@bytechain.cn/colamd-themes 0.3.0

package/dist/extractors/docx-extractor.js

@@ -0,0 +1,370 @@
+/**
+ * DOCX 文档样式提取器 — v2.0
+ *
+ * 增强功能：
+ * 1. 提取各级标题（H1-H6）的完整字体样式定义
+ * 2. 提取正文字体、字号、颜色、行间距
+ * 3. 提取表格样式（表头/单元格字体、背景色、边框）
+ * 4. 对无法提取的样式应用 DocxStyleParadigm 标准参考值
+ * 5. 自动验证 WCAG 对比度合规性
+ */
+import { readFileSync } from "node:fs";
+import { basename } from "node:path";
+import mammoth from "mammoth";
+import { makeThemeStyle, } from "../models.js";
+import { normalizeThemeStyle } from "../color-utils.js";
+import { getRecommendedPreset, validateDocumentStyle, } from "../docx-style-paradigm.js";
+export class DocxExtractor {
+    canHandle(source) {
+        return /\.docx$/i.test(source);
+    }
+    async extract(source) {
+        const buffer = readFileSync(source);
+        const name = basename(source).replace(/\.docx$/i, "");
+        const theme = makeThemeStyle(name, "docx", source);
+        const result = await mammoth.convertToHtml({ buffer }, {
+            styleMap: [
+                "p[style-name='Heading 1'] => h1:fresh",
+                "p[style-name='Heading 2'] => h2:fresh",
+                "p[style-name='Heading 3'] => h3:fresh",
+                "p[style-name='Heading 4'] => h4:fresh",
+                "p[style-name='Heading 5'] => h5:fresh",
+                "p[style-name='Heading 6'] => h6:fresh",
+                "r[style-name='Strong'] => strong",
+            ],
+        });
+        if (result.value) {
+            this.extractDetailedStyles(theme, result.value);
+        }
+        this.applyParadigmDefaults(theme);
+        normalizeThemeStyle(theme);
+        return theme;
+    }
+    /**
+     * 从 HTML 输出中提取详细的文档元素样式
+     */
+    extractDetailedStyles(theme, html) {
+        const p = theme.palette;
+        const t = theme.typography;
+        this.extractBodyStyles(t, html);
+        this.extractHeadingStyles(theme, html);
+        this.extractTableStyles(p, html);
+        this.extractColorPalette(p, html);
+        this.extractBlockStyles(p, html);
+    }
+    /**
+     * 提取正文基础样式（字体、字号、行高）
+     */
+    extractBodyStyles(typography, html) {
+        const fontMatch = html.match(/font-family:\s*['"]?([^;'"]+)['"]?/i);
+        if (fontMatch) {
+            typography.bodyFontFamily = fontMatch[1].trim();
+        }
+        const fontSizeMatch = html.match(/font-size:\s*([\d.]+)(pt|px)/i);
+        if (fontSizeMatch) {
+            const size = parseFloat(fontSizeMatch[1]);
+            const unit = fontSizeMatch[2];
+            typography.baseFontSize = unit === "pt" ? `${size * 1.333}px` : `${size}px`;
+        }
+    }
+    /**
+     * 提取各级标题的详细样式
+     */
+    extractHeadingStyles(theme, html) {
+        const headingLevels = ["h1", "h2", "h3", "h4", "h5", "h6"];
+        for (const level of headingLevels) {
+            const regex = new RegExp(`<${level}[^>]*style="[^"]*"`, "gi");
+            let match = regex.exec(html);
+            if (match) {
+                const styleStr = match[0];
+                const extracted = this.parseInlineStyle(styleStr);
+                if (extracted.fontSize) {
+                    const size = parseFloat(extracted.fontSize);
+                    const unit = extracted.fontSize.replace(/[\d.]/g, "") || "px";
+                    const pxSize = unit === "pt" ? size * 1.333 : size;
+                    switch (level) {
+                        case "h1":
+                            theme.seedTypography.fontScaleH1 = pxSize / 16;
+                            break;
+                        case "h2":
+                            theme.seedTypography.fontScaleH2 = pxSize / 16;
+                            break;
+                        case "h3":
