leptris 1.9.319.0-arm-linux → 1.9.321.0-arm-linux

data/vendor-src/libleptris/src/leptris/xsd/compile.c

@@ -13,13 +13,122 @@
 #define XSD_NS "http://www.w3.org/2001/XMLSchema"
 
 extern int xsd_simple_valid(const XsdSimple* simple, const char* v);
+extern struct xsd_validator* xsd_validator_new(
+    struct leptris_xsd_schema* s);
+extern void xsd_validator_free(struct xsd_validator* v);
+extern int xsd_validator_run(struct xsd_validator* v, LeptrisDocument doc);
+extern size_t xsd_validator_error_count(struct xsd_validator* v);
+extern const char* xsd_validator_error_at(struct xsd_validator* v,
+                                          size_t i);
+
+extern int xsd_content_valid(struct leptris_xsd_schema* s, XsdCm* model,
+                             const char* target_ns,
+                             const char* const* names,
+                             const char* const* ns_uris, size_t count);
 
 struct leptris_xsd_schema {
     size_t declaration_count;
     XsdSimple* simple_types;
+    XsdElementDecl* elements;
+    XsdCm* complex_types; /* name-carrying model roots */
+    XsdTypeAttrs* type_attrs;
+    char* target_ns;
+    void* validator; /* slice-4 parked error list */
     char* error;
 };
 
+/* ---- slice 3: content-model capture ------------------------------ */
+
+static char* xsd_strdup(const char* s);
+
+static XsdCm* xsd_cm_new(XsdCmKind kind) {
+    XsdCm* cm = (XsdCm*)calloc(1, sizeof(*cm));
+    if (!cm) return NULL;
+    cm->kind = kind;
+    cm->min = 1;
+    cm->max = 1;
+    return cm;
+}
+
+static void xsd_cm_free(XsdCm* cm) {
+    while (cm) {
+        XsdCm* next = cm->next;
+        xsd_cm_free(cm->first_child);
+        free(cm->name);
+        free(cm->type);
+        free(cm->ns);
+        free(cm->any_ns);
+        free(cm);
+        cm = next;
+    }
+}
+
+static void xsd_cm_occurrence(XsdCm* cm, LeptrisElement e) {
+    const char* mn = leptris_element_attribute(e, "minOccurs");
+    if (mn) cm->min = (strcmp(mn, "unbounded") == 0) ? 0 : atoi(mn);
+    const char* mx = leptris_element_attribute(e, "maxOccurs");
+    if (mx) cm->max = (strcmp(mx, "unbounded") == 0) ? -1 : atoi(mx);
+}
+
+/* Recursive model build over sequence | choice | all | element |
+ * any children. Returns the child list head; *tail_out receives
+ * the list tail for sibling chaining. */
+static XsdCm* xsd_cm_parse(LeptrisElement group, XsdCm** tail_out) {
+    XsdCm* head = NULL;
+    XsdCm* tail = NULL;
+    for (LeptrisNodeRef n =
+             leptris_node_first_child(leptris_element_as_node(group));
+         n; n = leptris_node_next_sibling(n)) {
+        if (leptris_node_get_type(n) != LEPTRIS_NODE_TYPE_ELEMENT)
+            continue;
+        LeptrisElement e = (LeptrisElement)n;
+        const char* local = NULL, *prefix = NULL, *uri = NULL;
+        leptris_element_expanded_name(e, &local, &prefix, &uri);
+        if (!local || (uri && strcmp(uri, XSD_NS) != 0)) continue;
+
+        XsdCm* cm = NULL;
+        if (strcmp(local, "sequence") == 0 || strcmp(local, "choice") == 0 ||
+            strcmp(local, "all") == 0) {
+            cm = xsd_cm_new(strcmp(local, "sequence") == 0 ? XSD_CM_SEQ
+                            : strcmp(local, "choice") == 0 ? XSD_CM_CHOICE
+                                                           : XSD_CM_ALL);
+            if (cm) cm->first_child = xsd_cm_parse(e, NULL);
+        } else if (strcmp(local, "element") == 0) {
+            cm = xsd_cm_new(XSD_CM_ELEMENT);
+            if (cm) {
+                const char* nm =
+                    leptris_element_attribute(e, "name");
+                if (!nm) nm = leptris_element_attribute(e, "ref");
+                cm->name = xsd_strdup(nm);
+                cm->type = xsd_strdup(
