recker 1.0.79 → 1.0.80-next.70ea84d
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/dist/browser/browser/mini.d.ts +2 -2
- package/dist/browser/browser/recker-mini.d.ts +8 -8
- package/dist/browser/browser/recker.d.ts +11 -8
- package/dist/browser/browser/recker.js +54 -6
- package/dist/browser/core/client.d.ts +15 -10
- package/dist/browser/core/client.js +54 -38
- package/dist/browser/index.iife.min.js +129 -130
- package/dist/browser/index.min.js +129 -130
- package/dist/browser/index.mini.iife.js +5697 -636
- package/dist/browser/index.mini.iife.min.js +47 -48
- package/dist/browser/index.mini.min.js +47 -48
- package/dist/browser/index.mini.umd.js +5697 -636
- package/dist/browser/index.mini.umd.min.js +47 -48
- package/dist/browser/index.umd.min.js +129 -130
- package/dist/browser/mini.d.ts +2 -2
- package/dist/browser/plugins/proxy-rotator.d.ts +2 -2
- package/dist/browser/plugins/proxy-rotator.js +6 -28
- package/dist/browser/recker-mini.d.ts +8 -8
- package/dist/browser/recker.d.ts +11 -8
- package/dist/browser/recker.js +54 -6
- package/dist/browser/scrape/document.js +2 -2
- package/dist/browser/scrape/element.js +7 -1
- package/dist/browser/scrape/parser/nodes/html.js +1 -1
- package/dist/browser/scrape/spider.d.ts +52 -0
- package/dist/browser/scrape/spider.js +620 -38
- package/dist/browser/scrape/types.d.ts +2 -0
- package/dist/browser/search/google.d.ts +26 -1
- package/dist/browser/search/google.js +427 -45
- package/dist/browser/seo/analyzer.d.ts +1 -0
- package/dist/browser/seo/analyzer.js +144 -1
- package/dist/browser/seo/index.d.ts +1 -1
- package/dist/browser/seo/keyword-campaign-analyzer.d.ts +2 -0
- package/dist/browser/seo/keyword-campaign-analyzer.js +538 -0
- package/dist/browser/seo/keyword-campaign-seed-advanced.d.ts +17 -0
- package/dist/browser/seo/keyword-campaign-seed-advanced.js +269 -0
- package/dist/browser/seo/keyword-campaign-seed-core.d.ts +29 -0
- package/dist/browser/seo/keyword-campaign-seed-core.js +525 -0
- package/dist/browser/seo/keyword-campaign-shared.d.ts +165 -0
- package/dist/browser/seo/keyword-campaign-shared.js +59 -0
- package/dist/browser/seo/keyword-campaign.d.ts +4 -107
- package/dist/browser/seo/keyword-campaign.js +2 -380
- package/dist/browser/seo/keywords.js +5 -22
- package/dist/browser/seo/types.d.ts +19 -0
- package/dist/browser/transport/curl.d.ts +5 -1
- package/dist/browser/transport/curl.js +207 -50
- package/dist/browser/transport/undici.d.ts +4 -3
- package/dist/browser/transport/undici.js +123 -49
- package/dist/browser/types/index.d.ts +9 -3
- package/dist/browser/utils/binary-manager.js +26 -3
- package/dist/browser/utils/block-detector.d.ts +8 -0
- package/dist/browser/utils/block-detector.js +542 -7
- package/dist/cli/commands/hls-runner.js +5 -4
- package/dist/cli/commands/live-runner.js +5 -4
- package/dist/cli/commands/loadtest-runner.js +3 -2
- package/dist/cli/commands/search.d.ts +2 -0
- package/dist/cli/commands/search.js +105 -0
- package/dist/cli/commands/seo-runner.js +9 -7
- package/dist/cli/commands/seo.js +140 -1
- package/dist/cli/commands/serve.js +75 -131
- package/dist/cli/commands/server-runner.js +59 -82
- package/dist/cli/commands/spider-runner.d.ts +37 -1
- package/dist/cli/commands/spider-runner.js +134 -10
- package/dist/cli/commands/spider.d.ts +18 -1
- package/dist/cli/commands/spider.js +457 -27
- package/dist/cli/events/handlers/cli.js +30 -1
- package/dist/cli/events/handlers/tui.js +26 -0
- package/dist/cli/events/types.d.ts +27 -0
- package/dist/cli/handler.d.ts +2 -12
- package/dist/cli/handler.js +20 -15
- package/dist/cli/handlers/protocols.js +39 -12
- package/dist/cli/handlers/search.d.ts +2 -0
- package/dist/cli/handlers/search.js +171 -0
- package/dist/cli/handlers/seo-analyze.d.ts +1 -0
- package/dist/cli/handlers/seo-analyze.js +666 -0
- package/dist/cli/handlers/seo-robots.d.ts +1 -0
- package/dist/cli/handlers/seo-robots.js +76 -0
- package/dist/cli/handlers/seo-serp.d.ts +54 -0
- package/dist/cli/handlers/seo-serp.js +243 -0
- package/dist/cli/handlers/seo-sitemap.d.ts +1 -0
- package/dist/cli/handlers/seo-sitemap.js +55 -0
- package/dist/cli/handlers/seo-spider.d.ts +1 -0
- package/dist/cli/handlers/seo-spider.js +334 -0
- package/dist/cli/handlers/seo.d.ts +8 -4
- package/dist/cli/handlers/seo.js +294 -442
- package/dist/cli/handlers/spider.js +94 -17
- package/dist/cli/handlers/streaming.js +5 -1
