimiter_length = delimiter.len();
     let mut next_part_start = 0;
 
-    for mat in line.match_indices(&delimiter) {
-        buffer.push(Range {
-            start: next_part_start,
-            end: mat.0,
-        });
-        next_part_start = mat.0 + delimiter_length;
+    for (idx, _) in line.match_indices(&delimiter) {
+        if !(greedy && idx == next_part_start) {
+            buffer.push(Range {
+                start: next_part_start,
+                end: idx,
+            });
+        }
+
+        next_part_start = idx + delimiter_length;
     }
 
     buffer.push(Range {
@@ -91,17 +99,15 @@ pub fn cut_str(
         return Ok(());
     }
 
-    bounds_as_ranges.clear();
-    build_ranges_vec(bounds_as_ranges, line, &opt.delimiter);
+    build_ranges_vec(bounds_as_ranges, line, &opt.delimiter, opt.greedy_delimiter);
 
     if opt.compress_delimiter
         && (opt.bounds_type == BoundsType::Fields || opt.bounds_type == BoundsType::Lines)
     {
         compressed_line_buf.clear();
         compress_delimiter(bounds_as_ranges, line, &opt.delimiter, compressed_line_buf);
         line = compressed_line_buf;
-        bounds_as_ranges.clear();
-        build_ranges_vec(bounds_as_ranges, line, &opt.delimiter);
+        build_ranges_vec(bounds_as_ranges, line, &opt.delimiter, opt.greedy_delimiter);
     }
 
     if opt.bounds_type == BoundsType::Characters && bounds_as_ranges.len() > 2 {
@@ -231,4 +237,64 @@ mod tests {
         assert_eq!(complement_std_range(2, &(0..1)), vec![1..2]);
         assert_eq!(complement_std_range(2, &(1..2)), vec![0..1]);
     }
+
+    #[test]
+    fn test_build_ranges_vec() {
+        let mut v_range: Vec<Range<usize>> = Vec::new();
+
+        // non greedy
+
+        v_range.clear();
+        build_ranges_vec(&mut v_range, "", "-", false);
+        assert_eq!(v_range, vec![]);
+
+        v_range.clear();
+        build_ranges_vec(&mut v_range, "a", "-", false);
+        assert_eq!(v_range, vec![Range { start: 0, end: 1 }]);
+
+        v_range.clear();
+        build_ranges_vec(&mut v_range, "a-b", "-", false);
+        assert_eq!(
+            v_range,
+            vec![Range { start: 0, end: 1 }, Range { start: 2, end: 3 }]
+        );
+
+        v_range.clear();
+        build_ranges_vec(&mut v_range, "-a-", "-", false);
+        assert_eq!(
+            v_range,
+            vec![
+                Range { start: 0, end: 0 },
+                Range { start: 1, end: 2 },
+                Range { start: 3, end: 3 }
+            ]
+        );
+
+        v_range.clear();
+        build_ranges_vec(&mut v_range, "a--", "-", false);
+        assert_eq!(
+            v_range,
+            vec![
+                Range { start: 0, end: 1 },
+                Range { start: 2, end: 2 },
+                Range { start: 3, end: 3 }
+            ]
+        );
+
+        // greedy
+
+        v_range.clear();
+        build_ranges_vec(&mut v_range, "a--b", "-", true);
+        assert_eq!(
+            v_range,
+            vec![Range { start: 0, end: 1 }, Range { start: 3, end: 4 }]
+        );
+
+        v_range.clear();
+        build_ranges_vec(&mut v_range, "a--", "-", true);
+        assert_eq!(
+            v_range,
+            vec![Range { start: 0, end: 1 }, Range { start: 3, end: 3 }]
+        );
+    }
 }
diff --git a/src/options.rs b/src/options.rs
@@ -16,6 +16,7 @@ pub struct Opt {
     pub bounds: UserBoundsList,
     pub bounds_type: BoundsType,
     pub only_delimited: bool,
+    pub greedy_delimiter: bool,
     pub compress_delimiter: bool,
     pub replace_delimiter: Option<String>,
     pub trim: Option<Trim>,

diff --git a/tests/cli.rs b/tests/cli.rs
@@ -243,3 +243,15 @@ fn it_format_fields() {
         .success()
         .stdout("Say hello to our world.\nJust {saying}\n");
 }
+
+#[test]
+fn it_cuts_using_a_greedy_delimiter() {
+    let mut cmd = Command::cargo_bin(env!("CARGO_PKG_NAME")).unwrap();
+
+    let assert = cmd
+        .args(&["-g", "-d", "-", "-f", "1,2"])
+        .write_stdin("a---b")
+        .assert();
+
+    assert.success().stdout("ab\n");
+}