@productivemark/snipcss 1.0.0

package/dist/extraction/extraction-pipeline.js

@@ -0,0 +1,200 @@
+import { ExtractionContext, } from '../types/index.js';
+import { ViewportManager } from '../browser/viewport-manager.js';
+import { StylesheetCollector } from './stylesheet-collector.js';
+import { DomLabeler } from './dom-labeler.js';
+import { StyleMatcher } from './style-matcher.js';
+import { PseudoStateHandler } from './pseudo-state-handler.js';
+import { KeyframeCollector } from './keyframe-collector.js';
+import { FontCollector } from './font-collector.js';
+import { CssVariableResolver } from './css-variable-resolver.js';
+import { SelectorFixer } from './selector-fixer.js';
+import { RuleDeduplicator } from './rule-deduplicator.js';
+import { ResultBuilder } from './result-builder.js';
+import { removeExtraAttributes } from './html-cleaner.js';
+import { getTailwindHtml, getTailwindBodyClasses } from '../tailwind/tailwind-converter.js';
+const BATCH_SIZE = 5; // Number of elements to process in parallel
+/**
+ * Main extraction pipeline orchestrator.
+ * Replaces the doSnipper() function from snipbackground.js.
+ */
+export class ExtractionPipeline {
+    browserManager;
+    viewportManager = new ViewportManager();
+    stylesheetCollector = new StylesheetCollector();
+    domLabeler = new DomLabeler();
+    styleMatcher = new StyleMatcher();
+    pseudoStateHandler = new PseudoStateHandler();
+    keyframeCollector = new KeyframeCollector();
+    fontCollector = new FontCollector();
+    variableResolver = new CssVariableResolver();
+    selectorFixer = new SelectorFixer();
+    ruleDeduplicator = new RuleDeduplicator();
+    resultBuilder = new ResultBuilder();
+    constructor(browserManager) {
+        this.browserManager = browserManager;
+    }
+    /**
+     * Extract CSS from a page element and optionally convert to Tailwind.
+     */
+    async extract(url, selector, options = {}) {
+        const ctx = new ExtractionContext();
+        ctx.siteUrl = url;
+        const resolveVariables = options.resolveVariables !== false;
+        const includeHoverStates = options.includeHoverStates !== false;
+        const doRemoveClasses = options.removeUnusedClasses ?? true;
+        const doRemoveAttrs = options.removeUnusedAttributes ?? true;
+        const viewportOption = options.viewport || 'all';
+        let bp = null;
+        try {
+            // STEP 1: Create page and CDP session
+            bp = await this.browserManager.createPage(url);
+            const { page, cdp } = bp;
+            // STEP 2: Start collecting stylesheets
+            await this.stylesheetCollector.startCollecting(cdp);
+            // Wait a moment for stylesheets to be collected
+            await page.waitForLoadState('networkidle');
+            // STEP 3: Get DOM document
+            const doc = await cdp.send('DOM.getDocument');
+            const docRootNodeId = doc.root.nodeId;
+            // STEP 4: Stop collecting and get stylesheets
+            const stylesheets = this.stylesheetCollector.stopCollecting();
+            ctx.stylesheetArr = stylesheets;
+            // STEP 5: Label elements in the DOM
+            const labelResult = await this.domLabeler.labelElements(page, selector);
+            ctx.allClassnamesArr = labelResult.allClassnamesArr;
+            console.log(`Labeled ${labelResult.allClassnamesArr.length} elements`);
+            // STEP 6: Parse fonts and CSS variables from stylesheets
+            await this.fontCollector.collectAll(cdp, stylesheets, url, ctx);
+            // STEP 7: Determine viewports to process
+            const viewports = this.viewportManager.getViewportsForOption(viewportOption, options.customWidth);
+            // STEP 8: For each viewport, extract matched styles
+            for (const viewport of viewports) {
+                console.log(`Processing viewport: ${viewport.name} (${viewport.width}x${viewport.height})`);
+                // Set viewport (skip for default)
+                if (viewport.name !== 'default') {
+                    await this.viewportManager.setViewport(cdp, viewport);
+                    // Wait for layout to stabilize
+                    await page.waitForTimeout(500);
+                }
+                // Refresh DOM document after viewport change
+                const vpDoc = await cdp.send('DOM.getDocument');
+                // Process elements in batches for parallelization
+                // NOTE: getMatchedStylesForNode is parallelizable, but
+                // forcePseudoState is NOT (it mutates global state)
+                for (let i = 0; i < ctx.allClassnamesArr.length; i += BATCH_SIZE) {
+                    const batch = ctx.allClassnamesArr.slice(i, i + BATCH_SIZE);
+                    // Parallel: Get matched styles for batch
+                    const batchResults = await Promise.all(batch.map(async (classname) => {
+                        try {
+                            // Query DOM for the element
+                            const node = await cdp.send('DOM.querySelector', {
+                                nodeId: vpDoc.root.nodeId,
+                                selector: '.' + classname,
+                            });
+                            if (!node.nodeId || node.nodeId === 0) {
+                                console.log(`Element not found: .${classname}`);
+                                return null;
+                            }
+                            // Get matched styles (THE core CDP call)
+                            const allMatchedStyles = await this.styleMatcher.getMatchedStyles(cdp, node.nodeId);
+                            return { classname, nodeId: node.nodeId, allMatchedStyles };
+                        }
+                        catch (e) {
+                            console.error(`Error matching styles for .${classname}:`, e);
+                            return null;
+                        }
+                    }));
+                    // Process results (sequential for state-dependent operations)
+                    for (const result of batchResults) {
+                        if (!result)
+                            continue;
+                        const { classname, nodeId, allMatchedStyles } = result;
+                        // Process matched rules
+                        const matchedRules = this.styleMatcher.processMatchedStyles(allMatchedStyles, classname, ctx, { resolveVariables, mediaQueriesOnly: viewport.name !== 'default' && viewport.name.startsWith('custom') });
+                        // Extract CSS variables
+                        this.styleMatcher.extractCssVariables(matchedRules, classname, ctx);
+                        // Collect keyframes
+                        this.keyframeCollector.collect(allMatchedStyles, ctx);
+                        // Convert to SnippedRules and add (deduplicated)
+                        for (const ruleMatch of matchedRules) {
+                            const snippedRule = this.styleMatcher.toSnippedRule(ruleMatch, classname, viewport.name, ctx);
+                            if (snippedRule) {
+                                this.ruleDeduplicator.addRule(snippedRule, ctx);
+                            }
+                        }
+                    }
+                    // Sequential: Extract hover/checked styles (mutates page state)
+                    if (includeHoverStates) {
+                        for (const result of batchResults) {
+                            if (!result)
+                                continue;
+                            const { classname, nodeId } = result;
+                            // Hover styles
+                            const hoverRules = await this.pseudoStateHandler.extractHoverStyles(cdp, page, nodeId, classname, vpDoc.root.nodeId);
+                            for (const ruleMatch of hoverRules) {
+                                const snippedRule = this.styleMatcher.toSnippedRule(ruleMatch, classname, viewport.name, ctx);
+                                if (snippedRule) {
+                                    snippedRule.is_hover = true;
+                                    this.ruleDeduplicator.addRule(snippedRule, ctx);
+                                }
+                            }
+                            // Checked styles (for checkboxes/radios)
+                            const checkedRules = await this.pseudoStateHandler.extractCheckedStyles(cdp, page, nodeId, classname, vpDoc.root.nodeId);
+                            for (const ruleMatch of checkedRules) {
+                                const snippedRule = this.styleMatcher.toSnippedRule(ruleMatch, classname, viewport.name, ctx);
+                                if (snippedRule) {
+                                    this.ruleDeduplicator.addRule(snippedRule, ctx);
+                                }
+                            }
+                        }
+                    }
+                }
+                // Reset viewport
+                if (viewport.name !== 'default') {
+                    await this.viewportManager.clearViewport(cdp);
+                }
+            }
+            console.log(`Extracted ${ctx.snippedArr.length} CSS rules`);
+            // STEP 9: Build final result
+            const result = this.resultBuilder.buildResult(ctx, labelResult.allElementOuterHtml, {
+                resolveVariables,
+            });
+            // STEP 10: Tailwind conversion
+            try {
+                const labeledHtml = labelResult.allElementOuterHtml;
+                result.tailwindHtml = getTailwindHtml(labeledHtml, result.css, ctx.snippedArr, false, // forceBreakpoints
+                resolveVariables, ctx);
+                result.tailwindBodyClasses = getTailwindBodyClasses(ctx.snippedArr, false, resolveVariables, [], // tailwindUltimateArr - populated inside getTailwindBodyClasses
+                ctx);
+            }
+            catch (e) {
+                console.error('Tailwind conversion error:', e);
+                // Non-fatal: CSS output is still valid
+            }
+            // STEP 11: Get clean HTML (without marker classes)
+            result.html = await this.domLabeler.getCleanHtml(page, selector);
+            // STEP 12: Remove unused attributes and classes from HTML
+            if (doRemoveClasses || doRemoveAttrs) {
+                const cleanOpts = {
+                    removeUnusedClasses: doRemoveClasses,
+                    removeUnusedAttributes: doRemoveAttrs,
+                    keepTailwindLabels: false,
+                };
+                result.html = removeExtraAttributes(result.html, ctx.snippedArr, cleanOpts);
+                if (result.tailwindHtml) {
+                    result.tailwindHtml = removeExtraAttributes(result.tailwindHtml, ctx.snippedArr, { ...cleanOpts, keepTailwindLabels: true });
+                }
+            }
+            // STEP 13: Remove labels from DOM
+            await this.domLabeler.removeLabels(page);
+            return result;
+        }
+        finally {
+            // Cleanup
+            if (bp) {
+                await this.browserManager.closePage(bp);
+            }
+        }
+    }
+}
+//# sourceMappingURL=extraction-pipeline.js.map