+                    leptris_element_attribute(e, "type"));
+            }
+        } else if (strcmp(local, "any") == 0) {
+            cm = xsd_cm_new(XSD_CM_ANY);
+            if (cm) {
+                cm->any_ns = xsd_strdup(
+                    leptris_element_attribute(e, "namespace"));
+                const char* pc = leptris_element_attribute(
+                    e, "processContents");
+                cm->process_skip =
+                    (pc && (strcmp(pc, "skip") == 0 ||
+                            strcmp(pc, "lax") == 0)) ? 1 : 0;
+            }
+        } else {
+            continue; /* annotation & friends */
+        }
+        if (!cm) continue;
+        xsd_cm_occurrence(cm, e);
+        if (tail)
+            tail->next = cm;
+        else
+            head = cm;
+        tail = cm;
+    }
+    if (tail_out) *tail_out = tail;
+    return head;
+}
+
 static char* xsd_strdup(const char* s) {
     if (!s) return NULL;
     size_t n = strlen(s) + 1;
@@ -161,6 +270,9 @@ LEPTRIS_API LeptrisXsdSchema leptris_xsd_compile(const char* xsd_text,
         return NULL;
     }
 
+    /* target namespace (xs:any ##other/##targetNamespace filters) */
+    s->target_ns = NULL;
+
     LeptrisDocument doc = leptris_parse_string(xsd_text, len, status);
     if (!doc) {
         xsd_set_error(s, "schema document is not well-formed XML");
@@ -172,6 +284,8 @@ LEPTRIS_API LeptrisXsdSchema leptris_xsd_compile(const char* xsd_text,
         xsd_set_error(s, "schema document has no root element");
         goto fail;
     }
+    s->target_ns = xsd_strdup(
+        leptris_element_attribute(root, "targetNamespace"));
 
     const char* local = NULL;
     const char* prefix = NULL;
@@ -195,7 +309,77 @@ LEPTRIS_API LeptrisXsdSchema leptris_xsd_compile(const char* xsd_text,
                                       &uri);
         if (!xsd_is_declaration(local, uri)) continue;
         s->declaration_count++;
-        if (strcmp(local, "simpleType") == 0 && uri && !strcmp(uri, XSD_NS)) {
+        if (strcmp(local, "element") == 0 && uri && !strcmp(uri, XSD_NS)) {
+            const char* nm = leptris_element_attribute((LeptrisElement)n,
+                                                       "name");
+            const char* ty = leptris_element_attribute((LeptrisElement)n,
+                                                       "type");
+            if (nm && ty) {
+                XsdElementDecl* d =
+                    (XsdElementDecl*)calloc(1, sizeof(*d));
+                if (d) {
+                    d->name = xsd_strdup(nm);
+                    d->type = xsd_strdup(ty);
+                    d->next = s->elements;
+                    s->elements = d;
+                }
+            }
+        } else if (strcmp(local, "complexType") == 0 && uri &&
+                   !strcmp(uri, XSD_NS)) {
+            const char* nm =
+                leptris_element_attribute((LeptrisElement)n, "name");
+            if (nm) {
+                XsdCm* root =
+                    xsd_cm_new(XSD_CM_SEQ); /* wrapper keeps the name */
+                if (root) {
+                    root->name = xsd_strdup(nm);
+                    root->first_child =
+                        xsd_cm_parse((LeptrisElement)n, NULL);
+                    root->next = s->complex_types;
+                    s->complex_types = root;
+                }
+                /* slice 4: attribute declarations of the type */
+                XsdAttrDecl* head = NULL;
+                XsdAttrDecl* tail = NULL;
+                for (LeptrisNodeRef ac = leptris_node_first_child(
+                         leptris_element_as_node((LeptrisElement)n));
+                     ac; ac = leptris_node_next_sibling(ac)) {
