notescraft/crates/index/tests/operators.rs

//! The search operators end to end over the public API: `section:`,
//! `block:`, `task:`, `task-todo:`, `task-done:` and the case switch, alone
//! and composed with `path:`, `file:`, `tag:`, `line:`, `-term` and `OR`.
//!
//! Every row is `(query, case_sensitive, notes it must hit)` over one small
//! vault, so a regression names the row that broke.

use notescraft_index::search::{MAX_CANDIDATES, SearchIndex, SearchOptions, SearchQuery};

/// One note per shape: sections, blocks, tasks, fences, CRLF and unicode.
fn vault() -> SearchIndex {
    let mut notes = SearchIndex::new();
    notes.update("n/sections.md", "# Alpha\n\nalpha text here\n\n## Beta\n\nbeta text with gamma\n\n# Delta\n\ndelta text\n").unwrap();
    notes.update("n/blocks.md", "alpha in one block\n\nbeta in another\n\nalpha beta together in a block\n").unwrap();
    notes.update("n/tasks.md", "- [ ] alpha todo #work\n\n- [x] beta done\n> - [ ] gamma quoted\n  - [ ] nested alpha\n\n1. [x] ordered delta\n").unwrap();
    notes.update("n/fenced.md", "```md\n# Fake heading\n\n- [ ] fake task\n```\n\n# Real heading\n\n- [x] real task\n").unwrap();
    notes.update("play/n/other.md", "alpha beta\n").unwrap();
    notes.update("u/unicode.md", "# 見出し\n\n漢字 ひらがな #日本語\n\n- [ ] 漢字 task\n").unwrap();
    // A CRLF note: the `\r` is no marker for headings, blocks or tasks.
    notes.update("work/n/crlf.md", "# Alpha\r\nalpha one\r\n\r\nbeta one\r\n\r\n- [ ] alpha task\r\n").unwrap();
    notes
}

fn search(query: &str, sensitive: bool) -> Vec<String> {
    let options = if sensitive { SearchOptions::matching_case() } else { SearchOptions::new() };
    let parsed = SearchQuery::parse_with(query, &options).unwrap_or_else(|error| panic!("{query:?} must parse: {error}"));
    let mut paths: Vec<String> = vault().search(&parsed, MAX_CANDIDATES).into_iter().map(|hit| hit.path).collect();
    paths.sort();
    paths
}

