change ignore to use a custom file

instead of cli arguments

Cargo.toml

@@ -7,3 +7,4 @@ edition = "2021"
 
 [dependencies]
 clap = { version = "4.4.3", features = ["derive"] }
+glob-match = "0.2.1"

src/args.rs

@@ -16,15 +16,13 @@ pub struct Args {
     /// where your backup will be stored
     #[arg()]
     pub target: Option<PathBuf>,
-    /// directories to ignore.
-    /// can be paths relative to <source> (like backups/) or paths starting with <source> (like my_source/backups/).
-    /// if <index> starts with <source>, it is automatically ignored and doesn't need to be specified.
-    #[arg(long, num_args(0..))]
-    pub ignore: Vec<PathBuf>,
     /// don't ask for confirmation, just apply the changes.
     #[arg(long)]
     pub noconfirm: bool,
 
+    #[arg(long)]
+    pub ignore: Option<PathBuf>,
+
     #[command(flatten)]
     pub settings: Settings,
 }

src/config.rs

@@ -0,0 +1,199 @@
+use std::{
+    iter::{Enumerate, Peekable},
+    path::{Path, PathBuf},
+    str::Lines,
+};
+
+#[derive(Clone, Copy, Debug)]
+pub struct FsEntry<'a> {
+    pub path: &'a Path,
+    pub is_directory: bool,
+}
+
+#[derive(Debug)]
+pub struct Ignore(pub Vec<Specifier>);
+#[derive(Debug)]
+pub enum Specifier {
+    /// Everything in this config-part will be explicitly ignored.
+    /// Files in ignored directories, even if they would later match,
+    /// will not be seen my rembackup.
+    Except(Ignore),
+    Entries(Match),
+    Files(Match),
+    /// Descend into a directory.
+    /// Contains an inner config, which uses paths relative to the matched directory.
+    InDir {
+        dir: Match,
+        inner: Ignore,
+    },
+}
+#[derive(Debug)]
+pub enum Match {
+    Any,
+    Eq(PathBuf),
+    Glob(String),
+}
+
+impl Match {
+    pub fn matches(&self, path: &Path) -> bool {
+        match self {
+            Self::Any => true,
+            Self::Eq(v) => v.as_path() == path,
+            Self::Glob(v) => path
+                .to_str()
+                .is_some_and(|path| glob_match::glob_match(v, path)),
+        }
+    }
+}
+
+impl Ignore {
+    /// If `self.matches(entry)` is `Some(v)`, returns `v`, if not, returns `false`.
+    pub fn matches_or_default(&self, entry: &FsEntry) -> bool {
+        self.matches(entry).unwrap_or(false)
+    }
+    pub fn matches(&self, entry: &FsEntry) -> Option<bool> {
+        self.0.iter().rev().filter_map(|v| v.matches(entry)).next()
+    }
+    /// applies each specifier to each element of the `entries()` iterator.
+    /// any specifier overrides all earlier ones,
+    /// but the first entry that produces true or false will determine the output.
+    pub fn matches_of<'a, I: Iterator<Item = &'a FsEntry<'a>>>(
+        &self,
+        entries: impl Fn() -> I,
+    ) -> Option<bool> {
+        self.0
+            .iter()
+            .rev()
+            .filter_map(|v| entries().filter_map(|entry| v.matches(entry)).next())
+            .next()
+    }
+}
+impl Specifier {
+    pub fn matches(&self, entry: &FsEntry) -> Option<bool> {
+        match self {
+            Self::Except(inner) => inner.matches(entry).map(std::ops::Not::not),
+            Self::Entries(path) => path.matches(entry.path).then_some(true),
+            Self::Files(path) => (!entry.is_directory && path.matches(entry.path)).then_some(true),
+            Self::InDir { dir, inner } => {
+                dbg!(&inner);
+                if dbg!(inner.0.is_empty()) {
+                    // this has no inner things, so we just check for this directory
+                    // if this is a directory and it matches, then return true
+                    (entry.is_directory && dir.matches(entry.path)).then_some(true)
+                } else {
+                    // this has inner things, so, for every matching parent,
+                    // get the relative path (by removing the parent), ...
+                    let mut path = entry.path;
+                    let mut paths = vec![];
+                    while let Some(p) = path.parent() {
+                        if dir.matches(p) {
+                            if let Ok(p) = entry.path.strip_prefix(p) {
+                                let mut e = *entry;
+                                e.path = p;
+                                paths.push(e);
+                            } else {
+                                eprintln!("[WARN] Parent {p:?} of path {:?} could not be removed... this is probably be a bug.", entry.path);
+                            }
+                        }
+                        path = p;
