feat(kdl): add DSL schema validation

This commit is contained in:
2026-08-28 21:37:22 -04:00
parent 467d10bd0c
commit 67bfcde386
4 changed files with 1359 additions and 0 deletions
+724
View File
@@ -0,0 +1,724 @@
/*
* schema.c - stupidtools DSL schema validation (todo 9).
*
* Validates a parsed KDL document (src/kdl/ast.h) against the stupidtools
* DSL. The grammar is pinned in schema.h - the contract for todos 10/16/23.
* This module only READS the AST; it never modifies or frees it.
*
* Type checks go through the value model (src/kdl/value.h): names, files,
* version strings and check arguments must be non-empty unannotated
* strings; an option's `default` must be an unannotated boolean. Value-
* model failures (bad escapes, overflow) are re-owned as schema errors.
*
* ERRORS AND SPANS
* ----------------
* Every failure returns an owned st_error of category ST_ERR_KDL_SCHEMA
* whose span is heap-allocated IN THE SAME BLOCK as the error struct
* (the parser.c pattern), so err->span stays valid until st_error_free()
* and is never dangling. The span is borrowed-by-value from the AST's
* token refs, whose `file` pointer in turn borrows the caller's source
* buffer - the same lifetime contract as the parser. The single span-less
* error is "missing project" on an empty document: there is no source
* position (or file name) to point at.
*
* Copyright (c) 2026 huntedbytheirs
* SPDX-License-Identifier: BSD-3-Clause
*/
#include "kdl/schema.h"
#include "error.h"
#include "kdl/ast.h"
#include "kdl/value.h"
#include <stdbool.h>
#include <stddef.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
/* ---- owned errors ----------------------------------------------------- */
/* Build an owned ST_ERR_KDL_SCHEMA error whose span lives in the same
* allocation as the error struct (aligned right after it). Returns NULL
* only on allocation failure. */
static struct st_error *
err_at_owned(struct st_span sp, const char *msg)
{
const size_t align = _Alignof(struct st_span);
const size_t esize =
(sizeof(struct st_error) + align - 1) & ~(align - 1);
struct st_error *e;
struct st_span *spc;
size_t mlen;
if (msg == NULL) {
msg = "";
}
e = malloc(esize + sizeof(struct st_span));
if (e == NULL) {
return NULL;
}
mlen = strlen(msg);
e->message = malloc(mlen + 1);
if (e->message == NULL) {
free(e);
return NULL;
}
memcpy(e->message, msg, mlen + 1);
e->category = ST_ERR_KDL_SCHEMA;
spc = (struct st_span *)((unsigned char *)e + esize);
*spc = sp;
e->span = spc;
return e;
}
/* Re-own a value-model (ST_ERR_KDL_PARSE) error as a schema error: copy
* the message and the span VALUE into a fresh block. The caller frees the
* original error. */
static struct st_error *
schema_err_from(const struct st_error *e)
{
struct st_span sp;
if (e->span != NULL) {
sp = *e->span;
} else {
sp = (struct st_span){ NULL, 0, 0 };
}
return err_at_owned(sp, e->message != NULL ? e->message : "");
}
/* Interpret a token ref through the value model, converting any failure
* into an owned schema error (the original is freed here). On success
* `*out` holds the interpreted value (caller frees via
* st_kdl_value_free). */
static struct st_error *
value_interpret(const struct st_kdl_token_ref *tok,
const struct st_kdl_token_ref *ann,
struct st_kdl_value *out)
{
struct st_error *e = st_kdl_value_from_token_annotated(tok, ann, out);
if (e != NULL) {
struct st_error *r = schema_err_from(e);
st_error_free(e);
return r;
}
return NULL;
}
/* ---- naming helpers --------------------------------------------------- */
/* Copy a node name / property key into a fixed buffer for MESSAGE text,
* truncating to fit. Identifiers are copied raw (the common case, no
* allocation); other string forms are interpreted via the value model,
* falling back to "?" when that fails (message-only, never fatal). */
static void
name_into(const struct st_kdl_token_ref *name, char *buf, size_t cap)
{
if (cap == 0) {
return;
}
if (name->kind == ST_TOK_IDENT) {
size_t n = name->len < cap - 1 ? name->len : cap - 1;
memcpy(buf, name->text, n);
buf[n] = '\0';
return;
}
{
struct st_kdl_value v;
struct st_error *e = st_kdl_value_from_token(name, &v);
if (e == NULL && v.kind == ST_KDL_VAL_STRING) {
snprintf(buf, cap, "%s", v.as.str);
} else {
snprintf(buf, cap, "?");
}
st_error_free(e);
st_kdl_value_free(&v);
}
}
/* Compare a node name against a literal, accepting every string form
* (identifiers compared directly; other forms via the value model).
