forgejo-todo-checker/src/scanner.rs

//
use std::{collections::HashSet, path::Path};

use crate::{
    issues::Issue,
    model::{Config, Line, Marker},
    printer::Printer,
};
use anyhow::Result;
use file_format::FileFormat;
use ignore::Walk;

pub trait FileScanner {
    fn scan_file(
        &self,
        path: &Path,
        config: &Config,
        printer: &impl Printer,
        issues: &HashSet<Issue>,
    ) -> Result<u32>;
}

pub fn find_markers(
    printer: &impl Printer,
    config: &Config,
    issues: HashSet<Issue>,
    file_scanner: &impl FileScanner,
) -> Result<u32, anyhow::Error> {
    let mut errors = 0;
    for file in Walk::new(config.fs().base()).flatten() {
        let path = file.path();
        if is_text_file(config, path)? {
            errors += file_scanner.scan_file(path, config, printer, &issues)?
        }
    }
    Ok(errors)
}

fn is_text_file(config: &Config, path: &Path) -> Result<bool> {
    Ok(config.fs().path_is_file(path)?
        && FileFormat::from_file(path)?
            .media_type()
            .starts_with("text/"))
}

pub struct DefaultFileScanner;
impl FileScanner for DefaultFileScanner {
    fn scan_file(
        &self,
        file: &Path,
        config: &Config,
        printer: &impl Printer,
        issues: &HashSet<Issue>,
    ) -> Result<u32> {
        let relative_path = file.strip_prefix(config.fs().base())?.to_path_buf();
        let mut errors = 0;
        config
            .fs()
            .file_read_to_string(file)? // tarpaulin uncovered okay
            .lines()
            .enumerate()
            .map(|(n, line)| {
                Line::builder()
                    .relative_path(relative_path.clone())
                    .num(n + 1) // line numbers are not 0-based, but enumerate is
                    .value(line.to_owned())
                    .build()
            })
            .filter_map(|line| line.into_marker().ok())
            .filter(|marker| !matches!(marker, Marker::Unmarked))
            .map(|marker| has_open_issue(marker, issues))
            .for_each(|marker| match marker {
                Marker::Invalid(_) => {
                    errors += 1;
                    printer.println(marker.to_string());
                }
                Marker::Closed(_, _) => {
                    errors += 1;
                    printer.println(marker.to_string());
                }
                _ => {}
            });
        if errors > 0 {
            printer.println(format!(
                ">> {errors} errors in {}\n",
                relative_path.to_string_lossy()
            ));
        }

        Ok(errors)
    }
}

fn has_open_issue(marker: Marker, issues: &HashSet<Issue>) -> Marker {
    if let Marker::Valid(_, ref issue) = marker {
        let has_open_issue = issues.contains(issue);
        if has_open_issue {
            marker
        } else {
            marker.into_closed()
        }
    } else {
        marker
    }
}