+                    if (leptris_node_get_type(ac) !=
+                        LEPTRIS_NODE_TYPE_ELEMENT)
+                        continue;
+                    LeptrisElement ae = (LeptrisElement)ac;
+                    const char* al = NULL, *ap = NULL, *au = NULL;
+                    leptris_element_expanded_name(ae, &al, &ap, &au);
+                    if (!al || (au && strcmp(au, XSD_NS) != 0)) continue;
+                    if (strcmp(al, "attribute") != 0) continue;
+                    const char* an = leptris_element_attribute(ae, "name");
+                    if (!an) continue; /* ref forms: slice 5 */
+                    XsdAttrDecl* d = (XsdAttrDecl*)calloc(1, sizeof(*d));
+                    if (!d) continue;
+                    d->name = xsd_strdup(an);
+                    d->type = xsd_strdup(
+                        leptris_element_attribute(ae, "type"));
+                    const char* use =
+                        leptris_element_attribute(ae, "use");
+                    d->required = (use && strcmp(use, "required") == 0);
+                    if (tail)
+                        tail->next = d;
+                    else
+                        head = d;
+                    tail = d;
+                }
+                if (head) {
+                    XsdTypeAttrs* ta =
+                        (XsdTypeAttrs*)calloc(1, sizeof(*ta));
+                    if (ta) {
+                        ta->type_name = xsd_strdup(nm);
+                        ta->attrs = head;
+                        ta->next = s->type_attrs;
+                        s->type_attrs = ta;
+                    }
+                }
+            }
+        } else if (strcmp(local, "simpleType") == 0 && uri && !strcmp(uri, XSD_NS)) {
             XsdSimple* captured = xsd_capture_simple((LeptrisElement)n);
             if (captured) {
                 captured->next = s->simple_types;
@@ -218,6 +402,33 @@ fail:
 LEPTRIS_API void leptris_xsd_free(LeptrisXsdSchema schema) {
     struct leptris_xsd_schema* s = (struct leptris_xsd_schema*)schema;
     if (!s) return;
+    xsd_cm_free(s->complex_types);
+    XsdTypeAttrs* ta = s->type_attrs;
+    while (ta) {
+        XsdTypeAttrs* tn = ta->next;
+        XsdAttrDecl* a = ta->attrs;
+        while (a) {
+            XsdAttrDecl* an = a->next;
+            free(a->name);
+            free(a->type);
+            free(a);
+            a = an;
+        }
+        free(ta->type_name);
+        free(ta);
+        ta = tn;
+    }
+    XsdElementDecl* d = s->elements;
+    while (d) {
+        XsdElementDecl* dn = d->next;
+        free(d->name);
+        free(d->type);
+        free(d);
+        d = dn;
+    }
+    free(s->target_ns);
+    if (s->validator)
+        xsd_validator_free((struct xsd_validator*)s->validator);
     XsdSimple* t = s->simple_types;
     while (t) {
         XsdSimple* next = t->next;
@@ -261,6 +472,103 @@ LEPTRIS_API int leptris_xsd_simple_valid(LeptrisXsdSchema schema,
     return xsd_valid_chain(s, t, lexical, 0);
 }
 
+/* ---- validator accessors (validate.c) --------------------------- */
+
+const XsdElementDecl* xsd_find_element(struct leptris_xsd_schema* s,
+                                       const char* name) {
+    if (!s || !name) return NULL;
+    for (XsdElementDecl* d = s->elements; d; d = d->next)
+        if (strcmp(d->name, name) == 0) return d;
+    return NULL;
+}
+
+XsdCm* xsd_find_complex(struct leptris_xsd_schema* s,
+                        const char* type_name) {
+    if (!s || !type_name) return NULL;
+    for (XsdCm* c = s->complex_types; c; c = c->next)
+        if (strcmp(c->name, type_name) == 0) return c;
+    return NULL;
+}
+
+const XsdSimple* xsd_find_simple_pub(struct leptris_xsd_schema* s,
