ADR-0021 implementation: per-command usage templates in parse errors

New `dsl::usage` module: registry pairing each command's
entry-keyword with a `parse.usage.*` catalog key.
`matched_entry()` resolves the entry keyword from the
consumed token prefix; multi-entry families (add, drop,
show) return all matching keys.

Catalog: new `parse.usage.<command>` keys (one per command),
`parse.token.{keyword,punct,...}` vocabulary (one per
Keyword/Punct variant + token-class labels + LexError
kinds), and `parse.available_commands` for the no-prefix
fallback. Catalog grows ~60 entries.

Validator: extended KEYS_AND_PLACEHOLDERS; new completeness
test asserts every Keyword and Punct variant has its
`parse.token.*` entry.

`app::dispatch_dsl` rewritten to compose three blocks per
ADR-0021 §2: caret + structural/custom error + usage block
(or available-commands fallback per §5). Caret math fixed
to use original-input byte position rather than
trimmed-input position (the lexer no longer trims before
lexing). Three pre-existing app tests adjusted to look
across all error lines instead of `output.back()` (the
usage block is now the last line).

`dsl::usage::matched_entry` uses `<=` rather than `<` for
position comparison so custom errors raised by `try_map`
(whose span starts at the first consumed token) still
resolve to the entry keyword.