+                            theme.seedTypography.fontScaleH3 = pxSize / 16;
+                            break;
+                        case "h4":
+                            theme.seedTypography.fontScaleH4 = pxSize / 16;
+                            break;
+                        case "h5":
+                            theme.seedTypography.fontScaleH5 = pxSize / 16;
+                            break;
+                        case "h6":
+                            theme.seedTypography.fontScaleH6 = pxSize / 16;
+                            break;
+                    }
+                }
+                if (extracted.fontWeight) {
+                    const weight = parseInt(extracted.fontWeight) || 700;
+                    if (weight > theme.typography.headingWeight) {
+                        theme.typography.headingWeight = weight;
+                    }
+                }
+                if (extracted.color && level === "h1") {
+                    theme.palette.headingColor = extracted.color;
+                }
+            }
+        }
+    }
+    /**
+     * 提取表格详细样式（表头、单元格、边框）
+     */
+    extractTableStyles(palette, html) {
+        const thRegex = /<th[^>]*style="([^"]*)"/gi;
+        let thMatch = thRegex.exec(html);
+        if (thMatch) {
+            const thStyle = this.parseInlineStyle(thMatch[0]);
+            if (thStyle.backgroundColor) {
+                palette.tableHeaderBg = thStyle.backgroundColor;
+            }
+            if (thStyle.color) {
+                palette.headingColor = palette.headingColor || thStyle.color;
+            }
+            if (thStyle.fontSize) {
+                const size = parseFloat(thStyle.fontSize);
+                palette.tableHeaderBg = palette.tableHeaderBg || "#f0f0f0";
+            }
+        }
+        const tdRegex = /<td[^>]*style="([^"]*)"/gi;
+        let tdMatch = tdRegex.exec(html);
+        if (tdMatch) {
+            const tdStyle = this.parseInlineStyle(tdMatch[0]);
+            if (tdStyle.backgroundColor) {
+                palette.codeBg = tdStyle.backgroundColor;
+            }
+        }
+        const tableBorderMatch = html.match(/<table[^>]*style="[^"]*border[^;]*:[^;]*([^;"'\s]+)/i);
+        if (tableBorderMatch) {
+            palette.borderColor = palette.borderColor || "#cccccc";
+        }
+        if (!palette.tableHeaderBg) {
+            palette.tableHeaderBg = "#f0f0f0";
+        }
+    }
+    /**
+     * 提取全局色彩方案
+     */
+    extractColorPalette(palette, html) {
+        const colorMatch = html.match(/color:\s*(#[0-9a-fA-F]{3,6})/i);
+        if (colorMatch) {
+            palette.textColor = colorMatch[1];
+        }
+        const bgMatch = html.match(/background(?:-color)?:\s*(#[0-9a-fA-F]{3,6})/i);
+        if (bgMatch && bgMatch[1].toLowerCase() !== "#ffffff") {
+            palette.bgColor = bgMatch[1];
+        }
+        const strongMatch = html.match(/<strong[^>]*style="[^"]*color:\s*(#[0-9a-fA-F]{3,6})/i);
+        if (strongMatch) {
+            palette.accentColor = strongMatch[1];
+        }
+        const linkMatch = html.match(/<a[^>]*style="[^"]*color:\s*(#[0-9a-fA-F]{3,6})/i);
+        if (linkMatch) {
+            palette.accentColor = linkMatch[1];
+        }
+    }
+    /**
+     * 提取引用块和代码块样式
+     */
+    extractBlockStyles(palette, html) {
+        const preMatch = html.match(/<pre[^>]*style="[^"]*background(?:-color)?:\s*(#[0-9a-fA-F]{3,6})/i);
+        if (preMatch) {
+            palette.codeBlockBg = preMatch[1];
+        }
+        const preColorMatch = html.match(/<pre[^>]*style="[^"]*color:\s*(#[0-9a-fA-F]{3,6})/i);
+        if (preColorMatch) {
+            palette.codeBlockText = preColorMatch[1];
+        }