package/dist/extraction/extraction-pipeline.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"extraction-pipeline.js","sourceRoot":"","sources":["../../src/extraction/extraction-pipeline.ts"],"names":[],"mappings":"AACA,OAAO,EACL,iBAAiB,GAElB,MAAM,mBAAmB,CAAC;AAE3B,OAAO,EAAE,eAAe,EAAE,MAAM,gCAAgC,CAAC;AACjE,OAAO,EAAE,mBAAmB,EAAE,MAAM,2BAA2B,CAAC;AAChE,OAAO,EAAE,UAAU,EAAE,MAAM,kBAAkB,CAAC;AAC9C,OAAO,EAAE,YAAY,EAAE,MAAM,oBAAoB,CAAC;AAClD,OAAO,EAAE,kBAAkB,EAAE,MAAM,2BAA2B,CAAC;AAC/D,OAAO,EAAE,iBAAiB,EAAE,MAAM,yBAAyB,CAAC;AAC5D,OAAO,EAAE,aAAa,EAAE,MAAM,qBAAqB,CAAC;AACpD,OAAO,EAAE,mBAAmB,EAAE,MAAM,4BAA4B,CAAC;AACjE,OAAO,EAAE,aAAa,EAAE,MAAM,qBAAqB,CAAC;AACpD,OAAO,EAAE,gBAAgB,EAAE,MAAM,wBAAwB,CAAC;AAC1D,OAAO,EAAE,aAAa,EAAE,MAAM,qBAAqB,CAAC;AACpD,OAAO,EAAE,qBAAqB,EAAE,MAAM,mBAAmB,CAAC;AAC1D,OAAO,EAAE,eAAe,EAAE,sBAAsB,EAAE,MAAM,mCAAmC,CAAC;AAE5F,MAAM,UAAU,GAAG,CAAC,CAAC,CAAC,4CAA4C;AAElE;;;GAGG;AACH,MAAM,OAAO,kBAAkB;IACrB,cAAc,CAAiB;IAC/B,eAAe,GAAG,IAAI,eAAe,EAAE,CAAC;IACxC,mBAAmB,GAAG,IAAI,mBAAmB,EAAE,CAAC;IAChD,UAAU,GAAG,IAAI,UAAU,EAAE,CAAC;IAC9B,YAAY,GAAG,IAAI,YAAY,EAAE,CAAC;IAClC,kBAAkB,GAAG,IAAI,kBAAkB,EAAE,CAAC;IAC9C,iBAAiB,GAAG,IAAI,iBAAiB,EAAE,CAAC;IAC5C,aAAa,GAAG,IAAI,aAAa,EAAE,CAAC;IACpC,gBAAgB,GAAG,IAAI,mBAAmB,EAAE,CAAC;IAC7C,aAAa,GAAG,IAAI,aAAa,EAAE,CAAC;IACpC,gBAAgB,GAAG,IAAI,gBAAgB,EAAE,CAAC;IAC1C,aAAa,GAAG,IAAI,aAAa,EAAE,CAAC;IAE5C,YAAY,cAA8B;QACxC,IAAI,CAAC,cAAc,GAAG,cAAc,CAAC;IACvC,CAAC;IAED;;OAEG;IACH,KAAK,CAAC,OAAO,CACX,GAAW,EACX,QAAgB,EAChB,UAA6B,EAAE;QAE/B,MAAM,GAAG,GAAG,IAAI,iBAAiB,EAAE,CAAC;QACpC,GAAG,CAAC,OAAO,GAAG,GAAG,CAAC;QAElB,MAAM,gBAAgB,GAAG,OAAO,CAAC,gBAAgB,KAAK,KAAK,CAAC;QAC5D,MAAM,kBAAkB,GAAG,OAAO,CAAC,kBAAkB,KAAK,KAAK,CAAC;QAChE,MAAM,eAAe,GAAG,OAAO,CAAC,mBAAmB,IAAI,IAAI,CAAC;QAC5D,MAAM,aAAa,GAAG,OAAO,CAAC,sBAAsB,IAAI,IAAI,CAAC;QAC7D,MAAM,cAAc,GAAG,OAAO,CAAC,QAAQ,IAAI,KAAK,CAAC;QAEjD,IAAI,EAAE,GAAuB,IAAI,CAAC;QAElC,IAAI,CAAC;YACH,sCAAsC;YACtC,EAAE,GAAG,MAAM,IAAI,CAAC,cAAc,CAAC,UAAU,CAAC,GAAG,CAAC,CAAC;YAC/C,MAAM,EAAE,IAAI,EAAE,GAAG,EAAE,GAAG,EAAE,CAAC;YAEzB,uCAAuC;YACvC,MAAM,IAAI,CAAC,mBAAmB,CAAC,eAAe,CAAC,GAAG,CAAC,CAAC;YAEpD,gDAAgD;YAChD,MAAM,IAAI,CAAC,gBAAgB,CAAC,aAAa,CAAC,CAAC;YAE3C,2BAA2B;YAC3B,MAAM,GAAG,GAAG,MAAM,GAAG,CAAC,IAAI,CAAC,iBAAiB,CAAC,CAAC;YAC9C,MAAM,aAAa,GAAG,GAAG,CAAC,IAAI,CAAC,MAAM,CAAC;YAEtC,8CAA8C;YAC9C,MAAM,WAAW,GAAG,IAAI,CAAC,mBAAmB,CAAC,cAAc,EAAE,CAAC;YAC9D,GAAG,CAAC,aAAa,GAAG,WAAW,CAAC;YAEhC,oCAAoC;YACpC,MAAM,WAAW,GAAG,MAAM,IAAI,CAAC,UAAU,CAAC,aAAa,CAAC,IAAI,EAAE,QAAQ,CAAC,CAAC;YACxE,GAAG,CAAC,gBAAgB,GAAG,WAAW,CAAC,gBAAgB,CAAC;YAEpD,OAAO,CAAC,GAAG,CAAC,WAAW,WAAW,CAAC,gBAAgB,CAAC,MAAM,WAAW,CAAC,CAAC;YAEvE,yDAAyD;YACzD,MAAM,IAAI,CAAC,aAAa,CAAC,UAAU,CAAC,GAAG,EAAE,WAAW,EAAE,GAAG,EAAE,GAAG,CAAC,CAAC;YAEhE,yCAAyC;YACzC,MAAM,SAAS,GAAG,IAAI,CAAC,eAAe,CAAC,qBAAqB,CAC1D,cAAc,EACd,OAAO,CAAC,WAAW,CACpB,CAAC;YAEF,oDAAoD;YACpD,KAAK,MAAM,QAAQ,IAAI,SAAS,EAAE,CAAC;gBACjC,OAAO,CAAC,GAAG,CAAC,wBAAwB,QAAQ,CAAC,IAAI,KAAK,QAAQ,CAAC,KAAK,IAAI,QAAQ,CAAC,MAAM,GAAG,CAAC,CAAC;gBAE5F,kCAAkC;gBAClC,IAAI,QAAQ,CAAC,IAAI,KAAK,SAAS,EAAE,CAAC;oBAChC,MAAM,IAAI,CAAC,eAAe,CAAC,WAAW,CAAC,GAAG,EAAE,QAAQ,CAAC,CAAC;oBACtD,+BAA+B;oBAC/B,MAAM,IAAI,CAAC,cAAc,CAAC,GAAG,CAAC,CAAC;gBACjC,CAAC;gBAED,6CAA6C;gBAC7C,MAAM,KAAK,GAAG,MAAM,GAAG,CAAC,IAAI,CAAC,iBAAiB,CAAC,CAAC;gBAEhD,kDAAkD;gBAClD,uDAAuD;gBACvD,oDAAoD;gBACpD,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,GAAG,CAAC,gBAAgB,CAAC,MAAM,EAAE,CAAC,IAAI,UAAU,EAAE,CAAC;oBACjE,MAAM,KAAK,GAAG,GAAG,CAAC,gBAAgB,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,GAAG,UAAU,CAAC,CAAC;oBAE5D,yCAAyC;oBACzC,MAAM,YAAY,GAAG,MAAM,OAAO,CAAC,GAAG,CACpC,KAAK,CAAC,GAAG,CAAC,KAAK,EAAE,SAAS,EAAE,EAAE;wBAC5B,IAAI,CAAC;4BACH,4BAA4B;4BAC5B,MAAM,IAAI,GAAG,MAAM,GAAG,CAAC,IAAI,CAAC,mBAAmB,EAAE;gCAC/C,MAAM,EAAE,KAAK,CAAC,IAAI,CAAC,MAAM;gCACzB,QAAQ,EAAE,GAAG,GAAG,SAAS;6BAC1B,CAAC,CAAC;4BAEH,IAAI,CAAC,IAAI,CAAC,MAAM,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;gCACtC,OAAO,CAAC,GAAG,CAAC,uBAAuB,SAAS,EAAE,CAAC,CAAC;gCAChD,OAAO,IAAI,CAAC;4BACd,CAAC;4BAED,yCAAyC;4BACzC,MAAM,gBAAgB,GAAG,MAAM,IAAI,CAAC,YAAY,CAAC,gBAAgB,CAC/D,GAAG,EAAE,IAAI,CAAC,MAAM,CACjB,CAAC;4BAEF,OAAO,EAAE,SAAS,EAAE,MAAM,EAAE,IAAI,CAAC,MAAM,EAAE,gBAAgB,EAAE,CAAC;wBAC9D,CAAC;wBAAC,OAAO,CAAC,EAAE,CAAC;4BACX,OAAO,CAAC,KAAK,CAAC,8BAA8B,SAAS,GAAG,EAAE,CAAC,CAAC,CAAC;4BAC7D,OAAO,IAAI,CAAC;wBACd,CAAC;oBACH,CAAC,CAAC,CACH,CAAC;oBAEF,8DAA8D;oBAC9D,KAAK,MAAM,MAAM,IAAI,YAAY,EAAE,CAAC;wBAClC,IAAI,CAAC,MAAM;4BAAE,SAAS;wBACtB,MAAM,EAAE,SAAS,EAAE,MAAM,EAAE,gBAAgB,EAAE,GAAG,MAAM,CAAC;wBAEvD,wBAAwB;wBACxB,MAAM,YAAY,GAAG,IAAI,CAAC,YAAY,CAAC,oBAAoB,CACzD,gBAAgB,EAChB,SAAS,EACT,GAAG,EACH,EAAE,gBAAgB,EAAE,gBAAgB,EAAE,QAAQ,CAAC,IAAI,KAAK,SAAS,IAAI,QAAQ,CAAC,IAAI,CAAC,UAAU,CAAC,QAAQ,CAAC,EAAE,CAC1G,CAAC;wBAEF,wBAAwB;wBACxB,IAAI,CAAC,YAAY,CAAC,mBAAmB,CAAC,YAAY,EAAE,SAAS,EAAE,GAAG,CAAC,CAAC;wBAEpE,oBAAoB;wBACpB,IAAI,CAAC,iBAAiB,CAAC,OAAO,CAAC,gBAAgB,EAAE,GAAG,CAAC,CAAC;wBAEtD,iDAAiD;wBACjD,KAAK,MAAM,SAAS,IAAI,YAAY,EAAE,CAAC;4BACrC,MAAM,WAAW,GAAG,IAAI,CAAC,YAAY,CAAC,aAAa,CACjD,SAAS,EAAE,SAAS,EAAE,QAAQ,CAAC,IAAI,EAAE,GAAG,CACzC,CAAC;4BACF,IAAI,WAAW,EAAE,CAAC;gCAChB,IAAI,CAAC,gBAAgB,CAAC,OAAO,CAAC,WAAW,EAAE,GAAG,CAAC,CAAC;4BAClD,CAAC;wBACH,CAAC;oBACH,CAAC;oBAED,gEAAgE;oBAChE,IAAI,kBAAkB,EAAE,CAAC;wBACvB,KAAK,MAAM,MAAM,IAAI,YAAY,EAAE,CAAC;4BAClC,IAAI,CAAC,MAAM;gCAAE,SAAS;4BACtB,MAAM,EAAE,SAAS,EAAE,MAAM,EAAE,GAAG,MAAM,CAAC;4BAErC,eAAe;4BACf,MAAM,UAAU,GAAG,MAAM,IAAI,CAAC,kBAAkB,CAAC,kBAAkB,CACjE,GAAG,EAAE,IAAI,EAAE,MAAM,EAAE,SAAS,EAAE,KAAK,CAAC,IAAI,CAAC,MAAM,CAChD,CAAC;4BACF,KAAK,MAAM,SAAS,IAAI,UAAU,EAAE,CAAC;gCACnC,MAAM,WAAW,GAAG,IAAI,CAAC,YAAY,CAAC,aAAa,CACjD,SAAS,EAAE,SAAS,EAAE,QAAQ,CAAC,IAAI,EAAE,GAAG,CACzC,CAAC;gCACF,IAAI,WAAW,EAAE,CAAC;oCAChB,WAAW,CAAC,QAAQ,GAAG,IAAI,CAAC;oCAC5B,IAAI,CAAC,gBAAgB,CAAC,OAAO,CAAC,WAAW,EAAE,GAAG,CAAC,CAAC;gCAClD,CAAC;4BACH,CAAC;4BAED,yCAAyC;4BACzC,MAAM,YAAY,GAAG,MAAM,IAAI,CAAC,kBAAkB,CAAC,oBAAoB,CACrE,GAAG,EAAE,IAAI,EAAE,MAAM,EAAE,SAAS,EAAE,KAAK,CAAC,IAAI,CAAC,MAAM,CAChD,CAAC;4BACF,KAAK,MAAM,SAAS,IAAI,YAAY,EAAE,CAAC;gCACrC,MAAM,WAAW,GAAG,IAAI,CAAC,YAAY,CAAC,aAAa,CACjD,SAAS,EAAE,SAAS,EAAE,QAAQ,CAAC,IAAI,EAAE,GAAG,CACzC,CAAC;gCACF,IAAI,WAAW,EAAE,CAAC;oCAChB,IAAI,CAAC,gBAAgB,CAAC,OAAO,CAAC,WAAW,EAAE,GAAG,CAAC,CAAC;gCAClD,CAAC;4BACH,CAAC;wBACH,CAAC;oBACH,CAAC;gBACH,CAAC;gBAED,iBAAiB;gBACjB,IAAI,QAAQ,CAAC,IAAI,KAAK,SAAS,EAAE,CAAC;oBAChC,MAAM,IAAI,CAAC,eAAe,CAAC,aAAa,CAAC,GAAG,CAAC,CAAC;gBAChD,CAAC;YACH,CAAC;YAED,OAAO,CAAC,GAAG,CAAC,aAAa,GAAG,CAAC,UAAU,CAAC,MAAM,YAAY,CAAC,CAAC;YAE5D,6BAA6B;YAC7B,MAAM,MAAM,GAAG,IAAI,CAAC,aAAa,CAAC,WAAW,CAAC,GAAG,EAAE,WAAW,CAAC,mBAAmB,EAAE;gBAClF,gBAAgB;aACjB,CAAC,CAAC;YAEH,+BAA+B;YAC/B,IAAI,CAAC;gBACH,MAAM,WAAW,GAAG,WAAW,CAAC,mBAAmB,CAAC;gBACpD,MAAM,CAAC,YAAY,GAAG,eAAe,CACnC,WAAW,EACX,MAAM,CAAC,GAAG,EACV,GAAG,CAAC,UAAU,EACd,KAAK,EAAE,mBAAmB;gBAC1B,gBAAgB,EAChB,GAAG,CACJ,CAAC;gBACF,MAAM,CAAC,mBAAmB,GAAG,sBAAsB,CACjD,GAAG,CAAC,UAAU,EACd,KAAK,EACL,gBAAgB,EAChB,EAAE,EAAE,gEAAgE;gBACpE,GAAG,CACJ,CAAC;YACJ,CAAC;YAAC,OAAO,CAAC,EAAE,CAAC;gBACX,OAAO,CAAC,KAAK,CAAC,4BAA4B,EAAE,CAAC,CAAC,CAAC;gBAC/C,uCAAuC;YACzC,CAAC;YAED,mDAAmD;YACnD,MAAM,CAAC,IAAI,GAAG,MAAM,IAAI,CAAC,UAAU,CAAC,YAAY,CAAC,IAAI,EAAE,QAAQ,CAAC,CAAC;YAEjE,0DAA0D;YAC1D,IAAI,eAAe,IAAI,aAAa,EAAE,CAAC;gBACrC,MAAM,SAAS,GAAG;oBAChB,mBAAmB,EAAE,eAAe;oBACpC,sBAAsB,EAAE,aAAa;oBACrC,kBAAkB,EAAE,KAAK;iBAC1B,CAAC;gBACF,MAAM,CAAC,IAAI,GAAG,qBAAqB,CAAC,MAAM,CAAC,IAAI,EAAE,GAAG,CAAC,UAAU,EAAE,SAAS,CAAC,CAAC;gBAC5E,IAAI,MAAM,CAAC,YAAY,EAAE,CAAC;oBACxB,MAAM,CAAC,YAAY,GAAG,qBAAqB,CACzC,MAAM,CAAC,YAAY,EACnB,GAAG,CAAC,UAAU,EACd,EAAE,GAAG,SAAS,EAAE,kBAAkB,EAAE,IAAI,EAAE,CAC3C,CAAC;gBACJ,CAAC;YACH,CAAC;YAED,kCAAkC;YAClC,MAAM,IAAI,CAAC,UAAU,CAAC,YAAY,CAAC,IAAI,CAAC,CAAC;YAEzC,OAAO,MAAM,CAAC;QAChB,CAAC;gBAAS,CAAC;YACT,UAAU;YACV,IAAI,EAAE,EAAE,CAAC;gBACP,MAAM,IAAI,CAAC,cAAc,CAAC,SAAS,CAAC,EAAE,CAAC,CAAC;YAC1C,CAAC;QACH,CAAC;IACH,CAAC;CACF"}

