initial table support

CMakeLists.txt

@@ -32,9 +32,9 @@ find_package(chmlib CONFIG REQUIRED)
 find_package(lexbor CONFIG REQUIRED)
 find_package(maddy CONFIG REQUIRED)
 find_package(nlohmann_json CONFIG REQUIRED)
-find_package(pdfium CONFIG REQUIRED)
+find_package(pdfium CONFIG REQUIRED )
+find_package(wxWidgets CONFIG REQUIRED COMPONENTS webview)
 find_package(Poco REQUIRED COMPONENTS Foundation XML)
-find_package(wxWidgets CONFIG REQUIRED)
 find_package(Gettext)
 
 if(GETTEXT_FOUND)
@@ -95,8 +95,11 @@ target_link_libraries(paperback PRIVATE
 	wx::base
 	wx::core
 	wx::net
+	wx::webview
 )
-
+if(WIN32)
+	target_link_libraries(paperback PRIVATE "${CMAKE_BINARY_DIR}/vcpkg_installed/${VCPKG_TARGET_TRIPLET}/lib/WebView2LoaderStatic.lib")
+endif()
 if(WIN32)
 	file(GLOB DLLS "${VCPKG_INSTALLED_DIR}/${VCPKG_TARGET_TRIPLET}/bin/*.dll")
 	set(LOCAL_DLLS "")
@@ -157,4 +160,4 @@ if(WIN32)
 	endif()
 else()
 	add_custom_target(release DEPENDS package)
-endif()
+endif()

src/constants.hpp

@@ -73,6 +73,8 @@ enum {
 	ID_NEXT_LINK,
 	ID_PREVIOUS_LINK,
 	ID_ACTIVATE_LINK,
+	ID_PREVIOUS_TABLE,
+	ID_NEXT_TABLE,
 	ID_PREVIOUS_LIST,
 	ID_NEXT_LIST,
 	ID_PREVIOUS_LIST_ITEM,

src/document.hpp

@@ -35,6 +35,12 @@ struct heading_info {
 	std::string text;
 };
 
+struct table_info {
+	size_t offset;
+	std::string text;
+	std::string ref;
+};
+
 struct document_stats {
 	size_t word_count{0};
 	size_t line_count{0};
@@ -48,6 +54,7 @@ struct document {
 	document_buffer buffer;
 	std::vector<std::unique_ptr<toc_item>> toc_items;
 	std::map<std::string, size_t> id_positions;
+	std::map<size_t, wxString> html_tables;
 	std::vector<std::string> spine_items;
 	std::map<std::string, std::string> manifest_items;
 	mutable document_stats stats;

src/document_buffer.cpp

@@ -58,10 +58,16 @@ void document_buffer::add_link(int pos, const wxString& text, const wxString& re
 	markers.emplace_back(pos, marker_type::link, text, ref, 0);
 }
 
+
 void document_buffer::finalize_markers() {
 	sort_markers();
 }
 
+void document_buffer::add_table(int pos, const wxString& text, const wxString& ref) {
+	markers.emplace_back(pos, marker_type::table, text, ref, 0);
+	sort_markers();
+}
+
 void document_buffer::add_marker(int pos, marker_type type, const wxString& text, const wxString& ref, int level) {
 	if (is_heading_marker(type) && level == 0) {
 		level = heading_level_from_type(type);

src/document_buffer.hpp

@@ -23,6 +23,7 @@ enum class marker_type {
 	section_break,
 	toc_item,
 	link,
+	table,
 	list,
 	list_item
 };
@@ -67,6 +68,7 @@ class document_buffer {
 	void add_section_break(const wxString& label = wxString());
 	void add_toc_marker(const wxString& text, const wxString& ref = wxString());
 	void add_link(int pos, const wxString& text, const wxString& ref);
+	void add_table(int pos, const wxString& text, const wxString& ref);
 	void add_marker(int pos, marker_type type, const wxString& text = wxString(), const wxString& ref = wxString(), int level = 0);
 	void finalize_markers();
 	void clear();

src/document_manager.cpp

@@ -13,6 +13,7 @@
 #include "dialogs.hpp"
 #include "document.hpp"
 #include "document_buffer.hpp"
+#include "table_dialog.hpp"
 #include "main_window.hpp"
 #include "parser.hpp"
 #include "utils.hpp"
@@ -441,6 +442,54 @@ void document_manager::go_to_next_link() const {
 	navigate_to_link(true);
 }
 
