recker 1.0.78 → 1.0.79-next.70f027c

package/dist/seo/analyzer.js

@@ -39,6 +39,7 @@ export class SeoAnalyzer {
         const socialLinkDetails = this.analyzeSocialLinks(links, socialDomains);
         const headings = this.analyzeHeadings();
         const content = this.analyzeContent(headings);
+        const contentSections = this.extractContentSectionSignals(headings);
         const linkAnalysis = this.buildLinkAnalysis(links);
         const imageAnalysis = this.buildImageAnalysis(images);
         const social = this.buildSocialAnalysis(og, twitter);
@@ -48,6 +49,16 @@ export class SeoAnalyzer {
         const feeds = this.analyzeFeeds();
         const conversion = this.analyzeConversionElements(links, visibleText);
         const pageType = this.detectPageType(jsonLd);
+        const linkAnchorTexts = links
+            .map((link) => link.text?.trim())
+            .filter((text) => Boolean(text))
+            .filter((text) => text.length <= 140)
+            .filter((text) => !text.match(/^\s*$/))
+            .slice(0, 80);
+        const linkUrlSamples = links
+            .map((link) => extractReadablePathFromUrl(link.href))
+            .filter((sample) => sample.length > 0)
+            .slice(0, 80);
         const context = this.buildRuleContext({
             pageType,
             meta,
@@ -71,7 +82,7 @@ export class SeoAnalyzer {
         const ruleResults = this.rulesEngine.evaluate(context);
         const checks = this.convertToCheckResults(ruleResults);
         const { score, grade } = this.calculateScore(ruleResults);
-        const summary = this.buildSummary(ruleResults, checks, {
+        const summary = this.buildSummary(checks, {
             content,
             imageAnalysis,
             linkAnalysis,
@@ -124,12 +135,15 @@ export class SeoAnalyzer {
                 items: jsonLd,
             },
             headings: headings,
+            contentSections,
             content,
             keywords,
             links: linkAnalysis,
             images: imageAnalysis,
             social,
             technical,
+            linkAnchorTexts,
+            linkUrlSamples,
         };
     }
     getMainBody() {
@@ -181,6 +195,97 @@ export class SeoAnalyzer {
             return 'other';
         }
     }
+    extractContentSectionSignals(headings) {
+        const body = this.getMainBody();
+        if (!body)
+            return [];
+        const candidateTags = 'h1,h2,h3,h4,h5,h6,p,li,ol,ul,article,section,main,figure,figcaption,blockquote,table,th,td';
+        const nodes = body.querySelectorAll(candidateTags);
+        const normalizeSignalText = (value) => value
+            .replace(/\s+/g, ' ')
+            .replace(/[,\.;:!?()[\]{}"“”']/g, '')
+            .trim()
+            .toLowerCase();
+        const headingStack = [];
+        const seen = new Set();
+        const results = [];
+        const pushSectionSignal = (entry) => {
+            const fingerprint = `${entry.tagName}|${entry.headingPath.join(' > ')}|${entry.text.slice(0, 120)}`;
+            if (seen.has(fingerprint))
+                return;
+            seen.add(fingerprint);
+            results.push(entry);
+        };
+        for (const node of nodes) {
+            const tagName = node.tagName.toLowerCase();
+            if (/^h[1-6]$/.test(tagName)) {
+                const level = Number.parseInt(tagName.slice(1), 10);
+                const text = normalizeSignalText(node.text);
+                if (!text || text.length < 4) {
+                    continue;
+                }
+                while (headingStack.length > 0 && headingStack[headingStack.length - 1].level >= level) {
+                    headingStack.pop();
+                }
+                headingStack.push({ level, text });
+                const headingPath = headingStack.map((entry) => entry.text);
+                const source = {
+                    headingPath,
+                    source: 'heading',
+                    tagName,
+                    headingLevel: level,
+                    text,
+                    wordCount: text.split(/\s+/).length,
+                    linkDensity: 0,
+                    weight: 1.45,
