Merge pull request #8787 from Misakait/fix/ptx-tex-format-bug

fix(ptx): Correct various output formatting bugs

Author: cakebaker

src/uu/ptx/src/ptx.rs

@@ -6,6 +6,7 @@
 // spell-checker:ignore (ToDOs) corasick memchr Roff trunc oset iset CHARCLASS
 
 use std::cmp;
+use std::cmp::PartialEq;
 use std::collections::{BTreeSet, HashMap, HashSet};
 use std::ffi::{OsStr, OsString};
 use std::fmt::Write as FmtWrite;
@@ -22,7 +23,7 @@ use uucore::error::{FromIo, UError, UResult, UUsageError};
 use uucore::format_usage;
 use uucore::translate;
 
-#[derive(Debug)]
+#[derive(Debug, PartialEq)]
 enum OutFormat {
     Dumb,
     Roff,
@@ -483,7 +484,18 @@ fn get_output_chunks(
     let tail_end = trim_broken_word_right(all_after, tail_beg, tail_end);
 
     // trim away whitespace again.
-    let (tail_beg, tail_end) = trim_idx(all_after, tail_beg, tail_end);
+    let (tail_beg, mut tail_end) = trim_idx(all_after, tail_beg, tail_end);
+    // Fix: Manually trim trailing char (like "a") that are preceded by a space.
+    // This handles cases like "is a" which are not correctly trimmed by the
+    // preceding functions.
+    if tail_end >= 2
+        && (tail_end - 2) > tail_beg
+        && all_after[tail_end - 2].is_whitespace()
+        && !all_after[tail_end - 1].is_whitespace()
+    {
+        tail_end -= 1;
+        (_, tail_end) = trim_idx(all_after, tail_beg, tail_end);
+    }
 
     // and get the string
     let tail_str: String = all_after[tail_beg..tail_end].iter().collect();
@@ -511,19 +523,21 @@ fn get_output_chunks(
     // and get the string.
     let head_str: String = all_before[head_beg..head_end].iter().collect();
     head.push_str(&head_str);
+    //The TeX mode does not output truncation characters.
+    if config.format != OutFormat::Tex {
+        // put right context truncation string if needed
+        if after_end != all_after.len() && tail_beg == tail_end {
+            after.push_str(&config.trunc_str);
+        } else if after_end != all_after.len() && tail_end != all_after.len() {
+            tail.push_str(&config.trunc_str);
+        }
 
-    // put right context truncation string if needed
-    if after_end != all_after.len() && tail_beg == tail_end {
-        after.push_str(&config.trunc_str);
-    } else if after_end != all_after.len() && tail_end != all_after.len() {
-        tail.push_str(&config.trunc_str);
-    }
-
-    // put left context truncation string if needed
-    if before_beg != 0 && head_beg == head_end {
-        before = format!("{}{before}", config.trunc_str);
-    } else if before_beg != 0 && head_beg != 0 {
-        head = format!("{}{head}", config.trunc_str);
+        // put left context truncation string if needed
+        if before_beg != 0 && head_beg == head_end {
+            before = format!("{}{before}", config.trunc_str);
+        } else if before_beg != 0 && head_beg != 0 {
+            head = format!("{}{head}", config.trunc_str);
+        }
     }
 
     (tail, before, after, head)

tests/by-util/test_ptx.rs

@@ -11,6 +11,25 @@ fn test_invalid_arg() {
     new_ucmd!().arg("--definitely-invalid").fails_with_code(1);
 }
 
+#[test]
+fn test_tex_format_no_truncation_markers() {
+    let input = "Hello world Rust is a fun language";
+    new_ucmd!()
+        .args(&["-G", "-w", "30", "--format=tex"])
+        .pipe_in(input)
+        .succeeds()
+        .stdout_only_fixture("test_tex_format_no_truncation_markers.expected");
+}
+#[test]
+fn gnu_ext_disabled_chunk_no_over_reading() {
+    let input = "Hello World Rust is a fun language";
+    new_ucmd!()
+        .args(&["-G", "-w", "30"])
+        .pipe_in(input)
+        .succeeds()
+        .stdout_only_fixture("gnu_ext_disabled_chunk_no_over_reading.expected");
+}
+
 #[test]
 fn test_truncation_no_extra_space_in_after() {
     new_ucmd!()

tests/fixtures/ptx/gnu_ext_disabled_chunk_no_over_reading.expected

@@ -0,0 +1,7 @@
+.xx "Rust is/" "" "Hello World" ""
+.xx "" "Hello World" "Rust is a/" ""
+.xx "is/" "Hello" "World Rust" ""
+.xx "" "/Rust is" "a fun/" ""
+.xx "" "/Rust is a" "fun language" ""
+.xx "" "/World Rust" "is a fun/" ""
+.xx "" "/is a fun" "language" ""

tests/fixtures/ptx/test_tex_format_no_truncation_markers.expected

@@ -0,0 +1,7 @@
+\xx {Rust is}{}{Hello}{ world}{}
+\xx {}{Hello world}{Rust}{ is a}{}
+\xx {}{Rust is}{a}{ fun}{}
+\xx {}{Rust is a}{fun}{ language}{}
+\xx {}{world Rust}{is}{ a fun}{}
+\xx {}{is a fun}{language}{}{}
+\xx {is}{Hello}{world}{ Rust}{}