package/dist/extraction/font-collector.d.ts

@@ -0,0 +1,26 @@
+import { CDPSession } from 'playwright';
+import { FontData, StylesheetInfo, ExtractionContext } from '../types/index.js';
+/**
+ * Parses @font-face declarations and CSS variables from stylesheets.
+ * Port of snipbackground.js:1343-1627
+ */
+export declare class FontCollector {
+    /**
+     * Parse @font-face rules from stylesheet text.
+     */
+    parseFontFaces(cssText: string, baseUrl: string): FontData[];
+    /**
+     * Extract Google Font imports from stylesheet text.
+     */
+    parseGoogleFontImports(cssText: string): string[];
+    /**
+     * Extract global CSS variables from stylesheet text.
+     * Looks for :root, :host, *, html, body selectors.
+     */
+    parseGlobalCssVariables(cssText: string, ctx: ExtractionContext): void;
+    /**
+     * Collect all font data and CSS variables from all stylesheets.
+     */
+    collectAll(cdp: CDPSession, stylesheets: StylesheetInfo[], siteUrl: string, ctx: ExtractionContext): Promise<void>;
+}
+//# sourceMappingURL=font-collector.d.ts.map

package/dist/extraction/font-collector.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"font-collector.d.ts","sourceRoot":"","sources":["../../src/extraction/font-collector.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,UAAU,EAAE,MAAM,YAAY,CAAC;AACxC,OAAO,EAAE,QAAQ,EAAE,cAAc,EAAE,iBAAiB,EAAoB,MAAM,mBAAmB,CAAC;AAGlG;;;GAGG;AACH,qBAAa,aAAa;IACxB;;OAEG;IACH,cAAc,CAAC,OAAO,EAAE,MAAM,EAAE,OAAO,EAAE,MAAM,GAAG,QAAQ,EAAE;IA6D5D;;OAEG;IACH,sBAAsB,CAAC,OAAO,EAAE,MAAM,GAAG,MAAM,EAAE;IAcjD;;;OAGG;IACH,uBAAuB,CACrB,OAAO,EAAE,MAAM,EACf,GAAG,EAAE,iBAAiB,GACrB,IAAI;IA0CP;;OAEG;IACG,UAAU,CACd,GAAG,EAAE,UAAU,EACf,WAAW,EAAE,cAAc,EAAE,EAC7B,OAAO,EAAE,MAAM,EACf,GAAG,EAAE,iBAAiB,GACrB,OAAO,CAAC,IAAI,CAAC;CAqDjB"}