* Sets *out and returns NULL on success, or an owned schema error when
* the name is a string the value model cannot interpret. */
static struct st_error *
name_equals(const struct st_kdl_token_ref *name, const char *s, bool *out)
{
if (name->kind == ST_TOK_IDENT) {
*out = name->len == strlen(s) &&
memcmp(name->text, s, name->len) == 0;
return NULL;
}
{
struct st_kdl_value v;
struct st_error *e = st_kdl_value_from_token(name, &v);
if (e != NULL) {
struct st_error *r = schema_err_from(e);
st_error_free(e);
*out = false;
return r;
}
*out = v.kind == ST_KDL_VAL_STRING && strcmp(v.as.str, s) == 0;
st_kdl_value_free(&v);
return NULL;
}
}
/* ---- argument helpers ------------------------------------------------- */
/* Interpret `tok` (with optional annotation) as the non-empty unannotated
* string a DSL slot requires; on success *out owns the string (caller
* frees). `ctx` names the slot in messages, e.g. "target name". */
static struct st_error *
require_string_arg(const struct st_kdl_token_ref *tok,
const struct st_kdl_token_ref *ann,
const char *ctx, char **out)
{
struct st_kdl_value v;
struct st_error *e;
char msg[192];
*out = NULL;
e = value_interpret(tok, ann, &v);
if (e != NULL) {
return e;
}
if (v.annotation != NULL) {
snprintf(msg, sizeof msg, "%s must not carry a type annotation",
ctx);
e = err_at_owned(tok->span, msg);
st_kdl_value_free(&v);
return e;
}
if (v.kind != ST_KDL_VAL_STRING) {
snprintf(msg, sizeof msg, "%s must be a non-empty string (got %s)",
ctx, st_kdl_value_kind_name(v.kind));
e = err_at_owned(tok->span, msg);
st_kdl_value_free(&v);
return e;
}
if (v.as.str[0] == '\0') {
snprintf(msg, sizeof msg, "%s must be a non-empty string", ctx);
e = err_at_owned(tok->span, msg);
st_kdl_value_free(&v);
return e;
}
*out = v.as.str;
v.as.str = NULL;
st_kdl_value_free(&v); /* frees the annotation only */
return NULL;
}
/* The shared name shape of target/feature/option: exactly one non-empty
* string argument. `ctx` is the node kind ("target"). On success *out
* owns the name (caller frees). */
static struct st_error *
one_name_arg(const struct st_kdl_node *n, const char *ctx, char **out)
{
char msg[192];
char cctx[96];
*out = NULL;
if (n->args == NULL) {
snprintf(msg, sizeof msg, "%s requires a name argument", ctx);
return err_at_owned(n->name.span, msg);
}
if (n->args->next != NULL) {
char nm[65];
name_into(&n->args->value, nm, sizeof nm);
snprintf(msg, sizeof msg, "%s '%s' takes exactly one argument",
ctx, nm);
return err_at_owned(n->args->next->value.span, msg);
}
snprintf(cctx, sizeof cctx, "%s name", ctx);
return require_string_arg(&n->args->value, n->args->annotation, cctx,
out);
}
/* ---- per-node validators ---------------------------------------------- */
/* project "name" version "semver": the name is the first positional
* argument; the version string follows the LITERAL `version` keyword.