- package/dist/cli/index.js +11 -2
- package/dist/cli/presets.d.ts +1 -1
- package/dist/cli/presets.js +15 -4
- package/dist/cli/tui/app.js +6 -1
- package/dist/cli/tui/components/rich-response.d.ts +72 -0
- package/dist/cli/tui/components/rich-response.js +117 -0
- package/dist/cli/tui/executor-commands/background.js +30 -24
- package/dist/cli/tui/executor-commands/testing.js +3 -2
- package/dist/cli/tui/hooks/useDomains.d.ts +17 -0
- package/dist/cli/tui/hooks/useHelp.js +15 -2
- package/dist/cli/tui/job-manager.d.ts +4 -4
- package/dist/cli/tui/job-manager.js +5 -1
- package/dist/cli/tui/spider-tui.d.ts +63 -0
- package/dist/cli/tui/spider-tui.js +120 -2
- package/dist/cli/types.d.ts +12 -0
- package/dist/cli/types.js +1 -0
- package/dist/cli/utils/option-helpers.d.ts +10 -0
- package/dist/cli/utils/option-helpers.js +63 -0
- package/dist/cli/utils/score-color.d.ts +11 -0
- package/dist/cli/utils/score-color.js +11 -0
- package/dist/cli/utils/serp-campaign.d.ts +53 -0
- package/dist/cli/utils/serp-campaign.js +53 -0
- package/dist/cli/utils/serp-config.d.ts +26 -0
- package/dist/cli/utils/serp-config.js +125 -0
- package/dist/core/client.d.ts +15 -10
- package/dist/core/client.js +54 -38
- package/dist/index.d.ts +1 -0
- package/dist/index.js +1 -0
- package/dist/mcp/cli.js +35 -34
- package/dist/mcp/client.d.ts +2 -2
- package/dist/mcp/client.js +20 -2
- package/dist/mcp/contract.d.ts +1 -1
- package/dist/mcp/profiles.js +5 -1
- package/dist/mcp/prompts/index.js +8 -4
- package/dist/mcp/resources/index.js +46 -23
- package/dist/mcp/server.js +9 -6
- package/dist/mcp/tools/protocols.js +9 -2
- package/dist/mcp/tools/registry.js +13 -3
- package/dist/mcp/tools/seo.js +427 -2
- package/dist/mcp/types.d.ts +5 -1
- package/dist/plugins/proxy-rotator.d.ts +2 -2
- package/dist/plugins/proxy-rotator.js +6 -28
- package/dist/raffel/client.d.ts +38 -0
- package/dist/raffel/client.js +282 -0
- package/dist/raffel/index.d.ts +2 -0
- package/dist/raffel/index.js +2 -0
- package/dist/raffel/types.d.ts +40 -0
- package/dist/raffel/types.js +14 -0
- package/dist/recker.d.ts +13 -7
- package/dist/recker.js +58 -6
- package/dist/scrape/document.js +2 -2
- package/dist/scrape/element.js +7 -1
- package/dist/scrape/parser/nodes/html.js +1 -1
- package/dist/scrape/spider.d.ts +52 -0
- package/dist/scrape/spider.js +620 -38
- package/dist/scrape/types.d.ts +2 -0
- package/dist/search/google.d.ts +26 -1
- package/dist/search/google.js +427 -45
- package/dist/search/index.d.ts +1 -1
- package/dist/seo/analyzer.d.ts +1 -0
- package/dist/seo/analyzer.js +144 -1
- package/dist/seo/index.d.ts +1 -1
- package/dist/seo/keyword-campaign-analyzer.d.ts +2 -0
- package/dist/seo/keyword-campaign-analyzer.js +538 -0
- package/dist/seo/keyword-campaign-seed-advanced.d.ts +17 -0
- package/dist/seo/keyword-campaign-seed-advanced.js +269 -0
- package/dist/seo/keyword-campaign-seed-core.d.ts +29 -0
- package/dist/seo/keyword-campaign-seed-core.js +525 -0
- package/dist/seo/keyword-campaign-shared.d.ts +165 -0
- package/dist/seo/keyword-campaign-shared.js +59 -0
- package/dist/seo/keyword-campaign.d.ts +4 -107
- package/dist/seo/keyword-campaign.js +2 -380
- package/dist/seo/keywords.js +5 -22
- package/dist/seo/types.d.ts +19 -0
- package/dist/template/index.d.ts +1 -1
- package/dist/template/types.d.ts +0 -2
- package/dist/testing/index.d.ts +1 -22
- package/dist/testing/index.js +1 -11
- package/dist/transport/curl.d.ts +5 -1
- package/dist/transport/curl.js +207 -50
- package/dist/transport/undici.d.ts +4 -3
- package/dist/transport/undici.js +123 -49
- package/dist/types/index.d.ts +9 -3
- package/dist/utils/binary-manager.js +26 -3
- package/dist/utils/block-detector.d.ts +8 -0
- package/dist/utils/block-detector.js +542 -7
- package/dist/version.js +1 -1
- package/package.json +12 -1
- package/dist/testing/mock-dns-server.d.ts +0 -69
- package/dist/testing/mock-dns-server.js +0 -269
- package/dist/testing/mock-ftp-server.d.ts +0 -89
- package/dist/testing/mock-ftp-server.js +0 -562
- package/dist/testing/mock-hls-server.d.ts +0 -80
- package/dist/testing/mock-hls-server.js +0 -381
- package/dist/testing/mock-http-server.d.ts +0 -124
- package/dist/testing/mock-http-server.js +0 -343
- package/dist/testing/mock-proxy-server.d.ts +0 -108
- package/dist/testing/mock-proxy-server.js +0 -615
- package/dist/testing/mock-sse-server.d.ts +0 -76
- package/dist/testing/mock-sse-server.js +0 -291