+        if (!palette.blockquoteBorder) {
+            palette.blockquoteBorder = palette.accentColor;
+        }
+    }
+    /**
+     * 解析内联 style 字符串为键值对对象
+     */
+    parseInlineStyle(styleTag) {
+        const styleMatch = styleTag.match(/style="([^"]*)"/i);
+        if (!styleMatch)
+            return {};
+        const styles = {};
+        const declarations = styleMatch[1].split(";");
+        for (const decl of declarations) {
+            const [prop, ...valueParts] = decl.split(":").map((s) => s.trim());
+            if (prop && valueParts.length > 0) {
+                const camelProp = prop.replace(/-([a-z])/g, (_m, c) => c.toUpperCase());
+                styles[camelProp] = valueParts.join(":").trim();
+            }
+        }
+        return styles;
+    }
+    /**
+     * 应用 DocxStyleParadigm 标准参考值填充缺失字段
+     */
+    applyParadigmDefaults(theme) {
+        const preset = getRecommendedPreset();
+        const p = theme.palette;
+        const t = theme.typography;
+        const st = theme.seedTypography;
+        if (!p.textColor || p.textColor === "#000000") {
+            p.textColor = preset.body.color;
+        }
+        if (!p.headingColor) {
+            p.headingColor = preset.headings.h1.color;
+        }
+        if (!p.bgColor || p.bgColor === "#ffffff") {
+            p.bgColor = preset.pageBackground;
+        }
+        if (!p.textMuted) {
+            p.textMuted = preset.mutedTextColor;
+        }
+        if (!p.accentColor || p.accentColor === "#0000ff") {
+            p.accentColor = preset.accentColor;
+        }
+        if (!p.tableHeaderBg) {
+            p.tableHeaderBg = preset.table.headerBgColor;
+        }
+        if (!p.codeBg) {
+            p.codeBg = preset.inlineCode.bgColor;
+        }
+        if (!p.codeBlockBg) {
+            p.codeBlockBg = preset.codeBlock.bgColor;
+        }
+        if (!p.blockquoteBg) {
+            p.blockquoteBg = preset.blockquote.bgColor;
+        }
+        if (!p.selectionBg) {
+            p.selectionBg = preset.selectionBg;
+        }
+        if (!t.bodyFontFamily || t.bodyFontFamily === "Georgia, serif") {
+            t.bodyFontFamily = preset.body.fontFamily;
+        }
+        if (!t.headingFontFamily) {
+            t.headingFontFamily = preset.headings.h1.fontFamily;
+        }
+        if (!t.baseFontSize) {
+            t.baseFontSize = `${preset.body.fontSize}px`;
+        }
+        if (t.baseLineHeight < 1.5) {
+            t.baseLineHeight = preset.body.lineHeight;
+        }
+        st.fontBody = preset.body.fontFamily;
+        st.fontHeading = preset.headings.h1.fontFamily;
+        st.lineHeightBody = preset.body.lineHeight;
+        st.lineHeightHeading = preset.headings.h1.lineHeight;
+        const issues = validateDocumentStyle(this.themeToParadigm(theme));
+        for (const issue of issues) {
+            if (issue.level === "error") {
+                console.warn(`[DocxExtractor] ${issue.element}: ${issue.message}`);
+            }
+        }
+    }
+    /**
+     * 将 ThemeStyle 转换为 DocumentStyleParadigm 用于验证
+     */
+    themeToParadigm(theme) {
+        return {
+            pageBackground: theme.palette.bgColor,
+            headings: {
+                h1: {
+                    fontFamily: theme.seedTypography.fontHeading,
+                    fontSize: 16 * theme.seedTypography.fontScaleH1,
+                    fontWeight: theme.typography.headingWeight,
+                    color: theme.palette.headingColor,
+                    lineHeight: theme.seedTypography.lineHeightHeading,
+                },
+                h2: {
+                    fontFamily: theme.seedTypography.fontHeading,
+                    fontSize: 16 * theme.seedTypography.fontScaleH2,
