feat(pipeline): the pure pipeline, checked against its goldens
The modules that turn a frozen range into passes and passes into a verdict: :Schema, :Tools (the DENY, git and base lists, a leaf, so the config and the pool both read them), :Plan, :Artifact, :Inventory, :Dispositions, :State, :Normalize, :Stream, :Prompt, :Config, and the URL half of :PgWire. Units, jobs, results, the config and the claude stream stay JSON from end to end, so every key keeps its place; :Text gives JSON values truthiness, get, ==, membership and Str(), and throws Dyn::Error under a name for each kind of unexpected data. Exactness where the bytes are seen: - the stream classifiers are hand-written scanners with case-insensitive folds (the Kelvin sign is a k), Unicode \b and \s, and `.` stopping at \n; the SQL call detection takes `<exe> sql` with any path before it; - config problems carry the JSON scanner's and the UTF-8 decoder's own texts, positions included; - a float total is Neumaier's compensated sum, and the build passes -ffp-contract=off: -march=native would fuse a*b+c; - `crafter-review inventory` and `config` print what the goldens record, byte for byte (the fixture and 14 configs, broken ones included); a message names this binary as the command to run. tests/Goldens: the recorded answers over the fixtures next to them (the commits fixture, a rich versioned series with CRLF, a folded subject, Markdown headings in and out of fences, the code-key rewrites); the suite compares 2304 results as DumpFile writes them, so a key out of order fails too. All agree. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
parent
b02de3ceff
commit
741e3a656d
GPG key ID:
2937E59CDCC1BCFB
70 changed files with 30565 additions and 21 deletions
|
|
@ -33,6 +33,19 @@ export namespace Crafter::Review::Dyn {
|
|||
// The code points from start to end: a negative index counts from the
|
||||
// end, and both are clipped to the text.
|
||||
std::string Slice(std::string_view text, std::int64_t start, std::optional<std::int64_t> end = std::nullopt);
|
||||
// The code point at byte `at` and the bytes it takes: U+FFFD and the
|
||||
// length of the invalid sequence where the bytes are no UTF-8.
|
||||
std::pair<char32_t, std::size_t> CodePointAt(std::string_view text, std::size_t at);
|
||||
// [start, end) of a slice of `size` items: a negative index counts from
|
||||
// the end, and both are clipped to the sequence.
|
||||
std::pair<std::size_t, std::size_t> SliceBounds(std::size_t size, std::int64_t start, std::optional<std::int64_t> end = std::nullopt);
|
||||
|
||||
// ---- regex classes, for the hand-written scanners -----------------------
|
||||
// The bytes of the run at `at` of regex \d (Unicode decimal digits), \s
|
||||
// (Unicode whitespace) or \S characters.
|
||||
std::size_t DecimalRun(std::string_view text, std::size_t at);
|
||||
std::size_t SpaceRun(std::string_view text, std::size_t at);
|
||||
std::size_t NonSpaceRun(std::string_view text, std::size_t at);
|
||||
|
||||
// ---- characters ----------------------------------------------------------
|
||||
bool IsSpace(char32_t c);
|
||||
|
|
@ -41,6 +54,8 @@ export namespace Crafter::Review::Dyn {
|
|||
// The digit's value, -1 for a character that is not a decimal digit.
|
||||
std::int32_t DecimalValue(char32_t c);
|
||||
bool IsPrintable(char32_t c);
|
||||
// regex \w: what \b takes for a word character.
|
||||
bool IsWord(char32_t c);
|
||||
// The lowercase ASCII letter `c` matches case-insensitively, or '\0':
|
||||
// ASCII letters, and the four code points outside ASCII that fold onto
|
||||
// one (İ ı → i, ſ → s, K → k).
|
||||
|
|
@ -70,14 +85,27 @@ export namespace Crafter::Review::Dyn {
|
|||
std::size_t Count(std::string_view text, std::string_view sub);
|
||||
std::string Replace(std::string_view text, std::string_view from, std::string_view to);
|
||||
std::string Join(std::span<const std::string> parts, std::string_view separator);
|
||||
// f'{text:<width}' / text.ljust(width): padded with spaces to `width`
|
||||
// code points, where std::format would count display columns.
|
||||
std::string LJust(std::string_view text, std::size_t width);
|
||||
|
||||
// ---- reading text files ----------------------------------------------------
|
||||
// A text file, read whole: strict UTF-8 — an invalid byte throws
|
||||
// std::runtime_error naming the file and the offset — and universal
|
||||
// newlines.
|
||||
// A text file, read whole: strict UTF-8 — invalid bytes throw Error
|
||||
// ("UnicodeDecodeError") with Utf8DecodeError()'s message — and universal
|
||||
// newlines. A file that cannot be opened throws Error ("FileNotFoundError").
|
||||
std::string ReadText(const std::filesystem::path& path);
|
||||
// \r\n and a lone \r become \n, as text mode reads them.
|
||||
std::string UniversalNewlines(std::string_view text);
|
||||
// The message for bytes that are not UTF-8 — "'utf-8' codec can't decode
|
||||
// byte 0xff in position 3: invalid start byte" — nullopt for valid UTF-8.
