lmocr 0.1.2

pub fn is_page_number_line(line: &str) -> bool {
    if line.chars().all(|c| c.is_ascii_digit()) {
        return true;
    }
    let lower = line.to_ascii_lowercase();
    if lower.len() > 8 {
        return false;
    }
    lower
        .chars()
        .all(|c| matches!(c, 'i' | 'v' | 'x' | 'l' | 'c' | 'd' | 'm'))
}

pub fn contains_page_number_token(line: &str) -> bool {
    let tokens: Vec<&str> = line.split_whitespace().collect();
    if tokens.len() > 5 {
        return false;
    }
    tokens.iter().any(|token| {
        let trimmed = token
            .trim_matches(|c: char| !c.is_ascii_alphanumeric())
            .to_string();
        if trimmed.is_empty() {
            return false;
        }
        is_page_number_line(trimmed.as_str())
    })
}

pub fn looks_like_running_header(line: &str) -> bool {
    let trimmed = line.trim();
    if trimmed.len() < 3 || trimmed.len() > 60 {
        return false;
    }
    let mut letters = 0usize;
    let mut uppercase = 0usize;
    for ch in trimmed.chars() {
        if ch.is_ascii_alphabetic() {
            letters += 1;
            if ch.is_ascii_uppercase() {
                uppercase += 1;
            }
        }
    }
    letters > 0 && uppercase == letters
}

pub fn looks_like_numbered_running_header(line: &str) -> bool {
    let tokens: Vec<&str> = line.split_whitespace().collect();
    if tokens.len() < 3 || tokens.len() > 10 {
        return false;
    }

    let page_token = tokens[tokens.len() - 1].trim_matches(|c: char| !c.is_ascii_alphanumeric());
    if !is_page_number_line(page_token) {
        return false;
    }

    let mut letters = 0usize;
    let mut uppercase = 0usize;
    for token in &tokens[..tokens.len() - 1] {
        for ch in token.chars() {
            if ch.is_ascii_alphabetic() {
                letters += 1;
                if ch.is_ascii_uppercase() {
                    uppercase += 1;
                }
            }
        }
    }
    letters > 0 && uppercase * 100 / letters >= 80
}

pub fn looks_like_boundary_header(line: &str) -> bool {
    let trimmed = line.trim();
    if trimmed.len() < 3 || trimmed.len() > 60 || starts_with_block(trimmed) {
        return false;
    }
    let tokens: Vec<&str> = trimmed.split_whitespace().collect();
    if tokens.is_empty() || tokens.len() > 6 {
        return false;
    }
    if tokens
        .iter()
        .any(|token| token.chars().any(|ch| ch.is_ascii_digit()))
    {
        return false;
    }
    if tokens.iter().any(|token| {
        token
            .chars()
            .any(|ch| matches!(ch, '.' | ',' | ';' | ':' | '!' | '?'))
    }) {
        return false;
    }
    tokens.iter().all(|token| is_title_or_upper_word(token))
}

pub fn is_title_or_upper_word(token: &str) -> bool {
    let word = token.trim_matches(|c: char| !c.is_ascii_alphabetic());
    if word.is_empty() {
        return false;
    }
    if word.chars().all(|ch| ch.is_ascii_uppercase()) {
        return true;
    }
    let mut chars = word.chars();
    let Some(first) = chars.next() else {
        return false;
    };
    first.is_ascii_uppercase() && chars.all(|ch| ch.is_ascii_lowercase())
}

pub fn starts_with_block(s: &str) -> bool {
    let trimmed = s.trim_start();
    trimmed.starts_with('#')
        || trimmed.starts_with("```")
        || trimmed.starts_with("<pre>")
        || trimmed.starts_with('>')
        || trimmed.starts_with("- ")
        || trimmed.starts_with("* ")
        || trimmed.starts_with("+ ")
        || is_numbered_list(trimmed)
}

pub fn ends_with_block(s: &str) -> bool {
    let trimmed = s.trim_end();
    trimmed.ends_with("</pre>")
        || trimmed.ends_with("</table>")
        || trimmed.ends_with("</blockquote>")
        || trimmed.ends_with("***")
        || trimmed.ends_with("---")
}