+                };
+                pushSectionSignal(source);
+                continue;
+            }
+            if (tagName === 'ul' || tagName === 'ol') {
+                continue;
+            }
+            const text = normalizeSignalText(node.text);
+            if (!text || text.split(/\s+/).length < 4) {
+                continue;
+            }
+            const words = text.split(/\s+/);
+            const wordCount = words.length;
+            if (wordCount < 4)
+                continue;
+            const anchorCount = node.querySelectorAll('a').length;
+            const linkDensity = wordCount > 0
+                ? Math.min(1, anchorCount / Math.max(1, wordCount))
+                : 0;
+            if (linkDensity > 0.85)
+                continue;
+            const headingPath = headingStack.length > 0
+                ? headingStack.map((entry) => entry.text)
+                : headings.h1Count > 0
+                    ? ['conteúdo']
+                    : ['home'];
+            const sourceLabel = tagName === 'li' ? 'list-item'
+                : tagName === 'figure' || tagName === 'figcaption' ? 'figure'
+                    : (tagName === 'table' || tagName === 'th' || tagName === 'td') ? 'table'
+                        : 'paragraph';
+            const headingLevel = headingStack.length > 0 ? headingStack[headingStack.length - 1].level : 1;
+            const baseWeight = sourceLabel === 'paragraph' ? 1.25 : 1.05;
+            const adjustedWeight = baseWeight * (1 - (linkDensity * 0.4));
+            pushSectionSignal({
+                headingPath,
+                source: sourceLabel,
+                tagName,
+                headingLevel,
+                text,
+                wordCount,
+                linkDensity,
+                weight: Math.max(0.5, adjustedWeight),
+            });
+        }
+        return results
+            .filter((entry) => entry.text.length >= 20)
+            .slice(0, 160);
+    }
     getVisibleText() {
         const body = this.getMainBody();
         if (!body)
@@ -951,6 +1056,27 @@ export class SeoAnalyzer {
         }
         return undefined;
     }
+    classifyInfoCheck(result) {
+        if (result.status !== 'info')
+            return undefined;
+        const message = (result.message || '').toLowerCase();
+        const evidenceIssue = (result.evidence?.issue || '').toLowerCase();
+        const recommendation = (result.recommendation || '').toLowerCase();
+        const context = `${message} ${evidenceIssue} ${recommendation}`;
+        if (/not\s+applicable/.test(context) ||
+            /not\s+available/.test(context) ||
+            /data\s+unavailable/.test(context) ||
+            /context\s+unavailable/.test(context) ||
+            /unable\s+to\s+check/.test(context) ||
+            /cannot\s+verify/.test(context) ||
+            /cannot\s+be\s+determined/.test(context) ||
+            /\bunavailable\b/.test(context) ||
+            /not\s+set/.test(context) ||
+            /not\s+present/.test(context)) {
+            return 'not_applicable';
+        }
+        return 'suggestion';
+    }
     convertToCheckResults(results) {
         return results.map((r) => ({
             id: r.id,
@@ -959,26 +1085,36 @@ export class SeoAnalyzer {
             severity: r.severity,
             status: r.status,
             message: r.message,
+            infoType: this.classifyInfoCheck(r),
             value: r.value,
             recommendation: r.recommendation,
             evidence: r.evidence,
         }));
     }
-    buildSummary(ruleResults, checks, data) {
+    buildSummary(checks, data) {
         const pageType = data.pageType;
         const timings = data.timings;
         const passed = checks.filter((c) => c.status === 'pass').length;
         const warnings = checks.filter((c) => c.status === 'warn').length;
         const errors = checks.filter((c) => c.status === 'fail').length;
         const infos = checks.filter((c) => c.status === 'info').length;
+        const notApplicable = checks.filter((c) => c.status === 'info' && c.infoType === 'not_applicable').length;