package/dist/extraction/font-collector.js

@@ -0,0 +1,160 @@
+import { resolveUrl } from '../utils/helpers.js';
+/**
+ * Parses @font-face declarations and CSS variables from stylesheets.
+ * Port of snipbackground.js:1343-1627
+ */
+export class FontCollector {
+    /**
+     * Parse @font-face rules from stylesheet text.
+     */
+    parseFontFaces(cssText, baseUrl) {
+        const fonts = [];
+        // Regex to find @font-face blocks
+        const fontFaceRegex = /@font-face\s*\{([^}]+)\}/gi;
+        let match;
+        while ((match = fontFaceRegex.exec(cssText)) !== null) {
+            const block = match[1];
+            const font = {
+                font_family: '',
+                font_url: '',
+                full_rule: '@font-face {' + block + '}',
+            };
+            // Extract font-family
+            const familyMatch = block.match(/font-family\s*:\s*(['"]?)([^;'"]+)\1/i);
+            if (familyMatch) {
+                font.font_family = familyMatch[2].trim();
+            }
+            // Extract src URLs
+            const srcMatch = block.match(/src\s*:\s*([^;]+)/i);
+            if (srcMatch) {
+                const urlMatch = srcMatch[1].match(/url\(['"]?([^'")\s]+)['"]?\)/);
+                if (urlMatch) {
+                    font.font_url = resolveUrl(urlMatch[1], baseUrl);
+                }
+                // Extract format
+                const formatMatch = srcMatch[1].match(/format\(['"]?([^'")\s]+)['"]?\)/);
+                if (formatMatch) {
+                    font.font_format = formatMatch[1];
+                }
+            }
+            // Extract font-weight
+            const weightMatch = block.match(/font-weight\s*:\s*([^;]+)/i);
+            if (weightMatch) {
+                font.font_weight = weightMatch[1].trim();
+            }
+            // Extract font-style
+            const styleMatch = block.match(/font-style\s*:\s*([^;]+)/i);
+            if (styleMatch) {
+                font.font_style = styleMatch[1].trim();
+            }
+            // Extract font-display
+            const displayMatch = block.match(/font-display\s*:\s*([^;]+)/i);
+            if (displayMatch) {
+                font.font_display = displayMatch[1].trim();
+            }
+            if (font.font_family) {
+                fonts.push(font);
+            }
+        }
+        return fonts;
+    }
+    /**
+     * Extract Google Font imports from stylesheet text.
+     */
+    parseGoogleFontImports(cssText) {
+        const imports = [];
+        const importRegex = /@import\s+url\(['"]?(https?:\/\/fonts\.googleapis\.com[^'")\s]+)['"]?\)/gi;
+        let match;
+        while ((match = importRegex.exec(cssText)) !== null) {
+            if (!imports.includes(match[1])) {
+                imports.push(match[1]);
+            }
+        }
+        return imports;
+    }
+    /**
+     * Extract global CSS variables from stylesheet text.
+     * Looks for :root, :host, *, html, body selectors.
+     */
+    parseGlobalCssVariables(cssText, ctx) {
+        // Match :root, :host, *, html, body blocks
+        const globalBlockRegex = /(?::root|:host|\*|html|body)\s*\{([^}]+)\}/gi;
+        let match;
+        while ((match = globalBlockRegex.exec(cssText)) !== null) {
+            const block = match[1];
+            const selector = match[0].split('{')[0].trim();
+            // Find all --var declarations
+            const varRegex = /(--[a-zA-Z0-9_-]+)\s*:\s*([^;]+)/g;
+            let varMatch;
+            while ((varMatch = varRegex.exec(block)) !== null) {
+                const prop = varMatch[1];
+                const val = varMatch[2].trim();
+                ctx.cssvarAllArr[prop] = val + ';';
+                if (!ctx.cssvarDefinedArr[prop]) {
+                    ctx.cssvarDefinedArr[prop] = [];
+                }
+                const globalKey = `__global__${selector}${prop}`;
+                const exists = ctx.cssvarDefinedArr[prop].some((item) => item.key === globalKey);
+                if (!exists) {
+                    ctx.cssvarDefinedArr[prop].push({
+                        key: globalKey,
+                        label: '__global__',
+                        value: val,
+                        media: '',
+                        selector,
+                        source: 'stylesheet',
+                    });
+                }
+            }
+        }
+    }
+    /**
+     * Collect all font data and CSS variables from all stylesheets.
+     */
+    async collectAll(cdp, stylesheets, siteUrl, ctx) {
+        // Fetch all stylesheet texts in parallel
+        const results = await Promise.all(stylesheets
+            .filter(ss => ss.origin !== 'user-agent')
+            .map(async (ss) => {
+            try {
+                const result = await cdp.send('CSS.getStyleSheetText', {
+                    styleSheetId: ss.stylesheet_id,
+                });
+                return { ss, text: result.text || '' };
+            }
+            catch {
+                return { ss, text: '' };
+            }
+        }));
+        for (const { ss, text } of results) {
+            if (!text)
+                continue;
+            const baseUrl = ss.source_url || siteUrl;
+            // Parse @font-face rules
+            if (text.includes('@font-face')) {
+                const fonts = this.parseFontFaces(text, baseUrl);
+                for (const font of fonts) {
+                    // Deduplicate by font_family + font_weight + font_style
+                    const exists = ctx.customfontsArr.some((f) => f.font_family === font.font_family &&
+                        f.font_weight === font.font_weight &&
+                        f.font_style === font.font_style);
+                    if (!exists) {
+                        ctx.customfontsArr.push(font);
+                    }
+                }
+            }
+            // Parse Google Font imports
+            const imports = this.parseGoogleFontImports(text);
+            for (const imp of imports) {
+                if (!ctx.importfontsArr.includes(imp)) {
+                    ctx.importfontsArr.push(imp);
+                }
+            }
+            // Parse global CSS variables
+            if (text.includes('--')) {
+                this.parseGlobalCssVariables(text, ctx);
+            }
+        }
+    }
+}
+//# sourceMappingURL=font-collector.js.map