+                    fontWeight: theme.typography.headingWeight - 100 || 600,
+                    color: theme.palette.headingColor,
+                    lineHeight: theme.seedTypography.lineHeightHeading,
+                },
+                h3: {
+                    fontFamily: theme.seedTypography.fontHeading,
+                    fontSize: 16 * theme.seedTypography.fontScaleH3,
+                    fontWeight: 600,
+                    color: theme.palette.headingColor,
+                    lineHeight: theme.seedTypography.lineHeightHeading + 0.05,
+                },
+                h4: {
+                    fontFamily: theme.seedTypography.fontHeading,
+                    fontSize: 16 * theme.seedTypography.fontScaleH4,
+                    fontWeight: 600,
+                    color: theme.palette.textMuted,
+                    lineHeight: theme.seedTypography.lineHeightHeading + 0.05,
+                },
+                h5: {
+                    fontFamily: theme.seedTypography.fontHeading,
+                    fontSize: 16 * theme.seedTypography.fontScaleH5,
+                    fontWeight: 500,
+                    color: theme.palette.textMuted,
+                    lineHeight: theme.seedTypography.lineHeightBody - 0.25,
+                },
+                h6: {
+                    fontFamily: theme.seedTypography.fontHeading,
+                    fontSize: 16 * theme.seedTypography.fontScaleH6,
+                    fontWeight: 500,
+                    color: theme.palette.textMuted,
+                    lineHeight: theme.seedTypography.lineHeightBody - 0.2,
+                },
+            },
+            body: {
+                fontFamily: theme.seedTypography.fontBody,
+                fontSize: parseInt(theme.typography.baseFontSize) || 16,
+                fontWeight: 400,
+                color: theme.palette.textColor,
+                lineHeight: theme.typography.baseLineHeight,
+            },
+            blockquote: {
+                fontFamily: theme.seedTypography.fontBody,
+                fontSize: Math.max(13, (parseInt(theme.typography.baseFontSize) || 16) - 2),
+                fontWeight: 400,
+                color: theme.palette.textMuted,
+                lineHeight: theme.typography.baseLineHeight - 0.15,
+                borderLeftColor: theme.palette.blockquoteBorder,
+                bgColor: theme.palette.blockquoteBg,
+            },
+            codeBlock: {
+                fontFamily: theme.typography.codeFontFamily,
+                fontSize: Math.max(12, (parseInt(theme.typography.baseFontSize) || 16) - 3),
+                fontWeight: 400,
+                color: theme.palette.codeBlockText || theme.palette.textColor,
+                lineHeight: theme.seedTypography.lineHeightCode,
+                bgColor: theme.palette.codeBlockBg,
+            },
+            inlineCode: {
+                fontSize: Math.max(12, (parseInt(theme.typography.baseFontSize) || 16) - 3),
+                color: theme.palette.accentColor,
+                bgColor: theme.palette.codeBg,
+            },
+            table: {
+                headerFontFamily: theme.seedTypography.fontHeading,
+                headerFontSize: Math.max(12, (parseInt(theme.typography.baseFontSize) || 16) - 3),
+                headerFontWeight: 700,
+                headerColor: "#ffffff",
+                headerBgColor: theme.palette.tableHeaderBg,
+                headerLineHeight: 1.4,
+                cellFontFamily: theme.seedTypography.fontBody,
+                cellFontSize: Math.max(11, (parseInt(theme.typography.baseFontSize) || 16) - 4),
+                cellColor: theme.palette.textColor,