/// `(query, case_sensitive, notes it must hit)`.
const ROWS: &[(&str, bool, &[&str])] = &[
    // section:
    ("section:(alpha text)", false, &["n/sections.md"]),
    ("section:(alpha gamma)", false, &["n/sections.md", "n/tasks.md"]),
    ("section:(alpha delta)", false, &["n/tasks.md"]),
    ("section:(gamma delta)", false, &["n/tasks.md"]),
    ("section:(delta text)", false, &["n/sections.md"]),
    ("section:(alpha one)", false, &["n/blocks.md", "work/n/crlf.md"]),
    ("section:(alpha beta)", false, &["n/blocks.md", "n/sections.md", "n/tasks.md", "play/n/other.md", "work/n/crlf.md"]),
    ("section:alpha", false, &["n/blocks.md", "n/sections.md", "n/tasks.md", "play/n/other.md", "work/n/crlf.md"]),
    ("section:nosuch", false, &[]),
    ("section:(見出し 漢字)", false, &["u/unicode.md"]),
    ("section:(漢字 ひらがな)", false, &["u/unicode.md"]),
    ("section:(fake heading)", false, &["n/fenced.md"]),
    ("section:(fake real)", false, &[]),
    ("section:(real task)", false, &["n/fenced.md"]),
    ("section:(todo nested)", false, &["n/tasks.md"]),
    ("section:alpha path:n/sections.md", false, &["n/sections.md"]),
    ("section:alpha -path:n/", false, &["play/n/other.md", "work/n/crlf.md"]),
    ("section:alpha tag:#work", false, &["n/tasks.md"]),
    ("section:alpha -alpha", false, &[]),
    ("section:alpha file:Other", false, &["play/n/other.md"]),
    ("section:(alpha gamma) OR section:(real task)", false, &["n/fenced.md", "n/sections.md", "n/tasks.md"]),
    ("(section:alpha OR section:fake)", false, &["n/blocks.md", "n/fenced.md", "n/sections.md", "n/tasks.md", "play/n/other.md", "work/n/crlf.md"]),
    ("section:alpha line:(nested alpha)", false, &["n/tasks.md"]),
    // block:
    ("block:(alpha beta)", false, &["n/blocks.md", "n/tasks.md", "play/n/other.md"]),
    ("block:(alpha in one block)", false, &["n/blocks.md"]),
    ("block:together", false, &["n/blocks.md"]),
    ("block:(together in)", false, &["n/blocks.md"]),
    ("block:(alpha one)", false, &["n/blocks.md", "work/n/crlf.md"]),
    ("block:(alpha one beta)", false, &[]),
    ("block:fake", false, &["n/fenced.md"]),
    ("block:real", false, &["n/fenced.md"]),
    ("block:(fake real)", false, &[]),
    ("block:fake real", false, &["n/fenced.md"]),
    ("block:(fake task)", false, &["n/fenced.md"]),
    ("block:漢字", false, &["u/unicode.md"]),
    ("block:alpha path:work/", false, &["work/n/crlf.md"]),
    ("block:alpha -block", false, &["n/sections.md", "n/tasks.md", "play/n/other.md", "work/n/crlf.md"]),
    ("block:alpha -one", false, &["n/sections.md", "n/tasks.md", "play/n/other.md"]),
    ("block:(alpha beta) tag:#work", false, &["n/tasks.md"]),
    ("block:(delta text)", false, &["n/sections.md"]),
    ("block:gamma", false, &["n/sections.md", "n/tasks.md"]),
    ("block:(quoted nested)", false, &["n/tasks.md"]),
    ("block:(beta done quoted)", false, &["n/tasks.md"]),
    // task:
    ("task:alpha", false, &["n/tasks.md", "work/n/crlf.md"]),
    ("task:(alpha nested)", false, &["n/tasks.md"]),
    ("task:(alpha beta)", false, &[]),
    ("task:gamma", false, &["n/tasks.md"]),
    ("task:quoted", false, &["n/tasks.md"]),
    ("task:ordered", false, &["n/tasks.md"]),
    ("task:fake", false, &[]),
    ("task:real", false, &["n/fenced.md"]),
    ("task:漢字", false, &["u/unicode.md"]),
    ("task:todo", false, &["n/tasks.md"]),
    ("task:alpha tag:#work", false, &["n/tasks.md"]),
    ("task:alpha -path:play/", false, &["n/tasks.md", "work/n/crlf.md"]),
    ("task:alpha path:work/", false, &["work/n/crlf.md"]),
    ("task:alpha file:Tasks", false, &["n/tasks.md"]),
    ("task:alpha OR task:gamma", false, &["n/tasks.md", "work/n/crlf.md"]),
    ("task:alpha -task-done:", false, &["work/n/crlf.md"]),
    ("task-todo:", false, &["n/tasks.md", "u/unicode.md", "work/n/crlf.md"]),
    ("task-done:", false, &["n/fenced.md", "n/tasks.md"]),
    ("task-todo: -path:work/", false, &["n/tasks.md", "u/unicode.md"]),
    ("task-done:delta", false, &["n/tasks.md"]),
    ("task-todo:delta", false, &[]),
    ("task-done:(ordered delta)", false, &["n/tasks.md"]),
    // The structural operators stay case-insensitive, the terms do not.
    ("match-case: task-done: real", false, &["n/fenced.md"]),
    ("match-case: task-done: REAL", false, &[]),
    // match-case: and ignore-case:
    ("match-case: alpha", false, &["n/blocks.md", "n/sections.md", "n/tasks.md", "play/n/other.md", "work/n/crlf.md"]),
    ("match-case: Alpha", false, &["n/sections.md", "work/n/crlf.md"]),
    ("match-case: ALPHA", false, &[]),
    ("ignore-case: ALPHA", false, &["n/blocks.md", "n/sections.md", "n/tasks.md", "play/n/other.md", "work/n/crlf.md"]),
    ("match-case: ALPHA ignore-case:", false, &["n/blocks.md", "n/sections.md", "n/tasks.md", "play/n/other.md", "work/n/crlf.md"]),
    ("ignore-case: ALPHA match-case:", false, &[]),
    ("match-case: \"alpha text\"", false, &["n/sections.md"]),
    ("match-case: \"Alpha text\"", false, &[]),
    ("match-case: \"Alpha text\" OR ignore-case: \"alpha text\"", false, &["n/sections.md"]),
    ("match-case: alpha -path:n/", false, &["play/n/other.md", "work/n/crlf.md"]),
    ("match-case: alpha path:n/sections.md", false, &["n/sections.md"]),
    ("match-case: alpha tag:#work", false, &["n/tasks.md"]),
    ("match-case: alpha -ALPHA", false, &["n/blocks.md", "n/sections.md", "n/tasks.md", "play/n/other.md", "work/n/crlf.md"]), // the negation is cased too: `-ALPHA` subtracts nothing
    ("match-case: crlf", false, &["work/n/crlf.md"]),
    ("match-case: Tasks", false, &[]),
    ("match-case: 漢字", false, &["u/unicode.md"]),
    ("match-case: 漢字の", false, &[]),
    ("match-case: section:(ALPHA)", false, &["n/blocks.md", "n/sections.md", "n/tasks.md", "play/n/other.md", "work/n/crlf.md"]),
    // The options parameter says the same as the in-query flag.
    ("alpha", true, &["n/blocks.md", "n/sections.md", "n/tasks.md", "play/n/other.md", "work/n/crlf.md"]),
    ("Alpha", true, &["n/sections.md", "work/n/crlf.md"]),
    ("ALPHA", true, &[]),
    ("alpha", false, &["n/blocks.md", "n/sections.md", "n/tasks.md", "play/n/other.md", "work/n/crlf.md"]),
    ("\"alpha text\"", true, &["n/sections.md"]),
    ("section:(ALPHA text)", true, &["n/sections.md"]),
];

#[test]
fn every_operator_row_behaves() {
    assert!(ROWS.len() >= 50, "the table keeps {} rows", ROWS.len());
    let mut broken: Vec<String> = Vec::new();
    for (query, sensitive, want) in ROWS {
        let mut expected: Vec<String> = want.iter().map(|path| (*path).to_string()).collect();
        expected.sort();
        let got = search(query, *sensitive);
        if got != expected {
            broken.push(format!("{query:?} (case_sensitive={sensitive}) got {got:?}, want {expected:?}"));
        }
    }
    assert_eq!(broken, Vec::<String>::new());
}

#[test]
fn the_case_switch_needs_a_term_to_search_for() {
    for query in ["match-case:", "ignore-case:", "match-case: ignore-case:", "-match-case:"] {
        assert!(SearchQuery::parse(query).is_err(), "{query:?} searches for nothing");
    }
    assert!(SearchQuery::parse_with("match-case:", &SearchOptions::matching_case()).is_err());
}