- package/dist/testing/mock-telnet-server.d.ts +0 -59
- package/dist/testing/mock-telnet-server.js +0 -274
- package/dist/testing/mock-udp-server.d.ts +0 -43
- package/dist/testing/mock-udp-server.js +0 -188
- package/dist/testing/mock-websocket-server.d.ts +0 -76
- package/dist/testing/mock-websocket-server.js +0 -334
- package/dist/testing/mock-whois-server.d.ts +0 -56
- package/dist/testing/mock-whois-server.js +0 -234
- package/dist/testing/proxy-certs.d.ts +0 -19
- package/dist/testing/proxy-certs.js +0 -208
|
@@ -0,0 +1,538 @@
|
|
|
1
|
+
import { searchGoogleAdvanced } from '../search/google.js';
|
|
2
|
+
import { clampNumber, normalizeHost, normalizeKeyword, normalizePlacement, toSeedArray, DEFAULT_SERP_MIN_KEYWORD_LENGTH, } from './keyword-campaign-shared.js';
|
|
3
|
+
import { isLikelyBlocked, isLikelyCaptcha, sleep, buildSerpBlockReason, DEFAULT_SERP_MAX_CONSECUTIVE_BLOCKS, DEFAULT_SERP_SEARCH_CAPTCHA_COOLDOWN_MS, DEFAULT_SERP_SEARCH_DELAY_MS, DEFAULT_SERP_SEARCH_JITTER_MS, DEFAULT_SERP_SEARCH_RETRY_COUNT, DEFAULT_SERP_SEARCH_RETRY_DELAY_MS, } from './keyword-campaign-seed-core.js';
|
|
4
|
+
import { mergeKeywordSeeds } from './keyword-campaign-seed-advanced.js';
|
|
5
|
+
function normalizePageUrl(value) {
|
|
6
|
+
return value && value.trim().length > 0 ? value.trim() : 'unknown-page';
|
|
7
|
+
}
|
|
8
|
+
function ensureCompetitorTracker(trackers, domain) {
|
|
9
|
+
const normalizedDomain = domain.toLowerCase();
|
|
10
|
+
const existing = trackers.get(normalizedDomain);
|
|
11
|
+
if (existing) {
|
|
12
|
+
return existing;
|
|
13
|
+
}
|
|
14
|
+
const created = {
|
|
15
|
+
organicQueries: 0,
|
|
16
|
+
paidQueries: 0,
|
|
17
|
+
bestOrganicRank: null,
|
|
18
|
+
bestPaidRank: null,
|
|
19
|
+
keywords: new Set(),
|
|
20
|
+
outperformedQueries: 0,
|
|
21
|
+
outperformedGapTotal: 0,
|
|
22
|
+
outperformedTargetPositionTotal: 0,
|
|
23
|
+
outperformedCompetitorPositionTotal: 0,
|
|
24
|
+
outperformedComparableQueries: 0,
|
|
25
|
+
outperformedByPage: new Map(),
|
|
26
|
+
};
|
|
27
|
+
trackers.set(normalizedDomain, created);
|
|
28
|
+
return created;
|
|
29
|
+
}
|
|
30
|
+
function trackOutperformance(tracker, sourcePage, targetPosition, competitorPosition) {
|
|
31
|
+
tracker.outperformedQueries += 1;
|
|
32
|
+
const bucket = tracker.outperformedByPage.get(sourcePage) ?? {
|
|
33
|
+
outperformedQueries: 0,
|
|
34
|
+
gapTotal: 0,
|
|
35
|
+
targetTotal: 0,
|
|
36
|
+
competitorTotal: 0,
|
|
37
|
+
comparableQueries: 0,
|
|
38
|
+
};
|
|
39
|
+
bucket.outperformedQueries += 1;
|
|
40
|
+
if (targetPosition !== null) {
|
|
41
|
+
const gap = targetPosition - competitorPosition;
|
|
42
|
+
bucket.gapTotal += gap;
|
|
43
|
+
bucket.targetTotal += targetPosition;
|
|
44
|
+
bucket.competitorTotal += competitorPosition;
|
|
45
|
+
bucket.comparableQueries += 1;
|
|
46
|
+
tracker.outperformedGapTotal += gap;
|
|
47
|
+
tracker.outperformedTargetPositionTotal += targetPosition;
|
|
48
|
+
tracker.outperformedCompetitorPositionTotal += competitorPosition;
|
|
49
|
+
tracker.outperformedComparableQueries += 1;
|
|
50
|
+
}
|
|
51
|
+
tracker.outperformedByPage.set(sourcePage, bucket);
|
|
52
|
+
}
|
|
53
|
+
function buildEmptyReport(targetUrl, targetDomain) {
|
|
54
|
+
return {
|
|
55
|
+
targetUrl,
|
|
56
|
+
targetDomain,
|
|
57
|
+
results: [],
|
|
58
|
+
summary: {
|
|
59
|
+
queriesRequested: 0,
|
|
60
|
+
queriesExecuted: 0,
|
|
61
|
+
queriesFound: 0,
|
|
62
|
+
queriesBlocked: 0,
|
|
63
|
+
queriesCaptcha: 0,
|
|
64
|
+
campaignStopped: false,
|
|
65
|
+
avgTopPosition: null,
|
|
66
|
+
top3Count: 0,
|
|
67
|
+
top10Count: 0,
|
|
68
|
+
topOrganicCompetitors: [],
|
|
69
|
+
topPaidCompetitors: [],
|
|
70
|
+
competitorCoverage: {
|
|
71
|
+
organicUniqueDomains: 0,
|
|
72
|
+
paidUniqueDomains: 0,
|
|
73
|
+
},
|
|
74
|
+
},
|
|
75
|
+
pageComparison: [],
|
|
76
|
+
campaign: {
|
|
77
|
+
active: false,
|
|
78
|
+
confidence: 'low',
|
|
79
|
+
evidence: [],
|
|
80
|
+
},
|
|
81
|
+
};
|
|
82
|
+
}
|
|
83
|
+
function dedupeDomainResults(results, targetDomain) {
|
|
84
|
+
const seen = new Set();
|
|
85
|
+
const domainResults = [];
|
|
86
|
+
for (const result of results) {
|
|
87
|
+
if (!result.domain || result.domain === targetDomain)
|
|
88
|
+
continue;
|
|
89
|
+
if (seen.has(result.domain))
|
|
90
|
+
continue;
|
|
91
|
+
seen.add(result.domain);
|
|
92
|
+
