nokolexbor 0.2.0

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

@@ -0,0 +1,431 @@
+/*
+ * Copyright (C) 2018 Alexander Borisov
+ *
+ * Author: Alexander Borisov <borisov@lexbor.com>
+ */
+
+#ifndef LEXBOR_HTML_TREE_H
+#define LEXBOR_HTML_TREE_H
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include "lexbor/dom/interfaces/node.h"
+#include "lexbor/dom/interfaces/attr.h"
+
+#include "lexbor/html/base.h"
+#include "lexbor/html/node.h"
+#include "lexbor/html/tokenizer.h"
+#include "lexbor/html/interfaces/document.h"
+#include "lexbor/html/tag.h"
+#include "lexbor/html/tree/error.h"
+
+
+typedef bool
+(*lxb_html_tree_insertion_mode_f)(lxb_html_tree_t *tree,
+                                  lxb_html_token_t *token);
+
+typedef lxb_status_t
+(*lxb_html_tree_append_attr_f)(lxb_html_tree_t *tree,
+                               lxb_dom_attr_t *attr, void *ctx);
+
+typedef struct {
+    lexbor_array_obj_t *text_list;
+    bool               have_non_ws;
+}
+lxb_html_tree_pending_table_t;
+
+struct lxb_html_tree {
+    lxb_html_tokenizer_t           *tkz_ref;
+
+    lxb_html_document_t            *document;
+    lxb_dom_node_t                 *fragment;
+
+    lxb_html_form_element_t        *form;
+
+    lexbor_array_t                 *open_elements;
+    lexbor_array_t                 *active_formatting;
+    lexbor_array_obj_t             *template_insertion_modes;
+
+    lxb_html_tree_pending_table_t  pending_table;
+
+    lexbor_array_obj_t             *parse_errors;
+
+    bool                           foster_parenting;
+    bool                           frameset_ok;
+    bool                           scripting;
+
+    lxb_html_tree_insertion_mode_f mode;
+    lxb_html_tree_insertion_mode_f original_mode;
+    lxb_html_tree_append_attr_f    before_append_attr;
+
+    lxb_status_t                   status;
+
+    size_t                         ref_count;
+};
+
+typedef enum {
+    LXB_HTML_TREE_INSERTION_POSITION_CHILD  = 0x00,
+    LXB_HTML_TREE_INSERTION_POSITION_BEFORE = 0x01
+}
+lxb_html_tree_insertion_position_t;
+
+
+LXB_API lxb_html_tree_t *
+lxb_html_tree_create(void);
+
+LXB_API lxb_status_t
+lxb_html_tree_init(lxb_html_tree_t *tree, lxb_html_tokenizer_t *tkz);
+
+LXB_API lxb_html_tree_t *
+lxb_html_tree_ref(lxb_html_tree_t *tree);
+
+LXB_API lxb_html_tree_t *
+lxb_html_tree_unref(lxb_html_tree_t *tree);
+
+LXB_API void
+lxb_html_tree_clean(lxb_html_tree_t *tree);
+
+LXB_API lxb_html_tree_t *
+lxb_html_tree_destroy(lxb_html_tree_t *tree);
+
+LXB_API lxb_status_t
+lxb_html_tree_stop_parsing(lxb_html_tree_t *tree);
+
+LXB_API bool
+lxb_html_tree_process_abort(lxb_html_tree_t *tree);
+
+LXB_API void
+lxb_html_tree_parse_error(lxb_html_tree_t *tree, lxb_html_token_t *token,
+                          lxb_html_tree_error_id_t id);
+
+LXB_API bool
+lxb_html_tree_construction_dispatcher(lxb_html_tree_t *tree,
+                                      lxb_html_token_t *token);
+
+LXB_API lxb_dom_node_t *
+lxb_html_tree_appropriate_place_inserting_node(lxb_html_tree_t *tree,
+                                      lxb_dom_node_t *override_target,
+                                      lxb_html_tree_insertion_position_t *ipos);
+
+LXB_API lxb_html_element_t *