Tests: 668 passing, 0 failing, 1 ignored (650 baseline →
+18: 8 usage + 1 token-vocab completeness + 9 new
integration tests in tests/parse_error_pedagogy.rs
covering create/add/drop/show/frobulate/update/insert
cases). Clippy clean.
This commit is contained in:
claude@clouddev1
2026-05-10 14:41:32 +00:00
parent fdaf7e3e0e
commit 11071ae164
6 changed files with 776 additions and 19 deletions
+260
View File
@@ -0,0 +1,260 @@
//! Per-command usage template registry (ADR-0021 §1).
//!
//! Each registered entry pairs a `Keyword` (the command's entry
//! token) with a catalog key under `parse.usage.*`. The renderer
//! in `app.rs::dispatch_dsl` looks up matching entries when a
//! parse error has consumed at least one keyword token; entries
//! whose `entry` matches the consumed keyword are rendered as
//! the "usage:" block.
//!
//! For `add` and `drop` (multi-entry families), every matching
//! entry renders — the user gets the full family of options,
//! which is the most pedagogically useful behaviour at the
//! moment of confusion.
//!
//! Adding a new command means: (1) the parser combinator,
//! (2) one entry in `REGISTRY`, (3) one YAML key under
//! `parse.usage.*` in `src/friendly/strings/en-US.yaml`. The
//! catalog validator catches a missing YAML entry; a per-command
//! unit test (`every_command_has_a_registry_entry`) catches a
//! missing registry entry.
use crate::dsl::keyword::Keyword;
use crate::dsl::lexer::{Token, TokenKind};
#[derive(Debug, Clone, Copy)]
pub struct UsageEntry {
/// First keyword that distinguishes this command. Used as
/// the registry-lookup key.
pub entry: Keyword,
/// Catalog key under `parse.usage.*` (ADR-0021 §1). The
/// renderer translates this through the catalog at render
/// time.
pub catalog_key: &'static str,
}
/// One `UsageEntry` per command. Multi-entry families (`add`,
/// `drop`, `show`) appear multiple times.
pub const REGISTRY: &[UsageEntry] = &[
UsageEntry {
entry: Keyword::Create,
catalog_key: "parse.usage.create_table",
},
UsageEntry {
entry: Keyword::Drop,
catalog_key: "parse.usage.drop_table",
},
UsageEntry {
entry: Keyword::Drop,
catalog_key: "parse.usage.drop_column",
},
UsageEntry {
entry: Keyword::Drop,
catalog_key: "parse.usage.drop_relationship",
},
UsageEntry {
entry: Keyword::Add,
catalog_key: "parse.usage.add_column",
},
UsageEntry {
entry: Keyword::Add,
catalog_key: "parse.usage.add_relationship",
},
UsageEntry {
entry: Keyword::Rename,
catalog_key: "parse.usage.rename_column",
},
UsageEntry {
entry: Keyword::Change,
catalog_key: "parse.usage.change_column",
},
UsageEntry {
entry: Keyword::Show,
catalog_key: "parse.usage.show_data",
},
UsageEntry {
entry: Keyword::Show,
catalog_key: "parse.usage.show_table",
},
UsageEntry {
entry: Keyword::Insert,
catalog_key: "parse.usage.insert",
},
UsageEntry {
entry: Keyword::Update,
catalog_key: "parse.usage.update",
},
UsageEntry {
entry: Keyword::Delete,
catalog_key: "parse.usage.delete",
},
UsageEntry {
entry: Keyword::Replay,
catalog_key: "parse.usage.replay",
},
];
/// Find the entry-keyword whose grammar to illustrate.
///
/// `failure_position` is a byte offset in the source pointing
/// at where the parser stopped. Returns the keyword and the
/// catalog keys for every matching usage entry, or `None` if no
/// keyword was consumed before the failure — in which case the
/// caller falls back to the available-commands list per
/// ADR-0021 §5.
#[must_use]
pub fn matched_entry(
tokens: &[Token],
failure_position: usize,
) -> Option<(Keyword, Vec<&'static str>)> {
// Tokens covered by the failure span: their start byte is at
// or before `failure_position`. `<=` (rather than `<`) lets
// custom errors raised by `try_map` — whose span starts at
// the first consumed token — find that first token as the
// entry keyword. Structural errors (whose span points at the
// unexpected token) still find the entry keyword consumed
// before that point.
let entry = tokens
.iter()
.take_while(|t| t.span.0 <= failure_position)
.find_map(|t| match &t.kind {
TokenKind::Keyword(kw) => Some(*kw),
_ => None,
})?;
let matches: Vec<&'static str> = REGISTRY
.iter()
.filter(|e| e.entry == entry)
.map(|e| e.catalog_key)
.collect();
if matches.is_empty() {
None
} else {
Some((entry, matches))
}
}
/// The full set of command-entry keywords, alphabetised by their
/// canonical literal. Used by the "available commands:" fallback
/// (ADR-0021 §5) when no keyword was consumed.
#[must_use]
pub fn entry_keywords_alphabetised() -> Vec<Keyword> {
let mut seen = std::collections::HashSet::new();
let mut out: Vec<Keyword> = REGISTRY
.iter()
.filter_map(|e| if seen.insert(e.entry) { Some(e.entry) } else { None })
.collect();
out.sort_by_key(|k| k.as_str());
out
}
#[cfg(test)]
mod tests {
use super::*;
use crate::dsl::lexer::lex;
use pretty_assertions::assert_eq;
#[test]
fn every_command_has_a_registry_entry() {
// The parser recognises ten command-entry keywords
// (ADR-0009 + ADR-0006 + ADR-0014). Each MUST be
// represented in the registry — otherwise a parse error
// for that command renders no usage block and the H1a
// pedagogy gap reopens for that family.
for entry in [
Keyword::Create,
Keyword::Drop,
Keyword::Add,
Keyword::Rename,
Keyword::Change,
Keyword::Show,
Keyword::Insert,
Keyword::Update,
Keyword::Delete,
Keyword::Replay,
] {
assert!(
REGISTRY.iter().any(|e| e.entry == entry),
"no usage entry for `{}`",
entry.as_str(),
);
}
}
#[test]
fn matched_entry_returns_none_when_no_keyword_consumed() {
let tokens = lex("frobulate Customers");
assert!(matched_entry(&tokens, 0).is_none());
}
#[test]
fn matched_entry_finds_entry_when_failure_position_equals_first_token_start() {
// Custom errors raised by `try_map` carry the matched
// span — whose `start` is the first consumed token's
// byte offset. For `create table Customers` (incomplete,
// raises the "tables need at least one column" custom
// error), failure position == first token start == 0.
// The entry keyword must still resolve.
let tokens = lex("create table Customers");
assert_eq!(tokens.first().unwrap().span.0, 0);
let (kw, keys) = matched_entry(&tokens, 0).expect("should match Create");
assert_eq!(kw, Keyword::Create);
assert_eq!(keys, vec!["parse.usage.create_table"]);
}
#[test]
fn matched_entry_finds_single_entry_command() {
let tokens = lex("create");
let pos = tokens.last().expect("non-empty").span.1;
let (kw, keys) = matched_entry(&tokens, pos).expect("should match");
assert_eq!(kw, Keyword::Create);
assert_eq!(keys, vec!["parse.usage.create_table"]);
}
#[test]
fn matched_entry_returns_all_family_members_for_add() {
let tokens = lex("add");
let pos = tokens.last().expect("non-empty").span.1;
let (kw, keys) = matched_entry(&tokens, pos).expect("should match");
assert_eq!(kw, Keyword::Add);
// Order matches REGISTRY declaration order. Both add-*
// commands surface.
assert!(keys.contains(&"parse.usage.add_column"));
assert!(keys.contains(&"parse.usage.add_relationship"));
}
#[test]
fn matched_entry_returns_all_family_members_for_drop() {
let tokens = lex("drop");
let pos = tokens.last().expect("non-empty").span.1;
let (kw, keys) = matched_entry(&tokens, pos).expect("should match");
assert_eq!(kw, Keyword::Drop);
assert!(keys.contains(&"parse.usage.drop_table"));
assert!(keys.contains(&"parse.usage.drop_column"));
assert!(keys.contains(&"parse.usage.drop_relationship"));
}
#[test]
fn matched_entry_resolves_to_first_keyword_for_partial_command() {
// `update Customers set` consumed all three tokens; the
// entry keyword is `update` (the first), not `set` (the
// last).
let tokens = lex("update Customers set");
let pos = tokens.last().expect("non-empty").span.1;
let (kw, keys) = matched_entry(&tokens, pos).expect("should match");
assert_eq!(kw, Keyword::Update);
assert_eq!(keys, vec!["parse.usage.update"]);
}
#[test]
fn entry_keywords_alphabetised_returns_ten_unique_sorted_commands() {
let keys = entry_keywords_alphabetised();
let names: Vec<&str> = keys.iter().map(|k| k.as_str()).collect();
assert_eq!(
names,
vec![
"add", "change", "create", "delete", "drop", "insert",
"rename", "replay", "show", "update",
],
);
}
}