pub fn is_numbered_list(s: &str) -> bool {
    let mut chars = s.chars();
    let mut saw_digit = false;
    for ch in &mut chars {
        if ch.is_ascii_digit() {
            saw_digit = true;
            continue;
        }
        if saw_digit && (ch == '.' || ch == ')') {
            return true;
        }
        break;
    }
    false
}

pub fn ends_with_hyphen(s: &str) -> bool {
    s.ends_with('-') || s.ends_with('‐') || s.ends_with('‑')
}

pub fn last_line_is_list_item(s: &str) -> bool {
    let last_line = s.rsplit('\n').next().unwrap_or(s).trim_start();
    last_line.starts_with("- ") || last_line.starts_with("* ") || last_line.starts_with("+ ")
}

pub fn is_terminal_punct(ch: char) -> bool {
    matches!(ch, '.' | '!' | '?' | '…')
}

pub fn last_significant_char(s: &str) -> Option<char> {
    for ch in s.trim_end().chars().rev() {
        if ch.is_whitespace() {
            continue;
        }
        if matches!(ch, '"' | '\'' | '\u{201D}' | '\u{2019}' | ')' | ']' | '}') {
            continue;
        }
        return Some(ch);
    }
    None
}

#[cfg(test)]
mod tests {
    use super::*;

    // ── is_page_number_line ──

    #[test]
    fn page_number_pure_digits() {
        assert!(is_page_number_line("42"));
        assert!(is_page_number_line("1"));
        assert!(is_page_number_line("999"));
    }

    #[test]
    fn page_number_roman_numerals() {
        assert!(is_page_number_line("iv"));
        assert!(is_page_number_line("xii"));
        assert!(is_page_number_line("VII"));
        assert!(is_page_number_line("MCMLXIV")); // 7 chars, under limit
    }

    #[test]
    fn page_number_roman_too_long() {
        // 9 chars of roman-like letters should be rejected
        assert!(!is_page_number_line("mmmmmmmmm"));
    }

    #[test]
    fn page_number_false_positive_short_roman_words() {
        // "civil" (5 chars) is all roman-numeral chars and under the 8-char limit,
        // so it's a known false positive. Acceptable because these short words
        // rarely appear as standalone lines in real PDFs.
        assert!(is_page_number_line("civil"));
        assert!(is_page_number_line("dim"));
        // But longer words are rejected by the length guard
        assert!(!is_page_number_line("civilization"));
    }

    #[test]
    fn page_number_empty_string() {
        // empty string: all(digit) on empty iterator is true
        assert!(is_page_number_line(""));
    }

    #[test]
    fn page_number_mixed_content() {
        assert!(!is_page_number_line("page 5"));
        assert!(!is_page_number_line("42a"));
    }

    // ── contains_page_number_token ──

    #[test]
    fn page_token_with_punctuation() {
        assert!(contains_page_number_token("(7)"));
        assert!(contains_page_number_token("—42—"));
        assert!(contains_page_number_token("CHAPTER iv"));
    }

    #[test]
    fn page_token_too_many_tokens() {
        assert!(!contains_page_number_token("one two three four five 42"));
    }

    #[test]
    fn page_token_at_five_tokens() {
        assert!(contains_page_number_token("A B C D 42"));
    }

    #[test]
    fn page_token_all_punctuation_stripped_empty() {
        // token "---" trims to empty → not a page number
        assert!(!contains_page_number_token("---"));
    }

    // ── looks_like_running_header ──

    #[test]
    fn running_header_all_caps() {
        assert!(looks_like_running_header("CHAPTER ONE"));
        assert!(looks_like_running_header("THE GREAT WAR"));
    }

    #[test]
    fn running_header_mixed_case_rejected() {
        assert!(!looks_like_running_header("Chapter One"));
        assert!(!looks_like_running_header("the great war"));
    }

    #[test]
    fn running_header_too_short() {
        assert!(!looks_like_running_header("AB")); // len 2 < 3
    }

    #[test]
    fn running_header_at_boundary_length() {
        assert!(looks_like_running_header("ABC")); // len 3, exactly at min
        let long = "A".repeat(60);
        assert!(looks_like_running_header(&long)); // len 60, exactly at max
        let too_long = "A".repeat(61);
        assert!(!looks_like_running_header(&too_long));
    }