+
+
+void document_manager::go_to_previous_table() {
+	document* doc = get_active_document();
+	wxTextCtrl* text_ctrl = get_active_text_ctrl();
+	if (!doc || !text_ctrl) {
+		return;
+	}
+	if (doc->buffer.count_markers_by_type(marker_type::table) == 0) {
+		speak(_("No tables."));
+		return;
+	}
+	int current_pos = text_ctrl->GetInsertionPoint();
+	int prev_index = doc->buffer.previous_marker_index(current_pos, marker_type::table);
+	if (prev_index == -1) {
+		speak(_("No previous table."));
+		return;
+	}
+	const marker* table_marker = doc->buffer.get_marker(prev_index);
+	if (table_marker) {
+		go_to_position(table_marker->pos);
+		speak(table_marker->text);
+	}
+}
+
+void document_manager::go_to_next_table() {
+	document* doc = get_active_document();
+	wxTextCtrl* text_ctrl = get_active_text_ctrl();
+	if (!doc || !text_ctrl) {
+		return;
+	}
+	if (doc->buffer.count_markers_by_type(marker_type::table) == 0) {
+		speak(_("No tables."));
+		return;
+	}
+	int current_pos = text_ctrl->GetInsertionPoint();
+	int next_index = doc->buffer.next_marker_index(current_pos, marker_type::table);
+	if (next_index == -1) {
+		speak(_("No next table."));
+		return;
+	}
+	const marker* table_marker = doc->buffer.get_marker(next_index);
+	if (table_marker) {
+		go_to_position(table_marker->pos);
+		speak(table_marker->text);
+	}
+}
+
 void document_manager::activate_current_link() const {
 	document* doc = get_active_document();
 	const wxTextCtrl* text_ctrl = get_active_text_ctrl();
@@ -522,6 +571,28 @@ void document_manager::activate_current_link() const {
 	}
 }
 
+void document_manager::activate_current_table() {
+	document* doc = get_active_document();
+	wxTextCtrl* text_ctrl = get_active_text_ctrl();
+	if (!doc || !text_ctrl) {
+		return;
+	}
+	int current_pos = text_ctrl->GetInsertionPoint();
+	int table_index = doc->buffer.current_marker_index(current_pos, marker_type::table);
+	if (table_index == -1) {
+		return;
+	}
+	const marker* table_marker = doc->buffer.get_marker(table_index);
+	if (!table_marker) {
+		return;
+	}
+	if (current_pos < table_marker->pos || current_pos > (table_marker->pos + table_marker->text.length())) {
+		return;
+	}
+	table_dialog dlg(&main_win, _("Table"), table_marker->ref);
+	dlg.ShowModal();
+}
+
 void document_manager::navigate_to_list(bool next) const {
 	const document* doc = get_active_document();
 	wxTextCtrl* text_ctrl = get_active_text_ctrl();
@@ -665,7 +736,7 @@ void document_manager::toggle_bookmark() const {
 	std::vector<bookmark> bookmarks = config.get_bookmarks(tab->file_path);
 	bookmark to_toggle(bookmark_start, bookmark_end);
 	bool was_bookmarked = false;
-	for (const auto& bm : bookmarks) {
+    	for (const auto& bm : bookmarks) {
 		if (bm == to_toggle) {
 			was_bookmarked = true;
 			break;

src/document_manager.hpp

@@ -77,6 +77,9 @@ class document_manager {
 	void go_to_next_bookmark() const;
 	void go_to_previous_link() const;
 	void go_to_next_link() const;
+	void go_to_previous_table();
+	void go_to_next_table();
+	void activate_current_table();
 	void go_to_previous_list() const;
 	void go_to_next_list() const;
 	void go_to_previous_list_item() const;

src/docx_parser.cpp

@@ -129,8 +129,13 @@ void docx_parser::traverse(Node* node, wxString& text, std::vector<heading_info>
 		}
 		if (local_name == "p") {
 			process_paragraph(element, text, headings, doc, rels);
-			return; // process_paragraph handles its children
-		}
+			return; 
+		} else if (local_name == "tbl") {
+            auto table_data = process_table(element);
+            doc->buffer.add_table(text.length(), table_data.first, table_data.second);
+            text += table_data.first + "\n";
+            return;
+        }
 	}
 	Node* child = node->firstChild();
 	while (child != nullptr) {
@@ -139,6 +144,67 @@ void docx_parser::traverse(Node* node, wxString& text, std::vector<heading_info>
 	}
 }
 
