comparison contrib/merge-lists/src/main.rs @ 49011:b999edb15f8c

merge-lists: make it possible to specify pattern to match

The `merge-lists` tool doesn't know anything about Python other than
its regex that attempts to match import lines. Let's make it possible
to pass in a custom regex so it's easy to use the tool for e.g. C/C++
`#include` lines or Rust `use` lines (given the limited).

Differential Revision: https://phab.mercurial-scm.org/D12392

49010:681b25ea579e

use clap::Parser;
use itertools::Itertools;
use regex::bytes::Regex;
use similar::ChangeTag;
use std::cmp::{max, min, Ordering};
use std::collections::HashSet;

/// A tool that performs a 3-way merge, resolving conflicts in sorted lists and
/// leaving other conflicts unchanged. This is useful with Mercurial's support
/// for partial merge tools (configured in `[partial-merge-tools]`).
#[derive(Parser, Debug)]
#[clap(version, about, long_about = None)]
struct Args {
    /// Path to the file's content in the "local" side
    local: OsString,

    /// Path to the file's content in the base
    base: OsString,

    /// Path to the file's content in the "other" side
    other: OsString,
}

fn main() {
    let args: Args = Args::parse();

    let base_bytes = std::fs::read(&base_path).unwrap();
    let local_bytes = std::fs::read(&local_path).unwrap();
    let other_bytes = std::fs::read(&other_path).unwrap();

    let regex =
        regex::bytes::Regex::new(r"import \w+(\.\w+)*( +#.*)?\n|from (\w+(\.\w+)* import \w+( as \w+)?(, \w+( as \w+)?)*( +#.*)?)\r?\n?").unwrap();
    let (new_base_bytes, new_local_bytes, new_other_bytes) =
        resolve(&base_bytes, &local_bytes, &other_bytes, &regex);

    // Write out the result if anything changed
    if new_base_bytes != base_bytes {

49011:b999edb15f8c

use clap::{ArgGroup, Parser};
use itertools::Itertools;
use regex::bytes::Regex;
use similar::ChangeTag;
use std::cmp::{max, min, Ordering};
use std::collections::HashSet;

/// A tool that performs a 3-way merge, resolving conflicts in sorted lists and
/// leaving other conflicts unchanged. This is useful with Mercurial's support
/// for partial merge tools (configured in `[partial-merge-tools]`).
#[derive(Parser, Debug)]
#[clap(version, about, long_about = None)]
#[clap(group(ArgGroup::new("match").required(true).args(&["pattern", "python-imports"])))]
struct Args {
    /// Path to the file's content in the "local" side
    local: OsString,

    /// Path to the file's content in the base
    base: OsString,

    /// Path to the file's content in the "other" side
    other: OsString,

    /// Regular expression to use
    #[clap(long, short)]
    pattern: Option<String>,

    /// Use built-in regular expression for Python imports
    #[clap(long)]
    python_imports: bool,
}

fn get_regex(args: &Args) -> Regex {
    let pattern = if args.python_imports {
        r"import \w+(\.\w+)*( +#.*)?\n|from (\w+(\.\w+)* import \w+( as \w+)?(, \w+( as \w+)?)*( +#.*)?)"
    } else if let Some(pattern) = &args.pattern {
        pattern
    } else {
        ".*"
    };
    let pattern = format!(r"{}\r?\n?", pattern);
    regex::bytes::Regex::new(&pattern).unwrap()
}

fn main() {
    let args: Args = Args::parse();

    let base_bytes = std::fs::read(&base_path).unwrap();
    let local_bytes = std::fs::read(&local_path).unwrap();
    let other_bytes = std::fs::read(&other_path).unwrap();

    let regex = get_regex(&args);
    let (new_base_bytes, new_local_bytes, new_other_bytes) =
        resolve(&base_bytes, &local_bytes, &other_bytes, &regex);

    // Write out the result if anything changed
    if new_base_bytes != base_bytes {