package/dist/extraction/font-collector.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"font-collector.js","sourceRoot":"","sources":["../../src/extraction/font-collector.ts"],"names":[],"mappings":"AAEA,OAAO,EAAE,UAAU,EAAE,MAAM,qBAAqB,CAAC;AAEjD;;;GAGG;AACH,MAAM,OAAO,aAAa;IACxB;;OAEG;IACH,cAAc,CAAC,OAAe,EAAE,OAAe;QAC7C,MAAM,KAAK,GAAe,EAAE,CAAC;QAE7B,kCAAkC;QAClC,MAAM,aAAa,GAAG,4BAA4B,CAAC;QACnD,IAAI,KAAK,CAAC;QAEV,OAAO,CAAC,KAAK,GAAG,aAAa,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,KAAK,IAAI,EAAE,CAAC;YACtD,MAAM,KAAK,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC;YACvB,MAAM,IAAI,GAAa;gBACrB,WAAW,EAAE,EAAE;gBACf,QAAQ,EAAE,EAAE;gBACZ,SAAS,EAAE,cAAc,GAAG,KAAK,GAAG,GAAG;aACxC,CAAC;YAEF,sBAAsB;YACtB,MAAM,WAAW,GAAG,KAAK,CAAC,KAAK,CAAC,uCAAuC,CAAC,CAAC;YACzE,IAAI,WAAW,EAAE,CAAC;gBAChB,IAAI,CAAC,WAAW,GAAG,WAAW,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;YAC3C,CAAC;YAED,mBAAmB;YACnB,MAAM,QAAQ,GAAG,KAAK,CAAC,KAAK,CAAC,oBAAoB,CAAC,CAAC;YACnD,IAAI,QAAQ,EAAE,CAAC;gBACb,MAAM,QAAQ,GAAG,QAAQ,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,8BAA8B,CAAC,CAAC;gBACnE,IAAI,QAAQ,EAAE,CAAC;oBACb,IAAI,CAAC,QAAQ,GAAG,UAAU,CAAC,QAAQ,CAAC,CAAC,CAAC,EAAE,OAAO,CAAC,CAAC;gBACnD,CAAC;gBACD,iBAAiB;gBACjB,MAAM,WAAW,GAAG,QAAQ,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,iCAAiC,CAAC,CAAC;gBACzE,IAAI,WAAW,EAAE,CAAC;oBAChB,IAAI,CAAC,WAAW,GAAG,WAAW,CAAC,CAAC,CAAC,CAAC;gBACpC,CAAC;YACH,CAAC;YAED,sBAAsB;YACtB,MAAM,WAAW,GAAG,KAAK,CAAC,KAAK,CAAC,4BAA4B,CAAC,CAAC;YAC9D,IAAI,WAAW,EAAE,CAAC;gBAChB,IAAI,CAAC,WAAW,GAAG,WAAW,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;YAC3C,CAAC;YAED,qBAAqB;YACrB,MAAM,UAAU,GAAG,KAAK,CAAC,KAAK,CAAC,2BAA2B,CAAC,CAAC;YAC5D,IAAI,UAAU,EAAE,CAAC;gBACf,IAAI,CAAC,UAAU,GAAG,UAAU,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;YACzC,CAAC;YAED,uBAAuB;YACvB,MAAM,YAAY,GAAG,KAAK,CAAC,KAAK,CAAC,6BAA6B,CAAC,CAAC;YAChE,IAAI,YAAY,EAAE,CAAC;gBACjB,IAAI,CAAC,YAAY,GAAG,YAAY,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;YAC7C,CAAC;YAED,IAAI,IAAI,CAAC,WAAW,EAAE,CAAC;gBACrB,KAAK,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YACnB,CAAC;QACH,CAAC;QAED,OAAO,KAAK,CAAC;IACf,CAAC;IAED;;OAEG;IACH,sBAAsB,CAAC,OAAe;QACpC,MAAM,OAAO,GAAa,EAAE,CAAC;QAC7B,MAAM,WAAW,GAAG,2EAA2E,CAAC;QAChG,IAAI,KAAK,CAAC;QAEV,OAAO,CAAC,KAAK,GAAG,WAAW,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,KAAK,IAAI,EAAE,CAAC;YACpD,IAAI,CAAC,OAAO,CAAC,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;gBAChC,OAAO,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;YACzB,CAAC;QACH,CAAC;QAED,OAAO,OAAO,CAAC;IACjB,CAAC;IAED;;;OAGG;IACH,uBAAuB,CACrB,OAAe,EACf,GAAsB;QAEtB,2CAA2C;QAC3C,MAAM,gBAAgB,GAAG,8CAA8C,CAAC;QACxE,IAAI,KAAK,CAAC;QAEV,OAAO,CAAC,KAAK,GAAG,gBAAgB,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,KAAK,IAAI,EAAE,CAAC;YACzD,MAAM,KAAK,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC;YACvB,MAAM,QAAQ,GAAG,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;YAE/C,8BAA8B;YAC9B,MAAM,QAAQ,GAAG,mCAAmC,CAAC;YACrD,IAAI,QAAQ,CAAC;YAEb,OAAO,CAAC,QAAQ,GAAG,QAAQ,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,KAAK,IAAI,EAAE,CAAC;gBAClD,MAAM,IAAI,GAAG,QAAQ,CAAC,CAAC,CAAC,CAAC;gBACzB,MAAM,GAAG,GAAG,QAAQ,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;gBAE/B,GAAG,CAAC,YAAY,CAAC,IAAI,CAAC,GAAG,GAAG,GAAG,GAAG,CAAC;gBAEnC,IAAI,CAAC,GAAG,CAAC,gBAAgB,CAAC,IAAI,CAAC,EAAE,CAAC;oBAChC,GAAG,CAAC,gBAAgB,CAAC,IAAI,CAAC,GAAG,EAAE,CAAC;gBAClC,CAAC;gBAED,MAAM,SAAS,GAAG,aAAa,QAAQ,GAAG,IAAI,EAAE,CAAC;gBACjD,MAAM,MAAM,GAAG,GAAG,CAAC,gBAAgB,CAAC,IAAI,CAAC,CAAC,IAAI,CAC5C,CAAC,IAAsB,EAAE,EAAE,CAAC,IAAI,CAAC,GAAG,KAAK,SAAS,CACnD,CAAC;gBAEF,IAAI,CAAC,MAAM,EAAE,CAAC;oBACZ,GAAG,CAAC,gBAAgB,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC;wBAC9B,GAAG,EAAE,SAAS;wBACd,KAAK,EAAE,YAAY;wBACnB,KAAK,EAAE,GAAG;wBACV,KAAK,EAAE,EAAE;wBACT,QAAQ;wBACR,MAAM,EAAE,YAAY;qBACrB,CAAC,CAAC;gBACL,CAAC;YACH,CAAC;QACH,CAAC;IACH,CAAC;IAED;;OAEG;IACH,KAAK,CAAC,UAAU,CACd,GAAe,EACf,WAA6B,EAC7B,OAAe,EACf,GAAsB;QAEtB,yCAAyC;QACzC,MAAM,OAAO,GAAG,MAAM,OAAO,CAAC,GAAG,CAC/B,WAAW;aACR,MAAM,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,CAAC,MAAM,KAAK,YAAY,CAAC;aACxC,GAAG,CAAC,KAAK,EAAE,EAAE,EAAE,EAAE;YAChB,IAAI,CAAC;gBACH,MAAM,MAAM,GAAG,MAAM,GAAG,CAAC,IAAI,CAAC,uBAAuB,EAAE;oBACrD,YAAY,EAAE,EAAE,CAAC,aAAa;iBAC/B,CAAC,CAAC;gBACH,OAAO,EAAE,EAAE,EAAE,IAAI,EAAE,MAAM,CAAC,IAAI,IAAI,EAAE,EAAE,CAAC;YACzC,CAAC;YAAC,MAAM,CAAC;gBACP,OAAO,EAAE,EAAE,EAAE,IAAI,EAAE,EAAE,EAAE,CAAC;YAC1B,CAAC;QACH,CAAC,CAAC,CACL,CAAC;QAEF,KAAK,MAAM,EAAE,EAAE,EAAE,IAAI,EAAE,IAAI,OAAO,EAAE,CAAC;YACnC,IAAI,CAAC,IAAI;gBAAE,SAAS;YAEpB,MAAM,OAAO,GAAG,EAAE,CAAC,UAAU,IAAI,OAAO,CAAC;YAEzC,yBAAyB;YACzB,IAAI,IAAI,CAAC,QAAQ,CAAC,YAAY,CAAC,EAAE,CAAC;gBAChC,MAAM,KAAK,GAAG,IAAI,CAAC,cAAc,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC;gBACjD,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;oBACzB,wDAAwD;oBACxD,MAAM,MAAM,GAAG,GAAG,CAAC,cAAc,CAAC,IAAI,CACpC,CAAC,CAAW,EAAE,EAAE,CACd,CAAC,CAAC,WAAW,KAAK,IAAI,CAAC,WAAW;wBAClC,CAAC,CAAC,WAAW,KAAK,IAAI,CAAC,WAAW;wBAClC,CAAC,CAAC,UAAU,KAAK,IAAI,CAAC,UAAU,CACnC,CAAC;oBACF,IAAI,CAAC,MAAM,EAAE,CAAC;wBACZ,GAAG,CAAC,cAAc,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;oBAChC,CAAC;gBACH,CAAC;YACH,CAAC;YAED,4BAA4B;YAC5B,MAAM,OAAO,GAAG,IAAI,CAAC,sBAAsB,CAAC,IAAI,CAAC,CAAC;YAClD,KAAK,MAAM,GAAG,IAAI,OAAO,EAAE,CAAC;gBAC1B,IAAI,CAAC,GAAG,CAAC,cAAc,CAAC,QAAQ,CAAC,GAAG,CAAC,EAAE,CAAC;oBACtC,GAAG,CAAC,cAAc,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;gBAC/B,CAAC;YACH,CAAC;YAED,6BAA6B;YAC7B,IAAI,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,EAAE,CAAC;gBACxB,IAAI,CAAC,uBAAuB,CAAC,IAAI,EAAE,GAAG,CAAC,CAAC;YAC1C,CAAC;QACH,CAAC;IACH,CAAC;CACF"}

package/dist/extraction/html-cleaner.d.ts

@@ -0,0 +1,16 @@
+import type { SnippedRule } from '../types/index.js';
+export interface CleanupOptions {
+    removeUnusedClasses?: boolean;
+    removeUnusedAttributes?: boolean;
+    keepTailwindLabels?: boolean;
+}
+/**
+ * Remove unused attributes and CSS classes from extracted HTML.
+ *
+ * @param html         The extracted HTML string
+ * @param snippedArr   The array of extracted CSS rules
+ * @param options      What to clean up
+ * @returns            Cleaned HTML string
+ */
+export declare function removeExtraAttributes(html: string, snippedArr: SnippedRule[], options?: CleanupOptions): string;
+//# sourceMappingURL=html-cleaner.d.ts.map

package/dist/extraction/html-cleaner.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"html-cleaner.d.ts","sourceRoot":"","sources":["../../src/extraction/html-cleaner.ts"],"names":[],"mappings":"AAKA,OAAO,KAAK,EAAE,WAAW,EAAE,MAAM,mBAAmB,CAAC;AA8CrD,MAAM,WAAW,cAAc;IAC7B,mBAAmB,CAAC,EAAE,OAAO,CAAC;IAC9B,sBAAsB,CAAC,EAAE,OAAO,CAAC;IACjC,kBAAkB,CAAC,EAAE,OAAO,CAAC;CAC9B;AAED;;;;;;;GAOG;AACH,wBAAgB,qBAAqB,CACnC,IAAI,EAAE,MAAM,EACZ,UAAU,EAAE,WAAW,EAAE,EACzB,OAAO,GAAE,cAAmB,GAC3B,MAAM,CAoGR"}