+lxb_html_tree_insert_foreign_element(lxb_html_tree_t *tree,
+                                     lxb_html_token_t *token, lxb_ns_id_t ns);
+
+LXB_API lxb_html_element_t *
+lxb_html_tree_create_element_for_token(lxb_html_tree_t *tree,
+                                       lxb_html_token_t *token, lxb_ns_id_t ns,
+                                       lxb_dom_node_t *parent);
+
+LXB_API lxb_status_t
+lxb_html_tree_append_attributes(lxb_html_tree_t *tree,
+                                lxb_dom_element_t *element,
+                                lxb_html_token_t *token, lxb_ns_id_t ns);
+
+LXB_API lxb_status_t
+lxb_html_tree_append_attributes_from_element(lxb_html_tree_t *tree,
+                                             lxb_dom_element_t *element,
+                                             lxb_dom_element_t *from,
+                                             lxb_ns_id_t ns);
+
+LXB_API lxb_status_t
+lxb_html_tree_adjust_mathml_attributes(lxb_html_tree_t *tree,
+                                       lxb_dom_attr_t *attr, void *ctx);
+
+LXB_API lxb_status_t
+lxb_html_tree_adjust_svg_attributes(lxb_html_tree_t *tree,
+                                    lxb_dom_attr_t *attr, void *ctx);
+
+LXB_API lxb_status_t
+lxb_html_tree_adjust_foreign_attributes(lxb_html_tree_t *tree,
+                                        lxb_dom_attr_t *attr, void *ctx);
+
+LXB_API lxb_status_t
+lxb_html_tree_insert_character(lxb_html_tree_t *tree, lxb_html_token_t *token,
+                               lxb_dom_node_t **ret_node);
+
+LXB_API lxb_status_t
+lxb_html_tree_insert_character_for_data(lxb_html_tree_t *tree,
+                                        lexbor_str_t *str,
+                                        lxb_dom_node_t **ret_node);
+
+LXB_API lxb_dom_comment_t *
+lxb_html_tree_insert_comment(lxb_html_tree_t *tree,
+                             lxb_html_token_t *token, lxb_dom_node_t *pos);
+
+LXB_API lxb_dom_document_type_t *
+lxb_html_tree_create_document_type_from_token(lxb_html_tree_t *tree,
+                                              lxb_html_token_t *token);
+
+LXB_API void
+lxb_html_tree_node_delete_deep(lxb_html_tree_t *tree, lxb_dom_node_t *node);
+
+LXB_API lxb_html_element_t *
+lxb_html_tree_generic_rawtext_parsing(lxb_html_tree_t *tree,
+                                      lxb_html_token_t *token);
+
+LXB_API lxb_html_element_t *
+lxb_html_tree_generic_rcdata_parsing(lxb_html_tree_t *tree,
+                                     lxb_html_token_t *token);
+
+LXB_API void
+lxb_html_tree_generate_implied_end_tags(lxb_html_tree_t *tree,
+                                        lxb_tag_id_t ex_tag, lxb_ns_id_t ex_ns);
+
+LXB_API void
+lxb_html_tree_generate_all_implied_end_tags_thoroughly(lxb_html_tree_t *tree,
+                                                       lxb_tag_id_t ex_tag,
+                                                       lxb_ns_id_t ex_ns);
+
+LXB_API void
+lxb_html_tree_reset_insertion_mode_appropriately(lxb_html_tree_t *tree);
+
+LXB_API lxb_dom_node_t *
+lxb_html_tree_element_in_scope(lxb_html_tree_t *tree, lxb_tag_id_t tag_id,
+                               lxb_ns_id_t ns, lxb_html_tag_category_t ct);
+