+                cellBgColor: theme.palette.bgColor,
+                cellLineHeight: 1.5,
+                cellPadding: "8px 12px",
+                borderColor: theme.palette.borderColor,
+                borderWidth: "1px",
+            },
+            accentColor: theme.palette.accentColor,
+            mutedTextColor: theme.palette.textMuted,
+            selectionBg: theme.palette.selectionBg,
+        };
+    }
+}
+//# sourceMappingURL=docx-extractor.js.map

package/dist/extractors/docx-extractor.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"docx-extractor.js","sourceRoot":"","sources":["../../src/extractors/docx-extractor.ts"],"names":[],"mappings":"AAAA;;;;;;;;;GASG;AAEH,OAAO,EAAE,YAAY,EAAE,MAAM,SAAS,CAAC;AACvC,OAAO,EAAE,QAAQ,EAAE,MAAM,WAAW,CAAC;AACrC,OAAO,OAAO,MAAM,SAAS,CAAC;AAE9B,OAAO,EACL,cAAc,GAIf,MAAM,cAAc,CAAC;AACtB,OAAO,EAAE,mBAAmB,EAAuC,MAAM,mBAAmB,CAAC;AAC7F,OAAO,EAEL,oBAAoB,EACpB,qBAAqB,GAKtB,MAAM,2BAA2B,CAAC;AAEnC,MAAM,OAAO,aAAa;IACxB,SAAS,CAAC,MAAc;QACtB,OAAO,UAAU,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;IACjC,CAAC;IAED,KAAK,CAAC,OAAO,CAAC,MAAc;QAC1B,MAAM,MAAM,GAAG,YAAY,CAAC,MAAM,CAAC,CAAC;QACpC,MAAM,IAAI,GAAG,QAAQ,CAAC,MAAM,CAAC,CAAC,OAAO,CAAC,UAAU,EAAE,EAAE,CAAC,CAAC;QACtD,MAAM,KAAK,GAAG,cAAc,CAAC,IAAI,EAAE,MAAM,EAAE,MAAM,CAAC,CAAC;QAEnD,MAAM,MAAM,GAAG,MAAM,OAAO,CAAC,aAAa,CACxC,EAAE,MAAM,EAAE,EACV;YACE,QAAQ,EAAE;gBACR,uCAAuC;gBACvC,uCAAuC;gBACvC,uCAAuC;gBACvC,uCAAuC;gBACvC,uCAAuC;gBACvC,uCAAuC;gBACvC,kCAAkC;aACnC;SACF,CACF,CAAC;QAEF,IAAI,MAAM,CAAC,KAAK,EAAE,CAAC;YACjB,IAAI,CAAC,qBAAqB,CAAC,KAAK,EAAE,MAAM,CAAC,KAAK,CAAC,CAAC;QAClD,CAAC;QAED,IAAI,CAAC,qBAAqB,CAAC,KAAK,CAAC,CAAC;QAClC,mBAAmB,CAAC,KAAK,CAAC,CAAC;QAC3B,OAAO,KAAK,CAAC;IACf,CAAC;IAED;;OAEG;IACK,qBAAqB,CAAC,KAAiB,EAAE,IAAY;QAC3D,MAAM,CAAC,GAAG,KAAK,CAAC,OAAO,CAAC;QACxB,MAAM,CAAC,GAAG,KAAK,CAAC,UAAU,CAAC;QAE3B,IAAI,CAAC,iBAAiB,CAAC,CAAC,EAAE,IAAI,CAAC,CAAC;QAChC,IAAI,CAAC,oBAAoB,CAAC,KAAK,EAAE,IAAI,CAAC,CAAC;QACvC,IAAI,CAAC,kBAAkB,CAAC,CAAC,EAAE,IAAI,CAAC,CAAC;QACjC,IAAI,CAAC,mBAAmB,CAAC,CAAC,EAAE,IAAI,CAAC,CAAC;QAClC,IAAI,CAAC,kBAAkB,CAAC,CAAC,EAAE,IAAI,CAAC,CAAC;IACnC,CAAC;IAED;;OAEG;IACK,iBAAiB,CAAC,UAAsB,EAAE,IAAY;QAC5D,MAAM,SAAS,GAAG,IAAI,CAAC,KAAK,CAAC,qCAAqC,CAAC,CAAC;QACpE,IAAI,SAAS,EAAE,CAAC;YACd,UAAU,CAAC,cAAc,GAAG,SAAS,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;QAClD,CAAC;QAED,MAAM,aAAa,GAAG,IAAI,CAAC,KAAK,CAAC,+BAA+B,CAAC,CAAC;QAClE,IAAI,aAAa,EAAE,CAAC;YAClB,MAAM,IAAI,GAAG,UAAU,CAAC,aAAa,CAAC,CAAC,CAAC,CAAC,CAAC;YAC1C,MAAM,IAAI,GAAG,aAAa,CAAC,CAAC,CAAC,CAAC;YAC9B,UAAU,CAAC,YAAY,GAAG,IAAI,KAAK,IAAI,CAAC,CAAC,CAAC,GAAG,IAAI,GAAG,KAAK,IAAI,CAAC,CAAC,CAAC,GAAG,IAAI,IAAI,CAAC;QAC9E,CAAC;IACH,CAAC;IAED;;OAEG;IACK,oBAAoB,CAAC,KAAiB,EAAE,IAAY;QAC1D,MAAM,aAAa,GAAG,CAAC,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,EAAE,IAAI,CAAC,CAAC;QAE3D,KAAK,MAAM,KAAK,IAAI,aAAa,EAAE,CAAC;YAClC,MAAM,KAAK,GAAG,IAAI,MAAM,CACtB,IAAI,KAAK,oBAAoB,EAC7B,IAAI,CACL,CAAC;YACF,IAAI,KAAK,GAAG,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YAE7B,IAAI,KAAK,EAAE,CAAC;gBACV,MAAM,QAAQ,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC;gBAC1B,MAAM,SAAS,GAAG,IAAI,CAAC,gBAAgB,CAAC,QAAQ,CAAC,CAAC;gBAElD,IAAI,SAAS,CAAC,QAAQ,EAAE,CAAC;oBACvB,MAAM,IAAI,GAAG,UAAU,CAAC,SAAS,CAAC,QAAQ,CAAC,CAAC;oBAC5C,MAAM,IAAI,GAAG,SAAS,CAAC,QAAQ,CAAC,OAAO,CAAC,QAAQ,EAAE,EAAE,CAAC,IAAI,IAAI,CAAC;oBAC9D,MAAM,MAAM,GAAG,IAAI,KAAK,IAAI,CAAC,CAAC,CAAC,IAAI,GAAG,KAAK,CAAC,CAAC,CAAC,IAAI,CAAC;oBAEnD,QAAQ,KAAK,EAAE,CAAC;wBACd,KAAK,IAAI;4BAAE,KAAK,CAAC,cAAc,CAAC,WAAW,GAAG,MAAM,GAAG,EAAE,CAAC;4BAAC,MAAM;wBACjE,KAAK,IAAI;4BAAE,KAAK,CAAC,cAAc,CAAC,WAAW,GAAG,MAAM,GAAG,EAAE,CAAC;4BAAC,MAAM;wBACjE,KAAK,IAAI;4BAAE,KAAK,CAAC,cAAc,CAAC,WAAW,GAAG,MAAM,GAAG,EAAE,CAAC;4BAAC,MAAM;wBACjE,