nokolexbor 0.2.0

data/vendor/lexbor/source/lexbor/html/tokenizer.h

@@ -0,0 +1,343 @@
+/*
+ * Copyright (C) 2018-2020 Alexander Borisov
+ *
+ * Author: Alexander Borisov <borisov@lexbor.com>
+ */
+
+#ifndef LEXBOR_HTML_TOKENIZER_H
+#define LEXBOR_HTML_TOKENIZER_H
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include "lexbor/core/sbst.h"
+#include "lexbor/core/array_obj.h"
+
+#include "lexbor/html/base.h"
+#include "lexbor/html/token.h"
+
+#include "lexbor/tag/tag.h"
+#include "lexbor/ns/ns.h"
+
+
+/* State */
+typedef const lxb_char_t *
+(*lxb_html_tokenizer_state_f)(lxb_html_tokenizer_t *tkz,
+                              const lxb_char_t *data, const lxb_char_t *end);
+
+typedef lxb_html_token_t *
+(*lxb_html_tokenizer_token_f)(lxb_html_tokenizer_t *tkz,
+                              lxb_html_token_t *token, void *ctx);
+
+
+struct lxb_html_tokenizer {
+    lxb_html_tokenizer_state_f       state;
+    lxb_html_tokenizer_state_f       state_return;
+
+    lxb_html_tokenizer_token_f       callback_token_done;
+    void                             *callback_token_ctx;
+
+    lexbor_hash_t                    *tags;
+    lexbor_hash_t                    *attrs;
+    lexbor_mraw_t                    *attrs_mraw;
+
+    /* For a temp strings and other templary data */
+    lexbor_mraw_t                    *mraw;
+
+    /* Current process token */
+    lxb_html_token_t                 *token;
+
+    /* Memory for token and attr */
+    lexbor_dobject_t                 *dobj_token;
+    lexbor_dobject_t                 *dobj_token_attr;
+
+    /* Parse error */
+    lexbor_array_obj_t               *parse_errors;
+
+    /*
+     * Leak abstractions.
+     * The only place where the specification causes mixing Tree Builder
+     * and Tokenizer. We kill all beauty.
+     * Current Tree parser. This is not ref (not ref count).
+     */
+    lxb_html_tree_t                  *tree;
+
+    /* Temp */
+    const lxb_char_t                 *markup;
+    const lxb_char_t                 *temp;
+    lxb_tag_id_t                     tmp_tag_id;
+
+    lxb_char_t                       *start;
+    lxb_char_t                       *pos;
+    const lxb_char_t                 *end;
+    const lxb_char_t                 *begin;
+    const lxb_char_t                 *last;
+
+    /* Entities */
+    const lexbor_sbst_entry_static_t *entity;
+    const lexbor_sbst_entry_static_t *entity_match;
+    uintptr_t                        entity_start;
+    uintptr_t                        entity_end;
+    uint32_t                         entity_length;
+    uint32_t                         entity_number;
+    bool                             is_attribute;
+
+    /* Process */
+    lxb_html_tokenizer_opt_t         opt;
