support named capture groups

Author: yaa110

Cargo.lock

@@ -8,7 +8,7 @@ license = "MIT OR Apache-2.0"
 name = "nomino"
 readme = "README.md"
 repository = "https://github.com/yaa110/nomino"
-version = "1.5.2"
+version = "1.6.0"
 
 [dependencies]
 anyhow = "1.0"

Cargo.toml

@@ -56,9 +56,37 @@ Options:
   -V, --version             Print version
   -w, --overwrite           Overwrites output files, otherwise, a '_' is prepended to filename
 
-OUTPUT pattern accepts placeholders that have the format of '{I:P}' where 'I' is the index of captured group and 'P' is the padding of digits with `0`. Please refer to https://github.com/yaa110/nomino for more information.
+OUTPUT pattern accepts placeholders that have the format of '{G:P}' where 'G' is the captured group and 'P' is the padding of digits with `0`. Please refer to https://github.com/yaa110/nomino for more information.
 ```
 
+### Placeholders
+
+1. Placeholders have the format of `{G:P}` where `G` is the captured group and `P` is the padding of digits with `0`. For example, `{2:3}` means the third captured group with a padding of 3, i.e. `1` is formatted as `001`.
+1. Indices start from `0`, and `{0}` means the filename.
+1. The capture group `G` could be dropped, i.e. `{}` or `{:3}`. In this case an auto incremental index is used which starts from `1`. For example, `{} {}` equals `{1} {2}`.
+1. `{` and `}` characters could be escaped using `\` character, i.e. `\\{` and `\\}` in cli.
+1. Padding is only used for positive numbers, e.g. the formatted result of `{:3}` for `1` is `001`, for `-1` is `-1` and for `a` is `a`.
+1. If `--sort` option is used, the first index `{0}` is the filename and the second index `{1}` or first occurrence of `{}` is the enumerator index.
+
+### Capture Groups
+
+The accepted syntax of regex pattern is [Rust Regex](https://docs.rs/regex/latest/regex/).
+
+Consider this example:
+
+```regex
+(?<first>\w)(\w)\w(?<last>\w)
+```
+
+This regular expression defines 4 capture groups:
+
+- The group at index `0` corresponds to the overall match. It is always present in every match and never has a name: `{0}`.
+- The group at index `1` with name `first` corresponding to the first letter: `{1}`, `{first}` or the first occurrence of `{}`.
+- The group at index `2` with no name corresponding to the second letter: `{2}` or the second occurrence of `{}`.
+- The group at index `3` with name `last` corresponding to the fourth and last letter: `{3}`, `{last}` or the third occurrence of `{}`.
+
+`?<first>` and `?<last>` are named capture groups.
+
 ### Windows
 
 On Windows, `\\` must be used to separate path components in file paths because `\` is a special character in regular expressions.
@@ -73,23 +101,6 @@ On Windows, `\\` must be used to separate path components in file paths because
 }
 ```
 
-## Output
-
-The output is necessary when using `--sort` or `--regex` options.
-
-### Regex
-
-The accepted syntax of regex pattern is [Rust Regex](https://docs.rs/regex/latest/regex/).
-
-### Placeholders
-
-1. Placeholders have the format of `{I:P}` where `I` is the index of captured group and `P` is the padding of digits with `0`. For example, `{2:3}` means the third captured group with a padding of 3, i.e. `1` is formatted as `001`.
-1. Indices start from `0`, and `{0}` means the filename.
-1. The index `I` could be dropped, i.e. `{}` or `{:3}`. In this case an auto incremental index is used which starts from `1`. For example, `{} {}` equals `{1} {2}`.
-1. `{` and `}` characters could be escaped using `\` character, i.e. `\\{` and `\\}` in cli.
-1. Padding is only used for positive numbers, e.g. the formatted result of `{:3}` for `1` is `001`, for `-1` is `-1` and for `a` is `a`.
-1. If `--sort` option is used, the first index `{0}` is the filename and the second index `{1}` or first occurrence of `{}` is the enumerator index.
-
 ## Wiki
 
 - **[Examples](https://github.com/yaa110/nomino/wiki/Examples)** learn nomino by examples

README.md

@@ -6,8 +6,8 @@ use std::path::PathBuf;
     about,
     author,
     version,
-    after_help = "OUTPUT pattern accepts placeholders that have the format of '{I:P}' where 'I' \
-    is the index of captured group and 'P' is the padding of digits with `0`. Please refer to \
+    after_help = "OUTPUT pattern accepts placeholders that have the format of '{G:P}' where 'G' \
+    is the captured group and 'P' is the padding of digits with `0`. Please refer to \
     https://github.com/yaa110/nomino for more information.",
     next_display_order = None,
 )]

src/cli.rs

@@ -1,10 +1,11 @@
+use super::{provider::Capture, Provider};
 use crate::errors::FormatError;
 