domainResults.push({
|
|
93
|
+
domain: result.domain,
|
|
94
|
+
rank: result.rank,
|
|
95
|
+
placement: result.placement,
|
|
96
|
+
placementHint: result.placementHint,
|
|
97
|
+
url: result.url,
|
|
98
|
+
title: result.title,
|
|
99
|
+
});
|
|
100
|
+
}
|
|
101
|
+
return domainResults;
|
|
102
|
+
}
|
|
103
|
+
function trackCompetitors(candidates, keyword, targetDomain, trackers) {
|
|
104
|
+
for (const competitor of candidates) {
|
|
105
|
+
if (competitor.domain === targetDomain)
|
|
106
|
+
continue;
|
|
107
|
+
const current = ensureCompetitorTracker(trackers, competitor.domain);
|
|
108
|
+
if (competitor.placement === 'ad') {
|
|
109
|
+
current.paidQueries += 1;
|
|
110
|
+
if (current.bestPaidRank === null || competitor.rank < current.bestPaidRank) {
|
|
111
|
+
current.bestPaidRank = competitor.rank;
|
|
112
|
+
}
|
|
113
|
+
}
|
|
114
|
+
else {
|
|
115
|
+
current.organicQueries += 1;
|
|
116
|
+
if (current.bestOrganicRank === null || competitor.rank < current.bestOrganicRank) {
|
|
117
|
+
current.bestOrganicRank = competitor.rank;
|
|
118
|
+
}
|
|
119
|
+
}
|
|
120
|
+
current.keywords.add(keyword);
|
|
121
|
+
}
|
|
122
|
+
}
|
|
123
|
+
function makeSummaryEntry(domain, tracker) {
|
|
124
|
+
const outperformedPages = [...tracker.outperformedByPage.entries()]
|
|
125
|
+
.map(([pageUrl, pageStats]) => ({
|
|
126
|
+
pageUrl,
|
|
127
|
+
outperformedQueries: pageStats.outperformedQueries,
|
|
128
|
+
avgTargetPosition: pageStats.comparableQueries > 0
|
|
129
|
+
? Math.round(pageStats.targetTotal / pageStats.comparableQueries)
|
|
130
|
+
: null,
|
|
131
|
+
avgCompetitorPosition: pageStats.comparableQueries > 0
|
|
132
|
+
? Math.round(pageStats.competitorTotal / pageStats.comparableQueries)
|
|
133
|
+
: null,
|
|
134
|
+
avgGap: pageStats.comparableQueries > 0
|
|
135
|
+
? Math.round(pageStats.gapTotal / pageStats.comparableQueries)
|
|
136
|
+
: null,
|
|
137
|
+
}))
|
|
138
|
+
.sort((a, b) => b.outperformedQueries - a.outperformedQueries || a.pageUrl.localeCompare(b.pageUrl));
|
|
139
|
+
return {
|
|
140
|
+
domain,
|
|
141
|
+
organicQueries: tracker.organicQueries,
|
|
142
|
+
paidQueries: tracker.paidQueries,
|
|
143
|
+
bestOrganicRank: tracker.bestOrganicRank,
|
|
144
|
+
bestPaidRank: tracker.bestPaidRank,
|
|
145
|
+
matchedKeywords: tracker.keywords.size,
|
|
146
|
+
totalOutperformedQueries: tracker.outperformedQueries,
|
|
147
|
+
avgOutperformedGap: tracker.outperformedComparableQueries > 0
|
|
148
|
+
? Math.round(tracker.outperformedGapTotal / tracker.outperformedComparableQueries)
|
|
149
|
+
: null,
|
|
150
|
+
avgTargetPositionWhenOutperformed: tracker.outperformedComparableQueries > 0
|
|
151
|
+
? Math.round(tracker.outperformedTargetPositionTotal / tracker.outperformedComparableQueries)
|
|
152
|
+
: null,
|
|
153
|
+
avgCompetitorPositionWhenOutperformed: tracker.outperformedComparableQueries > 0
|
|
154
|
+
? Math.round(tracker.outperformedCompetitorPositionTotal / tracker.outperformedComparableQueries)
|
|
155
|
+
: null,
|
|
156
|
+
outperformedPages,
|
|
157
|
+
};
|
|
158
|
+
}
|
|
159
|
+
function finalizeCompetitorSummary(trackers) {
|
|
160
|
+
const ranked = [...trackers.entries()].map(([domain, tracker]) => ({
|
|
161
|
+
domain,
|
|
162
|
+
tracker,
|
|
163
|
+
organicQueries: tracker.organicQueries,
|
|
164
|
+
paidQueries: tracker.paidQueries,
|
|
165
|
+
bestOrganicRank: tracker.bestOrganicRank,
|
|
166
|
+
bestPaidRank: tracker.bestPaidRank,
|
|
167
|
+
matchedKeywords: tracker.keywords.size,
|
|
168
|
+
score: (tracker.organicQueries * 2) + (tracker.paidQueries * 3),
|
|
169
|
+
}));
|
|
170
|
+
const scoreSort = (a, b) => b.score - a.score || a.domain.localeCompare(b.domain);
|
|
171
|
+
const topOrganic = ranked
|
|
172
|
+
.filter((entry) => entry.organicQueries > 0)
|
|
173
|
+
.sort(scoreSort)
|
|
174
|
+
.map(({ domain, tracker }) => makeSummaryEntry(domain, tracker))
|
|
175
|
+
.slice(0, 10);
|
|
176
|
+
const topPaid = ranked
|
|
177
|
+
.filter((entry) => entry.paidQueries > 0)
|
|
178
|
+
.sort(scoreSort)
|
|
179
|
+
.map(({ domain, tracker }) => makeSummaryEntry(domain, tracker))
|
|
180
|
+
.slice(0, 10);
|
|
181
|
+
return {
|
|
182
|
+
topOrganic,
|
|
183
|
+
topPaid,
|
|
184
|
+
coverage: {
|
|
185
|
+
organicUniqueDomains: ranked.filter((entry) => entry.organicQueries > 0).length,
|
|
186
|
+
paidUniqueDomains: ranked.filter((entry) => entry.paidQueries > 0).length,