* No properties, no children, exactly three arguments. */
static struct st_error *
validate_project(const struct st_kdl_node *n)
{
struct st_kdl_arg *a = n->args;
struct st_kdl_arg *ver_kw;
char msg[192];
char name[65];
char ctx[96];
struct st_error *e;
char *name_str = NULL;
char *ver_str = NULL;
if (a == NULL) {
return err_at_owned(n->name.span,
"project requires a name argument");
}
e = require_string_arg(&a->value, a->annotation, "project name",
&name_str);
if (e != NULL) {
return e;
}
name_into(&a->value, name, sizeof name);
a = a->next;
if (a == NULL) {
snprintf(msg, sizeof msg,
"project '%s' requires a version (expected: project \"name\" "
"version \"semver\")", name);
e = err_at_owned(n->name.span, msg);
goto done;
}
if (a->annotation != NULL ||
!(a->value.kind == ST_TOK_IDENT && a->value.len == 7 &&
memcmp(a->value.text, "version", 7) == 0)) {
snprintf(msg, sizeof msg,
"project '%s': expected the keyword 'version' as the second "
"argument", name);
e = err_at_owned(a->value.span, msg);
goto done;
}
ver_kw = a;
a = a->next;
if (a == NULL) {
snprintf(msg, sizeof msg,
"project '%s': 'version' requires the version string argument",
name);
e = err_at_owned(ver_kw->value.span, msg);
goto done;
}
snprintf(ctx, sizeof ctx, "project '%s' version", name);
e = require_string_arg(&a->value, a->annotation, ctx, &ver_str);
if (e != NULL) {
goto done;
}
if (a->next != NULL) {
snprintf(msg, sizeof msg,
"project '%s' takes exactly three arguments: the name, the "
"'version' keyword, and the version string", name);
e = err_at_owned(a->next->value.span, msg);
goto done;
}
if (n->props != NULL) {
char key[65];
name_into(&n->props->key, key, sizeof key);
snprintf(msg, sizeof msg, "project '%s' has unexpected property "
"'%s'", name, key);
e = err_at_owned(n->props->key.span, msg);
goto done;
}
if (n->children != NULL) {
snprintf(msg, sizeof msg, "project '%s' takes no children", name);
e = err_at_owned(n->children->name.span, msg);
goto done;
}
e = NULL;
done:
free(name_str);
free(ver_str);
return e;
}
/* One `src "<file>"` or `feature "<name>"` child of a target: exactly one
* non-empty string argument, no properties, no children. */
static struct st_error *
validate_target_child(const struct st_kdl_node *c, const char *ctx,
bool is_src)
{
const char *kind = is_src ? "src" : "feature";
const char *slot = is_src ? "file" : "name";
char cctx[192];
char msg[192];
if (c->args == NULL) {
snprintf(msg, sizeof msg, "%s '%s' requires a %s argument", ctx,
kind, slot);
return err_at_owned(c->name.span, msg);
}
if (c->args->next != NULL) {
snprintf(msg, sizeof msg, "%s '%s' takes exactly one argument",
ctx, kind);
return err_at_owned(c->args->next->value.span, msg);
}
if (c->props != NULL) {
snprintf(msg, sizeof msg, "%s '%s' takes no properties", ctx, kind);
return err_at_owned(c->props->key.span, msg);
}
if (c->children != NULL) {
snprintf(msg, sizeof msg, "%s '%s' takes no children", ctx, kind);
return err_at_owned(c->children->name.span, msg);
}
{
char *tmp = NULL;
struct st_error *e;
snprintf(cctx, sizeof cctx, "%s '%s' %s", ctx, kind, slot);
e = require_string_arg(&c->args->value, c->args->annotation, cctx,
&tmp);
free(tmp);
return e;
}
}
/* target "name" { src ...; feature ...; }: the name shape plus a children
* block whose nodes are only `src` (a file) or `feature` (a reference to
* a top-level feature by name). Presence-checking the reference against
* the document is OPTIONAL for this todo - only the shape is enforced. */
static struct st_error *
validate_target(const struct st_kdl_node *n)
{
char msg[256];
char name[65];
char ctx[96];
struct st_error *e;
char *name_str = NULL;