    #[test]
    fn running_header_with_spaces_and_numbers() {
        // numbers and spaces are not ascii_alphabetic, so they're ignored
        // as long as all letters are uppercase it matches
        assert!(looks_like_running_header("CHAPTER 3"));
    }

    // ── looks_like_numbered_running_header ──

    #[test]
    fn numbered_header_typical() {
        assert!(looks_like_numbered_running_header("FREDERICK THE GREAT 9"));
    }

    #[test]
    fn numbered_header_too_few_tokens() {
        assert!(!looks_like_numbered_running_header("TITLE 9")); // 2 tokens < 3
    }

    #[test]
    fn numbered_header_no_page_number_at_end() {
        assert!(!looks_like_numbered_running_header("CHAPTER THE THIRD end"));
    }

    #[test]
    fn numbered_header_below_80_percent_uppercase() {
        // "ABc DEF 9": before last token: A,B,c,D,E,F = 6 letters, 5 upper = 83% ≥ 80%
        assert!(looks_like_numbered_running_header("ABc DEF 9"));
        // To actually get below 80%: need >20% lowercase
        // "Abcd EF 9": A,b,c,d,E,F = 6 letters, 3 upper = 50% < 80%
        assert!(!looks_like_numbered_running_header("Abcd EF 9"));
    }

    #[test]
    fn numbered_header_at_80_percent_uppercase() {
        // "ABCd EFG 9": before last token: A,B,C,d,E,F,G = 7 letters, 6 upper = 85% ≥ 80%
        assert!(looks_like_numbered_running_header("ABCd EFG 9"));
    }

    #[test]
    fn numbered_header_eleven_tokens_rejected() {
        let line = "A B C D E F G H I J 9";
        let tokens: Vec<&str> = line.split_whitespace().collect();
        assert_eq!(tokens.len(), 11);
        assert!(!looks_like_numbered_running_header(line));
    }

    // ── looks_like_boundary_header ──

    #[test]
    fn boundary_header_title_case() {
        assert!(looks_like_boundary_header("Introduction"));
        assert!(looks_like_boundary_header("Chapter One"));
        assert!(looks_like_boundary_header("CONCLUSION"));
    }

    #[test]
    fn boundary_header_rejects_digits() {
        assert!(!looks_like_boundary_header("Chapter 3"));
    }

    #[test]
    fn boundary_header_rejects_punctuation() {
        assert!(!looks_like_boundary_header("Hello, World"));
        assert!(!looks_like_boundary_header("Section:"));
    }

    #[test]
    fn boundary_header_rejects_block_start() {
        assert!(!looks_like_boundary_header("# Introduction"));
        assert!(!looks_like_boundary_header("> Quote"));
        assert!(!looks_like_boundary_header("- List item"));
    }

    #[test]
    fn boundary_header_rejects_too_many_tokens() {
        assert!(!looks_like_boundary_header(
            "One Two Three Four Five Six Seven"
        )); // 7 tokens
    }

    #[test]
    fn boundary_header_rejects_lowercase_word() {
        assert!(!looks_like_boundary_header("hello world"));
    }

    #[test]
    fn boundary_header_too_short_or_long() {
        assert!(!looks_like_boundary_header("AB")); // 2 chars
        let long = format!("{} End", "Abcdefghij".repeat(6)); // >60 chars
        assert!(!looks_like_boundary_header(&long));
    }

    // ── is_title_or_upper_word ──

    #[test]
    fn title_word_cases() {
        assert!(is_title_or_upper_word("Hello"));
        assert!(is_title_or_upper_word("HELLO"));
        assert!(!is_title_or_upper_word("hello"));
        assert!(!is_title_or_upper_word("hELLO"));
        assert!(!is_title_or_upper_word("HeLLO"));
    }

    #[test]
    fn title_word_with_surrounding_punctuation() {
        assert!(is_title_or_upper_word("(Hello)"));
        assert!(is_title_or_upper_word("\"WORLD\""));
    }

    #[test]
    fn title_word_empty_or_no_alpha() {
        assert!(!is_title_or_upper_word(""));
        assert!(!is_title_or_upper_word("123"));
        assert!(!is_title_or_upper_word("---"));
    }