|
|
187
|
+
},
|
|
188
|
+
};
|
|
189
|
+
}
|
|
190
|
+
export async function analyzeKeywordCampaign(options) {
|
|
191
|
+
const targetUrl = options.targetUrl.trim();
|
|
192
|
+
const targetDomain = normalizeHost(targetUrl);
|
|
193
|
+
const queryLimit = clampNumber(options.maxQueries, 20);
|
|
194
|
+
const defaultCampaignResultLimit = 25;
|
|
195
|
+
const searchResultLimit = Math.max(clampNumber(options.maxResultsPerQuery, defaultCampaignResultLimit), 20);
|
|
196
|
+
const minKeywordLength = options.minKeywordLength ?? DEFAULT_SERP_MIN_KEYWORD_LENGTH;
|
|
197
|
+
const minWeight = Math.max(1, options.minWeight ?? 1);
|
|
198
|
+
const searchConcurrency = clampNumber(options.searchConcurrency, 4);
|
|
199
|
+
const searchDelayMs = Math.max(0, options.searchDelayMs ?? DEFAULT_SERP_SEARCH_DELAY_MS);
|
|
200
|
+
const searchDelayJitterMs = Math.max(0, options.searchDelayJitterMs ?? DEFAULT_SERP_SEARCH_JITTER_MS);
|
|
201
|
+
const searchCaptchaCooldownMs = Math.max(0, options.searchCaptchaCooldownMs ?? DEFAULT_SERP_SEARCH_CAPTCHA_COOLDOWN_MS);
|
|
202
|
+
const searchRetryCount = Math.max(0, options.searchRetryCount ?? DEFAULT_SERP_SEARCH_RETRY_COUNT);
|
|
203
|
+
const searchRetryDelayMs = Math.max(0, options.searchRetryDelayMs ?? DEFAULT_SERP_SEARCH_RETRY_DELAY_MS);
|
|
204
|
+
const searchMaxConsecutiveBlocks = Math.max(0, options.searchMaxConsecutiveBlocks ?? DEFAULT_SERP_MAX_CONSECUTIVE_BLOCKS);
|
|
205
|
+
const discovered = toSeedArray(options.discoveredKeywords ?? []);
|
|
206
|
+
const preset = (options.presetKeywords ?? []).map((keyword) => ({
|
|
207
|
+
keyword,
|
|
208
|
+
normalizedKeyword: normalizeKeyword(keyword),
|
|
209
|
+
source: 'preset',
|
|
210
|
+
weight: 1,
|
|
211
|
+
}));
|
|
212
|
+
const seeds = mergeKeywordSeeds(discovered, preset, minKeywordLength, minWeight).slice(0, queryLimit);
|
|
213
|
+
const report = buildEmptyReport(targetUrl, targetDomain);
|
|
214
|
+
if (seeds.length === 0 || !targetDomain) {
|
|
215
|
+
report.summary.queriesRequested = 0;
|
|
216
|
+
return report;
|
|
217
|
+
}
|
|
218
|
+
const pageBuckets = new Map();
|
|
219
|
+
const competitorTrackers = new Map();
|
|
220
|
+
const campaignEvidence = [];
|
|
221
|
+
const campaignControl = {
|
|
222
|
+
consecutiveBlockSignals: 0,
|
|
223
|
+
limitConsecutiveBlockSignals: searchMaxConsecutiveBlocks,
|
|
224
|
+
};
|
|
225
|
+
const resultsByIndex = new Array(seeds.length);
|
|
226
|
+
const requestTiming = {
|
|
227
|
+
nextRequestAt: 0,
|
|
228
|
+
delayMultiplier: 1,
|
|
229
|
+
backoffMultiplier: 1,
|
|
230
|
+
cooldownBonus: 1,
|
|
231
|
+
lastFailurePenaltyAppliedAt: 0,
|
|
232
|
+
lastFailureAt: 0,
|
|
233
|
+
};
|
|
234
|
+
let requestGate = Promise.resolve();
|
|
235
|
+
const runInGate = async (task) => {
|
|
236
|
+
const previous = requestGate;
|
|
237
|
+
let release;
|
|
238
|
+
requestGate = new Promise((resolve) => {
|
|
239
|
+
release = resolve;
|
|
240
|
+
});
|
|
241
|
+
await previous;
|
|
242
|
+
try {
|
|
243
|
+
await task();
|
|
244
|
+
}
|
|
245
|
+
finally {
|
|
246
|
+
release?.();
|
|
247
|
+
}
|
|
248
|
+
};
|
|
249
|
+
const gate = async () => {
|
|
250
|
+
await runInGate(async () => {
|
|
251
|
+
const now = Date.now();
|
|
252
|
+
const targetDelay = Math.round(Math.max(0, searchDelayMs * requestTiming.delayMultiplier));
|
|
253
|
+
const jitter = searchDelayJitterMs > 0
|
|
254
|
+
? Math.floor(Math.random() * (searchDelayJitterMs + 1))
|
|
255
|
+
: 0;
|
|
256
|
+
const availableAt = Math.max(requestTiming.nextRequestAt, now);
|
|
257
|
+
const waitMs = Math.max(0, availableAt - now);
|
|
258
|
+
requestTiming.nextRequestAt = availableAt + targetDelay + jitter;
|
|
259
|
+
if (waitMs > 0) {
|
|
260
|
+
await sleep(waitMs);
|
|
261
|
+
}
|
|
262
|
+
});
|
|
263
|
+
};
|
|
264
|
+
const applySignalCooldown = (signalType) => {
|
|
265
|
+
const now = Date.now();
|
|
266
|
+
if (signalType === 'success') {
|
|
267
|
+
requestTiming.delayMultiplier = Math.max(1, requestTiming.delayMultiplier * 0.75);
|
|
268
|
+
requestTiming.backoffMultiplier = Math.max(1, requestTiming.backoffMultiplier * 0.85);
|
|
269
|
+
requestTiming.lastFailureAt = 0;
|
|
270
|
+
return;
|
|
271
|
+
}
|
|
272
|
+
requestTiming.lastFailureAt = now;
|
|
273
|
+
requestTiming.delayMultiplier = Math.min(8, requestTiming.delayMultiplier * 1.8);
|
|
274
|
+
requestTiming.backoffMultiplier = Math.min(4, requestTiming.backoffMultiplier + 1);