+LXB_API lxb_dom_node_t *
+lxb_html_tree_element_in_scope_by_node(lxb_html_tree_t *tree,
+                                       lxb_dom_node_t *by_node,
+                                       lxb_html_tag_category_t ct);
+
+LXB_API lxb_dom_node_t *
+lxb_html_tree_element_in_scope_h123456(lxb_html_tree_t *tree);
+
+LXB_API lxb_dom_node_t *
+lxb_html_tree_element_in_scope_tbody_thead_tfoot(lxb_html_tree_t *tree);
+
+LXB_API lxb_dom_node_t *
+lxb_html_tree_element_in_scope_td_th(lxb_html_tree_t *tree);
+
+LXB_API bool
+lxb_html_tree_check_scope_element(lxb_html_tree_t *tree);
+
+LXB_API void
+lxb_html_tree_close_p_element(lxb_html_tree_t *tree, lxb_html_token_t *token);
+
+LXB_API bool
+lxb_html_tree_adoption_agency_algorithm(lxb_html_tree_t *tree,
+                                        lxb_html_token_t *token,
+                                        lxb_status_t *status);
+
+LXB_API bool
+lxb_html_tree_html_integration_point(lxb_dom_node_t *node);
+
+LXB_API lxb_status_t
+lxb_html_tree_adjust_attributes_mathml(lxb_html_tree_t *tree,
+                                       lxb_dom_attr_t *attr, void *ctx);
+
+LXB_API lxb_status_t
+lxb_html_tree_adjust_attributes_svg(lxb_html_tree_t *tree,
+                                    lxb_dom_attr_t *attr, void *ctx);
+
+
+/*
+ * Inline functions
+ */
+lxb_inline lxb_status_t
+lxb_html_tree_begin(lxb_html_tree_t *tree, lxb_html_document_t *document)
+{
+    tree->document = document;
+
+    return lxb_html_tokenizer_begin(tree->tkz_ref);
+}
+
+lxb_inline lxb_status_t
+lxb_html_tree_chunk(lxb_html_tree_t *tree, const lxb_char_t *html, size_t size)
+{
+    return lxb_html_tokenizer_chunk(tree->tkz_ref, html, size);
+}
+
+lxb_inline lxb_status_t
+lxb_html_tree_end(lxb_html_tree_t *tree)
+{
+    return lxb_html_tokenizer_end(tree->tkz_ref);
+}
+
+lxb_inline lxb_status_t
+lxb_html_tree_build(lxb_html_tree_t *tree, lxb_html_document_t *document,
+                    const lxb_char_t *html, size_t size)
+{
+    tree->status = lxb_html_tree_begin(tree, document);
+    if (tree->status != LXB_STATUS_OK) {
+        return tree->status;
+    }
+
+    tree->status = lxb_html_tree_chunk(tree, html, size);
+    if (tree->status != LXB_STATUS_OK) {
+        return tree->status;
+    }
+
+    return lxb_html_tree_end(tree);
+}
+
+lxb_inline lxb_dom_node_t *
+lxb_html_tree_create_node(lxb_html_tree_t *tree,
+                          lxb_tag_id_t tag_id, lxb_ns_id_t ns)
+{
+    return (lxb_dom_node_t *) lxb_html_interface_create(tree->document,
+                                                        tag_id, ns);
+}
+
+lxb_inline bool
+lxb_html_tree_node_is(lxb_dom_node_t *node, lxb_tag_id_t tag_id)
+{
+    return node->local_name == tag_id && node->ns == LXB_NS_HTML;
+}
+
+lxb_inline lxb_dom_node_t *
+lxb_html_tree_current_node(lxb_html_tree_t *tree)
+{
+    if (tree->open_elements->length == 0) {
+        return NULL;
+    }
+
+    return (lxb_dom_node_t *)
+        tree->open_elements->list[ (tree->open_elements->length - 1) ];
+}
+
+lxb_inline lxb_dom_node_t *
+lxb_html_tree_adjusted_current_node(lxb_html_tree_t *tree)
+{
+    if(tree->fragment != NULL && tree->open_elements->length == 1) {
+        return lxb_dom_interface_node(tree->fragment);
