colophon: the front matter of a book nobody typeset carefully

Read against five thousand real books, the title page is often not the
first thing in the file, and what is there instead has shape. Six kinds
of it were being read as titles:

  vi   CONTENTS                                  a folioed heading
  Copyright (c) 2013 by Barry Davies, BEM        a notice
  Translated from the French by Willard R. Trask a credit
  Vol. 11. HENRY JAMES, JR.                      a collected edition's list
  [PART ONE:                                     a part heading
  Piper Verlag GmbH, Munich<span style=...       a converter's leavings

Rejecting one line only moves the misread down one, because a page that
converted badly converted badly throughout, so the whole shape goes:
markdown links, stylesheet declarations, HTML entities, epub filenames,
and the sentences a copyright page sets.

The notice pattern wants the symbol or the year, never the word alone.
The first draft rejected 'Copyright and Its Discontents', which its own
test caught.

On sixty books this drops five claimed titles, all of them wrong, and
fixes six that were reading one homily's heading for every other
homily's title.

crates/colophon/src/title.rs

@@ -34,6 +34,11 @@ static LIFESPAN: LazyLock<Regex> = LazyLock::new(|| {
 /// `350 BC`, `1971` — a date alone on a line, which is not a title.
 static BARE_DATE: LazyLock<Regex> =
     LazyLock::new(|| Regex::new(r"(?i)^\d{1,4}\s*(bc|ad|bce|ce)?\.?$").expect("a valid pattern"));
+/// `Copyright © 2013 by`, `\u{a9} 2013 Richard Swinburne`, `Copyright 1971 by`.
+static COPYRIGHT_NOTICE: LazyLock<Regex> = LazyLock::new(|| {
+    Regex::new(r"(?i)(\u{a9}|\(c\)|copyright)\s*(\u{a9}|\(c\))?\s*(1[0-9]|20)\d{2}")
+        .expect("a valid pattern")
+});
 /// `Edited by`, with whoever it was set on the line beneath.
 static ROLE_ONLY: LazyLock<Regex> = LazyLock::new(|| {
     Regex::new(r"(?i)^(edited|translated|written|selected|compiled)\s+by:?$")
@@ -107,7 +112,111 @@ fn is_structural(line: &str) -> bool {
         .collect::<String>()
         .trim()
         .to_lowercase();
-    STRUCTURAL.binary_search(&norm.as_str()).is_ok()
+    let norm = strip_folio(&norm);
+    STRUCTURAL.binary_search(&norm).is_ok()
+}
+
+/// A heading with its page number taken off either end.
+///
+/// Front matter is folioed in roman, and the folio sits on the same line as
+/// the heading often enough to matter: `vi   CONTENTS` and `viii CONTENTS`
+/// are contents pages, and were being read as titles because they are not the
+/// bare word.
+fn strip_folio(norm: &str) -> &str {
+    let is_folio = |w: &str| {
+        !w.is_empty()
+            && (w.chars().all(|c| c.is_ascii_digit())
+                || w.chars()
+                    .all(|c| matches!(c, 'i' | 'v' | 'x' | 'l' | 'c' | 'd' | 'm')))
+    };
+    let mut out = norm.trim();
+    loop {
+        let trimmed = match out.split_once(' ') {
+            Some((head, rest)) if is_folio(head) => rest.trim_start(),
+            _ => match out.rsplit_once(' ') {
+                Some((rest, tail)) if is_folio(tail) => rest.trim_end(),
+                _ => return out,
+            },
+        };
+        if trimmed == out {
+            return out;
+        }
+        out = trimmed;
+    }
+}
+
+/// Lines that are printed matter rather than anything a book is called.
+///
+/// Every one of these was read as a title off a real book in a real library:
+/// a copyright notice, a translator's credit, a part heading in brackets, a
+/// span of HTML a converter left behind, and a volume entry from a collected
+/// edition's own list of volumes.
+fn is_furniture(line: &str) -> bool {
+    let t = line.trim();
+    if t.starts_with('[') || t.starts_with('\\') {
+        return true;
+    }
+    // A converter that left its markup in. Nothing printed on a title page is
+    // a tag, a stylesheet declaration, a markdown link or a filename, and a
+    // title that appears to be one is a fragment of the conversion rather than
+    // of the book. Rejecting only the first such line is no use: the next line
+    // of a mangled page is mangled too, so the whole shape has to go.
+    if t.contains("<span")
+        || t.contains("<div")
+        || t.contains("<p ")
+        || t.contains("</")
+        || t.contains("](")          // a markdown link the converter split
+        || t.contains("&quot;")
+        || t.contains("&amp;")
+        || t.contains("mso-")        // Word's stylesheet, via HTML export
+        || t.contains("font-family")
+        || t.contains("font-size")
+        || t.contains("style=")
+        || t.contains("_epub_")
+        || t.contains(".html")
+        || t.contains(".xhtml")
+    {
+        return true;
+    }
+    // Sentences a copyright page sets, which are not notices and so not caught
+    // by the notice pattern.
+    let lower = t.to_lowercase();
+    for phrase in [
+        "moral rights of the author",
+        "no part of this publication",
+        "british library cataloguing",
+        "library of congress cataloging",
+        "printed in the united",
+        "typeset by",
+        "a cip catalogue record",
+    ] {
+        if lower.contains(phrase) {
+            return true;
+        }
+    }
+    let low = t.to_lowercase();
+    // The notice, not the word. "Copyright and Its Discontents" is a book;
+    // "Copyright © 2013 by" is the page it would be printed on. What tells
+    // them apart is the symbol or the year, never the word alone.
+    if COPYRIGHT_NOTICE.is_match(t)
+        || low.contains("all rights reserved")
+        || low.starts_with("first published")
+        || low.starts_with("reprinted")
+    {
+        return true;
+    }
+    // "Translated from the French by Willard R. Trask" — a credit, not a name
+    // and not a title. ROLE_ONLY catches the bare "Translated by"; this is the
+    // whole sentence, which a title page sets on one line.
+    if low.starts_with("translated from") || low.starts_with("translated by") {
+        return true;
+    }
+    // "Vol. 11. HENRY JAMES, JR." — an entry in a collected edition's list of
+    // its own volumes, printed in the front matter of every volume.
+    if low.starts_with("vol.") || low.starts_with("volume ") && low.contains('.') {
+        return true;
+    }
+    false
 }
 