|
|
275
|
+
const isConsecutiveCaptcha = signalType === 'captcha'
|
|
276
|
+
&& now - requestTiming.lastFailurePenaltyAppliedAt < searchCaptchaCooldownMs;
|
|
277
|
+
const consecutivePenaltyMultiplier = isConsecutiveCaptcha
|
|
278
|
+
? 1 + Math.min(5, Math.floor((requestTiming.backoffMultiplier - 1) * 2))
|
|
279
|
+
: 1;
|
|
280
|
+
if (signalType === 'captcha') {
|
|
281
|
+
requestTiming.lastFailurePenaltyAppliedAt = now;
|
|
282
|
+
}
|
|
283
|
+
const cooldown = Math.round(searchCaptchaCooldownMs * requestTiming.backoffMultiplier * consecutivePenaltyMultiplier);
|
|
284
|
+
requestTiming.nextRequestAt = Math.max(requestTiming.nextRequestAt, now + cooldown);
|
|
285
|
+
};
|
|
286
|
+
report.summary.queriesRequested = seeds.length;
|
|
287
|
+
const registerSignal = (wasBlocked, keyword, source, sourcePage) => {
|
|
288
|
+
if (report.summary.campaignStopped)
|
|
289
|
+
return;
|
|
290
|
+
if (!wasBlocked) {
|
|
291
|
+
campaignControl.consecutiveBlockSignals = 0;
|
|
292
|
+
return;
|
|
293
|
+
}
|
|
294
|
+
const hasBlockLimit = campaignControl.limitConsecutiveBlockSignals > 0;
|
|
295
|
+
if (!hasBlockLimit) {
|
|
296
|
+
return;
|
|
297
|
+
}
|
|
298
|
+
campaignControl.consecutiveBlockSignals += 1;
|
|
299
|
+
if (campaignControl.consecutiveBlockSignals >= campaignControl.limitConsecutiveBlockSignals) {
|
|
300
|
+
report.summary.campaignStopped = true;
|
|
301
|
+
report.summary.campaignStopReason = `Parada antecipada por ${campaignControl.consecutiveBlockSignals} sinais consecutivos de bloqueio/captcha para evitar bloqueio adicional`;
|
|
302
|
+
campaignEvidence.push(`${source}: "${keyword}" ${sourcePage ? `(${sourcePage}) ` : ''}alcançou limite de bloqueio; pausa de segurança ativa para reduzir WAF/challenge.`);
|
|
303
|
+
}
|
|
304
|
+
};
|
|
305
|
+
const executeSearch = async (seed, pageKey, normalizedKeyword) => {
|
|
306
|
+
const query = seed.keyword;
|
|
307
|
+
let response;
|
|
308
|
+
let lastError;
|
|
309
|
+
let attempts = 0;
|
|
310
|
+
if (report.summary.campaignStopped) {
|
|
311
|
+
return undefined;
|
|
312
|
+
}
|
|
313
|
+
while (attempts <= searchRetryCount) {
|
|
314
|
+
if (report.summary.campaignStopped) {
|
|
315
|
+
return undefined;
|
|
316
|
+
}
|
|
317
|
+
attempts += 1;
|
|
318
|
+
if (report.summary.campaignStopped) {
|
|
319
|
+
return undefined;
|
|
320
|
+
}
|
|
321
|
+
await gate();
|
|
322
|
+
if (report.summary.campaignStopped) {
|
|
323
|
+
return undefined;
|
|
324
|
+
}
|
|
325
|
+
try {
|
|
326
|
+
response = await searchGoogleAdvanced(query, {
|
|
327
|
+
num: searchResultLimit,
|
|
328
|
+
maxResults: searchResultLimit,
|
|
329
|
+
transport: options.transport,
|
|
330
|
+
timeout: options.timeout,
|
|
331
|
+
country: options.country,
|
|
332
|
+
gl: options.gl,
|
|
333
|
+
hl: options.hl,
|
|
334
|
+
...(options.searchParams ?? {}),
|
|
335
|
+
});
|
|
336
|
+
}
|
|
337
|
+
catch (error) {
|
|
338
|
+
lastError = error;
|
|
339
|
+
}
|
|
340
|
+
if (!response) {
|
|
341
|
+
if (attempts <= searchRetryCount) {
|
|
342
|
+
const retryDelay = Math.min(searchRetryDelayMs * attempts, 6_000) + Math.floor(Math.random() * (searchDelayJitterMs + 1));
|
|
343
|
+
requestTiming.nextRequestAt = Math.max(requestTiming.nextRequestAt, Date.now() + retryDelay);
|
|
344
|
+
continue;
|
|
345
|
+
}
|
|
346
|
+
break;
|
|
347
|
+
}
|
|
348
|
+
const wasCaptcha = isLikelyCaptcha(response.captcha);
|
|
349
|
+
const wasBlocked = isLikelyBlocked(response.block);
|
|
350
|
+
if ((wasCaptcha || wasBlocked) && attempts <= searchRetryCount) {
|
|
351
|
+
const retryDelay = Math.min(searchRetryDelayMs * attempts, 6_000) + Math.floor(Math.random() * (searchDelayJitterMs + 1));
|
|
352
|
+
applySignalCooldown(wasCaptcha ? 'captcha' : 'blocked');
|
|
353
|
+
requestTiming.nextRequestAt = Math.max(requestTiming.nextRequestAt, Date.now() + retryDelay + (wasCaptcha ? searchCaptchaCooldownMs : 0));
|
|
354
|
+
response = undefined;
|
|
355
|
+
continue;
|
|
356
|
+
}
|
|
357
|
+
applySignalCooldown('success');
|
|
358
|
+
break;
|
|
359
|
+
}
|
|
360
|
+
if (!response) {
|
|
361
|
+
return {
|
|
362
|
+
result: {
|
|
363
|
+
keyword: query,
|
|
364
|
+
source: seed.source,
|
|
365
|
+
sourcePage: seed.sourcePage,
|
|
366
|
+
sourceWeight: seed.weight,
|
|
367
|
+
blocked: true,
|
|
368
|
+
blockReason: 'request failed',
|
|
369
|
+
found: false,
|
|
370
|
+
bestPosition: null,