+                                     const char* name) {
+    return xsd_find_simple(s, name);
+}
+
+const char* xsd_target_ns(struct leptris_xsd_schema* s) {
+    return s ? s->target_ns : NULL;
+}
+
+const XsdTypeAttrs* xsd_find_type_attrs(struct leptris_xsd_schema* s,
+                                        const char* type_name) {
+    if (!s || !type_name) return NULL;
+    for (XsdTypeAttrs* t = s->type_attrs; t; t = t->next)
+        if (strcmp(t->type_name, type_name) == 0) return t;
+    return NULL;
+}
+
+/* ---- slice 4: instance validation (validate.c engine) ------------ */
+
+LEPTRIS_API int leptris_xsd_validate(LeptrisXsdSchema schema,
+                                     LeptrisDocument doc) {
+    struct leptris_xsd_schema* s = (struct leptris_xsd_schema*)schema;
+    if (!s || !doc) return -1;
+    struct xsd_validator* v = xsd_validator_new(s);
+    if (!v) return -1;
+    int r = xsd_validator_run(v, doc);
+    /* park the run's errors on the schema for the accessors; a new
+     * validate run replaces them. */
+    struct xsd_validator* prev = (struct xsd_validator*)s->validator;
+    if (prev) xsd_validator_free(prev);
+    s->validator = v;
+    return r;
+}
+
+LEPTRIS_API size_t leptris_xsd_error_count(LeptrisXsdSchema schema) {
+    struct leptris_xsd_schema* s = (struct leptris_xsd_schema*)schema;
+    if (!s) return 0;
+    return xsd_validator_error_count(
+        (struct xsd_validator*)s->validator);
+}
+
+LEPTRIS_API const char* leptris_xsd_error_at(LeptrisXsdSchema schema,
+                                             size_t i) {
+    struct leptris_xsd_schema* s = (struct leptris_xsd_schema*)schema;
+    if (!s) return NULL;
+    return xsd_validator_error_at((struct xsd_validator*)s->validator,
+                                  i);
+}
+
+LEPTRIS_API int leptris_xsd_content_valid(
+    LeptrisXsdSchema schema, const char* element_name,
+    const char* const* child_names, const char* const* child_ns,
+    size_t child_count) {
+    struct leptris_xsd_schema* s = (struct leptris_xsd_schema*)schema;
+    if (!s || !element_name) return -1;
+
+    const XsdElementDecl* d = s->elements;
+    while (d && strcmp(d->name, element_name) != 0) d = d->next;
+    if (!d) return -1; /* unknown element */
+
+    if (!d->type) return 1; /* no type attr: anyType accepts all */
+
+    /* built-in or simpleType ref: element content is text-only —
+     * any element child fails */
+    if (strncmp(d->type, "xs:", 3) == 0 ||
+        xsd_find_simple(s, d->type))
+        return child_count == 0;
+
+    XsdCm* ct = s->complex_types;
+    while (ct && strcmp(ct->name, d->type) != 0) ct = ct->next;
+    if (!ct) return child_count == 0; /* unknown type: accept-empty */
+
+    XsdCm* model = ct->first_child;
+    if (!model) return child_count == 0;
+
+    return xsd_content_valid(s, model, s->target_ns, child_names,
+                             child_ns, child_count);
+}
+
 LEPTRIS_API size_t leptris_xsd_declaration_count(LeptrisXsdSchema schema) {
     if (!schema) return 0;
     return ((struct leptris_xsd_schema*)schema)->declaration_count;

data/vendor-src/libleptris/src/leptris/xsd/content_model.c

@@ -0,0 +1,232 @@
+/* xsd/content_model.c — #1075 slice 3: content-model evaluation.
+ *
+ * The compiled model tree lowers into a Thompson NFA (epsilon +
+ * symbol transitions, occurrence bounds via counting copies and a
+ * loop for unbounded) and the child sequence is consumed by
+ * simulating the live state set — linear in children x states, no
+ * backtracking blowup on nested choices.
+ *
+ * Namespace filters (xs:any/@namespace): "##any" (default),
+ * "##other" (not the target namespace, not unqualified),
+ * "##targetNamespace", "##local", or a space-separated URI list.