 fn people(who: &str) -> Vec<String> {
@@ -214,7 +323,7 @@ pub fn read(lines: &[String]) -> Option<TitlePage> {
                 continue;
             }
         }
-        if is_structural(line) || BARE_DATE.is_match(line) {
+        if is_structural(line) || is_furniture(line) || BARE_DATE.is_match(line) {
             if block.is_empty() {
                 continue;
             }
@@ -289,3 +398,97 @@ pub fn read(lines: &[String]) -> Option<TitlePage> {
     out.title.as_ref()?;
     Some(out)
 }
+
+#[cfg(test)]
+mod furniture_tests {
+    use super::*;
+
+    /// Every string here was read as a title off a book in a real library.
+    #[test]
+    fn printed_matter_is_not_a_title() {
+        for line in [
+            "vi   CONTENTS",
+            "viii   CONTENTS",
+            "xiv  ACKNOWLEDGEMENTS",
+            "12 INDEX",
+            "Translated from the French by Willard R. Trask",
+            "Copyright © 2013 by Barry Davies, BEM",
+            "[PART ONE:",
+            "Piper Verlag GmbH, Munich<span style=\"font-size\">",
+            "Vol. 11. HENRY JAMES, JR.",
+            "All rights reserved",
+            "First published 1971",
+        ] {
+            assert!(
+                is_structural(line) || is_furniture(line),
+                "read as a title: {line:?}"
+            );
+        }
+    }
+
+    #[test]
+    fn a_real_title_survives_all_of_that() {
+        for line in [
+            "The First World War",
+            "A Very Short Introduction",
+            "Christianity and Classical Culture",
+            "The Problems of Philosophy",
+            "Volume and Void",
+            "Contents of Consciousness",
+            "August 1914",
+            "Copyright and Its Discontents",
+        ] {
+            assert!(
+                !(is_structural(line) || is_furniture(line)),
+                "rejected a real title: {line:?}"
+            );
+        }
+    }
+
+    #[test]
+    fn a_folio_comes_off_either_end() {
+        assert_eq!(strip_folio("vi contents"), "contents");
+        assert_eq!(strip_folio("contents vi"), "contents");
+        assert_eq!(strip_folio("contents"), "contents");
+        // A title that happens to be roman numerals throughout is not a folio
+        // with nothing left; it keeps whatever it had.
+        assert_eq!(strip_folio("i"), "i");
+    }
+}
+
+#[cfg(test)]
+mod conversion_tests {
+    use super::*;
+
+    /// A mangled page is mangled on every line, so rejecting the first
+    /// leftover only moves the misread down one. Each of these is the line
+    /// colophon fell through to after the line above it was rejected.
+    #[test]
+    fn what_a_converter_left_behind_is_never_a_title() {
+        for line in [
+            "[PART ONE:",
+            "RULERS AND REDNECKS](Webb_9780767922951_epub_p01_r1.html)",
+            "[AFRICAN SLAVERY IN AMERICA - \\[1775\\]](index_split_004.html)",
+            "Piper Verlag GmbH, Munich<span style=\"font-family:&quot;Times\"",
+            "mso-bidi-font-family:&quot;Times New Roman&quot;;mso-ansi-language",
+            "The moral rights of the author have been asserted",
+            "No part of this publication may be reproduced",
+            "British Library Cataloguing in Publication Data",
+            "Typeset by SPI Publisher Services",
+        ] {
+            assert!(is_furniture(line), "read as a title: {line:?}");
+        }
+    }
+
+    #[test]
+    fn a_title_that_merely_sounds_like_the_copyright_page_survives() {
+        for line in [
+            "Copyright and Its Discontents",
+            "The Moral Rights of Man",
+            "Printed Matter",
+            "A Publication of the Institute",
+        ] {
+            assert!(!is_furniture(line), "rejected a real title: {line:?}");
+        }
+    }
+}