+std::pair<wxString, wxString> docx_parser::process_table(Element* table_element) {
+    wxString html = "<table border=\"1\" style=\"border-collapse: collapse; width: 100%;\">";
+    wxString placeholder = "table: ";
+
+    NodeList* rows = table_element->getElementsByTagNameNS(WORDML_NS, "tr");
+    for (unsigned long i = 0; i < rows->length(); ++i) {
+        html += "<tr>";
+        auto* row = dynamic_cast<Element*>(rows->item(i));
+        NodeList* cells = row->getElementsByTagNameNS(WORDML_NS, "tc");
+        for (unsigned long j = 0; j < cells->length(); ++j) {
+            auto* cell = dynamic_cast<Element*>(cells->item(j));
+            wxString cell_text = get_cell_text(cell);
+
+            if (i == 0) {
+                placeholder += cell_text + " ";
+            }
+
+            wxString style;
+            Element* tcPr = cell->getChildElementNS(WORDML_NS, "tcPr");
+            if (tcPr) {
+                Element* tcW = tcPr->getChildElementNS(WORDML_NS, "tcW");
+                if (tcW) {
+                    std::string width_str = tcW->getAttributeNS(WORDML_NS, "w");
+                    if (!width_str.empty()) {
+                        try {
+                            double width = std::stod(width_str);
+                            style += wxString::Format("width: %.2fpt;", width / 20.0);
+                        } catch (...) {
+                            // Ignore invalid width values
+                        }
+                    }
+                }
+            }
+
+            html += wxString::Format("<td style=\"%s\">%s</td>", style, cell_text);
+        }
+        html += "</tr>";
+    }
+    html += "</table>";
+
+    placeholder.Trim();
+    return {placeholder, html};
+}
+
+wxString docx_parser::get_cell_text(Element* cell_element) {
+    wxString cell_text;
+    NodeList* paragraphs = cell_element->getElementsByTagNameNS(WORDML_NS, "p");
+    for (unsigned long i = 0; i < paragraphs->length(); ++i) {
+        auto* p_element = dynamic_cast<Element*>(paragraphs->item(i));
+        NodeList* runs = p_element->getElementsByTagNameNS(WORDML_NS, "r");
+        for (unsigned long j = 0; j < runs->length(); ++j) {
+            auto* run_element = dynamic_cast<Element*>(runs->item(j));
+            cell_text += wxString::FromUTF8(get_run_text(run_element));
+        }
+        if (i < paragraphs->length() - 1) {
+            cell_text += "\n";
+        }
+    }
+    return cell_text;
+}
+
 void docx_parser::process_paragraph(Element* element, wxString& text, std::vector<heading_info>& headings, document* doc, const std::map<std::string, std::string>& rels) {
 	wxString paragraph_text;
 	int heading_level = 0;

src/docx_parser.hpp

@@ -44,6 +44,8 @@ class docx_parser : public parser {
 	static void process_hyperlink(Poco::XML::Element* element, wxString& text, document* doc, const std::map<std::string, std::string>& rels, size_t paragraph_start_offset);
 	static int get_heading_level(Poco::XML::Element* pr_element);
 	static std::string get_run_text(Poco::XML::Element* prun_element);
+	static std::pair<wxString, wxString> process_table(Poco::XML::Element* table_element);
+	static wxString get_cell_text(Poco::XML::Element* cell_element);
 	static std::string parse_hyperlink_instruction(const std::string& instruction);
 };
 

src/epub_parser.cpp

@@ -193,6 +193,7 @@ void epub_parser::process_section_content(conv& converter, const std::string& co
 		const auto& text = converter.get_text();
 		const auto& headings = converter.get_headings();
 		const auto& links = converter.get_links();
+		const auto& tables = converter.get_tables();
 		const auto& lists = converter.get_lists();
 		const auto& list_items = converter.get_list_items();
 		const auto& id_positions = converter.get_id_positions();
@@ -221,6 +222,9 @@ void epub_parser::process_section_content(conv& converter, const std::string& co
 			}
 			buffer.add_link(section_start + link.offset, wxString::FromUTF8(link.text), resolved_href);
 		}
+		for (const auto& table : tables) {
+			buffer.add_table(section_start + table.offset, wxString::FromUTF8(table.text), wxString::FromUTF8(table.ref));
+		}
 		for (const auto& list : lists) {
 			buffer.add_marker(section_start + list.offset, marker_type::list, wxString(), wxString(), list.item_count);
 		}

src/html_parser.cpp

@@ -55,6 +55,10 @@ std::unique_ptr<document> html_parser::load(const wxString& path) const {
 	for (const auto& link : links) {
 		doc->buffer.add_link(link.offset, wxString::FromUTF8(link.text), wxString::FromUTF8(link.ref));
 	}
+	const auto& tables = converter.get_tables();
+	for (const auto& table : tables) {
+		doc->buffer.add_table(table.offset, wxString::FromUTF8(table.text), wxString::FromUTF8(table.ref));
+	}
 	for (const auto& list : lists) {
 		doc->buffer.add_marker(list.offset, marker_type::list, wxString(), wxString(), list.item_count);
 	}