65 lines
2.0 KiB
C++
65 lines
2.0 KiB
C++
#pragma once
|
|
|
|
#include <Epub/blocks/TextBlock.h>
|
|
|
|
#include <functional>
|
|
#include <memory>
|
|
#include <string>
|
|
|
|
class GfxRenderer;
|
|
|
|
/**
|
|
* DictHtmlParser parses HTML dictionary definitions into ParsedText.
|
|
*
|
|
* Supports:
|
|
* - Bold: <b>, <strong>
|
|
* - Italic: <i>, <em>
|
|
* - Underline: <u>, <ins>
|
|
* - Lists: <ol>, <li> with numbering/bullets
|
|
* - Block elements: <p>, <br>, <hr>, </html> (entry separator)
|
|
* - HTML entities: numeric (&#NNN;, &#xHHH;) and named (&, etc.)
|
|
* - Superscript: <sup> rendered as ^text
|
|
*/
|
|
class DictHtmlParser {
|
|
public:
|
|
/**
|
|
* Parse HTML definition and populate ParsedText with styled words.
|
|
* Each paragraph/block creates a separate ParsedText via the callback.
|
|
*
|
|
* @param html The HTML definition text
|
|
* @param fontId Font ID for text width calculations
|
|
* @param renderer Reference to renderer for layout
|
|
* @param onParagraph Callback invoked for each paragraph/block of text
|
|
*/
|
|
static void parse(const std::string& html, int fontId, const GfxRenderer& renderer, uint16_t viewportWidth,
|
|
const std::function<void(std::shared_ptr<TextBlock>)>& onTextBlock);
|
|
|
|
private:
|
|
// Decode HTML entity at position i (starting with '&')
|
|
static std::string decodeEntity(const std::string& html, size_t& i);
|
|
|
|
// Extract tag name from position (after '<')
|
|
static std::string extractTagName(const std::string& html, size_t start, bool& isClosing);
|
|
|
|
// Check if tag is a block-level element
|
|
static bool isBlockTag(const std::string& tagName);
|
|
|
|
// Check if tag starts/ends bold
|
|
static bool isBoldTag(const std::string& tagName);
|
|
|
|
// Check if tag starts/ends italic
|
|
static bool isItalicTag(const std::string& tagName);
|
|
|
|
// Check if tag starts/ends underline
|
|
static bool isUnderlineTag(const std::string& tagName);
|
|
|
|
// Check if tag is superscript
|
|
static bool isSuperscriptTag(const std::string& tagName);
|
|
|
|
// Check if tag is list item
|
|
static bool isListItemTag(const std::string& tagName);
|
|
|
|
// Check if tag starts ordered list
|
|
static bool isOrderedListTag(const std::string& tagName);
|
|
};
|