    // ── starts_with_block / ends_with_block ──

    #[test]
    fn starts_with_block_variants() {
        assert!(starts_with_block("# Heading"));
        assert!(starts_with_block("```rust"));
        assert!(starts_with_block("<pre>code</pre>"));
        assert!(starts_with_block("> quote"));
        assert!(starts_with_block("- item"));
        assert!(starts_with_block("* item"));
        assert!(starts_with_block("+ item"));
        assert!(starts_with_block("1. numbered"));
        assert!(starts_with_block("12) numbered"));
        assert!(!starts_with_block("regular text"));
    }

    #[test]
    fn starts_with_block_leading_whitespace() {
        assert!(starts_with_block("  # Heading"));
        assert!(starts_with_block("\t- item"));
    }

    #[test]
    fn ends_with_block_variants() {
        assert!(ends_with_block("text</pre>"));
        assert!(ends_with_block("text</table>"));
        assert!(ends_with_block("text</blockquote>"));
        assert!(ends_with_block("***"));
        assert!(ends_with_block("---"));
        assert!(!ends_with_block("regular text"));
    }

    #[test]
    fn ends_with_block_trailing_whitespace() {
        assert!(ends_with_block("text</pre>  "));
        assert!(ends_with_block("---\n"));
    }

    // ── is_numbered_list ──

    #[test]
    fn numbered_list_cases() {
        assert!(is_numbered_list("1. item"));
        assert!(is_numbered_list("12. item"));
        assert!(is_numbered_list("1) item"));
        assert!(!is_numbered_list("a. item"));
        assert!(!is_numbered_list("text"));
        assert!(!is_numbered_list("5 text")); // digit but no dot/paren
    }

    // ── ends_with_hyphen ──

    #[test]
    fn hyphen_ascii_and_unicode() {
        assert!(ends_with_hyphen("word-"));
        assert!(ends_with_hyphen("word\u{2010}")); // ‐
        assert!(ends_with_hyphen("word\u{2011}")); // ‑
        assert!(!ends_with_hyphen("word"));
        assert!(!ends_with_hyphen("word—")); // em dash is NOT a hyphen
    }

    // ── last_line_is_list_item ──

    #[test]
    fn last_line_list_item_multiline() {
        assert!(last_line_is_list_item("paragraph\n- item"));
        assert!(last_line_is_list_item("paragraph\n* item"));
        assert!(last_line_is_list_item("paragraph\n+ item"));
        assert!(!last_line_is_list_item("paragraph\nnot a list"));
    }

    #[test]
    fn last_line_list_item_single_line() {
        assert!(last_line_is_list_item("- item"));
        assert!(!last_line_is_list_item("no dash"));
    }

    // ── is_terminal_punct ──

    #[test]
    fn terminal_punct() {
        assert!(is_terminal_punct('.'));
        assert!(is_terminal_punct('!'));
        assert!(is_terminal_punct('?'));
        assert!(is_terminal_punct('…'));
        assert!(!is_terminal_punct(','));
        assert!(!is_terminal_punct(';'));
        assert!(!is_terminal_punct('-'));
    }

    // ── last_significant_char ──

    #[test]
    fn last_significant_skips_quotes_and_brackets() {
        assert_eq!(last_significant_char("word.\""), Some('.'));
        assert_eq!(last_significant_char("word!)"), Some('!'));
        assert_eq!(last_significant_char("end']"), Some('d'));
    }

    #[test]
    fn last_significant_skips_smart_quotes() {
        assert_eq!(last_significant_char("word.\u{201D}"), Some('.'));
        assert_eq!(last_significant_char("word\u{2019}"), Some('d'));
    }

    #[test]
    fn last_significant_all_skippable() {
        assert_eq!(last_significant_char("\"')}]"), None);
        assert_eq!(last_significant_char("   "), None);
        assert_eq!(last_significant_char(""), None);
    }

    #[test]
    fn last_significant_trailing_whitespace() {
        assert_eq!(last_significant_char("word   "), Some('d'));
        assert_eq!(last_significant_char("end.\t\n"), Some('.'));
    }
}