+        const suggestions = checks.filter((c) => c.status === 'info' && c.infoType !== 'not_applicable').length;
         const totalChecks = checks.length;
         const scoringChecks = totalChecks - infos;
         const passRate = scoringChecks > 0 ? Math.round((passed / scoringChecks) * 100) : 100;
         const issuesByCategory = {};
-        for (const result of ruleResults) {
+        for (const result of checks) {
             const cat = result.category;
             if (!issuesByCategory[cat]) {
-                issuesByCategory[cat] = { passed: 0, warnings: 0, errors: 0 };
+                issuesByCategory[cat] = {
+                    passed: 0,
+                    warnings: 0,
+                    errors: 0,
+                    infos: 0,
+                    notApplicable: 0,
+                    suggestions: 0,
+                };
             }
             if (result.status === 'pass')
                 issuesByCategory[cat].passed++;
@@ -986,6 +1122,15 @@ export class SeoAnalyzer {
                 issuesByCategory[cat].warnings++;
             else if (result.status === 'fail')
                 issuesByCategory[cat].errors++;
+            else if (result.status === 'info') {
+                issuesByCategory[cat].infos++;
+                if (result.infoType === 'not_applicable') {
+                    issuesByCategory[cat].notApplicable++;
+                }
+                else {
+                    issuesByCategory[cat].suggestions++;
+                }
+            }
         }
         const topIssues = checks
             .filter((c) => c.status === 'fail' || c.status === 'warn')
@@ -1052,6 +1197,8 @@ export class SeoAnalyzer {
             warnings,
             errors,
             infos,
+            notApplicable,
+            suggestions,
             passRate,
             issuesByCategory,
             pageType: pageType,
@@ -1179,7 +1326,7 @@ export class SeoAnalyzer {
                 const level = parseInt(tagName.substring(1), 10);
                 const text = el.text.trim();
                 counts[level] = (counts[level] || 0) + 1;
-                structure.push({ level, text: text.slice(0, 80), count: 1 });
+                structure.push({ level, text: text.slice(0, 120), count: 1 });
                 if (level === 2) {
                     if (inSection) {
                         sectionWordCounts.push(currentSectionWordCount);
@@ -1219,8 +1366,24 @@ export class SeoAnalyzer {
         else if (counts[1] > 1) {
             issues.push('Multiple H1 tags');
         }
+        const tree = [];
+        const stack = [];
+        for (const heading of structure) {
+            const node = { level: heading.level, text: heading.text, children: [] };
+            while (stack.length > 0 && stack[stack.length - 1].level >= heading.level) {
+                stack.pop();
+            }
+            if (stack.length === 0) {
+                tree.push(node);
+            }
+            else {
+                stack[stack.length - 1].node.children.push(node);
+            }
+            stack.push({ node, level: heading.level });
+        }
         return {
             structure,
+            tree,
             h1Count: counts[1],
             hasProperHierarchy,
             issues,
@@ -1453,6 +1616,28 @@ export class SeoAnalyzer {
         return this.rulesEngine.getCategories();
     }
 }
+function extractReadablePathFromUrl(value) {
+    try {
+        const parsed = new URL(value);
+        const fragments = parsed.pathname
+            .split('/')
+            .filter((segment) => segment.length > 0)
+            .slice(0, 3)
+            .map((segment) => segment.replace(/[-_]/g, ' '))
+            .join(' ')
+            .replace(/\d+/g, ' ')
+            .replace(/\s+/g, ' ')
+            .trim();
+        const host = parsed.hostname.replace(/^www\./, '').replace(/\./g, ' ');
+        if (fragments.length > 0) {
+            return `${host} ${fragments}`.trim();
+        }
+        return host;
+    }
+    catch {
+        return '';
+    }
+}
 export async function analyzeSeo(html, options = {}) {
     const analyzer = await SeoAnalyzer.fromHtml(html, options);
     return analyzer.analyze();