KAAK,IAAI;4BAAE,KAAK,CAAC,cAAc,CAAC,WAAW,GAAG,MAAM,GAAG,EAAE,CAAC;4BAAC,MAAM;wBACjE,KAAK,IAAI;4BAAE,KAAK,CAAC,cAAc,CAAC,WAAW,GAAG,MAAM,GAAG,EAAE,CAAC;4BAAC,MAAM;wBACjE,KAAK,IAAI;4BAAE,KAAK,CAAC,cAAc,CAAC,WAAW,GAAG,MAAM,GAAG,EAAE,CAAC;4BAAC,MAAM;oBACnE,CAAC;gBACH,CAAC;gBAED,IAAI,SAAS,CAAC,UAAU,EAAE,CAAC;oBACzB,MAAM,MAAM,GAAG,QAAQ,CAAC,SAAS,CAAC,UAAU,CAAC,IAAI,GAAG,CAAC;oBACrD,IAAI,MAAM,GAAG,KAAK,CAAC,UAAU,CAAC,aAAa,EAAE,CAAC;wBAC5C,KAAK,CAAC,UAAU,CAAC,aAAa,GAAG,MAAM,CAAC;oBAC1C,CAAC;gBACH,CAAC;gBAED,IAAI,SAAS,CAAC,KAAK,IAAI,KAAK,KAAK,IAAI,EAAE,CAAC;oBACtC,KAAK,CAAC,OAAO,CAAC,YAAY,GAAG,SAAS,CAAC,KAAK,CAAC;gBAC/C,CAAC;YACH,CAAC;QACH,CAAC;IACH,CAAC;IAED;;OAEG;IACK,kBAAkB,CAAC,OAAqB,EAAE,IAAY;QAC5D,MAAM,OAAO,GAAG,2BAA2B,CAAC;QAC5C,IAAI,OAAO,GAAG,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QAEjC,IAAI,OAAO,EAAE,CAAC;YACZ,MAAM,OAAO,GAAG,IAAI,CAAC,gBAAgB,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,CAAC;YAClD,IAAI,OAAO,CAAC,eAAe,EAAE,CAAC;gBAC5B,OAAO,CAAC,aAAa,GAAG,OAAO,CAAC,eAAe,CAAC;YAClD,CAAC;YACD,IAAI,OAAO,CAAC,KAAK,EAAE,CAAC;gBAClB,OAAO,CAAC,YAAY,GAAG,OAAO,CAAC,YAAY,IAAI,OAAO,CAAC,KAAK,CAAC;YAC/D,CAAC;YACD,IAAI,OAAO,CAAC,QAAQ,EAAE,CAAC;gBACrB,MAAM,IAAI,GAAG,UAAU,CAAC,OAAO,CAAC,QAAQ,CAAC,CAAC;gBAC1C,OAAO,CAAC,aAAa,GAAG,OAAO,CAAC,aAAa,IAAI,SAAS,CAAC;YAC7D,CAAC;QACH,CAAC;QAED,MAAM,OAAO,GAAG,2BAA2B,CAAC;QAC5C,IAAI,OAAO,GAAG,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;QACjC,IAAI,OAAO,EAAE,CAAC;YACZ,MAAM,OAAO,GAAG,IAAI,CAAC,gBAAgB,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,CAAC;YAClD,IAAI,OAAO,CAAC,eAAe,EAAE,CAAC;gBAC5B,OAAO,CAAC,MAAM,GAAG,OAAO,CAAC,eAAe,CAAC;YAC3C,CAAC;QACH,CAAC;QAED,MAAM,gBAAgB,GAAG,IAAI,CAAC,KAAK,CAAC,sDAAsD,CAAC,CAAC;QAC5F,IAAI,gBAAgB,EAAE,CAAC;YACrB,OAAO,CAAC,WAAW,GAAG,OAAO,CAAC,WAAW,IAAI,SAAS,CAAC;QACzD,CAAC;QAED,IAAI,CAAC,OAAO,CAAC,aAAa,EAAE,CAAC;YAC3B,OAAO,CAAC,aAAa,GAAG,SAAS,CAAC;QACpC,CAAC;IACH,CAAC;IAED;;OAEG;IACK,mBAAmB,CAAC,OAAqB,EAAE,IAAY;QAC7D,MAAM,UAAU,GAAG,IAAI,CAAC,KAAK,CAAC,+BAA+B,CAAC,CAAC;QAC/D,IAAI,UAAU,EAAE,CAAC;YACf,OAAO,CAAC,SAAS,GAAG,UAAU,CAAC,CAAC,CAAC,CAAC;QACpC,CAAC;QAED,MAAM,OAAO,GAAG,IAAI,CAAC,KAAK,CAAC,+CAA+C,CAAC,CAAC;QAC5E,IAAI,OAAO,IAAI,OAAO,CAAC,CAAC,CAAC,CAAC,WAAW,EAAE,KAAK,SAAS,EAAE,CAAC;YACtD,OAAO,CAAC,OAAO,GAAG,OAAO,CAAC,CAAC,CAAC,CAAC;QAC/B,CAAC;QAED,MAAM,WAAW,GAAG,IAAI,CAAC,KAAK,CAAC,uDAAuD,CAAC,CAAC;QACxF,IAAI,WAAW,EAAE,CAAC;YAChB,OAAO,CAAC,WAAW,GAAG,WAAW,CAAC,CAAC,CAAC,CAAC;QACvC,CAAC;QAED,MAAM,SAAS,GAAG,IAAI,CAAC,KAAK,CAAC,kDAAkD,CAAC,CAAC;QACjF,IAAI,SAAS,EAAE,CAAC;YACd,OAAO,CAAC,WAAW,GAAG,SAAS,CAAC,CAAC,CAAC,CAAC;QACrC,CAAC;IACH,CAAC;IAED;;OAEG;IACK,kBAAkB,CAAC,OAAqB,EAAE,IAAY;QAC5D,MAAM,QAAQ,GAAG,IAAI,CAAC,KAAK,CAAC,oEAAoE,CAAC,CAAC;QAClG,IAAI,QAAQ,EAAE,CAAC;YACb,OAAO,CAAC,WAAW,GAAG,QAAQ,CAAC,CAAC,CAAC,CAAC;QACpC,CAAC;QAED,MAAM,aAAa,GAAG,IAAI,CAAC,KAAK,CAAC,oDAAoD,CAAC,CAAC;QACvF,IAAI,aAAa,EAAE,CAAC;YAClB,OAAO,CAAC,aAAa,GAAG,aAAa,CAAC,CAAC,CAAC,CAAC;QAC3C,CAAC;QAED,IAAI,CAAC,OAAO,CAAC,gBAAgB,EAAE,CAAC;YAC9B,OAAO,CAAC,gBAAgB,GAAG,OAAO,CAAC,WAAW,CAAC;QACjD,CAAC;IACH,CAAC;IAED;;OAEG;IACK,gBAAgB,CAAC,QAAgB;QACvC,MAAM,UAAU,GAAG,QAAQ,CAAC,KAAK,CAAC,kBAAkB,CAAC,CAAC;QACtD,IAAI,CAAC,UAAU;YAAE,OAAO,EAAE,CAAC;QAE3B,MAAM,MAAM,GAA2B,EAAE,CAAC;QAC1C,MAAM,YAAY,GAAG,UAAU,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC;QAE9C,KAAK,MAAM,IAAI,IAAI,YAAY,EAAE,CAAC;YAChC,MAAM,CAAC,IAAI,EAAE,GAAG,UAAU,CAAC,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC;YACnE,IAAI,IAAI,IAAI,UAAU,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;gBAClC,MAAM,SAAS,GAAG,IAAI,CAAC,OAAO,CAAC,WAAW,EAAE,CAAC,EAAE,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,WAAW,EAAE,CAAC,CAAC;gBACxE,MAAM,CAAC,SAAS,CAAC,GAAG,UAAU,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC;YAClD,