+ */
+#include <stdlib.h>
+#include <string.h>
+
+#include "xsd_internal.h"
+
+typedef struct {
+    int from;
+    int symbol;      /* -1 = epsilon; else a particle transition */
+    XsdCm* particle; /* symbol owner (element ref or wildcard) */
+    int to;
+} CmTrans;
+
+typedef struct {
+    CmTrans* trans;
+    size_t count, cap;
+    int state_count; /* explicit: state ids are NOT transition ids */
+    int start, accept;
+} CmNfa;
+
+static int nfa_state(CmNfa* n) { return n->state_count++; }
+
+static void nfa_trans(CmNfa* n, int from, int symbol, XsdCm* particle,
+                      int to) {
+    if (n->count == n->cap) {
+        size_t cap = n->cap ? n->cap * 2 : 16;
+        CmTrans* grown = (CmTrans*)realloc(n->trans, cap * sizeof(*grown));
+        if (!grown) return; /* OOM: walk treats the NFA as non-matching */
+        n->trans = grown;
+        n->cap = cap;
+    }
+    n->trans[n->count].from = from;
+    n->trans[n->count].symbol = symbol;
+    n->trans[n->count].particle = particle;
+    n->trans[n->count].to = to;
+    n->count++;
+}
+
+static void cm_build(CmNfa* n, XsdCm* node, int from, int to);
+static void cm_build_occurrence(CmNfa* n, XsdCm* node,
+                               int from, int to);
+
+/* chain the group children: first reads `from`, last writes `to`.
+ * xs:all is evaluated in sequence order for now (its order-free
+ * semantics arrive with slice 4's all-semantics pass). */
+static void cm_build_children(CmNfa* n, XsdCm* node, int from, int to) {
+    XsdCm* c = node->first_child;
+    if (!c) {
+        nfa_trans(n, from, -1, NULL, to);
+        return;
+    }
+    int cursor = from;
+    while (c->next) {
+        int mid = nfa_state(n);
+        cm_build_occurrence(n, c, cursor, mid);
+        cursor = mid;
+        c = c->next;
+    }
+    cm_build_occurrence(n, c, cursor, to);
+}
+
+static void cm_build(CmNfa* n, XsdCm* node, int from, int to) {
+    switch (node->kind) {
+        case XSD_CM_ELEMENT:
+        case XSD_CM_ANY:
+            nfa_trans(n, from, 0, node, to);
+            return;
+        case XSD_CM_SEQ:
+        case XSD_CM_ALL:
+            cm_build_children(n, node, from, to);
+            return;
+        case XSD_CM_CHOICE: {
+            XsdCm* c = node->first_child;
+            if (!c) {
+                nfa_trans(n, from, -1, NULL, to);
+                return;
+            }
+            for (; c; c = c->next) cm_build_occurrence(n, c, from, to);
+            return;
+        }
+    }
+}
+
+/* occurrence bounds: min copies, then (max-min) optionals, or an
+ * epsilon self-loop region for unbounded. */
+static void cm_build_occurrence(CmNfa* n, XsdCm* node, int from, int to) {
+    for (int i = 0; i < node->min; i++) {
+        int mid = nfa_state(n);
+        cm_build(n, node, from, mid);
+        from = mid;
+    }
+    if (node->max < 0) {
+        int loop = nfa_state(n);
+        nfa_trans(n, from, -1, NULL, loop);
+        cm_build(n, node, loop, loop);
+        nfa_trans(n, loop, -1, NULL, to);
+        return;
+    }
+    for (int i = node->min; i < node->max; i++) {
+        nfa_trans(n, from, -1, NULL, to);
+        int mid = nfa_state(n);
+        cm_build(n, node, from, mid);
+        from = mid;
+    }
+    nfa_trans(n, from, -1, NULL, to);
+}
+
+/* ---- namespace filter (xs:any/@namespace) ------------------------ */
+
+static int any_ns_matches(XsdCm* any, const char* uri,
+                          const char* target_ns) {
+    const char* f = any->any_ns ? any->any_ns : "##any";
