Detect man pages by contents

Not all man implementations set environment variables when invoking
the pager. For those that don't, we now detect man pages by contents.

Previously we looked at the existence of MANPATH, but that's *not* set
by "man" on macOS, I don't know where I got that from. Maybe it changed?

Author: walles

cmd/moor/moor.go

@@ -327,25 +327,23 @@ func russiaNotSupported() {
 	os.Exit(1)
 }
 
-// On man pages, disable line numbers by default.
+// For git output and man pages, disable line numbers by default.
 //
 // Before paging, "man" first checks the terminal width and formats the man page
-// to fit that width.
+// to fit that width. Git does that as well for git diff --stat.
 //
-// Then, if moor adds line numbers, the rightmost part of the man page won't be
-// visible.
+// Then, if moor adds line numbers, the rightmost part of the page will scroll
+// out of view.
 //
-// So we try to detect showing man pages, and in that case disable line numbers
-// so that the rightmost part of the page is visible by default.
+// So we try to this, and in that case disable line numbers so that the
+// rightmost part of the page is visible by default.
+//
+// See also internal/haveLoadedManPage(), where we try to detect man pages by
+// their contents.
 func noLineNumbersDefault() bool {
-	if os.Getenv("MANPATH") != "" {
-		// Set by "man" on macOS, skip line numbers in this case
-		return true
-	}
-
 	if os.Getenv("MAN_PN") != "" {
-		// Set by "man" on Ubuntu 22.04.4 when I tested it inside of Docker,
-		// skip line numbers in this case
+		// Set by "man" on Ubuntu 22.04.4 when I tested it inside of Docker.
+		log.Debug("MAN_PN is set, skipping line numbers for man page")
 		return true
 	}
 
@@ -354,6 +352,7 @@ func noLineNumbersDefault() bool {
 
 		// Neither logs nor diffs are helped by line numbers, turn them off by
 		// default.
+		log.Debug("GIT_EXEC_PATH is set, skipping line numbers when paging git output")
 		return true
 	}
 

internal/detectManPage.go

@@ -0,0 +1,19 @@
+package internal
+
+import (
+	"github.com/walles/moor/v2/internal/linemetadata"
+)
+
+func (p *Pager) haveLoadedManPage() bool {
+	reader := p.Reader()
+	if reader == nil {
+		return false
+	}
+
+	for _, line := range reader.GetLines(linemetadata.Index{}, 10).Lines {
+		if line.Line.HasManPageFormatting() {
+			return true
+		}
+	}
+	return false
+}

internal/pager.go

@@ -634,8 +634,17 @@ func (p *Pager) StartPaging(screen twin.Screen, chromaStyle *chroma.Style, chrom
 			}
 
 		case eventMaybeDone:
-			// Do nothing. We got this just so that we'll do the QuitIfOneScreen
-			// check (above) as soon as highlighting is done.
+			// Man pages come pre-formatted for the screen width, and line
+			// numbers will mess that up. So we disable line numbers if we
+			// detect a man page by its contents.
+			//
+			// See also noLineNumbersDefault() where we use environment
+			// variables to try to detect man paging.
+			if p.haveLoadedManPage() && len(p.readers) == 1 {
+				p.ShowLineNumbers = false
+				p.showLineNumbers = false
+				log.Info("man page detected by contents, disabling line numbers")
+			}
 
 		case eventSpinnerUpdate:
 			spinner = event.spinner

internal/reader/line.go

@@ -74,3 +74,7 @@ func (line *Line) Plain(lineIndex *linemetadata.Index) string {
 	}
 	return *line.plain
 }
+
+func (line *Line) HasManPageFormatting() bool {
+	return textstyles.HasManPageFormatting(line.raw)
+}

internal/textstyles/ansiTokenizer.go

@@ -368,7 +368,8 @@ func tokensFromStyledString(styledString _StyledString) []twin.StyledRune {
 		return tokens
 	}
 
-	// Special handling for man page formatted lines
+	// Special handling for man page formatted lines. If this is updated you
+	// must update HasManPageFormatting() as well.
 	for index := 0; index < len(runes); index++ {
 		nextIndex, token := consumeBullet(runes, index)
 		if nextIndex != index {
@@ -407,6 +408,34 @@ func tokensFromStyledString(styledString _StyledString) []twin.StyledRune {
 	return tokens
 }
 
+// Like tokensFromStyledString(), but only checks without building any formatting
+func HasManPageFormatting(s string) bool {
+	runes := []rune(s)
+	for index := range runes {
+		nextIndex, _ := consumeBullet(runes, index)
+		if nextIndex != index {
+			return true
+		}
+
+		nextIndex, _ = consumeBoldUnderline(runes, index)
+		if nextIndex != index {
+			return true
+		}
+
+		nextIndex, _ = consumeBold(runes, index)
+		if nextIndex != index {
+			return true
+		}
+
+		nextIndex, _ = consumeUnderline(runes, index)
+		if nextIndex != index {
+			return true
+		}
+	}
+
+	return false
+}
+
 type _StyledString struct {
 	String string
 	Style  twin.Style