 #[derive(Debug, PartialEq)]
 enum Segment {
     PlaceHolder {
         padding: Option<usize>,
-        index: usize,
+        capture: Capture,
     },
     String(String),
 }
@@ -16,12 +17,12 @@ impl Formatter {
     pub fn new(format: &str) -> Result<Self, FormatError> {
         let mut segments = Vec::new();
         let mut should_escape = false;
-        let mut is_parsing_index = false;
+        let mut is_parsing_capture = false;
         let mut is_parsing_padding = false;
         let mut current_segment = String::new();
-        let mut current_index: usize = 0;
+        let mut current_capture = Capture::Index(0);
         let mut current_padding: Option<usize> = None;
-        let mut incremental_index = 1;
+        let mut incremental_index: usize = 1;
         for (i, ch) in format.chars().enumerate() {
             if !should_escape && ch == '\\' {
                 should_escape = true;
@@ -31,29 +32,26 @@ impl Formatter {
                 return Err(FormatError::InvalidEscapeCharacter(i, ch));
             }
             match ch {
-                '{' if !should_escape && !is_parsing_index && !is_parsing_padding => {
+                '{' if !should_escape && !is_parsing_capture && !is_parsing_padding => {
                     if !current_segment.is_empty() {
                         segments.push(Segment::String(current_segment));
                         current_segment = String::new();
                     }
-                    is_parsing_index = true;
+                    is_parsing_capture = true;
                 }
                 '}' if !should_escape => {
-                    if !is_parsing_index && !is_parsing_padding {
+                    if !is_parsing_capture && !is_parsing_padding {
                         return Err(FormatError::UnopenedPlaceholder);
                     }
                     if current_segment.is_empty() {
-                        if is_parsing_index {
-                            current_index = incremental_index;
+                        if is_parsing_capture {
+                            current_capture = Capture::Index(incremental_index);
                             incremental_index += 1;
                         } else if is_parsing_padding {
                             current_padding = None;
                         }
-                    } else if is_parsing_index {
-                        current_index = current_segment
-                            .as_str()
-                            .parse()
-                            .map_err(|_| FormatError::InvalidIndex(current_segment.clone()))?;
+                    } else if is_parsing_capture {
+                        current_capture = current_segment.as_str().into();
                         current_padding = None;
                     } else if is_parsing_padding {
                         current_padding =
@@ -63,25 +61,22 @@ impl Formatter {
                     }
                     segments.push(Segment::PlaceHolder {
                         padding: current_padding,
-                        index: current_index,
+                        capture: current_capture,
                     });
                     current_segment.clear();
                     current_padding = None;
-                    current_index = 0;
-                    is_parsing_index = false;
+                    current_capture = Capture::Index(0);
+                    is_parsing_capture = false;
                     is_parsing_padding = false;
                 }
-                ':' if is_parsing_index => {
-                    is_parsing_index = false;
+                ':' if is_parsing_capture => {
+                    is_parsing_capture = false;
                     is_parsing_padding = true;
                     if current_segment.is_empty() {
-                        current_index = incremental_index;
+                        current_capture = Capture::Index(incremental_index);
                         incremental_index += 1;
                     } else {
-                        current_index = current_segment
-                            .as_str()
-                            .parse()
-                            .map_err(|_| FormatError::InvalidIndex(current_segment.clone()))?;
+                        current_capture = current_segment.as_str().into();
                         current_segment.clear();
                     }
                 }
@@ -91,7 +86,7 @@ impl Formatter {
                 }
             }
         }
-        if is_parsing_index || is_parsing_padding {
+        if is_parsing_capture || is_parsing_padding {
             return Err(FormatError::UnclosedPlaceholder);
         }
         if !current_segment.is_empty() {
@@ -100,12 +95,12 @@ impl Formatter {
         Ok(Self(segments))
     }
 
-    pub fn format(&self, vars: &[&str]) -> String {
+    pub fn format(&self, provider: impl Provider) -> String {
         let mut formatted = String::new();
         for segment in self.0.as_slice() {
             match segment {
-                Segment::PlaceHolder { padding, index } => {
-                    let Some(var) = vars.get(*index) else {
+                Segment::PlaceHolder { padding, capture } => {
+                    let Some(var) = provider.provide(capture) else {
                         continue;
                     };
                     if let Some((padding, digits)) =
@@ -187,7 +182,7 @@ mod tests {
         while let Some((format, vars, expected)) = format_vars_expected.pop() {
             let output = Formatter::new(format)
                 .expect(format!("unable to parse format '{}'", format).as_str());
-            let actual = output.format(vars.as_slice());
+            let actual = output.format(vars);
             assert_eq!(actual, expected);
         }
     }
@@ -200,8 +195,6 @@ mod tests {
             ("2:5}", FormatError::UnopenedPlaceholder),
             (r"\{2:5}", FormatError::UnopenedPlaceholder),
             (r"{2:5\}", FormatError::UnclosedPlaceholder),
-            ("{{2:5}}", FormatError::InvalidIndex("{2".to_string())),
-            ("{a}", FormatError::InvalidIndex("a".to_string())),
             ("{2:5a}", FormatError::InvalidPadding("5a".to_string())),
             ("init {2:5", FormatError::UnclosedPlaceholder),
             ("init {2:5 end", FormatError::UnclosedPlaceholder),

