Port Fish's Darwin parser

src/parse_man/darwin.rs

@@ -0,0 +1,46 @@
+use regex::{Regex, RegexBuilder};
+
+use super::util;
+use crate::Flag;
+
+/// For parsing Darwin man pages (ported from Fish)
+#[allow(
+ clippy::case_sensitive_file_extension_comparisons,
+ clippy::doc_markdown
+)]
+pub fn parse(cmd_name: &str, page_text: &str) -> Vec<Flag> {
+ let Some(start_ind) = page_text.find(".Sh DESCRIPTION") else {
+ return Vec::new();
+ };
+
+ let mut flags = Vec::new();
+
+ // Can't use util::get_section because we also want sections after DESCRIPTION
+ let content = &page_text[start_ind..];
+
+ // Replace '.It Fl' and the like with '.It -' (`Fl` is a dash)
+ let fl_re = Regex::new(r"(\...) Fl ").expect("Regex should be valid");
+ let content = fl_re.replace_all(content, "$1 -").replace(".Fl ", "-");
+
+ let desc_end = RegexBuilder::new(r"\n\.El.*+")
+ .dot_matches_new_line(true)
+ .build()
+ .expect("Regex should be valid");
+
+ let mut paras = content.split(".It");
+ paras.next(); // Discard the part before the first option
+ for para in paras {
+ let mut pieces = para.splitn(2, '\n');
+ if let Some(options) = pieces.next() {
+ let desc = pieces.next().map(|desc| {
+ let desc = desc.replace(".Nm", cmd_name);
+ desc_end.replace(&desc, "").to_string()
+ });
+ if let Some(flag) = util::make_flag(options, desc.as_deref()) {
+ flags.push(flag);
+ }
+ }
+ }
+
+ flags
+}

src/parse_man/mod.rs

@@ -1,4 +1,5 @@
 //! For parsing command information from man pages
+mod darwin;
 pub mod error;
 mod podman;
 mod scdoc;
@@ -68,6 +69,7 @@ pub fn parse_manpage_text(
  type4::parse(cmd_name, text),
  scdoc::parse(cmd_name, text),
  podman::parse(cmd_name, text),
+ darwin::parse(cmd_name, text),
  ]
  .into_iter()
  .flatten()

src/parse_man/util.rs

@@ -16,8 +16,7 @@ pub fn trim_desc(desc: &str) -> String {
 
 /// Get the contents of a section with the given title
 pub fn get_section(title: &str, text: &str) -> Option<String> {
- // [hH] is necessary for Darwin
- let re = RegexBuilder::new(&format!(r#"\.S[hH] {title}(.*?)(\.S[hH]|\z)"#))
+ let re = RegexBuilder::new(&format!(r#"\.SH {title}(.*?)(\.SH|\z)"#))
  .multi_line(true)
  .dot_matches_new_line(true)
  .build()
@@ -35,14 +34,15 @@ pub fn remove_groff_formatting(data: &str) -> String {
  .replace(r"\fB", "")
  .replace(r"\fR", "")
  .replace(r"\e", "");
+
  // TODO check if this one is necessary
- // also, fish uses a slightly different regex: `.PD( \d+)`, check if that's
- // fine
  let re = Regex::new(r"\.PD \d+").unwrap();
  let data = re.replace_all(&data, "");
+
  let data = data
  .replace(".BI", "")
  .replace(".BR", "")
+ .replace(".B", "")
  .replace("0.5i", "")
  .replace(".rb", "")
  .replace(r"\^", "")
@@ -51,12 +51,13 @@ pub fn remove_groff_formatting(data: &str) -> String {
  .replace(r"\ ", " ")
  .replace(r"\-", "-")
  .replace(r"\&", "")
- .replace(".B", "")
- .replace(r"\-", "-")
  // .replace(".I", "") // This breaks podman since it removes .IX
- .replace('\u{C}', "")
- .replace(r"\(cq", "'")
- .replace(r"\(aq", "'"); // Added by me, not from Fish. May need to remove all \(xx
+ .replace('\u{C}', "");
+
+ let quotes = Regex::new(r"\([ocadlr]q").unwrap();
+ let data = quotes.replace_all(&data, "'");
+
+ let data = data.replace(".Pp", ""); // Fish only replaces this one on MacOS
 
  // todo Fish doesn't do this, see how it handles .sp
  let re = Regex::new(&format!(r"\.sp( {NUM_RE}v?)?")).unwrap();

tests/man_integration_tests.rs

@@ -138,3 +138,8 @@ fn scdoc_sway_json() {
 fn podman_ncdu_json() {
  run_test("json", &["ncdu"], &["--cmds", "^ncdu"]);
 }
+
+#[test]
+fn darwin_bless() {
+ run_test("json", &["bless"], &["--cmds", "^bless"]);
+}