+    if (strcmp(f, "##any") == 0) return 1;
+    if (strcmp(f, "##local") == 0) return uri == NULL;
+    if (strcmp(f, "##targetNamespace") == 0)
+        return target_ns && uri && strcmp(uri, target_ns) == 0;
+    if (strcmp(f, "##other") == 0)
+        return uri != NULL &&
+               !(target_ns && strcmp(uri, target_ns) == 0);
+    /* space-separated URI list (+##local) */
+    const char* p = f;
+    while (*p) {
+        while (*p == ' ') p++;
+        const char* start = p;
+        while (*p && *p != ' ') p++;
+        size_t len = (size_t)(p - start);
+        if (len == 0) break;
+        if (strncmp(start, "##local", len) == 0 && len == 7) {
+            if (uri == NULL) return 1;
+            continue;
+        }
+        if (uri && strlen(uri) == len && strncmp(start, uri, len) == 0)
+            return 1;
+    }
+    return 0;
+}
+
+/* ---- simulation --------------------------------------------------- */
+
+typedef struct {
+    int* states;
+    size_t count, cap;
+} StateSet;
+
+static void ss_add(StateSet* ss, int st) {
+    for (size_t i = 0; i < ss->count; i++)
+        if (ss->states[i] == st) return;
+    if (ss->count == ss->cap) {
+        size_t cap = ss->cap ? ss->cap * 2 : 16;
+        int* grown = (int*)realloc(ss->states, cap * sizeof(int));
+        if (!grown) return;
+        ss->states = grown;
+        ss->cap = cap;
+    }
+    ss->states[ss->count++] = st;
+}
+
+static void ss_close(StateSet* ss, CmNfa* n) {
+    /* fixpoint over epsilon edges */
+    size_t i = 0;
+    while (i < ss->count) {
+        int st = ss->states[i];
+        for (size_t t = 0; t < n->count; t++)
+            if (n->trans[t].from == st && n->trans[t].symbol == -1)
+                ss_add(ss, n->trans[t].to);
+        i++;
+    }
+}
+
+int xsd_content_valid(struct leptris_xsd_schema* s, XsdCm* model,
+                      const char* target_ns, const char* const* names,
+                      const char* const* ns_uris, size_t count) {
+    (void)s;
+    if (!model) return -1;
+    CmNfa n;
+    memset(&n, 0, sizeof(n));
+    n.start = nfa_state(&n);
+    n.accept = nfa_state(&n);
+    cm_build_occurrence(&n, model, n.start, n.accept);
+
+    StateSet live;
+    memset(&live, 0, sizeof(live));
+    ss_add(&live, n.start);
+    ss_close(&live, &n);
+
+    for (size_t c = 0; c < count; c++) {
+        StateSet next;
+        memset(&next, 0, sizeof(next));
+        const char* uri = ns_uris ? ns_uris[c] : NULL;
+        for (size_t i = 0; i < live.count; i++) {
+            for (size_t t = 0; t < n.count; t++) {
+                CmTrans* tr = &n.trans[t];
+                if (tr->from != live.states[i] || tr->symbol == -1)
+                    continue;
+                XsdCm* p = tr->particle;
+                if (p->kind == XSD_CM_ELEMENT) {
+                    if (strcmp(p->name, names[c]) == 0 &&
+                        (!p->ns || (uri && strcmp(uri, p->ns) == 0)))
+                        ss_add(&next, tr->to);
+                } else if (p->kind == XSD_CM_ANY) {
+                    if (any_ns_matches(p, uri, target_ns))
+                        ss_add(&next, tr->to);
+                }
+            }
+        }
+        free(live.states);
+        live = next;
+        ss_close(&live, &n);
+        if (live.count == 0) {
+            free(n.trans);
+            return 0;
+        }
+    }
+
+    int accept = 0;
+    for (size_t i = 0; i < live.count; i++)
+        if (live.states[i] == n.accept) accept = 1;
+    free(live.states);
+    free(n.trans);
+    return accept;
+}