package/dist/extraction/html-cleaner.js

@@ -0,0 +1,149 @@
+/**
+ * Removes unused attributes and CSS classes from extracted HTML.
+ * Port of removeExtraAttributes() from snipbackground.js:5613-5800
+ */
+import * as cheerio from 'cheerio';
+const KEEP_ATTRS = new Set([
+    'align', 'for', 'type', 'value', 'valign', 'bgcolor', 'background',
+    'width', 'height', 'style', 'src', 'href', 'source', 'dir', 'viewbox',
+    'xmlns', 'placeholder', 'd', 'colspan', 'rowspan', 'span', 'headers',
+    'scope', 'alt', 'title', 'role', 'aria-label', 'aria-hidden',
+    'loading', 'decoding', 'srcset', 'sizes', 'media', 'rel', 'target',
+    'action', 'method', 'enctype', 'name',
+]);
+const SVG_TAGS = new Set([
+    'path', 'rect', 'circle', 'ellipse', 'line', 'polyline', 'polygon', 'svg',
+    'g', 'defs', 'use', 'symbol', 'clippath', 'mask', 'filter',
+]);
+const FORM_TAGS = new Set(['input', 'select', 'meter', 'progress', 'textarea', 'label']);
+const ICON_PREFIXES = [
+    'fa-', 'fa ', 'fas ', 'far ', 'fab ', 'fal ', 'fad ',
+    'bi-', 'bi ',
+    'icon-', 'icon ',
+    'ti-', 'ti ',
+    'material-icons',
+    'glyphicon',
+];
+function isIconFontClass(cls) {
+    const lower = cls.toLowerCase();
+    return ICON_PREFIXES.some(p => lower.startsWith(p)) || lower === 'fa';
+}
+/**
+ * Extract the base class name for selector matching.
+ * Strips pseudo-selectors, escapes, brackets, dots.
+ */
+function baseClassName(cls) {
+    let base = cls;
+    if (base.includes(':'))
+        base = base.split(':')[0];
+    if (base.includes('\\'))
+        base = base.split('\\')[0];
+    if (base.includes('/'))
+        base = base.split('/')[0];
+    if (base.includes('['))
+        base = base.split('[')[0];
+    if (base.includes('.'))
+        base = base.split('.')[0];
+    return base.trim();
+}
+/**
+ * Remove unused attributes and CSS classes from extracted HTML.
+ *
+ * @param html         The extracted HTML string
+ * @param snippedArr   The array of extracted CSS rules
+ * @param options      What to clean up
+ * @returns            Cleaned HTML string
+ */
+export function removeExtraAttributes(html, snippedArr, options = {}) {
+    const { removeUnusedClasses = true, removeUnusedAttributes = true, keepTailwindLabels = false, } = options;
+    if (!removeUnusedClasses && !removeUnusedAttributes)
+        return html;
+    // Collect all selectors for matching
+    const allSelectors = snippedArr.map(r => r.selector);
+    const $ = cheerio.load(html, { xml: { xmlMode: false, decodeEntities: false } }, false);
+    // Remove HTML comments
+    $.root().find('*').contents().filter((_i, node) => node.type === 'comment').remove();
+    // Also remove top-level comments
+    $.root().contents().filter((_i, node) => node.type === 'comment').remove();
+    $.root().find('*').each((_i, elem) => {
+        if (elem.type !== 'tag')
+            return;
+        const el = elem;
+        const tagName = (el.tagName || el.name || '').toLowerCase();
+        if (!tagName)
+            return;
+        // Skip SVG elements
+        if (SVG_TAGS.has(tagName))
+            return;
+        const attribs = el.attribs || {};
+        for (const attrName of Object.keys(attribs)) {
+            const lowerAttr = attrName.toLowerCase();
+            // Always keep essential attributes
+            if (KEEP_ATTRS.has(lowerAttr))
+                continue;
+            // Handle data-* attributes
+            if (lowerAttr.startsWith('data')) {
+                if (!removeUnusedAttributes)
+                    continue;
+                // Keep if any selector references this attribute
+                const inSelector = allSelectors.some(s => s.includes(attrName));
+                if (inSelector)
+                    continue;
+                // Remove it
+                $(elem).removeAttr(attrName);
+                continue;
+            }
+            // Handle class attribute
+            if (lowerAttr === 'class') {
+                if (!removeUnusedClasses)
+                    continue;
+                // In Tailwind mode, don't touch classes - they're utility classes, not CSS selectors
+                if (keepTailwindLabels)
+                    continue;
+                const classList = (attribs[attrName] || '').split(/\s+/).filter(Boolean);
+                const removeList = [];
+                for (const cls of classList) {
+                    // Keep icon font classes
+                    if (isIconFontClass(cls))
+                        continue;
+                    const base = baseClassName(cls);
+                    if (base === '')
+                        continue;
+                    // Check if this class appears in any selector
+                    const found = allSelectors.some(s => s.includes(base));
+                    if (!found) {
+                        removeList.push(cls);
+                    }
+                }
+                for (const cls of removeList) {
+                    $(elem).removeClass(cls);
+                }
+                continue;
+            }
+            // Handle id and name
+            if (lowerAttr === 'id' || lowerAttr === 'name') {
+                if (!removeUnusedAttributes)
+                    continue;
+                // Keep on form elements
+                if (FORM_TAGS.has(tagName))
+                    continue;
+                // Keep if any selector references this value
+                const attrVal = attribs[attrName];
+                const inSelector = allSelectors.some(s => s.includes(attrVal));
+                if (inSelector)
+                    continue;
+                $(elem).removeAttr(attrName);
+                continue;
+            }
+            // All other non-essential attributes
+            if (removeUnusedAttributes) {
+                $(elem).removeAttr(attrName);
+            }
+        }
+    });
+    let result = $.root().html() || '';
+    // Clean up empty class attributes
+    result = result.replace(/\s*class=""\s*/g, ' ').replace(/\s+>/g, '>');
+    return result;
+}
+//# sourceMappingURL=html-cleaner.js.map

