From 708bcb0b911431bf7fbdde5afc5296c7e196c19d Mon Sep 17 00:00:00 2001
From: Paul Campbell <campbjpa@amazon.com>
Date: Wed, 18 Sep 2024 07:44:34 +0100
Subject: [PATCH] feat: scan for TODO and FIXME markers

---
 Cargo.toml  |   1 +
 src/main.rs | 153 ++++++++++++++++++++++++++++++++++++++++++++++------
 2 files changed, 137 insertions(+), 17 deletions(-)

diff --git a/Cargo.toml b/Cargo.toml
index 844f56c..08ea434 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -7,3 +7,4 @@ edition = "2021"
 anyhow = "1.0"
 regex = "1.10"
 ureq = "2.10"
+kxio = "1.2"
diff --git a/src/main.rs b/src/main.rs
index c5a2ca7..94f3a39 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -1,24 +1,18 @@
-use std::path::PathBuf;
+use std::path::{Path, PathBuf};
 
-use anyhow::Context;
+use anyhow::{Context, Result};
+use kxio::fs::DirItem;
 use regex::Regex;
 
-struct Config {
-    workdir: PathBuf,
-    repo: String,
-    server: String,
-    auth_token: Option<String>,
-    prefix_pattern: Regex,
-    issue_pattern: Regex,
-}
-
-fn main() -> anyhow::Result<()> {
+fn main() -> Result<()> {
     println!("Forgejo TODO Checker!");
 
     let config = Config {
-        workdir: std::env::var("GITHUB_WORKSPACE")
-            .context("GITHUB_WORKSPACE")?
-            .into(),
+        fs: kxio::fs::new(
+            std::env::var("GITHUB_WORKSPACE")
+                .context("GITHUB_WORKSPACE")?
+                .into(),
+        ),
         repo: std::env::var("GITHUB_REPOSITORY").context("GITHUB_REPOSITORY")?,
         server: std::env::var("GITHUB_SERVER_URL").context("GITHUB_SERVER_URL")?,
         auth_token: std::env::var("REPO_TOKEN").ok(),
@@ -33,8 +27,12 @@ fn main() -> anyhow::Result<()> {
 
     // TODO: scan files in workdir
     // TODO: ignore files listed in .rgignore .ignore or .gitignore
-    // TODO: scan for malformed TODO and FIXME comments (e.g. no issue number) - add to error list
-    // TODO: build list of expected open issues with file locations where found
+
+    let mut found_markers = FoundMarkers::default();
+
+    scan_files(&config, &mut found_markers)?;
+
+    println!("{found_markers:?}");
 
     // list files in current directory to get a feel for what we have access to
     // std::fs::read_dir(workdir)?
@@ -60,3 +58,124 @@ fn main() -> anyhow::Result<()> {
 
     Ok(())
 }
+
+struct Config {
+    fs: kxio::fs::FileSystem,
+    repo: String,
+    server: String,
+    auth_token: Option<String>,
+    prefix_pattern: Regex,
+    issue_pattern: Regex,
+}
+
+#[derive(Debug, Default)]
+struct FoundMarkers {
+    markers: Vec<Marker>,
+    issue_markers: Vec<IssueMarker>,
+}
+impl FoundMarkers {
+    fn add_marker(&mut self, marker: Marker) {
+        self.markers.push(marker);
+    }
+    fn add_issue_marker(&mut self, issue_marker: IssueMarker) {
+        self.issue_markers.push(issue_marker);
+    }
+}
+
+fn scan_files(config: &Config, found_markers: &mut FoundMarkers) -> Result<()> {
+    scan_dir(config.fs.base(), config, found_markers)
+}
+
+fn scan_dir(dir: &Path, config: &Config, found_markers: &mut FoundMarkers) -> Result<()> {
+    let read_dir = config.fs.dir_read(dir)?;
+    for entry in read_dir {
+        match entry? {
+            DirItem::File(file) => scan_file(&file, config, found_markers),
+            DirItem::Dir(dir) => scan_dir(&dir, config, found_markers),
+            DirItem::SymLink(_) | DirItem::Fifo(_) | DirItem::Unsupported(_) => Ok(()),
+        }?;
+    }
+    Ok(())
+}
+
+fn scan_file(path: &Path, config: &Config, found_markers: &mut FoundMarkers) -> Result<()> {
+    config
+        .fs
+        .file_read_to_string(path)?
+        .lines()
+        .enumerate()
+        .filter_map(|(n, line)| prefix_match(n, line, path, config))
+        .for_each(|marker| {
+            if let Some(issue) = config
+                .issue_pattern
+                .find(&marker.line.value)
+                .map(|issue| issue.as_str())
+                .and_then(|issue| issue.parse::<usize>().ok())
+            {
+                found_markers.add_issue_marker(marker.into_issue_marker(issue));
+            } else {
+                found_markers.add_marker(marker);
+            }
+        });
+    Ok(())
+}
+
+fn prefix_match(num: usize, line: &str, path: &Path, config: &Config) -> Option<Marker> {
+    let find = config.prefix_pattern.find(line)?;
+    let marker_type = match find.as_str() {
+        "TODO" => Some(MarkerType::Todo),
+        "FIXME" => Some(MarkerType::Fixme),
+        _ => None,
+    }?;
+    Some(Marker {
+        marker_type,
+        file: path.to_path_buf(),
+        line: Line {
+            num,
+            value: line.to_owned(),
+        },
+    })
+}
+
+/// What type of comment
+#[derive(Debug)]
+enum MarkerType {
+    Todo,
+    Fixme,
+}
+
+#[derive(Debug)]
+struct Line {
+    num: usize,
+    value: String,
+}
+
+/// Represents a TODO or FIXME comment that doesn't have any issue number
+#[derive(Debug)]
+struct Marker {
+    /// What type of marker
+    marker_type: MarkerType,
+
+    /// Path of the file
+    file: PathBuf,
+
+    /// The line from the file
+    line: Line,
+}
+impl Marker {
+    fn into_issue_marker(self, issue: usize) -> IssueMarker {
+        IssueMarker {
+            marker: self,
+            issue,
+        }
+    }
+}
+
+#[derive(Debug)]
+struct IssueMarker {
+    /// The marker
+    marker: Marker,
+
+    /// The issue number
+    issue: usize,
+}