CAAC;QACH,CAAC;QAED,OAAO,MAAM,CAAC;IAChB,CAAC;IAED;;OAEG;IACK,qBAAqB,CAAC,KAAiB;QAC7C,MAAM,MAAM,GAAG,oBAAoB,EAAE,CAAC;QACtC,MAAM,CAAC,GAAG,KAAK,CAAC,OAAO,CAAC;QACxB,MAAM,CAAC,GAAG,KAAK,CAAC,UAAU,CAAC;QAC3B,MAAM,EAAE,GAAG,KAAK,CAAC,cAAc,CAAC;QAEhC,IAAI,CAAC,CAAC,CAAC,SAAS,IAAI,CAAC,CAAC,SAAS,KAAK,SAAS,EAAE,CAAC;YAC9C,CAAC,CAAC,SAAS,GAAG,MAAM,CAAC,IAAI,CAAC,KAAK,CAAC;QAClC,CAAC;QACD,IAAI,CAAC,CAAC,CAAC,YAAY,EAAE,CAAC;YACpB,CAAC,CAAC,YAAY,GAAG,MAAM,CAAC,QAAQ,CAAC,EAAE,CAAC,KAAK,CAAC;QAC5C,CAAC;QACD,IAAI,CAAC,CAAC,CAAC,OAAO,IAAI,CAAC,CAAC,OAAO,KAAK,SAAS,EAAE,CAAC;YAC1C,CAAC,CAAC,OAAO,GAAG,MAAM,CAAC,cAAc,CAAC;QACpC,CAAC;QACD,IAAI,CAAC,CAAC,CAAC,SAAS,EAAE,CAAC;YACjB,CAAC,CAAC,SAAS,GAAG,MAAM,CAAC,cAAc,CAAC;QACtC,CAAC;QACD,IAAI,CAAC,CAAC,CAAC,WAAW,IAAI,CAAC,CAAC,WAAW,KAAK,SAAS,EAAE,CAAC;YAClD,CAAC,CAAC,WAAW,GAAG,MAAM,CAAC,WAAW,CAAC;QACrC,CAAC;QACD,IAAI,CAAC,CAAC,CAAC,aAAa,EAAE,CAAC;YACrB,CAAC,CAAC,aAAa,GAAG,MAAM,CAAC,KAAK,CAAC,aAAa,CAAC;QAC/C,CAAC;QACD,IAAI,CAAC,CAAC,CAAC,MAAM,EAAE,CAAC;YACd,CAAC,CAAC,MAAM,GAAG,MAAM,CAAC,UAAU,CAAC,OAAO,CAAC;QACvC,CAAC;QACD,IAAI,CAAC,CAAC,CAAC,WAAW,EAAE,CAAC;YACnB,CAAC,CAAC,WAAW,GAAG,MAAM,CAAC,SAAS,CAAC,OAAO,CAAC;QAC3C,CAAC;QACD,IAAI,CAAC,CAAC,CAAC,YAAY,EAAE,CAAC;YACpB,CAAC,CAAC,YAAY,GAAG,MAAM,CAAC,UAAU,CAAC,OAAO,CAAC;QAC7C,CAAC;QACD,IAAI,CAAC,CAAC,CAAC,WAAW,EAAE,CAAC;YACnB,CAAC,CAAC,WAAW,GAAG,MAAM,CAAC,WAAW,CAAC;QACrC,CAAC;QAED,IAAI,CAAC,CAAC,CAAC,cAAc,IAAI,CAAC,CAAC,cAAc,KAAK,gBAAgB,EAAE,CAAC;YAC/D,CAAC,CAAC,cAAc,GAAG,MAAM,CAAC,IAAI,CAAC,UAAU,CAAC;QAC5C,CAAC;QACD,IAAI,CAAC,CAAC,CAAC,iBAAiB,EAAE,CAAC;YACzB,CAAC,CAAC,iBAAiB,GAAG,MAAM,CAAC,QAAQ,CAAC,EAAE,CAAC,UAAU,CAAC;QACtD,CAAC;QACD,IAAI,CAAC,CAAC,CAAC,YAAY,EAAE,CAAC;YACpB,CAAC,CAAC,YAAY,GAAG,GAAG,MAAM,CAAC,IAAI,CAAC,QAAQ,IAAI,CAAC;QAC/C,CAAC;QACD,IAAI,CAAC,CAAC,cAAc,GAAG,GAAG,EAAE,CAAC;YAC3B,CAAC,CAAC,cAAc,GAAG,MAAM,CAAC,IAAI,CAAC,UAAU,CAAC;QAC5C,CAAC;QAED,EAAE,CAAC,QAAQ,GAAG,MAAM,CAAC,IAAI,CAAC,UAAU,CAAC;QACrC,EAAE,CAAC,WAAW,GAAG,MAAM,CAAC,QAAQ,CAAC,EAAE,CAAC,UAAU,CAAC;QAC/C,EAAE,CAAC,cAAc,GAAG,MAAM,CAAC,IAAI,CAAC,UAAU,CAAC;QAC3C,EAAE,CAAC,iBAAiB,GAAG,MAAM,CAAC,QAAQ,CAAC,EAAE,CAAC,UAAU,CAAC;QAErD,MAAM,MAAM,GAAG,qBAAqB,CAAC,IAAI,CAAC,eAAe,CAAC,KAAK,CAAC,CAAC,CAAC;QAClE,KAAK,MAAM,KAAK,IAAI,MAAM,EAAE,CAAC;YAC3B,IAAI,KAAK,CAAC,KAAK,KAAK,OAAO,EAAE,CAAC;gBAC5B,OAAO,CAAC,IAAI,CAAC,mBAAmB,KAAK,CAAC,OAAO,KAAK,KAAK,CAAC,OAAO,EAAE,CAAC,CAAC;YACrE,CAAC;QACH,CAAC;IACH,CAAC;IAED;;OAEG;IACK,eAAe,CAAC,KAAiB;QACvC,OAAO;YACL,cAAc,EAAE,KAAK,CAAC,OAAO,CAAC,OAAO;YACrC,QAAQ,EAAE;gBACR,EAAE,EAAE;oBACF,UAAU,EAAE,KAAK,CAAC,cAAc,CAAC,WAAW;oBAC5C,QAAQ,EAAE,EAAE,GAAG,KAAK,CAAC,cAAc,CAAC,WAAW;oBAC/C,UAAU,EAAE,KAAK,CAAC,UAAU,CAAC,aAAa;oBAC1C,KAAK,EAAE,KAAK,CAAC,OAAO,CAAC,YAAY;oBACjC,UAAU,EAAE,KAAK,CAAC,cAAc,CAAC,iBAAiB;iBACnD;gBACD,EAAE,EAAE;oBACF,UAAU,EAAE,KAAK,CAAC,cAAc,CAAC,WAAW;oBAC5C,QAAQ,EAAE,EAAE,GAAG,KAAK,CAAC,cAAc,CAAC,WAAW;oBAC/C,UAAU,EAAE,KAAK,CAAC,UAAU,CAAC,aAAa,GAAG,GAAG,IAAI,GAAG;oBACvD,KAAK,EAAE,KAAK,CAAC,OAAO,CAAC,YAAY;oBACjC,UAAU,EAAE,KAAK,CAAC,cAAc,CAAC,iBAAiB;iBACnD;gBACD,EAAE,EAAE;oBACF,UAAU,EAAE,KAAK,CAAC,cAAc,CAAC,WAAW;oBAC5C,QAAQ,EAAE,EAAE,GAAG,KAAK,CAAC,cAAc,CAAC,WAAW;oBAC/C,UAAU,EAAE,GAAG;oBACf,KAAK,EAAE,KAAK,CAAC,OAAO,CAAC,YAAY;oBACjC,UAAU,EAAE,KAAK,CAAC,cAAc,CAAC,iBAAiB,GAAG,IAAI;iBAC1D;gBACD,EAAE,EAAE;oBACF,UAAU,EAAE,KAAK,CAAC,cAAc,CAAC,WAAW;oBAC5C,QAAQ,EAAE,EAAE,GAAG,KAAK,CAAC,cAAc,CAAC,WAAW;oBAC/C,UAAU,EAAE,GAAG;oBACf,KAAK,EAAE,KAAK,CAAC,OAAO,CAAC,SAAS;oBAC9B,UAAU,EAAE,KAAK,CAAC,cAAc,CAAC,iBAAiB,GAAG,IAAI;iBAC1D;gBACD,EAAE,EAAE;oBACF,UAAU,EAAE,KAAK,CAAC,cAAc,CAAC,WAAW;oBAC5C,QAAQ,EAAE,EAAE,GAAG,KAAK,CAAC,cAAc,CAAC,WAAW;oBAC/C,UAAU,EAAE,GAAG;oBACf,KAAK,EAAE,KAAK,CAAC,OAAO,CAAC,SAAS;oBAC9B,UAAU,EAAE,KAAK,CAAC,cAAc,CAAC,cAAc,GAAG,IAAI;iBACvD;gBACD,EAAE,EAAE;oBACF,UAAU,