fix: remove .br before empty lines to fix mandoc warnings

- Enhanced post-processing to remove .br macros that appear before empty lines
- This fixes 'WARNING: skipping paragraph macro: br before sp' from mandoc
- Also handles the common pattern of .br-empty-.br by removing both .br macros
- Added comprehensive test cases for the new patterns
- All mandoc 'br before sp' warnings are now resolved

Author: sylvestre

Cargo.lock

@@ -35,6 +35,7 @@ uudoc = [
   "dep:clap_complete",
   "dep:clap_mangen",
   "dep:fluent-syntax",
+  "dep:jiff",
   "dep:regex",
   "dep:zip",
 ]
@@ -481,6 +482,7 @@ clap_complete = { workspace = true, optional = true }
 clap_mangen = { workspace = true, optional = true }
 clap.workspace = true
 fluent-syntax = { workspace = true, optional = true }
+jiff = { workspace = true, optional = true }
 regex = { workspace = true, optional = true }
 itertools.workspace = true
 phf.workspace = true

Cargo.toml

@@ -17,8 +17,8 @@ use clap::{Arg, Command};
 use clap_complete::Shell;
 use clap_mangen::Man;
 use fluent_syntax::ast::{Entry, Message, Pattern};
-use jiff::Zoned;
 use fluent_syntax::parser;
+use jiff::Zoned;
 use regex::Regex;
 use textwrap::{fill, indent, termwidth};
 use zip::ZipArchive;
@@ -31,10 +31,13 @@ include!(concat!(env!("OUT_DIR"), "/uutils_map.rs"));
 /// Post-process a generated manpage to fix mandoc lint issues
 ///
 /// This function:
-/// - Fixes the TH header by uppercasing command names and removing invalid date formats
+/// - Fixes the TH header by uppercasing command names and adding a proper date
 /// - Removes trailing whitespace from all lines
 /// - Fixes redundant .br paragraph macros that cause mandoc warnings
-fn post_process_manpage(manpage: String) -> String {
+/// - Removes .br before empty lines to avoid "br before sp" warnings
+/// - Removes .br after empty lines to avoid "br after sp" warnings
+/// - Fixes escape sequences (e.g., \\\\0 to \\0) to avoid "undefined escape" warnings
+fn post_process_manpage(manpage: String, date: Option<&str>) -> String {
     // Only match TH headers that have at least a command name on the same line
     // Use [ \t] instead of \s to avoid matching newlines
     // Use a date format that satisfies mandoc (YYYY-MM-DD)
@@ -61,16 +64,22 @@ fn post_process_manpage(manpage: String) -> String {
         let line = lines[i].trim_end();
 
         if line == ".br" && !skip_indices.contains(&i) {
-            // Check for consecutive .br macros
-            if i > 0 && lines[i - 1].trim_end() == ".br" {
+            // Check for .br followed by empty line
+            if i + 1 < lines.len() && lines[i + 1].trim().is_empty() {
+                // Remove the .br when it's followed by an empty line
+                // This prevents "WARNING: skipping paragraph macro: br before sp"
                 skip_indices.insert(i);
+
+                // Also check if there's another .br after the empty line (common pattern)
+                if i + 2 < lines.len() && lines[i + 2].trim_end() == ".br" {
+                    skip_indices.insert(i + 2);
+                }
             }
-            // Check for .br, empty line, .br pattern
-            else if i + 2 < lines.len()
-                && lines[i + 1].trim().is_empty()
-                && lines[i + 2].trim_end() == ".br"
+            // Check for .br preceded by empty line or another .br
+            // This prevents "WARNING: skipping paragraph macro: br after sp" and consecutive .br
+            else if i > 0 && (lines[i - 1].trim().is_empty() || lines[i - 1].trim_end() == ".br")
             {
-                skip_indices.insert(i + 2);
+                skip_indices.insert(i);
             }
         }
     }
@@ -829,4 +838,44 @@ mod tests {
         let result4 = post_process_manpage(input4.to_string(), Some("2024-01-01"));
         assert_eq!(result4, expected4);
     }
+
+    #[test]
+    fn test_post_process_manpage_removes_br_before_empty_line() {
+        // Test that .br is removed when followed by empty line (which becomes .sp)
+        let input = ".TH TEST 1\nSome text\n.br\n\nMore text\n";
+        let expected = ".TH TEST 1 \"2024-01-01\"\nSome text\n\nMore text\n";
+
+        let result = post_process_manpage(input.to_string(), Some("2024-01-01"));
+        assert_eq!(result, expected);
+    }
+
+    #[test]
+    fn test_post_process_manpage_complex_br_before_empty() {
+        // Test multiple .br before empty line patterns
+        let input = ".TH TEST 1\nSection 1\n.br\n\nSection 2\n.br\n\nSection 3\n";
+        let expected = ".TH TEST 1 \"2024-01-01\"\nSection 1\n\nSection 2\n\nSection 3\n";
+
+        let result = post_process_manpage(input.to_string(), Some("2024-01-01"));
+        assert_eq!(result, expected);
+    }
+
+    #[test]
+    fn test_post_process_manpage_removes_br_after_empty_line() {
+        // Test that .br is removed when preceded by empty line (which becomes .sp)
+        let input = ".TH TEST 1\nSome text\n\n.br\nMore text\n";
+        let expected = ".TH TEST 1 \"2024-01-01\"\nSome text\n\nMore text\n";
+
+        let result = post_process_manpage(input.to_string(), Some("2024-01-01"));
+        assert_eq!(result, expected);
+    }
+
+    #[test]
+    fn test_post_process_manpage_fixes_escape_sequences() {
+        // Test that \\\\0 and \\0 are fixed to \e0 (literal backslash-zero)
+        let input = ".TH TEST 1\nText with \\\\\\\\0 and \\\\0 escape\n";
+        let expected = ".TH TEST 1 \"2024-01-01\"\nText with \\e0 and \\e0 escape\n";
+
+        let result = post_process_manpage(input.to_string(), Some("2024-01-01"));
+        assert_eq!(result, expected);
+    }
 }

src/bin/uudoc.rs

@@ -36,7 +36,7 @@ fn test_manpage_generation() {
     );
 
     let output_str = String::from_utf8_lossy(&output.stdout);
-    assert!(output_str.contains("\n.TH ls"), "{output_str}");
+    assert!(output_str.contains("\n.TH LS"), "{output_str}");
     assert!(output_str.contains('1'), "{output_str}");
     assert!(output_str.contains("\n.SH NAME\nls"), "{output_str}");
 }
@@ -62,7 +62,7 @@ fn test_manpage_coreutils() {
     );
 
     let output_str = String::from_utf8_lossy(&output.stdout);
-    assert!(output_str.contains("\n.TH coreutils"), "{output_str}");
+    assert!(output_str.contains("\n.TH COREUTILS"), "{output_str}");
     assert!(output_str.contains("coreutils"), "{output_str}");
     assert!(output_str.contains("\n.SH NAME\ncoreutils"), "{output_str}");
 }