feat: read .gz manpages

.gitattributes

@@ -1,2 +1,4 @@
 Cargo.lock linguist-generated=true
 flake.lock linguist-generated=true
+
+tests/** linguist-generated=true

cli/src/main.rs

@@ -27,6 +27,10 @@ struct CLI {
   #[arg(short, long)]
   out: PathBuf,
 
+  /// Turn on verbose output
+  #[arg(short, long)]
+  verbose: bool,
+
   /// Directories to exclude from search
   #[arg(short = 'D', long, value_delimiter = ',')]
   dirs_exclude: Option<Vec<PathBuf>>,
@@ -73,6 +77,9 @@ fn main() -> Result<()> {
 
       if let Some(cmd) = &args.cmd {
         if let Some(manpage) = man_completions::find_manpage(cmd, included) {
+          if args.verbose {
+            println!("Found manpage at {}", &manpage.display());
+          }
           let parsed = parse_manpage_at_path(cmd, manpage)?;
           let mut map = HashMap::new();
           map.insert(cmd.to_string(), parsed);

lib/src/parse/mod.rs

@@ -1,4 +1,10 @@
-use std::collections::HashMap;
+use std::{
+  collections::HashMap,
+  fs::File,
+  io::{BufReader, Read},
+};
+
+use flate2::bufread::GzDecoder;
 
 use crate::Result;
 use std::path::Path;
@@ -21,7 +27,19 @@ where
 {
   let path = manpage_path.as_ref();
   match path.extension() {
-    Some(ext) => todo!(),
+    Some(ext) => {
+      if ext == "gz" {
+        let file = File::open(path)?;
+        let reader = BufReader::new(file);
+        let mut decoder = GzDecoder::new(reader);
+        let mut str = String::new();
+        // TODO this only works with UTF-8
+        decoder.read_to_string(&mut str)?;
+        Ok(str)
+      } else {
+        todo!()
+      }
+    }
     None => todo!(),
   }
 }
@@ -34,5 +52,6 @@ where
 }
 
 pub fn parse_manpage_text(cmd_name: &str, text: &str) -> Result<CommandInfo> {
-  todo!()
+  let res = parse_man::parse(cmd_name, text)?;
+  Ok(res.unwrap())
 }

lib/src/parse/parse_man.rs

@@ -1,11 +1,26 @@
-use regex::Regex;
+use regex::{Regex, RegexBuilder};
 
-fn parse(s: &str) -> Option<()> {
-  match s.find("\nOPTIONS") {
-    Some(options_header) => {
-      let re = Regex::new("").unwrap();
+use crate::result::Result;
+
+use super::CommandInfo;
+
+/// Regex to get the contents of a section with the given title
+fn regex_for_section(title: &str) -> Regex {
+  RegexBuilder::new(&format!(r#"\.SH {title}(.*?)(\.SH|\z)"#))
+    .multi_line(true)
+    .dot_matches_new_line(true)
+    .build()
+    .unwrap()
+}
+
+pub fn parse(cmd_name: &str, page_text: &str) -> Result<Option<CommandInfo>> {
+  let re = regex_for_section(r#""OPTIONS""#);
+  match re.captures(page_text) {
+    Some(captures) => {
+      let content = captures.get(1).unwrap().as_str();
+      println!("{}", content);
       todo!()
     }
-    None => None
+    None => Ok(None),
   }
 }

lib/src/result.rs

@@ -4,8 +4,8 @@ pub type Result<T> = std::result::Result<T, Error>;
 
 #[derive(Debug, Error)]
 pub enum Error {
-  #[error("could not parse manpage")]
-  ParseError(String),
+  #[error("parse error in {cmd}: {msg}")]
+  ParseError { cmd: String, msg: String },
 
   #[error(transparent)]
   IoError(#[from] std::io::Error),