package/dist/seo/formatter.js

@@ -29,7 +29,16 @@ export function statusIcon(status) {
 export function formatCheck(check, showEvidence = false) {
     const lines = [];
     const icon = statusIcon(check.status);
-    lines.push(` ${icon} ${check.message}`);
+    let infoLabel = '';
+    if (check.status === 'info') {
+        if (check.infoType === 'not_applicable') {
+            infoLabel = colors.magenta(' [N/A]');
+        }
+        else if (check.infoType === 'suggestion') {
+            infoLabel = colors.cyan(' [SUGGESTION]');
+        }
+    }
+    lines.push(` ${icon} ${check.message}${infoLabel}`);
     if (check.value !== undefined) {
         lines.push(`    ${colors.gray('Value:')} ${check.value}`);
     }
@@ -239,7 +248,10 @@ export function formatSeoReport(report, url, options = {}) {
     lines.push(colors.gray(' ──────────────────────────────────────────────────'));
     lines.push('');
     lines.push(` ${colors.bold('Summary')}`);
-    lines.push(` ${colors.green('✔ Passed:')} ${s.passed}   ${colors.yellow('⚠ Warnings:')} ${s.warnings}   ${colors.red('✖ Errors:')} ${s.errors}   ${colors.blue('ℹ Info:')} ${s.infos}`);
+    const infoSummary = (s.notApplicable !== undefined || s.suggestions !== undefined)
+        ? ` (${s.notApplicable ?? 0} not applicable, ${s.suggestions ?? 0} suggestions)`
+        : '';
+    lines.push(` ${colors.green('✔ Passed:')} ${s.passed}   ${colors.yellow('⚠ Warnings:')} ${s.warnings}   ${colors.red('✖ Errors:')} ${s.errors}   ${colors.blue('ℹ Info:')} ${s.infos}${infoSummary}`);
     if (s.vitals) {
         lines.push(` ${colors.gray('Words:')} ${s.vitals.wordCount} | ${colors.gray('Images:')} ${s.vitals.imageCount} | ${colors.gray('Links:')} ${s.vitals.linkCount}`);
     }
@@ -381,12 +393,16 @@ function groupChecksByCategory(checks) {
         const warnings = catChecks.filter(c => c.status === 'warn').length;
         const errors = catChecks.filter(c => c.status === 'fail').length;
         const infos = catChecks.filter(c => c.status === 'info').length;
+        const notApplicable = catChecks.filter(c => c.status === 'info' && c.infoType === 'not_applicable').length;
+        const suggestions = catChecks.filter(c => c.status === 'info' && c.infoType !== 'not_applicable').length;
         const total = catChecks.length - infos;
         result[cat] = {
             passed,
             warnings,
             errors,
             infos,
+            notApplicable,
+            suggestions,
             passRate: total > 0 ? Math.round((passed / total) * 100) : 100,
             checks: catChecks,
         };
@@ -410,6 +426,8 @@ export function formatSeoReportJson(report, url, options) {
             warnings: report.checks.filter(c => c.status === 'warn').length,
             errors: report.checks.filter(c => c.status === 'fail').length,
             infos: report.checks.filter(c => c.status === 'info').length,
+            notApplicable: report.checks.filter(c => c.status === 'info' && c.infoType === 'not_applicable').length,
+            suggestions: report.checks.filter(c => c.status === 'info' && c.infoType !== 'not_applicable').length,
             passRate: report.summary.passRate,
             completeness: report.summary.completeness,
             topIssues: report.summary.topIssues,

package/dist/seo/index.d.ts

@@ -6,7 +6,7 @@ export type { SeoSpiderOptions, SeoPageResult, SiteWideIssue, SeoSpiderResult, }
 export { SeoRulesEngine, createRulesEngine, SEO_THRESHOLDS, ALL_SEO_RULES, } from './rules/index.js';
 export { generateSeoFilename, resolveOutputPath, writeReport, formatReportForJson, } from './output.js';
 export type { SeoOutputType, OutputOptions, WriteOptions } from './output.js';
-export type { SeoReport, SeoCheckResult, SeoStatus, SeoPageType, SeoTiming, HeadingAnalysis, HeadingInfo, ContentMetrics, LinkAnalysis, ImageAnalysis, SocialMetaAnalysis, TechnicalSeo, SeoAnalyzerOptions, } from './types.js';
+export type { SeoReport, SeoCheckResult, SeoStatus, SeoPageType, SeoTiming, HeadingAnalysis, HeadingInfo, HeadingNode, ContentMetrics, LinkAnalysis, ImageAnalysis, SocialMetaAnalysis, TechnicalSeo, SeoAnalyzerOptions, } from './types.js';
 export type { SeoRule, RuleContext, RuleResult, RuleEvidence, RuleCategory, RuleSeverity, RulesEngineOptions, } from './rules/index.js';
 export type { SeoAnalyzerFullOptions } from './analyzer.js';
 export { parseRobotsTxt, validateRobotsTxt, isPathAllowed, fetchAndValidateRobotsTxt, } from './validators/robots.js';

package/dist/seo/keyword-campaign-analyzer.d.ts

@@ -0,0 +1,2 @@
+import { type KeywordCampaignOptions, type KeywordCampaignReport } from './keyword-campaign-shared.js';
+export declare function analyzeKeywordCampaign(options: KeywordCampaignOptions): Promise<KeywordCampaignReport>;