package/dist/extraction/html-cleaner.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"html-cleaner.js","sourceRoot":"","sources":["../../src/extraction/html-cleaner.ts"],"names":[],"mappings":"AAAA;;;GAGG;AACH,OAAO,KAAK,OAAO,MAAM,SAAS,CAAC;AAGnC,MAAM,UAAU,GAAG,IAAI,GAAG,CAAC;IACzB,OAAO,EAAE,KAAK,EAAE,MAAM,EAAE,OAAO,EAAE,QAAQ,EAAE,SAAS,EAAE,YAAY;IAClE,OAAO,EAAE,QAAQ,EAAE,OAAO,EAAE,KAAK,EAAE,MAAM,EAAE,QAAQ,EAAE,KAAK,EAAE,SAAS;IACrE,OAAO,EAAE,aAAa,EAAE,GAAG,EAAE,SAAS,EAAE,SAAS,EAAE,MAAM,EAAE,SAAS;IACpE,OAAO,EAAE,KAAK,EAAE,OAAO,EAAE,MAAM,EAAE,YAAY,EAAE,aAAa;IAC5D,SAAS,EAAE,UAAU,EAAE,QAAQ,EAAE,OAAO,EAAE,OAAO,EAAE,KAAK,EAAE,QAAQ;IAClE,QAAQ,EAAE,QAAQ,EAAE,SAAS,EAAE,MAAM;CACtC,CAAC,CAAC;AAEH,MAAM,QAAQ,GAAG,IAAI,GAAG,CAAC;IACvB,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,SAAS,EAAE,MAAM,EAAE,UAAU,EAAE,SAAS,EAAE,KAAK;IACzE,GAAG,EAAE,MAAM,EAAE,KAAK,EAAE,QAAQ,EAAE,UAAU,EAAE,MAAM,EAAE,QAAQ;CAC3D,CAAC,CAAC;AAEH,MAAM,SAAS,GAAG,IAAI,GAAG,CAAC,CAAC,OAAO,EAAE,QAAQ,EAAE,OAAO,EAAE,UAAU,EAAE,UAAU,EAAE,OAAO,CAAC,CAAC,CAAC;AAEzF,MAAM,aAAa,GAAG;IACpB,KAAK,EAAE,KAAK,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM;IACpD,KAAK,EAAE,KAAK;IACZ,OAAO,EAAE,OAAO;IAChB,KAAK,EAAE,KAAK;IACZ,gBAAgB;IAChB,WAAW;CACZ,CAAC;AAEF,SAAS,eAAe,CAAC,GAAW;IAClC,MAAM,KAAK,GAAG,GAAG,CAAC,WAAW,EAAE,CAAC;IAChC,OAAO,aAAa,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,KAAK,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,IAAI,KAAK,KAAK,IAAI,CAAC;AACxE,CAAC;AAED;;;GAGG;AACH,SAAS,aAAa,CAAC,GAAW;IAChC,IAAI,IAAI,GAAG,GAAG,CAAC;IACf,IAAI,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC;QAAE,IAAI,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;IAClD,IAAI,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC;QAAE,IAAI,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC;IACpD,IAAI,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC;QAAE,IAAI,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;IAClD,IAAI,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC;QAAE,IAAI,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;IAClD,IAAI,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC;QAAE,IAAI,GAAG,IAAI,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;IAClD,OAAO,IAAI,CAAC,IAAI,EAAE,CAAC;AACrB,CAAC;AAQD;;;;;;;GAOG;AACH,MAAM,UAAU,qBAAqB,CACnC,IAAY,EACZ,UAAyB,EACzB,UAA0B,EAAE;IAE5B,MAAM,EACJ,mBAAmB,GAAG,IAAI,EAC1B,sBAAsB,GAAG,IAAI,EAC7B,kBAAkB,GAAG,KAAK,GAC3B,GAAG,OAAO,CAAC;IAEZ,IAAI,CAAC,mBAAmB,IAAI,CAAC,sBAAsB;QAAE,OAAO,IAAI,CAAC;IAEjE,qCAAqC;IACrC,MAAM,YAAY,GAAG,UAAU,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC;IAErD,MAAM,CAAC,GAAG,OAAO,CAAC,IAAI,CAAC,IAAI,EAAE,EAAE,GAAG,EAAE,EAAE,OAAO,EAAE,KAAK,EAAE,cAAc,EAAE,KAAK,EAAE,EAAE,EAAE,KAAK,CAAC,CAAC;IAExF,uBAAuB;IACvB,CAAC,CAAC,IAAI,EAAE,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,QAAQ,EAAE,CAAC,MAAM,CAAC,CAAC,EAAE,EAAE,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,KAAK,SAAS,CAAC,CAAC,MAAM,EAAE,CAAC;IACrF,iCAAiC;IACjC,CAAC,CAAC,IAAI,EAAE,CAAC,QAAQ,EAAE,CAAC,MAAM,CAAC,CAAC,EAAE,EAAE,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,IAAI,KAAK,SAAS,CAAC,CAAC,MAAM,EAAE,CAAC;IAE3E,CAAC,CAAC,IAAI,EAAE,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,IAAI,CAAC,CAAC,EAAE,EAAE,IAAI,EAAE,EAAE;QACnC,IAAI,IAAI,CAAC,IAAI,KAAK,KAAK;YAAE,OAAO;QAChC,MAAM,EAAE,GAAG,IAAW,CAAC;QACvB,MAAM,OAAO,GAAG,CAAC,EAAE,CAAC,OAAO,IAAI,EAAE,CAAC,IAAI,IAAI,EAAE,CAAC,CAAC,WAAW,EAAE,CAAC;QAC5D,IAAI,CAAC,OAAO;YAAE,OAAO;QAErB,oBAAoB;QACpB,IAAI,QAAQ,CAAC,GAAG,CAAC,OAAO,CAAC;YAAE,OAAO;QAElC,MAAM,OAAO,GAAG,EAAE,CAAC,OAAO,IAAI,EAAE,CAAC;QAEjC,KAAK,MAAM,QAAQ,IAAI,MAAM,CAAC,IAAI,CAAC,OAAO,CAAC,EAAE,CAAC;YAC5C,MAAM,SAAS,GAAG,QAAQ,CAAC,WAAW,EAAE,CAAC;YAEzC,mCAAmC;YACnC,IAAI,UAAU,CAAC,GAAG,CAAC,SAAS,CAAC;gBAAE,SAAS;YAExC,2BAA2B;YAC3B,IAAI,SAAS,CAAC,UAAU,CAAC,MAAM,CAAC,EAAE,CAAC;gBACjC,IAAI,CAAC,sBAAsB;oBAAE,SAAS;gBACtC,iDAAiD;gBACjD,MAAM,UAAU,GAAG,YAAY,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,QAAQ,CAAC,QAAQ,CAAC,CAAC,CAAC;gBAChE,IAAI,UAAU;oBAAE,SAAS;gBACzB,YAAY;gBACZ,CAAC,CAAC,IAAI,CAAC,CAAC,UAAU,CAAC,QAAQ,CAAC,CAAC;gBAC7B,SAAS;YACX,CAAC;YAED,yBAAyB;YACzB,IAAI,SAAS,KAAK,OAAO,EAAE,CAAC;gBAC1B,IAAI,CAAC,mBAAmB;oBAAE,SAAS;gBACnC,qFAAqF;gBACrF,IAAI,kBAAkB;oBAAE,SAAS;gBAEjC,MAAM,SAAS,GAAG,CAAC,OAAO,CAAC,QAAQ,CAAC,IAAI,EAAE,CAAC,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC;gBACzE,MAAM,UAAU,GAAa,EAAE,CAAC;gBAEhC,KAAK,MAAM,GAAG,IAAI,SAAS,EAAE,CAAC;oBAC5B,yBAAyB;oBACzB,IAAI,eAAe,CAAC,GAAG,CAAC;wBAAE,SAAS;oBAEnC,MAAM,IAAI,GAAG,aAAa,CAAC,GAAG,CAAC,CAAC;oBAChC,IAAI,IAAI,KAAK,EAAE;wBAAE,SAAS;oBAE1B,8CAA8C;oBAC9C,MAAM,KAAK,GAAG,YAAY,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC;oBACvD,IAAI,CAAC,KAAK,EAAE,CAAC;wBACX,UAAU,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;oBACvB,CAAC;gBACH,CAAC;gBAED,KAAK,MAAM,GAAG,IAAI,UAAU,EAAE,CAAC;oBAC7B,CAAC,CAAC,IAAI,CAAC,CAAC,WAAW,CAAC,GAAG,CAAC,CAAC;gBAC3B,CAAC;gBACD,SAAS;YACX,CAAC;YAED,qBAAqB;YACrB,IAAI,SAAS,KAAK,IAAI,IAAI,SAAS,KAAK,MAAM,EAAE,CAAC;gBAC/C,IAAI,CAAC,sBAAsB;oBAAE,SAAS;gBACtC,wBAAwB;gBACxB,IAAI,SAAS,CAAC,GAAG,CAAC,OAAO,CAAC;oBAAE,SAAS;gBACrC,6CAA6C;gBAC7C,MAAM,OAAO,GAAG,OAAO,CAAC,QAAQ,CAAC,CAAC;gBAClC,MAAM,UAAU,GAAG,YAAY,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,QAAQ,CAAC,OAAO,CAAC,CAAC,CAAC;gBAC/D,IAAI,UAAU;oBAAE,SAAS;gBACzB,CAAC,CAAC,IAAI,CAAC,CAAC,UAAU,CAAC,QAAQ,CAAC,CAAC;gBAC7B,SAAS;YACX,CAAC;YAED,qCAAqC;YACrC,IAAI,sBAAsB,EAAE,CAAC;gBAC3B,CAAC,CAAC,IAAI,CAAC,CAAC,UAAU,CAAC,QAAQ,CAAC,CAAC;YAC/B,CAAC;QACH,CAAC;IACH,CAAC,CAAC,CAAC;IAEH,IAAI,MAAM,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,IAAI,EAAE,IAAI,EAAE,CAAC;IACnC,kCAAkC;IAClC,MAAM,GAAG,MAAM,CAAC,OAAO,CAAC,iBAAiB,EAAE,GAAG,CAAC,CAAC,OAAO,CAAC,OAAO,EAAE,GAAG,CAAC,CAAC;IACtE,OAAO,MAAM,CAAC;AAChB,CAAC"}

package/dist/extraction/keyframe-collector.d.ts

@@ -0,0 +1,16 @@
+import { CDPMatchedStyles, ExtractionContext } from '../types/index.js';
+/**
+ * Collects @keyframes animation rules from matched styles.
+ * Port of snipbackground.js:1997-2035
+ */
+export declare class KeyframeCollector {
+    /**
+     * Extract keyframe animations from matched styles.
+     */
+    collect(allMatchedStyles: CDPMatchedStyles, ctx: ExtractionContext): void;
+    /**
+     * Generate CSS text for all collected keyframes.
+     */
+    generateCss(ctx: ExtractionContext): string;
+}
+//# sourceMappingURL=keyframe-collector.d.ts.map

package/dist/extraction/keyframe-collector.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"keyframe-collector.d.ts","sourceRoot":"","sources":["../../src/extraction/keyframe-collector.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,gBAAgB,EAAE,iBAAiB,EAAE,MAAM,mBAAmB,CAAC;AAExE;;;GAGG;AACH,qBAAa,iBAAiB;IAC5B;;OAEG;IACH,OAAO,CAAC,gBAAgB,EAAE,gBAAgB,EAAE,GAAG,EAAE,iBAAiB,GAAG,IAAI;IAuCzE;;OAEG;IACH,WAAW,CAAC,GAAG,EAAE,iBAAiB,GAAG,MAAM;CAgB5C"}