+    }
+
+    return lxb_html_tree_current_node(tree);
+}
+
+lxb_inline lxb_html_element_t *
+lxb_html_tree_insert_html_element(lxb_html_tree_t *tree,
+                                  lxb_html_token_t *token)
+{
+    return lxb_html_tree_insert_foreign_element(tree, token, LXB_NS_HTML);
+}
+
+lxb_inline void
+lxb_html_tree_insert_node(lxb_dom_node_t *to, lxb_dom_node_t *node,
+                          lxb_html_tree_insertion_position_t ipos)
+{
+    if (ipos == LXB_HTML_TREE_INSERTION_POSITION_BEFORE) {
+        lxb_dom_node_insert_before(to, node);
+        return;
+    }
+
+    lxb_dom_node_insert_child(to, node);
+}
+
+/* TODO: if we not need to save parse errors?! */
+lxb_inline void
+lxb_html_tree_acknowledge_token_self_closing(lxb_html_tree_t *tree,
+                                             lxb_html_token_t *token)
+{
+    if ((token->type & LXB_HTML_TOKEN_TYPE_CLOSE_SELF) == 0) {
+        return;
+    }
+
+    bool is_void = lxb_html_tag_is_void(token->tag_id);
+
+    if (is_void) {
+        lxb_html_tree_parse_error(tree, token,
+                                  LXB_HTML_RULES_ERROR_NOVOHTELSTTAWITRSO);
+    }
+}
+
+lxb_inline bool
+lxb_html_tree_mathml_text_integration_point(lxb_dom_node_t *node)
+{
+    if (node->ns == LXB_NS_MATH) {
+        switch (node->local_name) {
+            case LXB_TAG_MI:
+            case LXB_TAG_MO:
+            case LXB_TAG_MN:
+            case LXB_TAG_MS:
+            case LXB_TAG_MTEXT:
+                return true;
+        }
+    }
+
+    return false;
+}
+
+lxb_inline bool
+lxb_html_tree_scripting(lxb_html_tree_t *tree)
+{
+    return tree->scripting;
+}
+
+lxb_inline void
+lxb_html_tree_scripting_set(lxb_html_tree_t *tree, bool scripting)
+{
+    tree->scripting = scripting;
+}
+
+lxb_inline void
+lxb_html_tree_attach_document(lxb_html_tree_t *tree, lxb_html_document_t *doc)
+{
+    tree->document = doc;
+}
+
+/*
+ * No inline functions for ABI.
+ */
+LXB_API lxb_status_t
+lxb_html_tree_begin_noi(lxb_html_tree_t *tree, lxb_html_document_t *document);
+
+LXB_API lxb_status_t
+lxb_html_tree_chunk_noi(lxb_html_tree_t *tree, const lxb_char_t *html,
+                        size_t size);
+
+LXB_API lxb_status_t
+lxb_html_tree_end_noi(lxb_html_tree_t *tree);
+
+LXB_API lxb_status_t
+lxb_html_tree_build_noi(lxb_html_tree_t *tree, lxb_html_document_t *document,
+                        const lxb_char_t *html, size_t size);
+
+LXB_API lxb_dom_node_t *
+lxb_html_tree_create_node_noi(lxb_html_tree_t *tree,
+                              lxb_tag_id_t tag_id, lxb_ns_id_t ns);
+
+LXB_API bool
+lxb_html_tree_node_is_noi(lxb_dom_node_t *node, lxb_tag_id_t tag_id);
+
+LXB_API lxb_dom_node_t *
+lxb_html_tree_current_node_noi(lxb_html_tree_t *tree);
+
+LXB_API lxb_dom_node_t *
+lxb_html_tree_adjusted_current_node_noi(lxb_html_tree_t *tree);
+
+LXB_API lxb_html_element_t *
+lxb_html_tree_insert_html_element_noi(lxb_html_tree_t *tree,
+                                      lxb_html_token_t *token);
+
+LXB_API void
+lxb_html_tree_insert_node_noi(lxb_dom_node_t *to, lxb_dom_node_t *node,
+                              lxb_html_tree_insertion_position_t ipos);
+
+LXB_API void