|
|
371
|
+
totalChecked: searchResultLimit,
|
|
372
|
+
placement: 'unknown',
|
|
373
|
+
searchUrl: '',
|
|
374
|
+
searchTransport: options.transport ?? 'undici',
|
|
375
|
+
competitors: [],
|
|
376
|
+
},
|
|
377
|
+
wasBlocked: true,
|
|
378
|
+
wasCaptcha: false,
|
|
379
|
+
pageKey,
|
|
380
|
+
normalizedKeyword,
|
|
381
|
+
};
|
|
382
|
+
}
|
|
383
|
+
const wasBlocked = isLikelyBlocked(response.block);
|
|
384
|
+
const wasCaptcha = isLikelyCaptcha(response.captcha);
|
|
385
|
+
const parsedResults = response.results.map((result) => ({
|
|
386
|
+
...result,
|
|
387
|
+
domain: normalizeHost(result.url),
|
|
388
|
+
placement: normalizePlacement(result.placement),
|
|
389
|
+
}));
|
|
390
|
+
const matched = parsedResults.find((result) => {
|
|
391
|
+
return result.domain && result.domain === targetDomain;
|
|
392
|
+
});
|
|
393
|
+
const competitorCandidates = dedupeDomainResults(parsedResults.map((result) => ({
|
|
394
|
+
domain: result.domain,
|
|
395
|
+
rank: result.rank,
|
|
396
|
+
placement: result.placement,
|
|
397
|
+
placementHint: result.placementHint,
|
|
398
|
+
url: result.url,
|
|
399
|
+
title: result.title,
|
|
400
|
+
})), targetDomain);
|
|
401
|
+
if (wasCaptcha && searchCaptchaCooldownMs > 0) {
|
|
402
|
+
requestTiming.nextRequestAt = Math.max(requestTiming.nextRequestAt, Date.now() + searchCaptchaCooldownMs);
|
|
403
|
+
}
|
|
404
|
+
return {
|
|
405
|
+
result: {
|
|
406
|
+
keyword: query,
|
|
407
|
+
source: seed.source,
|
|
408
|
+
sourcePage: seed.sourcePage,
|
|
409
|
+
sourceWeight: seed.weight,
|
|
410
|
+
blocked: wasBlocked || wasCaptcha,
|
|
411
|
+
blockReason: buildSerpBlockReason(response.block, response.captcha),
|
|
412
|
+
found: Boolean(matched),
|
|
413
|
+
bestPosition: matched?.rank ?? null,
|
|
414
|
+
totalChecked: parsedResults.length,
|
|
415
|
+
matchedUrl: matched?.url,
|
|
416
|
+
matchedTitle: matched?.title,
|
|
417
|
+
matchedDisplayUrl: matched?.displayedUrl,
|
|
418
|
+
placement: matched ? normalizePlacement(matched.placement) : 'unknown',
|
|
419
|
+
placementHint: matched?.placementHint,
|
|
420
|
+
searchUrl: response.searchUrl,
|
|
421
|
+
searchTransport: response.transport.used,
|
|
422
|
+
competitors: competitorCandidates,
|
|
423
|
+
},
|
|
424
|
+
wasBlocked,
|
|
425
|
+
wasCaptcha,
|
|
426
|
+
pageKey,
|
|
427
|
+
normalizedKeyword,
|
|
428
|
+
};
|
|
429
|
+
};
|
|
430
|
+
const workers = Math.max(1, Math.min(searchConcurrency, seeds.length));
|
|
431
|
+
const workerState = { nextIndex: 0 };
|
|
432
|
+
const finalizeExecution = (execution) => {
|
|
433
|
+
const { result: searchResult, wasBlocked, wasCaptcha, pageKey, normalizedKeyword, } = execution;
|
|
434
|
+
const bucket = pageBuckets.get(pageKey) ?? {
|
|
435
|
+
tracked: 0,
|
|
436
|
+
found: 0,
|
|
437
|
+
totalPosition: 0,
|
|
438
|
+
top3: 0,
|
|
439
|
+
top10: 0,
|
|
440
|
+
};
|
|
441
|
+
bucket.tracked += 1;
|
|
442
|
+
report.summary.queriesExecuted += 1;
|
|
443
|
+
if (wasBlocked)
|
|
444
|
+
report.summary.queriesBlocked += 1;
|
|
445
|
+
if (wasCaptcha)
|
|
446
|
+
report.summary.queriesCaptcha += 1;
|
|
447
|
+
if (searchResult.placement === 'unknown' && searchResult.blocked) {
|
|
448
|
+
campaignEvidence.push(`Falha ao buscar keyword "${searchResult.keyword}"`);
|
|
449
|
+
}
|
|
450
|
+
trackCompetitors(searchResult.competitors, normalizedKeyword, targetDomain, competitorTrackers);
|
|
451
|
+
if (searchResult.found && searchResult.bestPosition !== null) {
|
|
452
|
+
const position = searchResult.bestPosition;
|
|
453
|
+
report.summary.queriesFound += 1;
|
|
454
|
+
if (position <= 3)
|
|
455
|
+
report.summary.top3Count += 1;
|
|
456
|
+
if (position <= 10)
|
|
457
|
+
report.summary.top10Count += 1;
|
|
458
|
+
if (searchResult.placement === 'ad') {
|
|
459
|
+
campaignEvidence.push(`${searchResult.keyword} aparece como anúncio em posição #${position}`);
|
|
460
|
+
}
|
|
461
|
+
else {
|
|
462
|
+
campaignEvidence.push(`${searchResult.keyword} aparece orgânico em posição #${position}`);
|
|
463
|
+
}
|
|
464
|
+
for (const competitor of searchResult.competitors) {
|
|
465
|
+
if (competitor.rank < position) {
|
|
466
|
+
const tracker = ensureCompetitorTracker(competitorTrackers, competitor.domain);
|
|
467
|
+
trackOutperformance(tracker, pageKey, position, competitor.rank);
|
|
468
|
+
}
|
|
469
|
+
}
|
|
470
|
+
bucket.found += 1;
|
|
471
|
+