+    lxb_status_t                     status;
+    bool                             is_eof;
+
+    lxb_html_tokenizer_t             *base;
+    size_t                           ref_count;
+};
+
+
+#include "lexbor/html/tokenizer/error.h"
+
+
+extern const lxb_char_t *lxb_html_tokenizer_eof;
+
+LXB_API lxb_html_tokenizer_t *
+lxb_html_tokenizer_create(void);
+
+LXB_API lxb_status_t
+lxb_html_tokenizer_init(lxb_html_tokenizer_t *tkz);
+
+LXB_API lxb_status_t
+lxb_html_tokenizer_inherit(lxb_html_tokenizer_t *tkz_to,
+                           lxb_html_tokenizer_t *tkz_from);
+
+LXB_API lxb_html_tokenizer_t *
+lxb_html_tokenizer_ref(lxb_html_tokenizer_t *tkz);
+
+LXB_API lxb_html_tokenizer_t *
+lxb_html_tokenizer_unref(lxb_html_tokenizer_t *tkz);
+
+LXB_API void
+lxb_html_tokenizer_clean(lxb_html_tokenizer_t *tkz);
+
+LXB_API lxb_html_tokenizer_t *
+lxb_html_tokenizer_destroy(lxb_html_tokenizer_t *tkz);
+
+LXB_API lxb_status_t
+lxb_html_tokenizer_tags_make(lxb_html_tokenizer_t *tkz, size_t table_size);
+
+LXB_API void
+lxb_html_tokenizer_tags_destroy(lxb_html_tokenizer_t *tkz);
+
+LXB_API lxb_status_t
+lxb_html_tokenizer_attrs_make(lxb_html_tokenizer_t *tkz, size_t table_size);
+
+LXB_API void
+lxb_html_tokenizer_attrs_destroy(lxb_html_tokenizer_t *tkz);
+
+LXB_API lxb_status_t
+lxb_html_tokenizer_begin(lxb_html_tokenizer_t *tkz);
+
+LXB_API lxb_status_t
+lxb_html_tokenizer_chunk(lxb_html_tokenizer_t *tkz,
+                         const lxb_char_t *data, size_t size);
+
+LXB_API lxb_status_t
+lxb_html_tokenizer_end(lxb_html_tokenizer_t *tkz);
+
+
+LXB_API const lxb_char_t *
+lxb_html_tokenizer_change_incoming(lxb_html_tokenizer_t *tkz,
+                                   const lxb_char_t *pos);
+
+LXB_API lxb_ns_id_t
+lxb_html_tokenizer_current_namespace(lxb_html_tokenizer_t *tkz);
+
+LXB_API void
+lxb_html_tokenizer_set_state_by_tag(lxb_html_tokenizer_t *tkz, bool scripting,
+                                    lxb_tag_id_t tag_id, lxb_ns_id_t ns);
+
+
+/*
+ * Inline functions
+ */
+lxb_inline void
+lxb_html_tokenizer_status_set(lxb_html_tokenizer_t *tkz, lxb_status_t status)
+{
+    tkz->status = status;
+}
+
+lxb_inline void
+lxb_html_tokenizer_tags_set(lxb_html_tokenizer_t *tkz, lexbor_hash_t *tags)
+{
+    tkz->tags = tags;
+}
+
+lxb_inline lexbor_hash_t *
+lxb_html_tokenizer_tags(lxb_html_tokenizer_t *tkz)
+{
+    return tkz->tags;
+}
+
+lxb_inline void
+lxb_html_tokenizer_attrs_set(lxb_html_tokenizer_t *tkz, lexbor_hash_t *attrs)
+{
+    tkz->attrs = attrs;
+}
+
+lxb_inline lexbor_hash_t *
+lxb_html_tokenizer_attrs(lxb_html_tokenizer_t *tkz)
+{
+    return tkz->attrs;
+}
+
+lxb_inline void
+lxb_html_tokenizer_attrs_mraw_set(lxb_html_tokenizer_t *tkz,
+                                  lexbor_mraw_t *mraw)
