@avocadostudio-ai/migration-sdk 0.1.0

package/dist/section-extractor.d.ts

@@ -0,0 +1,50 @@
+/**
+ * HTML section extractor — converts raw HTML into structured sections
+ * for LLM-based block mapping.
+ *
+ * Instead of sending 50KB of raw HTML to the LLM, this module identifies
+ * semantic section boundaries, classifies them by CSS class patterns,
+ * extracts structured content, and resolves lazy-loaded images.
+ */
+import type { ExtractedSection, NavExtraction, PageOutline, LayoutNode, VisualSection, RepeatGroup } from "./types.ts";
+/** Classify a section by its CSS classes and content */
+export declare function classifySection(openTag: string, innerHTML: string, preExtracted?: {
+    headings: Array<{
+        level: number;
+        text: string;
+    }>;
+}): {
+    classHints: string[];
+    suggestedBlockType?: string;
+};
+/**
+ * Extract structured sections from HTML.
+ *
+ * Identifies section boundaries using semantic tags, Elementor containers,
+ * and generic div analysis. Classifies each section by CSS class patterns
+ * and content heuristics. Resolves lazy-loaded images.
+ */
+export declare function extractSections(html: string, baseUrl: string): ExtractedSection[];
+/**
+ * Segment layout nodes into visual sections by detecting vertical gaps.
+ */
+export declare function segmentByVisualGaps(nodes: LayoutNode[], viewportWidth?: number): VisualSection[];
+/**
+ * Detect repeated structural patterns within a set of layout nodes.
+ * Groups nodes by structural signature and classifies repeat groups.
+ */
+export declare function detectRepeatedPatterns(nodes: LayoutNode[]): RepeatGroup[];
+/**
+ * Extract a compact page outline from HTML.
+ *
+ * Splits the page at heading boundaries (h1/h2) to produce a section-per-heading
+ * representation that captures the FULL page structure in ~2KB regardless of HTML size.
+ * This ensures the LLM sees every section even when the raw HTML is truncated.
+ */
+export declare function extractPageOutline(html: string, baseUrl: string, layoutNodes?: LayoutNode[]): PageOutline;
+export declare function extractNavigation(html: string, baseUrl: string): NavExtraction;
+/**
+ * Resolve lazy-loaded image sources in HTML.
+ * Replaces data-src, data-lazy-src, etc. with actual src attributes.
+ */
+export declare function resolveLazyImages(html: string): string;