+                    }
+                    // ... and then check if any match
+                    inner.matches_of(|| paths.iter())
+                }
+            }
+        }
+    }
+}
+
+impl Ignore {
+    pub fn parse(config: &str) -> Result<Self, String> {
+        Self::parsei(&mut config.lines().enumerate().peekable(), 0)
+    }
+    /// min_indent = parent_indent + 1, or 0 if no parent
+    fn parsei(lines: &mut Peekable<Enumerate<Lines>>, min_indent: usize) -> Result<Self, String> {
+        let mut indent = None;
+        let mut specifiers = vec![];
+        loop {
+            if let Some((line_nr, full_line)) = lines.peek() {
+                let line_nr = *line_nr;
+                let indent = {
+                    let line = full_line.trim_start();
+                    // check indentation
+                    let line_start_whitespace = &full_line[0..full_line.len() - line.len()];
+                    if let Some(c) = line_start_whitespace.chars().find(|c| *c != ' ') {
+                        return Err(format!(
+                        "Lines must start with any number of spaces, and no other whitespace character, but line {} contained the '{c:?}' character (Unicode {}).",
+                        line_nr + 1, c.escape_unicode()));
+                    }
+                    let line_indent = line_start_whitespace.len();
+                    if line_indent < min_indent {
+                        // less indentation than before, go up one layer of recursion
+                        break;
+                    }
+                    if let Some(indent) = indent {
+                        // check if we indent more/less than on the first line
+                        if line_indent != indent {
+                            return Err(format!(
+                                "Lines in one part of a config may must all have the same indentation! (expected {indent} spaces, but found {line_indent})"
+                            ));
+                        }
+                    } else {
+                        // store the first line's indent
+                        indent = Some(line_indent);
+                    }
+                    line_indent
+                };
+                // -- indentation is ok, this line belongs to us --
+                // because we only used `lines.peek()` until now
+                let line = lines.next().unwrap().1.trim_start();
+                if line.starts_with("#") {
+                    // comment, ignore
+                } else {
+                    let (line_type, args) =
+                        line.split_once(char::is_whitespace).unwrap_or((line, ""));
+                    specifiers.push(match line_type.to_lowercase().trim() {
+                        "except" => Specifier::Except(Ignore::parsei(lines, indent + 1)?),
+                        line_type => match (
+                            line_type.chars().next().unwrap_or(' '),
+                            line_type.chars().skip(1).next().unwrap_or(' '),
+                        ) {
+                            ('*', m) => Specifier::Entries(Match::parse_m(args, m, line_nr)?),
+                            ('+', m) => Specifier::Files(Match::parse_m(args, m, line_nr)?),
+                            ('/', m) => Specifier::InDir {
+                                dir: Match::parse_m(args, m, line_nr)?,
+                                inner: Ignore::parsei(lines, indent + 1)?,
+                            },
+                            _ => {
+                                return Err(format!(
+                                "Got '{line}' (Line {}), but expected one of [[*+/][a=*], except]",
+                                line_nr + 1
+                            ))
+                            }
+                        },
+                    });
+                }
+            } else {
+                break;
+            }
+        }
+        Ok(Self(specifiers))
+    }
+}
+impl Match {
+    fn parse_m(text: &str, m: char, line_nr: usize) -> Result<Self, String> {
+        Ok(match m {
+            'a' => Self::Any,
+            '=' => Self::Eq(text.into()),
+            '*' => Self::parse_glob(text, line_nr)?,
+            _ => {
+                return Err(format!(
+                    "[Line {}] unknown match-type '{m}', expected one of [a=*]",
+                    line_nr + 1
+                ))
+            }
+        })
+    }
+    fn parse_glob(text: &str, _line_nr: usize) -> Result<Self, String> {
+        Ok(Self::Glob(text.to_owned()))
+    }
+}

