note2any/src/core/content-processor.ts

/**
 * 文件：content-processor.ts
 * 作用：内容处理器，负责处理markdown内容的各种转换
 */

import { TFile, App } from 'obsidian';
import { ErrorHandler } from './error-handler';

export interface ProcessorOptions {
    enableImageToBase64?: boolean;
    enableLinkProcessing?: boolean;
    enableCodeHighlight?: boolean;
    enableMathProcessing?: boolean;
    platform?: string;
}

export class ContentProcessor {
    private app: App;

    constructor(app: App) {
        this.app = app;
    }

    /**
     * 处理图片链接，转换为base64或平台URL
     */
    async processImages(
        content: string, 
        file: TFile, 
        options: ProcessorOptions = {}
    ): Promise<string> {
        return await ErrorHandler.withErrorHandling(async () => {
            const { enableImageToBase64 = true } = options;
            
            if (!enableImageToBase64) {
                return content;
            }

            // WikiLink 图片处理: ![[image.png]]
            content = await this.processWikiLinkImages(content, file);
            
            // Markdown 图片处理: ![alt](image.png)
            content = await this.processMarkdownImages(content, file);
            
            return content;
        }, 'ContentProcessor.processImages', content) || content;
    }

    /**
     * 处理链接
     */
    processLinks(content: string, linkStyle: 'inline' | 'footnote' = 'inline'): string {
        return ErrorHandler.withErrorHandlingSync(() => {
            if (linkStyle === 'footnote') {
                return this.convertLinksToFootnotes(content);
            }
            return this.processInlineLinks(content);
        }, 'ContentProcessor.processLinks', content) || content;
    }

    /**
     * 处理代码块高亮
     */
    processCodeBlocks(content: string, highlightTheme: string = 'default'): string {
        return ErrorHandler.withErrorHandlingSync(() => {
            // 为代码块添加语法高亮类
            return content.replace(
                /```(\w+)?\n([\s\S]*?)```/g,
                (match, lang, code) => {
                    const language = lang || 'text';
                    return `<div class="code-section">
                        <pre><code class="language-${language}">${this.escapeHtml(code.trim())}</code></pre>
                    </div>`;
                }
            );
        }, 'ContentProcessor.processCodeBlocks', content) || content;
    }

    /**
     * 处理数学公式
     */
    processMath(content: string, mathEngine: 'latex' | 'asciimath' = 'latex'): string {
        return ErrorHandler.withErrorHandlingSync(() => {
            // 行内公式: $...$
            content = content.replace(/\$([^$]+)\$/g, (match, formula) => {
                return `<span class="math-inline" data-engine="${mathEngine}">${formula}</span>`;
            });

            // 块级公式: $$...$$
            content = content.replace(/\$\$([^$]+)\$\$/g, (match, formula) => {
                return `<div class="math-block" data-engine="${mathEngine}">${formula}</div>`;
            });

            return content;
        }, 'ContentProcessor.processMath', content) || content;
    }

    /**
     * 处理Gallery短代码
     */
    async processGalleryShortcode(content: string, galleryPath: string, numPics: number = 2): Promise<string> {
        return await ErrorHandler.withErrorHandling(async () => {
            const galleryRegex = /{{<gallery\s+dir="([^"]+)"(?:\s+figcaption="([^"]*)")?(?:\s+mppickall=(?:"(1|0)"|'(1|0)'|(1|0)))?\s*\/?>}}\s*{{<load-photoswipe>}}/g;
            
            return content.replace(galleryRegex, (match, dir, figcaption, q1, q2, unquoted) => {
                const pickAll = q1 === '1' || q2 === '1' || unquoted === '1';
                const maxPics = pickAll ? 999 : numPics;
                
                // 这里应该调用实际的图片列表获取逻辑
                // 为了简化，返回占位符
                return `<!-- Gallery: ${dir}, max: ${maxPics}, caption: ${figcaption || ''} -->`;
            });
        }, 'ContentProcessor.processGalleryShortcode', content) || content;
    }

    /**
     * 清理HTML标签
     */
    sanitizeHtml(content: string, allowedTags: string[] = []): string {
        return ErrorHandler.withErrorHandlingSync(() => {
            const allowedTagsSet = new Set(allowedTags);
            
            return content.replace(/<[^>]*>/g, (tag) => {
                const tagName = tag.match(/<\/?(\w+)/)?.[1]?.toLowerCase();
                if (tagName && allowedTagsSet.has(tagName)) {
                    return tag;
                }
                return '';
            });
        }, 'ContentProcessor.sanitizeHtml', content) || content;
    }

    /**
     * 处理自定义语法扩展
     */
    processCustomSyntax(content: string): string {
        return ErrorHandler.withErrorHandlingSync(() => {
            // 斜体标注: [fig 一段说明 /]
            content = content.replace(/\[fig\s+([^/]+)\s+\/\]/g, 
                '<span style="font-style:italic;color:#666;font-size:0.9em;">$1</span>');

            // 彩色提示块
            content = content.replace(/^\|\|([rgby]?)\s+(.+)$/gm, (match, color, text) => {
                const colorMap: Record<string, string> = {
                    'r': 'background:#8B4513;color:white',
                    'g': 'background:#9ACD32;color:black', 
                    'b': 'background:#D3D3D3;color:black',
                    'y': 'background:#FFFF99;color:black',
                    '': 'background:#F5F5F5;color:black'
                };
                const style = colorMap[color] || colorMap[''];
                return `<div style="padding:8px;margin:4px 0;border-radius:4px;${style}">${text}</div>`;
            });

            return content;
        }, 'ContentProcessor.processCustomSyntax', content) || content;
    }

    // 私有辅助方法

    private async processWikiLinkImages(content: string, file: TFile): Promise<string> {
        const wikiImageRegex = /!\[\[([^\]]+)\]\]/g;
        
        return content.replace(wikiImageRegex, (match, imagePath) => {
            // 这里应该实现实际的图片处理逻辑
            return `<img src="data:image/png;base64,placeholder" alt="${imagePath}">`;
        });
    }

    private async processMarkdownImages(content: string, file: TFile): Promise<string> {
        const markdownImageRegex = /!\[([^\]]*)\]\(([^)]+)\)/g;
        
        return content.replace(markdownImageRegex, (match, alt, src) => {
            // 这里应该实现实际的图片处理逻辑
            return `<img src="data:image/png;base64,placeholder" alt="${alt}">`;
        });
    }

    private convertLinksToFootnotes(content: string): string {
        const links: string[] = [];
        
        // 提取所有链接
        content = content.replace(/\[([^\]]+)\]\(([^)]+)\)/g, (match, text, url) => {
            links.push(url);
            return `${text}[${links.length}]`;
        });

        // 添加脚注
        if (links.length > 0) {
            content += '\n\n---\n\n';
            links.forEach((url, index) => {
                content += `[${index + 1}]: ${url}\n`;
            });
        }

        return content;
    }

    private processInlineLinks(content: string): string {
        return content.replace(/\[([^\]]+)\]\(([^)]+)\)/g, 
            '<a href="$2" style="color:#1e6bb8;text-decoration:none;">$1</a>');
    }

    private escapeHtml(text: string): string {
        const div = document.createElement('div');
        div.textContent = text;
        return div.innerHTML;
    }
}