struct st_kdl_node *c;
e = one_name_arg(n, "target", &name_str);
if (e != NULL) {
return e;
}
name_into(&n->args->value, name, sizeof name);
if (n->props != NULL) {
char key[65];
name_into(&n->props->key, key, sizeof key);
snprintf(msg, sizeof msg, "target '%s' has unexpected property "
"'%s'", name, key);
e = err_at_owned(n->props->key.span, msg);
goto done;
}
snprintf(ctx, sizeof ctx, "target '%s'", name);
for (c = n->children; c != NULL; c = c->next) {
bool is_src = false;
bool is_feat = false;
e = name_equals(&c->name, "src", &is_src);
if (e != NULL) {
goto done;
}
e = name_equals(&c->name, "feature", &is_feat);
if (e != NULL) {
goto done;
}
if (is_src || is_feat) {
e = validate_target_child(c, ctx, is_src);
if (e != NULL) {
goto done;
}
continue;
}
{
char cname[65];
name_into(&c->name, cname, sizeof cname);
snprintf(msg, sizeof msg,
"%s has unknown child node '%s' (expected 'src' or "
"'feature')", ctx, cname);
e = err_at_owned(c->name.span, msg);
goto done;
}
}
e = NULL;
done:
free(name_str);
return e;
}
/* feature "name" { <checks> }: the name shape (no properties) plus a
* children block of CHECK nodes, validated STRUCTURALLY ONLY. The 8
* check kinds and their exact argument/property shapes are todo 10's job;
* until its registry lands, every child is accepted provided it has at
* least one non-empty string argument and no children (its properties
* are deliberately left for todo 10 to validate). */
static struct st_error *
validate_feature(const struct st_kdl_node *n)
{
char msg[192];
char name[65];
struct st_error *e;
char *name_str = NULL;
struct st_kdl_node *c;
e = one_name_arg(n, "feature", &name_str);
if (e != NULL) {
return e;
}
name_into(&n->args->value, name, sizeof name);
if (n->props != NULL) {
char key[65];
name_into(&n->props->key, key, sizeof key);
snprintf(msg, sizeof msg, "feature '%s' has unexpected property "
"'%s'", name, key);
e = err_at_owned(n->props->key.span, msg);
goto done;
}
for (c = n->children; c != NULL; c = c->next) {
/*
* HOOK (todo 10): dispatch on the check node NAME through the
* feature-check registry (src/detect/check_registry.h) to
* validate the 8 kinds - header, function, library, type,
* sizeof, program, compiler_flag, pkg_config - and their
* argument/property shapes. Until then, structural checks only:
*/
if (c->args == NULL) {
char cname[65];
name_into(&c->name, cname, sizeof cname);
snprintf(msg, sizeof msg,
"feature '%s': check '%s' requires an argument", name,
cname);
e = err_at_owned(c->name.span, msg);
goto done;
}
{
char cname[65];
char cctx[192];
char *tmp = NULL;
name_into(&c->name, cname, sizeof cname);
snprintf(cctx, sizeof cctx, "feature '%s': check '%s' argument",
name, cname);
e = require_string_arg(&c->args->value, c->args->annotation,
cctx, &tmp);
free(tmp);
if (e != NULL) {
goto done;
}
}
if (c->children != NULL) {
char cname[65];
name_into(&c->name, cname, sizeof cname);
snprintf(msg, sizeof msg,
"feature '%s': check '%s' takes no children", name, cname);
e = err_at_owned(c->children->name.span, msg);
goto done;
}
}
e = NULL;
done:
free(name_str);
return e;
}
/* option "name" (default=<bool>)?: the name shape, no children, and at
* most one `default` property whose value must be an unannotated boolean
* (#true/#false) once interpreted through the value model. */
static struct st_error *
validate_option(const struct st_kdl_node *n)
{
char msg[256];
char name[65];
struct st_error *e;
char *name_str = NULL;
struct st_kdl_prop *p;
bool seen_default = false;
e = one_name_arg(n, "option", &name_str);
if (e != NULL) {
return e;
}
name_into(&n->args->value, name, sizeof name);
if (n->children != NULL) {