EAAE,KAAK,CAAC,cAAc,CAAC,WAAW;oBAC5C,QAAQ,EAAE,EAAE,GAAG,KAAK,CAAC,cAAc,CAAC,WAAW;oBAC/C,UAAU,EAAE,GAAG;oBACf,KAAK,EAAE,KAAK,CAAC,OAAO,CAAC,SAAS;oBAC9B,UAAU,EAAE,KAAK,CAAC,cAAc,CAAC,cAAc,GAAG,GAAG;iBACtD;aACF;YACD,IAAI,EAAE;gBACJ,UAAU,EAAE,KAAK,CAAC,cAAc,CAAC,QAAQ;gBACzC,QAAQ,EAAE,QAAQ,CAAC,KAAK,CAAC,UAAU,CAAC,YAAY,CAAC,IAAI,EAAE;gBACvD,UAAU,EAAE,GAAG;gBACf,KAAK,EAAE,KAAK,CAAC,OAAO,CAAC,SAAS;gBAC9B,UAAU,EAAE,KAAK,CAAC,UAAU,CAAC,cAAc;aAC5C;YACD,UAAU,EAAE;gBACV,UAAU,EAAE,KAAK,CAAC,cAAc,CAAC,QAAQ;gBACzC,QAAQ,EAAE,IAAI,CAAC,GAAG,CAAC,EAAE,EAAE,CAAC,QAAQ,CAAC,KAAK,CAAC,UAAU,CAAC,YAAY,CAAC,IAAI,EAAE,CAAC,GAAG,CAAC,CAAC;gBAC3E,UAAU,EAAE,GAAG;gBACf,KAAK,EAAE,KAAK,CAAC,OAAO,CAAC,SAAS;gBAC9B,UAAU,EAAE,KAAK,CAAC,UAAU,CAAC,cAAc,GAAG,IAAI;gBAClD,eAAe,EAAE,KAAK,CAAC,OAAO,CAAC,gBAAgB;gBAC/C,OAAO,EAAE,KAAK,CAAC,OAAO,CAAC,YAAY;aACpC;YACD,SAAS,EAAE;gBACT,UAAU,EAAE,KAAK,CAAC,UAAU,CAAC,cAAc;gBAC3C,QAAQ,EAAE,IAAI,CAAC,GAAG,CAAC,EAAE,EAAE,CAAC,QAAQ,CAAC,KAAK,CAAC,UAAU,CAAC,YAAY,CAAC,IAAI,EAAE,CAAC,GAAG,CAAC,CAAC;gBAC3E,UAAU,EAAE,GAAG;gBACf,KAAK,EAAE,KAAK,CAAC,OAAO,CAAC,aAAa,IAAI,KAAK,CAAC,OAAO,CAAC,SAAS;gBAC7D,UAAU,EAAE,KAAK,CAAC,cAAc,CAAC,cAAc;gBAC/C,OAAO,EAAE,KAAK,CAAC,OAAO,CAAC,WAAW;aACnC;YACD,UAAU,EAAE;gBACV,QAAQ,EAAE,IAAI,CAAC,GAAG,CAAC,EAAE,EAAE,CAAC,QAAQ,CAAC,KAAK,CAAC,UAAU,CAAC,YAAY,CAAC,IAAI,EAAE,CAAC,GAAG,CAAC,CAAC;gBAC3E,KAAK,EAAE,KAAK,CAAC,OAAO,CAAC,WAAW;gBAChC,OAAO,EAAE,KAAK,CAAC,OAAO,CAAC,MAAM;aAC9B;YACD,KAAK,EAAE;gBACL,gBAAgB,EAAE,KAAK,CAAC,cAAc,CAAC,WAAW;gBAClD,cAAc,EAAE,IAAI,CAAC,GAAG,CAAC,EAAE,EAAE,CAAC,QAAQ,CAAC,KAAK,CAAC,UAAU,CAAC,YAAY,CAAC,IAAI,EAAE,CAAC,GAAG,CAAC,CAAC;gBACjF,gBAAgB,EAAE,GAAG;gBACrB,WAAW,EAAE,SAAS;gBACtB,aAAa,EAAE,KAAK,CAAC,OAAO,CAAC,aAAa;gBAC1C,gBAAgB,EAAE,GAAG;gBAErB,cAAc,EAAE,KAAK,CAAC,cAAc,CAAC,QAAQ;gBAC7C,YAAY,EAAE,IAAI,CAAC,GAAG,CAAC,EAAE,EAAE,CAAC,QAAQ,CAAC,KAAK,CAAC,UAAU,CAAC,YAAY,CAAC,IAAI,EAAE,CAAC,GAAG,CAAC,CAAC;gBAC/E,SAAS,EAAE,KAAK,CAAC,OAAO,CAAC,SAAS;gBAClC,WAAW,EAAE,KAAK,CAAC,OAAO,CAAC,OAAO;gBAClC,cAAc,EAAE,GAAG;gBACnB,WAAW,EAAE,UAAU;gBACvB,WAAW,EAAE,KAAK,CAAC,OAAO,CAAC,WAAW;gBACtC,WAAW,EAAE,KAAK;aACnB;YAED,WAAW,EAAE,KAAK,CAAC,OAAO,CAAC,WAAW;YACtC,cAAc,EAAE,KAAK,CAAC,OAAO,CAAC,SAAS;YACvC,WAAW,EAAE,KAAK,CAAC,OAAO,CAAC,WAAW;SACvC,CAAC;IACJ,CAAC;CACF"}

package/dist/extractors/pdf-extractor.d.ts

@@ -0,0 +1,66 @@
+/**
+ * PDF 文档样式提取器 — v2.0
+ *
+ * 增强功能：
+ * 1. 基于文本结构启发式分析提取各级标题样式
+ * 2. 检测正文字体、字号、行间距特征
+ * 3. 识别表格、代码块等特殊元素
+ * 4. 集成 DocxStyleParadigm 标准参考值填充缺失字段
+ * 5. 自动验证 WCAG 对比度合规性
+ *
+ * 技术说明：
+ * pdf-parse 默认不提供字体元数据，因此采用多层启发式策略：
+ * - 层级1：文本结构分析（短行/大写/编号 = 标题候选）
+ * - 层级2：格式特征检测（等宽文本 = 代码块）
+ * - 层级3：DocxStyleParadigm 标准参考值 fallback
+ */
+import type { Extractor } from "./base.js";
+import { type ThemeStyle } from "../models.js";
+export declare class PdfExtractor implements Extractor {
+    canHandle(source: string): boolean;
+    extract(source: string): Promise<ThemeStyle>;
+    /**
+     * 分析 PDF 文本结构，推断文档元素样式
+     */
+    private analyzeTextStructure;
+    /**
+     * 单行文本特征分析
+     */
+    private analyzeLine;
+    /**
+     * 基于启发式规则检测标题层级
+     */
+    private detectHeadings;
+    /**
+     * 推断正文样式特征
+     */
+    private detectBodyText;
+    /**
+     * 检测代码块区域（等宽文本特征）
+     */
+    private detectCodeBlocks;
+    /**
+     * 检测表格结构（多列对齐文本）
+     */
+    private detectTables;
+    /**
+     * 基于文档特征推断配色方案
+     */
+    private inferColorPalette;
+    /**
+     * 提取 PDF 元数据信息
+     */
+    private populateFromMetadata;
+    /**
+     * 基于内容特征判断文档类型并应用对应预设
+     */
+    private detectDocumentType;
+    /**
+     * 应用 DocxStyleParadigm 标准参考值填充所有缺失字段
+     */
+    private applyParadigmDefaults;
+    /**
+     * 将 ThemeStyle 转换为 DocumentStyleParadigm 用于 WCAG 验证
+     */
+    private themeToParadigm;
+}