src/input/formatter.rs

@@ -47,7 +47,7 @@ impl InputIterator {
             });
             for (i, input) in inputs.into_iter().enumerate() {
                 let index = (i + 1).to_string();
-                let mut output = formatter.format(vec![input.as_str(), index.as_str()].as_slice());
+                let mut output = formatter.format(vec![input.as_str(), index.as_str()]);
                 if preserve_extension {
                     if let Some(extension) = Path::new(input.as_str()).extension() {
                         output.push('.');
@@ -94,14 +94,10 @@ impl Iterator for InputIterator {
                     };
                     let path = entry.path();
                     let input = path.strip_prefix("./").unwrap_or(path).to_string_lossy();
-                    let Some(cap) = re.captures(input.as_ref()) else {
+                    let Some(captures) = re.captures(input.as_ref()) else {
                         continue;
                     };
-                    let vars: Vec<&str> = cap
-                        .iter()
-                        .map(|c| c.map(|c| c.as_str()).unwrap_or_default())
-                        .collect();
-                    let mut output = formatter.format(vars.as_slice());
+                    let mut output = formatter.format(captures);
                     if *preserve_extension {
                         if let Some(extension) = Path::new(input.as_ref()).extension() {
                             output.push('.');

src/input/iterator.rs

@@ -0,0 +1,40 @@
+use regex::Captures;
+
+#[derive(Debug, PartialEq)]
+pub enum Capture {
+    Index(usize),
+    Name(String),
+}
+
+pub trait Provider {
+    fn provide(&self, cap: &Capture) -> Option<&str>;
+}
+
+impl Provider for Captures<'_> {
+    fn provide(&self, cap: &Capture) -> Option<&str> {
+        match cap {
+            Capture::Index(index) => self.get(*index),
+            Capture::Name(name) => self.name(name.as_str()),
+        }
+        .map(|m| m.as_str())
+    }
+}
+
+impl Provider for Vec<&'_ str> {
+    fn provide(&self, cap: &Capture) -> Option<&str> {
+        match cap {
+            Capture::Index(index) => self.get(*index).copied(),
+            _ => None,
+        }
+    }
+}
+
+impl From<&str> for Capture {
+    fn from(value: &str) -> Self {
+        if let Ok(index) = value.parse() {
+            Capture::Index(index)
+        } else {
+            Capture::Name(value.into())
+        }
+    }
+}

src/input/provider.rs

@@ -3,10 +3,12 @@ pub mod cli;
 pub mod input {
     mod formatter;
     mod iterator;
+    mod provider;
     mod separator;
     mod source;
     pub use self::formatter::*;
     pub use self::iterator::*;
+    pub use self::provider::*;
     pub use self::separator::*;
     pub use self::source::*;
 }

src/lib.rs

@@ -49,6 +49,51 @@ fn test_regex() {
     dir.close().unwrap();
 }
 
+#[test]
+fn test_named_regex() {
+    let dir = tempfile::tempdir().unwrap();
+
+    let inputs = vec![
+        "Nomino (2020) S1.E1.1080p.mkv",
+        "Nomino (2020) S1.E2.1080p.mkv",
+        "Nomino (2020) S1.E3.1080p.mkv",
+        "Nomino (2020) S1.E4.1080p.mkv",
+        "Nomino (2020) S1.E5.1080p.mkv",
+    ];
+
+    let mut outputs = vec!["01.mkv", "02.mkv", "03.mkv", "04.mkv", "05.mkv"];
+
+    for input in inputs {
+        let _ = File::create(dir.path().join(input)).unwrap();
+    }
+
+    let cmd = Command::cargo_bin(env!("CARGO_PKG_NAME"))
+        .unwrap()
+        .args(&[
+            "-E",
+            "-d",
+            dir.path().to_str().unwrap(),
+            "-r",
+            r".*E(?<episode>\d+).*",
+            "{episode:2}.mkv",
+        ])
+        .unwrap();
+
+    let mut files: Vec<String> = read_dir(dir.path())
+        .unwrap()
+        .map(|entry| entry.unwrap().file_name().to_str().unwrap().to_string())
+        .collect();
+
+    files.sort();
+    outputs.sort();
+
+    assert!(cmd.status.success());
+    assert_eq!(files.len(), outputs.len());
+    assert!(outputs.iter().zip(files.iter()).all(|(a, b)| a == b));
+
+    dir.close().unwrap();
+}
+
 #[test]
 fn test_regex_not_overwrite() {
     let dir = tempfile::tempdir().unwrap();