src/indexfile.rs

@@ -29,8 +29,8 @@ impl IndexFile {
                             false
                         }
                     }
-                    (Some(new), None) => !settings.dont_replace_if_timestamp_found,
-                    (None, Some(old)) => settings.replace_if_timestamp_lost,
+                    (Some(_), None) => !settings.dont_replace_if_timestamp_found,
+                    (None, Some(_)) => settings.replace_if_timestamp_lost,
                     (None, None) => settings.replace_if_timestamp_unknown,
                 })
     }

src/main.rs

@@ -3,34 +3,85 @@ use std::process::exit;
 use clap::Parser;
 
 use crate::{
-    apply_indexchanges::apply_indexchanges, indexchanges::IndexChange,
+    apply_indexchanges::apply_indexchanges, config::Ignore, indexchanges::IndexChange,
     update_index::perform_index_diff,
 };
 
 mod apply_indexchanges;
 mod args;
+mod config;
 mod indexchanges;
 mod indexfile;
 mod repr_file;
 mod update_index;
 
+const EXIT_IGNORE_FAILED: u8 = 200;
+const EXIT_DIFF_FAILED: u8 = 20;
+const EXIT_APPLY_FAILED: u8 = 30;
+
 fn main() {
     // get args
     let args = args::Args::parse();
     // index diff
     eprintln!("performing index diff...");
-    let source = &args.source;
-    let index = &args.index;
-    let ignore_subdirs = args
-        .ignore
-        .iter()
-        .map(|path| path.strip_prefix(source).unwrap_or(path))
-        .collect();
-    let changes = match perform_index_diff(source, index, ignore_subdirs, &args.settings) {
+    let cwd = match std::env::current_dir() {
+        Ok(v) => Some(v),
+        Err(e) => {
+            eprintln!("[WARN] Couldn't get current directory (CWD): {e}");
+            None
+        }
+    };
+    let source = if args.source.is_absolute() {
+        args.source.clone()
+    } else {
+        cwd.as_ref()
+            .expect("tried to use a relative path when there is no valid CWD")
+            .join(&args.source)
+    };
+    let index = if args.index.is_absolute() {
+        args.index.clone()
+    } else {
+        cwd.as_ref()
+            .expect("tried to use a relative path when there is no valid CWD")
+            .join(&args.index)
+    };
+    let target = args.target.as_ref().map(|target| {
+        if target.is_absolute() {
+            target.clone()
+        } else {
+            cwd.as_ref()
+                .expect("tried to use a relative path when there is no valid CWD")
+                .join(target)
+        }
+    });
+    let ignore = if let Some(path) = &args.ignore {
+        match std::fs::read_to_string(path) {
+            Ok(text) => match Ignore::parse(&text) {
+                Ok(config) => config,
+                Err(e) => {
+                    eprintln!("Couldn't parse ignore-file {path:?}: {e}");
+                    exit(EXIT_IGNORE_FAILED as _);
+                }
+            },
+            Err(e) => {
+                eprintln!("Couldn't load ignore-file {path:?}: {e}");
+                exit(EXIT_IGNORE_FAILED as _);
+            }
+        }
+    } else {
+        Ignore(vec![])
+    };
+    let changes = match perform_index_diff(
+        &source,
+        &index,
+        target.as_ref().map(|v| v.as_path()),
+        ignore,
+        &args.settings,
+    ) {
         Ok(c) => c,
         Err(e) => {
             eprintln!("Failed to generate index diff:\n    {e}");
-            exit(20);
+            exit(EXIT_DIFF_FAILED as _);
         }
     };
     if changes.is_empty() {
@@ -98,7 +149,7 @@ fn main() {
             Ok(()) => {}
             Err(e) => {
                 eprintln!("Failed to apply: {e}");
-                exit(30);
+                exit(EXIT_APPLY_FAILED as _);
             }
         }
     }