+lxb_html_tree_acknowledge_token_self_closing_noi(lxb_html_tree_t *tree,
+                                             lxb_html_token_t *token);
+
+LXB_API bool
+lxb_html_tree_mathml_text_integration_point_noi(lxb_dom_node_t *node);
+
+LXB_API bool
+lxb_html_tree_scripting_noi(lxb_html_tree_t *tree);
+
+LXB_API void
+lxb_html_tree_scripting_set_noi(lxb_html_tree_t *tree, bool scripting);
+
+LXB_API void
+lxb_html_tree_attach_document_noi(lxb_html_tree_t *tree,
+                                  lxb_html_document_t *doc);
+
+
+#ifdef __cplusplus
+} /* extern "C" */
+#endif
+
+#endif /* LEXBOR_HTML_TREE_H */

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

@@ -0,0 +1,111 @@
+/*
+ * Copyright (C) 2018 Alexander Borisov
+ *
+ * Author: Alexander Borisov <borisov@lexbor.com>
+ */
+
+#ifndef LEXBOR_HTML_TREE_RES_H
+#define LEXBOR_HTML_TREE_RES_H
+
+
+typedef struct {
+    const char *from;
+    const char *to;
+    size_t     len;
+}
+lxb_html_tree_res_attr_adjust_t;
+
+typedef struct {
+    const char       *name;
+    const char       *prefix;
+    const char       *local_name;
+
+    size_t           name_len;
+    size_t           prefix_len;
+
+    lxb_ns_id_t      ns;
+}
+lxb_html_tree_res_attr_adjust_foreign_t;
+
+
+static const lxb_html_tree_res_attr_adjust_t
+lxb_html_tree_res_attr_adjust_svg_map[] =
+{
+    {"attributename", "attributeName", 13},
+    {"attributetype", "attributeType", 13},
+    {"basefrequency", "baseFrequency", 13},
+    {"baseprofile", "baseProfile", 11},
+    {"calcmode", "calcMode", 8},
+    {"clippathunits", "clipPathUnits", 13},
+    {"diffuseconstant", "diffuseConstant", 15},
+    {"edgemode", "edgeMode", 8},
+    {"filterunits", "filterUnits", 11},
+    {"glyphref", "glyphRef", 8},
+    {"gradienttransform", "gradientTransform", 17},
+    {"gradientunits", "gradientUnits", 13},
+    {"kernelmatrix", "kernelMatrix", 12},
+    {"kernelunitlength", "kernelUnitLength", 16},
+    {"keypoints", "keyPoints", 9},
+    {"keysplines", "keySplines", 10},
+    {"keytimes", "keyTimes", 8},
+    {"lengthadjust", "lengthAdjust", 12},
+    {"limitingconeangle", "limitingConeAngle", 17},
+    {"markerheight", "markerHeight", 12},
+    {"markerunits", "markerUnits", 11},
+    {"markerwidth", "markerWidth", 11},
+    {"maskcontentunits", "maskContentUnits", 16},
+    {"maskunits", "maskUnits", 9},
+    {"numoctaves", "numOctaves", 10},
+    {"pathlength", "pathLength", 10},
+    {"patterncontentunits", "patternContentUnits", 19},
+    {"patterntransform", "patternTransform", 16},
+    {"patternunits", "patternUnits", 12},
+    {"pointsatx", "pointsAtX", 9},
+    {"pointsaty", "pointsAtY", 9},
+    {"pointsatz", "pointsAtZ", 9},
+    {"preservealpha", "preserveAlpha", 13},
+    {"preserveaspectratio", "preserveAspectRatio", 19},
+    {"primitiveunits", "primitiveUnits", 14},
+    {"refx", "refX", 4},
+    {"refy", "refY", 4},
+    {"repeatcount", "repeatCount", 11},
+    {"repeatdur", "repeatDur", 9},
+    {"requiredextensions", "requiredExtensions", 18},
+    {"requiredfeatures", "requiredFeatures", 16},
+    {"specularconstant", "specularConstant", 16},
+    {"specularexponent", "specularExponent", 16},
+    {"spreadmethod", "spreadMethod", 12},
+    {"startoffset", "startOffset", 11},