|
||||
std::optional<std::string> Utf8DecodeError(std::string_view bytes);
|
||||
// The message for text that is no JSON document, with the scanner's
|
||||
// reason and position ("Expecting ',' delimiter: line 3 column 5 (char
|
||||
// 21)"); nullopt for a document LoadJsonFile() takes.
|
||||
std::optional<std::string> JsonDecodeError(std::string_view text);
|
||||
// A JSON file: read as text, then parsed. Throws Error with the message a
|
||||
// "not JSON (...)" problem carries: the decode error, or the JSON error.
|
||||
Json LoadJsonFile(const std::filesystem::path& path);
|
||||
|
||||
// ---- numbers ------------------------------------------------------------------
|
||||
// A float as its shortest digits that round-trip; an exponent below 1e-4
|
||||
|
|
@ -127,6 +155,9 @@ export namespace Crafter::Review::Dyn {
|
|||
// std::invalid_argument with a ValueError message for unbalanced or
|
||||
// invalid brackets.
|
||||
SplitUrl UrlSplit(std::string_view url);
|
||||
// The five parts joined back into a URL: an empty query or fragment gets
|
||||
// no '?' or '#'.
|
||||
std::string UrlUnsplit(std::string_view scheme, std::string_view netloc, std::string_view path, std::string_view query, std::string_view fragment);
|
||||
// unquote(text): %XX sequences decoded as UTF-8, bad bytes replaced.
|
||||
std::string Unquote(std::string_view text);
|
||||
|
||||
|
|
@ -152,4 +183,63 @@ export namespace Crafter::Review::Dyn {
|
|||
// ---- errno texts -------------------------------------------------------------------------
|
||||
// "[Errno 111] Connection refused": an errno as a message prints it.
|
||||
std::string ErrnoText(std::int32_t code);
|
||||
|
||||
// ---- exceptions ------------------------------------------------------------------------------
|
||||
// Data the code does not expect — a key that is absent, a value of the
|
||||
// wrong type, a lookup in something that is no object — throws Error
|
||||
// named for the case (KeyError, TypeError, AttributeError): a caller
|
||||
// catches the names it expects, and anything else fails loudly.
|
||||
class Error : public std::runtime_error {
|
||||
public:
|
||||
Error(std::string_view kind, const std::string& message);
|
||||
const std::string& Kind() const;
|
||||
|
||||
private:
|
||||
std::string kind;
|
||||
};
|
||||
|
||||
// ---- JSON values ---------------------------------------------------------------------------------
|
||||
// The driver's data — units, jobs, results, dispositions, the config, the
|
||||
// claude stream — stays JSON from end to end, so every key keeps its place
|
||||
// and its unknown neighbours. These are the operations on it.
|
||||
//
|
||||
// Truthiness: null, false, 0, 0.0, "", [] and {} are false.
|
||||
bool Truthy(const Json& value);
|
||||
// `value or fallback`. A reference to one of the two, so neither may be a
|
||||
// temporary: name the fallback, or use EmptyDict()/EmptyList().
|
||||
const Json& Or(const Json& value, const Json& fallback);
|
||||
const Json& Or(Json&& value, const Json& fallback) = delete;
|
||||
const Json& Or(const Json& value, Json&& fallback) = delete;
|
||||
// The shared {} and [] that `x or {}` and `x or []` fall back to.
|
||||
const Json& EmptyDict();
|
||||
const Json& EmptyList();
|
||||
// Whether an object has the key; false for anything that is no object.
|
||||
bool Has(const Json& dict, std::string_view key);
|
||||
// The key's value, null when it is absent; Error (AttributeError) on
|
||||
// something that is no object.
|
||||
const Json& Get(const Json& dict, std::string_view key);
|
||||
// The key's value; Error (KeyError) when it is absent.
|
||||
const Json& At(const Json& dict, std::string_view key);
|
||||
// The type name messages use: dict, list, str, int, float, bool, NoneType.
|
||||
std::string_view TypeName(const Json& value);
|
||||
// An integer, a bool included.
|
||||
bool IsInt(const Json& value);
|
||||
// An integer or a float, a bool included.
|
||||
bool IsNumber(const Json& value);
|
||||
// A string as it is, anything else as ReprValue() prints it.
|
||||
std::string Str(const Json& value);
|
||||
// Str() of the value, or of the fallback when the value is falsy.
|
||||
std::string StrOr(const Json& value, std::string_view fallback);
|
||||
// a == b: numbers by value (True == 1 == 1.0), dicts whatever their key
|
||||
// order, a str never equal to a number.
|
||||
bool Equal(const Json& a, const Json& b);
|
||||
// Membership: an element of a list, a key of an object, a substring of a
|
||||
// string (Error when the item is no string there).
|
||||
bool In(const Json& item, const Json& container);
|
||||
// a + b on numbers: int + int stays an int, a float makes a float, a bool
|
||||
// counts as 0 or 1; Error (TypeError) for anything else.
|
||||
Json Add(const Json& a, const Json& b);
|
||||
// A total of floats by Neumaier's compensated summation, so it comes out
|
||||
// to the same last bit as the goldens record it.
|
||||
double Sum(std::span<const double> values);
|
||||
}
|
||||
|
|
|
|||
Loading…
Reference in a new issue