src/update_index.rs

@@ -2,7 +2,11 @@ use std::{collections::HashMap, fs, io, path::Path};
 
 use clap::Args;
 
-use crate::{indexchanges::IndexChange, indexfile::IndexFile};
+use crate::{
+    config::{FsEntry, Ignore, Match, Specifier},
+    indexchanges::IndexChange,
+    indexfile::IndexFile,
+};
 
 #[derive(Clone, Default, Args)]
 pub struct Settings {
@@ -26,20 +30,33 @@ pub struct Settings {
 pub fn perform_index_diff<'a>(
     source: &Path,
     index: &'a Path,
-    mut ignore_paths: Vec<&'a Path>,
+    target: Option<&'a Path>,
+    mut ignore: Ignore,
     settings: &Settings,
 ) -> io::Result<Vec<IndexChange>> {
     let mut changes = Vec::new();
     if let Ok(inner_index) = index.strip_prefix(source) {
-        eprintln!("[info] source contains index, but index will not be part of the backup.");
-        ignore_paths.push(inner_index);
+        eprintln!("[info] source contains index at {inner_index:?}, but index will not be part of the backup.");
+        ignore.0.push(Specifier::InDir {
+            dir: Match::Eq(inner_index.to_owned()),
+            inner: Ignore(vec![]),
+        });
+    }
+    if let Some(target) = target {
+        if let Ok(inner_target) = target.strip_prefix(source) {
+            eprintln!("[info] source contains target at {inner_target:?}, but target will not be part of the backup.");
+            ignore.0.push(Specifier::InDir {
+                dir: Match::Eq(inner_target.to_owned()),
+                inner: Ignore(vec![]),
+            });
+        }
     }
     rec(
         source.as_ref(),
         Path::new(""),
         index,
         &mut changes,
-        &ignore_paths,
+        &ignore,
         settings,
     )?;
     Ok(changes)
@@ -53,7 +70,7 @@ fn rec(
     index_files: &Path,
     // list of changes to be made
     changes: &mut Vec<IndexChange>,
-    ignore_paths: &Vec<&Path>,
+    ignore: &Ignore,
     settings: &Settings,
 ) -> Result<(), io::Error> {
     // used to find removals
@@ -75,26 +92,24 @@ fn rec(
     for entry in source_files {
         let entry = entry?;
         let rel_path = rel_path.join(entry.file_name());
+        let metadata = entry.metadata()?;
+
         // ignore entries
-        if ignore_paths.iter().any(|ii| &rel_path == ii) {
+        let fs_entry = FsEntry {
+            path: &rel_path,
+            is_directory: metadata.is_dir(),
+        };
+        if ignore.matches_or_default(&fs_entry) {
             continue;
         }
 
-        let metadata = entry.metadata()?;
         let in_index_and_is_dir = index_entries.remove(&entry.file_name());
         if metadata.is_dir() {
             if let Some(false) = in_index_and_is_dir {
                 // is dir, but was file -> remove file
                 changes.push(IndexChange::RemoveFile(rel_path.clone()));
             }
-            rec(
-                source,
-                &rel_path,
-                index_files,
-                changes,
-                ignore_paths,
-                settings,
-            )?;
+            rec(source, &rel_path, index_files, changes, ignore, settings)?;
         } else {
             if let Some(true) = in_index_and_is_dir {
                 // is file, but was dir -> remove dir