+{
+    tkz->attrs_mraw = mraw;
+}
+
+lxb_inline lexbor_mraw_t *
+lxb_html_tokenizer_attrs_mraw(lxb_html_tokenizer_t *tkz)
+{
+    return tkz->attrs_mraw;
+}
+
+lxb_inline void
+lxb_html_tokenizer_callback_token_done_set(lxb_html_tokenizer_t *tkz,
+                                           lxb_html_tokenizer_token_f call_func,
+                                           void *ctx)
+{
+    tkz->callback_token_done = call_func;
+    tkz->callback_token_ctx = ctx;
+}
+
+lxb_inline void *
+lxb_html_tokenizer_callback_token_done_ctx(lxb_html_tokenizer_t *tkz)
+{
+    return tkz->callback_token_ctx;
+}
+
+lxb_inline void
+lxb_html_tokenizer_state_set(lxb_html_tokenizer_t *tkz,
+                             lxb_html_tokenizer_state_f state)
+{
+    tkz->state = state;
+}
+
+lxb_inline void
+lxb_html_tokenizer_tmp_tag_id_set(lxb_html_tokenizer_t *tkz,
+                                  lxb_tag_id_t tag_id)
+{
+    tkz->tmp_tag_id = tag_id;
+}
+
+lxb_inline lxb_html_tree_t *
+lxb_html_tokenizer_tree(lxb_html_tokenizer_t *tkz)
+{
+    return tkz->tree;
+}
+
+lxb_inline void
+lxb_html_tokenizer_tree_set(lxb_html_tokenizer_t *tkz, lxb_html_tree_t *tree)
+{
+    tkz->tree = tree;
+}
+
+lxb_inline lexbor_mraw_t *
+lxb_html_tokenizer_mraw(lxb_html_tokenizer_t *tkz)
+{
+    return tkz->mraw;
+}
+
+lxb_inline lxb_status_t
+lxb_html_tokenizer_temp_realloc(lxb_html_tokenizer_t *tkz, size_t size)
+{
+    size_t length = tkz->pos - tkz->start;
+    size_t new_size = (tkz->end - tkz->start) + size + 4096;
+
+    tkz->start = (lxb_char_t *)lexbor_realloc(tkz->start, new_size);
+    if (tkz->start == NULL) {
+        tkz->status = LXB_STATUS_ERROR_MEMORY_ALLOCATION;
+        return tkz->status;
+    }
+
+    tkz->pos = tkz->start + length;
+    tkz->end = tkz->start + new_size;
+
+    return LXB_STATUS_OK;
+}
+
+lxb_inline lxb_status_t
+lxb_html_tokenizer_temp_append_data(lxb_html_tokenizer_t *tkz,
+                                    const lxb_char_t *data)
+{
+    size_t size = data - tkz->begin;
+
+    if ((tkz->pos + size) > tkz->end) {
+        if(lxb_html_tokenizer_temp_realloc(tkz, size)) {
+            return tkz->status;
+        }
+    }
+
+    tkz->pos = (lxb_char_t *) memcpy(tkz->pos, tkz->begin, size) + size;
+
+    return LXB_STATUS_OK;
+}
+
+lxb_inline lxb_status_t
+lxb_html_tokenizer_temp_append(lxb_html_tokenizer_t *tkz,
+                               const lxb_char_t *data, size_t size)
+{
+    if ((tkz->pos + size) > tkz->end) {
+        if(lxb_html_tokenizer_temp_realloc(tkz, size)) {
+            return tkz->status;
+        }
+    }
+
+    tkz->pos = (lxb_char_t *) memcpy(tkz->pos, data, size) + size;
+
+    return LXB_STATUS_OK;
+}
+
+
+/*
+ * No inline functions for ABI.