bucket.totalPosition += position;
|
|
472
|
+
if (position <= 3)
|
|
473
|
+
bucket.top3 += 1;
|
|
474
|
+
if (position <= 10)
|
|
475
|
+
bucket.top10 += 1;
|
|
476
|
+
}
|
|
477
|
+
pageBuckets.set(pageKey, bucket);
|
|
478
|
+
};
|
|
479
|
+
const runWorker = async () => {
|
|
480
|
+
while (true) {
|
|
481
|
+
if (report.summary.campaignStopped)
|
|
482
|
+
break;
|
|
483
|
+
if (workerState.nextIndex >= seeds.length)
|
|
484
|
+
break;
|
|
485
|
+
const currentIndex = workerState.nextIndex++;
|
|
486
|
+
const seed = seeds[currentIndex];
|
|
487
|
+
const pageKey = normalizePageUrl(seed.sourcePage ?? (seed.source === 'preset' ? 'preset-queries' : 'unknown'));
|
|
488
|
+
const execution = await executeSearch(seed, pageKey, seed.normalizedKeyword);
|
|
489
|
+
if (!execution) {
|
|
490
|
+
if (report.summary.campaignStopped)
|
|
491
|
+
break;
|
|
492
|
+
continue;
|
|
493
|
+
}
|
|
494
|
+
resultsByIndex[currentIndex] = execution;
|
|
495
|
+
registerSignal(execution.wasBlocked || execution.wasCaptcha, execution.result.keyword, execution.result.source, execution.result.sourcePage);
|
|
496
|
+
finalizeExecution(execution);
|
|
497
|
+
}
|
|
498
|
+
};
|
|
499
|
+
await Promise.all(new Array(workers).fill(null).map(() => runWorker()));
|
|
500
|
+
report.results = resultsByIndex
|
|
501
|
+
.filter((entry) => Boolean(entry))
|
|
502
|
+
.map(({ result }) => result);
|
|
503
|
+
const foundPositions = report.results
|
|
504
|
+
.filter((entry) => entry.found && entry.bestPosition !== null)
|
|
505
|
+
.map((entry) => entry.bestPosition);
|
|
506
|
+
if (foundPositions.length > 0) {
|
|
507
|
+
report.summary.avgTopPosition = Math.round(foundPositions.reduce((acc, pos) => acc + pos, 0) / foundPositions.length);
|
|
508
|
+
}
|
|
509
|
+
const hasTop3 = foundPositions.some((position) => position <= 3);
|
|
510
|
+
const hasTop10 = foundPositions.some((position) => position <= 10);
|
|
511
|
+
report.campaign = {
|
|
512
|
+
active: foundPositions.some((position) => position <= 10),
|
|
513
|
+
confidence: hasTop3 ? 'high' : hasTop10 ? 'medium' : foundPositions.length > 0 ? 'low' : 'low',
|
|
514
|
+
evidence: campaignEvidence,
|
|
515
|
+
};
|
|
516
|
+
const competitorSummary = finalizeCompetitorSummary(competitorTrackers);
|
|
517
|
+
report.summary.topOrganicCompetitors = competitorSummary.topOrganic;
|
|
518
|
+
report.summary.topPaidCompetitors = competitorSummary.topPaid;
|
|
519
|
+
report.summary.competitorCoverage = competitorSummary.coverage;
|
|
520
|
+
report.pageComparison = [...pageBuckets.entries()].map(([pageUrl, bucket]) => ({
|
|
521
|
+
pageUrl,
|
|
522
|
+
tracked: bucket.tracked,
|
|
523
|
+
found: bucket.found,
|
|
524
|
+
appearanceRate: bucket.tracked > 0
|
|
525
|
+
? Number(((bucket.found / bucket.tracked) * 100).toFixed(1))
|
|
526
|
+
: 0,
|
|
527
|
+
avgPosition: bucket.found > 0 ? Math.round(bucket.totalPosition / bucket.found) : null,
|
|
528
|
+
top3: bucket.top3,
|
|
529
|
+
top10: bucket.top10,
|
|
530
|
+
})).sort((a, b) => {
|
|
531
|
+
if (a.avgPosition === null)
|
|
532
|
+
return 1;
|
|
533
|
+
if (b.avgPosition === null)
|
|
534
|
+
return -1;
|
|
535
|
+
return a.avgPosition - b.avgPosition;
|
|
536
|
+
});
|
|
537
|
+
return report;
|
|
538
|
+
}
|
|
@@ -0,0 +1,17 @@
|
|
|
1
|
+
import type { SeoReport } from './types.js';
|
|
2
|
+
import type { KeywordCampaignSeed } from './keyword-campaign-shared.js';
|
|
3
|
+
export declare function deduplicateKeywordPermutations(seeds: KeywordCampaignSeed[]): KeywordCampaignSeed[];
|
|
4
|
+
export declare function getSeedWordCount(seed: KeywordCampaignSeed): number;
|
|
5
|
+
export declare function extractSchemaOrgLongTailSeeds(report: SeoReport, options: {
|
|
6
|
+
minKeywordLength: number;
|
|
7
|
+
maxPhrases: number;
|
|
8
|
+
sourcePage?: string;
|
|
9
|
+
}): KeywordCampaignSeed[];
|
|
10
|
+
export declare function extractHeadingPathCompositionSeeds(report: SeoReport, options: {
|
|
11
|
+
minKeywordLength: number;
|
|
12
|
+
maxPhrases: number;
|
|
13
|
+
topKeywordSet: Set<string>;
|
|
14
|
+
sourcePage?: string;
|
|
15
|
+
}): KeywordCampaignSeed[];
|
|
16
|
+
export declare function sortSeedsByWeight(a: KeywordCampaignSeed, b: KeywordCampaignSeed): number;
|
|
17
|
+
export declare function mergeKeywordSeeds(discovered: KeywordCampaignSeed[], preset: KeywordCampaignSeed[], minKeywordLength?: number, minWeight?: number): KeywordCampaignSeed[];
|