+    {"stddeviation", "stdDeviation", 12},
+    {"stitchtiles", "stitchTiles", 11},
+    {"surfacescale", "surfaceScale", 12},
+    {"systemlanguage", "systemLanguage", 14},
+    {"tablevalues", "tableValues", 11},
+    {"targetx", "targetX", 7},
+    {"targety", "targetY", 7},
+    {"textlength", "textLength", 10},
+    {"viewbox", "viewBox", 7},
+    {"viewtarget", "viewTarget", 10},
+    {"xchannelselector", "xChannelSelector", 16},
+    {"ychannelselector", "yChannelSelector", 16},
+    {"zoomandpan", "zoomAndPan", 10},
+};
+
+static const lxb_html_tree_res_attr_adjust_foreign_t
+lxb_html_tree_res_attr_adjust_foreign_map[] =
+{
+    {"xlink:actuate", "xlink", "actuate", 13, 5, LXB_NS_XLINK},
+    {"xlink:arcrole", "xlink", "arcrole", 13, 5, LXB_NS_XLINK},
+    {"xlink:href", "xlink", "href", 10, 5, LXB_NS_XLINK},
+    {"xlink:role", "xlink", "role", 10, 5, LXB_NS_XLINK},
+    {"xlink:show", "xlink", "show", 10, 5, LXB_NS_XLINK},
+    {"xlink:title", "xlink", "title", 11, 5, LXB_NS_XLINK},
+    {"xlink:type", "xlink", "type", 10, 5, LXB_NS_XLINK},
+    {"xml:lang", "xml", "lang", 8, 3, LXB_NS_XML},
+    {"xml:space", "xml", "space", 9, 3, LXB_NS_XML},
+    {"xmlns", "", "xmlns", 5, 0, LXB_NS_XMLNS},
+    {"xmlns:xlink", "xmlns", "xlink", 11, 5, LXB_NS_XMLNS}
+};
+
+
+#endif /* LEXBOR_HTML_TREE_RES_H */

data/vendor/lexbor/source/lexbor/ns/base.h

@@ -0,0 +1,32 @@
+/*
+ * Copyright (C) 2019 Alexander Borisov
+ *
+ * Author: Alexander Borisov <borisov@lexbor.com>
+ */
+
+#ifndef LEXBOR_NS_BASE_H
+#define LEXBOR_NS_BASE_H
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+
+#include "lexbor/core/base.h"
+
+
+#define LXB_NS_VERSION_MAJOR 1
+#define LXB_NS_VERSION_MINOR 2
+#define LXB_NS_VERSION_PATCH 0
+
+#define LXB_NS_VERSION_STRING                                                  \
+    LEXBOR_STRINGIZE(LXB_NS_VERSION_MAJOR) "."                                 \
+    LEXBOR_STRINGIZE(LXB_NS_VERSION_MINOR) "."                                 \
+    LEXBOR_STRINGIZE(LXB_NS_VERSION_PATCH)
+
+
+#ifdef __cplusplus
+} /* extern "C" */
+#endif
+
+#endif /* LEXBOR_NS_BASE_H */

data/vendor/lexbor/source/lexbor/ns/config.cmake

@@ -0,0 +1,2 @@
+set(DEPENDENCIES "core")
+set(DESCRIPTION "DOM/HTML namespace module. It is helper module for parsing HTML.")

data/vendor/lexbor/source/lexbor/ns/const.h

@@ -0,0 +1,37 @@
+/*
+ * Copyright (C) 2018 Alexander Borisov
+ *
+ * Author: Alexander Borisov <borisov@lexbor.com>
+ */
+
+/*
+ * Caution!
+ * This file generated by the script "utils/lexbor/tag_ns/tags.py"!
+ * Do not change this file!
+ */
+
+
+#ifndef LXB_NS_CONST_H
+#define LXB_NS_CONST_H
+
+#define LXB_NS_CONST_VERSION "253D4AFDA959234B48A478B956C3C777"
+#define LXB_NS_CONST_VERSION_253D4AFDA959234B48A478B956C3C777
+
+typedef uintptr_t lxb_ns_id_t;
+typedef uintptr_t lxb_ns_prefix_id_t;
+
+typedef enum {
+    LXB_NS__UNDEF      = 0x00,
+    LXB_NS__ANY        = 0x01,
+    LXB_NS_HTML        = 0x02,
+    LXB_NS_MATH        = 0x03,
+    LXB_NS_SVG         = 0x04,
+    LXB_NS_XLINK       = 0x05,
+    LXB_NS_XML         = 0x06,
+    LXB_NS_XMLNS       = 0x07,
+    LXB_NS__LAST_ENTRY = 0x08
+}
+lxb_ns_id_enum_t;
+
+
+#endif /* LXB_NS_CONST_H */