+ */
+LXB_API void
+lxb_html_tokenizer_status_set_noi(lxb_html_tokenizer_t *tkz,
+                                  lxb_status_t status);
+
+LXB_API void
+lxb_html_tokenizer_callback_token_done_set_noi(lxb_html_tokenizer_t *tkz,
+                                               lxb_html_tokenizer_token_f call_func,
+                                               void *ctx);
+
+LXB_API void *
+lxb_html_tokenizer_callback_token_done_ctx_noi(lxb_html_tokenizer_t *tkz);
+
+LXB_API void
+lxb_html_tokenizer_state_set_noi(lxb_html_tokenizer_t *tkz,
+                                 lxb_html_tokenizer_state_f state);
+
+LXB_API void
+lxb_html_tokenizer_tmp_tag_id_set_noi(lxb_html_tokenizer_t *tkz,
+                                      lxb_tag_id_t tag_id);
+
+LXB_API lxb_html_tree_t *
+lxb_html_tokenizer_tree_noi(lxb_html_tokenizer_t *tkz);
+
+LXB_API void
+lxb_html_tokenizer_tree_set_noi(lxb_html_tokenizer_t *tkz,
+                                lxb_html_tree_t *tree);
+
+LXB_API lexbor_mraw_t *
+lxb_html_tokenizer_mraw_noi(lxb_html_tokenizer_t *tkz);
+
+LXB_API lexbor_hash_t *
+lxb_html_tokenizer_tags_noi(lxb_html_tokenizer_t *tkz);
+
+
+#ifdef __cplusplus
+} /* extern "C" */
+#endif
+
+#endif /* LEXBOR_HTML_TOKENIZER_H */

data/vendor/lexbor/source/lexbor/html/tree/active_formatting.c

@@ -0,0 +1,241 @@
+/*
+ * Copyright (C) 2018 Alexander Borisov
+ *
+ * Author: Alexander Borisov <borisov@lexbor.com>
+ */
+
+#include "lexbor/dom/interfaces/node.h"
+
+#include "lexbor/html/tree/active_formatting.h"
+#include "lexbor/html/tree/open_elements.h"
+#include "lexbor/html/interfaces/element.h"
+
+
+static lxb_html_element_t lxb_html_tree_active_formatting_marker_static;
+
+static lxb_dom_node_t *lxb_html_tree_active_formatting_marker_node_static =
+    (lxb_dom_node_t *) &lxb_html_tree_active_formatting_marker_static;
+
+
+lxb_html_element_t *
+lxb_html_tree_active_formatting_marker(void)
+{
+    return &lxb_html_tree_active_formatting_marker_static;
+}
+
+void
+lxb_html_tree_active_formatting_up_to_last_marker(lxb_html_tree_t *tree)
+{
+    void **list = tree->active_formatting->list;
+
+    while (tree->active_formatting->length != 0) {
+        tree->active_formatting->length--;
+
+        if (list[tree->active_formatting->length]
+            == &lxb_html_tree_active_formatting_marker_static)
+        {
+            break;
+        }
+    }
+}
+
+void
+lxb_html_tree_active_formatting_remove_by_node(lxb_html_tree_t *tree,
+                                               lxb_dom_node_t *node)
+{
+    size_t delta;
+    void **list = tree->active_formatting->list;
+    size_t idx = tree->active_formatting->length;
+
+    while (idx != 0) {
+        idx--;
+
+        if (list[idx] == node) {
+            delta = tree->active_formatting->length - idx - 1;
+
+            memmove(list + idx, list + idx + 1, sizeof(void *) * delta);
+
+            tree->active_formatting->length--;
+
+            break;
+        }
+    }
+}
+
+bool
+lxb_html_tree_active_formatting_find_by_node(lxb_html_tree_t *tree,
+                                             lxb_dom_node_t *node,
+                                             size_t *return_pos)
+{
+    void **list = tree->active_formatting->list;
+
+    for (size_t i = 0; i < tree->active_formatting->length; i++) {
+        if (list[i] == node) {
+            if (return_pos) {
+                *return_pos = i;
+            }
+
+            return true;
+        }
+    }
+
+    if (return_pos) {
+        *return_pos = 0;
+    }
+
+    return false;
+}
+
+bool
+lxb_html_tree_active_formatting_find_by_node_reverse(lxb_html_tree_t *tree,
+                                                     lxb_dom_node_t *node,