snprintf(msg, sizeof msg, "option '%s' takes no children", name);
e = err_at_owned(n->children->name.span, msg);
goto done;
}
for (p = n->props; p != NULL; p = p->next) {
bool is_def = false;
e = name_equals(&p->key, "default", &is_def);
if (e != NULL) {
goto done;
}
if (!is_def) {
char key[65];
name_into(&p->key, key, sizeof key);
snprintf(msg, sizeof msg,
"option '%s' has unexpected property '%s' (only 'default' "
"is allowed)", name, key);
e = err_at_owned(p->key.span, msg);
goto done;
}
if (seen_default) {
snprintf(msg, sizeof msg,
"option '%s' has a duplicate 'default' property", name);
e = err_at_owned(p->key.span, msg);
goto done;
}
seen_default = true;
{
struct st_kdl_value v;
e = value_interpret(&p->value, p->annotation, &v);
if (e != NULL) {
goto done;
}
if (v.annotation != NULL) {
snprintf(msg, sizeof msg,
"option '%s': 'default' must not carry a type "
"annotation", name);
e = err_at_owned(p->value.span, msg);
st_kdl_value_free(&v);
goto done;
}
if (v.kind != ST_KDL_VAL_BOOL) {
snprintf(msg, sizeof msg,
"option '%s': 'default' must be a boolean (got %s)",
name, st_kdl_value_kind_name(v.kind));
e = err_at_owned(p->value.span, msg);
st_kdl_value_free(&v);
goto done;
}
st_kdl_value_free(&v);
}
}
e = NULL;
done:
free(name_str);
return e;
}
/* ---- top level -------------------------------------------------------- */
/* Classify a top-level node name. */
enum top_kind {
TOP_PROJECT = 0,
TOP_TARGET,
TOP_FEATURE,
TOP_OPTION,
TOP_UNKNOWN,
};
static struct st_error *
classify_top(const struct st_kdl_node *n, enum top_kind *out)
{
static const char *const names[] = { "project", "target", "feature",
"option" };
static const enum top_kind kinds[] = { TOP_PROJECT, TOP_TARGET,
TOP_FEATURE, TOP_OPTION };
size_t i;
for (i = 0; i < sizeof(names) / sizeof(names[0]); i++) {
bool eq = false;
struct st_error *e = name_equals(&n->name, names[i], &eq);
if (e != NULL) {
return e;
}
if (eq) {
*out = kinds[i];
return NULL;
}
}
*out = TOP_UNKNOWN;
return NULL;
}
/* ---- public API ------------------------------------------------------- */
struct st_error *
st_kdl_validate(const struct st_kdl_document *doc)
{
struct st_kdl_node *n;
bool saw_goal = false; /* saw at least one target or feature */
struct st_error *e;
if (doc == NULL || doc->nodes == NULL) {
/* No source position (or file name) exists to point at. */
return st_error_kdl_schema("missing required 'project' node");
}
n = doc->nodes;
{
enum top_kind kind;
e = classify_top(n, &kind);
if (e != NULL) {
return e;
}
if (kind != TOP_PROJECT) {
char nm[65];
char msg[192];
name_into(&n->name, nm, sizeof nm);
snprintf(msg, sizeof msg,
"expected 'project' as the first top-level node, got '%s'",
nm);
return err_at_owned(n->name.span, msg);
}
e = validate_project(n);
if (e != NULL) {
return e;
}
}
for (n = n->next; n != NULL; n = n->next) {
enum top_kind kind;
e = classify_top(n, &kind);
if (e != NULL) {
return e;
}
switch (kind) {
case TOP_PROJECT:
return err_at_owned(n->name.span, "duplicate 'project' node");
case TOP_TARGET:
e = validate_target(n);
if (e != NULL) {
return e;
}
saw_goal = true;
break;
case TOP_FEATURE:
e = validate_feature(n);
if (e != NULL) {
return e;
}
saw_goal = true;
break;
case TOP_OPTION:
e = validate_option(n);
if (e != NULL) {
return e;
}
break;
case TOP_UNKNOWN: {
char nm[65];
char msg[192];
name_into(&n->name, nm, sizeof nm);
snprintf(msg, sizeof msg,
"unknown top-level node '%s' (expected 'target', "
"'feature', or 'option')", nm);
return err_at_owned(n->name.span, msg);
}
}
}
if (!saw_goal) {
return err_at_owned(doc->nodes->name.span,
"missing 'target' or 'feature' node after 'project'");
}
return NULL;
}