+                                                     size_t *return_pos)
+{
+    void **list = tree->active_formatting->list;
+    size_t len = tree->active_formatting->length;
+
+    while (len != 0) {
+        len--;
+
+        if (list[len] == node) {
+            if (return_pos) {
+                *return_pos = len;
+            }
+
+            return true;
+        }
+    }
+
+    if (return_pos) {
+        *return_pos = 0;
+    }
+
+    return false;
+}
+
+lxb_status_t
+lxb_html_tree_active_formatting_reconstruct_elements(lxb_html_tree_t *tree)
+{
+    /* Step 1 */
+    if (tree->active_formatting->length == 0) {
+        return LXB_STATUS_OK;
+    }
+
+    lexbor_array_t *af = tree->active_formatting;
+    void **list = af->list;
+
+    /* Step 2-3 */
+    size_t af_idx = af->length - 1;
+
+    if(list[af_idx] == &lxb_html_tree_active_formatting_marker_static
+       || lxb_html_tree_open_elements_find_by_node_reverse(tree, list[af_idx],
+                                                           NULL))
+    {
+        return LXB_STATUS_OK;
+    }
+
+    /*
+     * Step 4-6
+     * Rewind
+     */
+    while (af_idx != 0) {
+        af_idx--;
+
+        if(list[af_idx] == &lxb_html_tree_active_formatting_marker_static ||
+           lxb_html_tree_open_elements_find_by_node_reverse(tree, list[af_idx],
+                                                            NULL))
+        {
+            /* Step 7 */
+            af_idx++;
+
+            break;
+        }
+    }
+
+    /*
+     * Step 8-10
+     * Create
+     */
+    lxb_dom_node_t *node;
+    lxb_html_element_t *element;
+    lxb_html_token_t fake_token = {0};
+
+    while (af_idx < af->length) {
+        node = list[af_idx];
+
+        fake_token.tag_id = node->local_name;
+        fake_token.base_element = node;
+
+        element = lxb_html_tree_insert_html_element(tree, &fake_token);
+        if (element == NULL) {
+            return LXB_STATUS_ERROR_MEMORY_ALLOCATION;
+        }
+
+        /* Step 9 */
+        list[af_idx] = lxb_dom_interface_node(element);
+
+        /* Step 10 */
+        af_idx++;
+    }
+
+    return LXB_STATUS_OK;
+}
+
+lxb_dom_node_t *
+lxb_html_tree_active_formatting_between_last_marker(lxb_html_tree_t *tree,
+                                                    lxb_tag_id_t tag_idx,
+                                                    size_t *return_idx)
+{
+    lxb_dom_node_t **list = (lxb_dom_node_t **) tree->active_formatting->list;
+    size_t idx = tree->active_formatting->length;
+
+    while (idx) {
+        idx--;
+
+        if (list[idx] == lxb_html_tree_active_formatting_marker_node_static) {
+            return NULL;
+        }
+
+        if (list[idx]->local_name == tag_idx && list[idx]->ns == LXB_NS_HTML) {
+            if (return_idx) {
+                *return_idx = idx;
+            }
+
+            return list[idx];
+        }
+    }
+
+    return NULL;
+}
+
+void
+lxb_html_tree_active_formatting_push_with_check_dupl(lxb_html_tree_t *tree,
+                                                     lxb_dom_node_t *node)
+{
+    lxb_dom_node_t **list = (lxb_dom_node_t **) tree->active_formatting->list;
+    size_t idx = tree->active_formatting->length;
+    size_t earliest_idx = (idx ? (idx - 1) : 0);
+    size_t count = 0;
+
+    while (idx) {
+        idx--;
+
+        if (list[idx] == lxb_html_tree_active_formatting_marker_node_static) {
+            break;
+        }
+
+        if(list[idx]->local_name == node->local_name && list[idx]->ns == node->ns
+            && lxb_dom_element_compare(lxb_dom_interface_element(list[idx]),
+                                       lxb_dom_interface_element(node)))
+        {
+            count++;
+            earliest_idx = idx;
+        }
+    }
+
+    if(count >= 3) {
+        lxb_html_tree_active_formatting_remove(tree, earliest_idx);
+    }
+
+    lxb_html_tree_active_formatting_push(tree, node);
+}

data/vendor/lexbor/source/lexbor/html/tree/active_formatting.h

@@ -0,0 +1,117 @@
+/*
+ * Copyright (C) 2018 Alexander Borisov
+ *
+ * Author: Alexander Borisov <borisov@lexbor.com>
+ */
+
+#ifndef LEXBOR_HTML_ACTIVE_FORMATTING_H
+#define LEXBOR_HTML_ACTIVE_FORMATTING_H
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include "lexbor/core/array.h"
+
+#include "lexbor/html/tree.h"
+
+
+lxb_html_element_t *
+lxb_html_tree_active_formatting_marker(void);
+
+void
+lxb_html_tree_active_formatting_up_to_last_marker(lxb_html_tree_t *tree);
+
+void
+lxb_html_tree_active_formatting_remove_by_node(lxb_html_tree_t *tree,
+                                               lxb_dom_node_t *node);
+
+bool
+lxb_html_tree_active_formatting_find_by_node(lxb_html_tree_t *tree,
+                                             lxb_dom_node_t *node,
+                                             size_t *return_pos);
+
+bool
+lxb_html_tree_active_formatting_find_by_node_reverse(lxb_html_tree_t *tree,
+                                                     lxb_dom_node_t *node,
+                                                     size_t *return_pos);
+
+lxb_status_t
+lxb_html_tree_active_formatting_reconstruct_elements(lxb_html_tree_t *tree);
+
+lxb_dom_node_t *
+lxb_html_tree_active_formatting_between_last_marker(lxb_html_tree_t *tree,
+                                                    lxb_tag_id_t tag_idx,
+                                                    size_t *return_idx);
+
+void
+lxb_html_tree_active_formatting_push_with_check_dupl(lxb_html_tree_t *tree,
+                                                     lxb_dom_node_t *node);
+
+
+/*
+ * Inline functions
+ */
+lxb_inline lxb_dom_node_t *
+lxb_html_tree_active_formatting_current_node(lxb_html_tree_t *tree)
+{
+    if (tree->active_formatting->length == 0) {
+        return NULL;
+    }
+
+    return (lxb_dom_node_t *) tree->active_formatting->list
+        [ (tree->active_formatting->length - 1) ];
+}
+
+lxb_inline lxb_dom_node_t *
+lxb_html_tree_active_formatting_first(lxb_html_tree_t *tree)
+{
+    return (lxb_dom_node_t *) lexbor_array_get(tree->active_formatting, 0);
+}
+
+lxb_inline lxb_dom_node_t *
+lxb_html_tree_active_formatting_get(lxb_html_tree_t *tree, size_t idx)
+{
+    return (lxb_dom_node_t *) lexbor_array_get(tree->active_formatting, idx);
+}
+
+lxb_inline lxb_status_t
+lxb_html_tree_active_formatting_push(lxb_html_tree_t *tree,
+                                     lxb_dom_node_t *node)
+{
+    return lexbor_array_push(tree->active_formatting, node);
+}
+
+lxb_inline lxb_dom_node_t *
+lxb_html_tree_active_formatting_pop(lxb_html_tree_t *tree)
+{
+    return (lxb_dom_node_t *) lexbor_array_pop(tree->active_formatting);
+}
+
+lxb_inline lxb_status_t
+lxb_html_tree_active_formatting_push_marker(lxb_html_tree_t *tree)
+{
+    return lexbor_array_push(tree->active_formatting,
+                             lxb_html_tree_active_formatting_marker());
+}
+
+lxb_inline lxb_status_t
+lxb_html_tree_active_formatting_insert(lxb_html_tree_t *tree,
+                                       lxb_dom_node_t *node, size_t idx)
+{
+    return lexbor_array_insert(tree->active_formatting, idx, node);
+}
+
+lxb_inline void
+lxb_html_tree_active_formatting_remove(lxb_html_tree_t *tree, size_t idx)
+{
+    lexbor_array_delete(tree->active_formatting, idx, 1);
+}
+
+
+#ifdef __cplusplus
+} /* extern "C" */
+#endif
+
+#endif /* LEXBOR_HTML_ACTIVE_FORMATTING_H */
+