Code Coverage |
||||||||||
Lines |
Functions and Methods |
Classes and Traits |
||||||||
| Total | |
80.00% |
48 / 60 |
|
50.00% |
3 / 6 |
CRAP | |
0.00% |
0 / 1 |
| CompanyResearchService | |
80.00% |
48 / 60 |
|
50.00% |
3 / 6 |
13.15 | |
0.00% |
0 / 1 |
| __construct | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getOrCreateForUrl | |
94.74% |
18 / 19 |
|
0.00% |
0 / 1 |
6.01 | |||
| runResearch | |
72.22% |
26 / 36 |
|
0.00% |
0 / 1 |
2.09 | |||
| getTtlDays | |
100.00% |
2 / 2 |
|
100.00% |
1 / 1 |
1 | |||
| isFresh | |
100.00% |
1 / 1 |
|
100.00% |
1 / 1 |
1 | |||
| getByDomain | |
0.00% |
0 / 1 |
|
0.00% |
0 / 1 |
2 | |||
| 1 | <?php |
| 2 | |
| 3 | namespace App\Http\Services\RolePlay; |
| 4 | |
| 5 | use App\Http\Models\CompanyResearch; |
| 6 | use App\Http\Models\Parameter; |
| 7 | use App\Http\Services\NodeJsAIBridgeService; |
| 8 | use App\Http\Services\WebScraperService; |
| 9 | use App\Jobs\RefreshCompanyResearchJob; |
| 10 | use Illuminate\Support\Facades\Log; |
| 11 | |
| 12 | /** |
| 13 | * Manages the lifecycle of cached company research records. |
| 14 | * |
| 15 | * Coordinates website scraping (via {@see WebScraperService}) and deep |
| 16 | * AI research (via {@see NodeJsAIBridgeService} with Google Search |
| 17 | * grounding) to build comprehensive company intelligence used by the |
| 18 | * roleplay auto-populate flow. |
| 19 | * |
| 20 | * Records are keyed by normalized domain and cached for a configurable |
| 21 | * TTL (default 30 days, controlled by the `company_research_ttl_days` |
| 22 | * system parameter). |
| 23 | */ |
| 24 | class CompanyResearchService |
| 25 | { |
| 26 | /** |
| 27 | * The research prompt template. |
| 28 | * |
| 29 | * The `{domain}` placeholder is replaced at runtime with the actual domain. |
| 30 | */ |
| 31 | private const RESEARCH_PROMPT = <<<'PROMPT' |
| 32 | You are a business intelligence analyst. Conduct an exhaustive deep-dive research on the company at {domain}. |
| 33 | |
| 34 | Search broadly: their main website, /about, /about-us, /products, /services, /solutions, /pricing, /team, /careers, /blog, /press, /news, /customers, /case-studies, and any public third-party sources (press releases, industry articles, LinkedIn, Crunchbase, Glassdoor, G2, etc.). |
| 35 | |
| 36 | Your goal is to extract and document the MAXIMUM amount of factual information. Do NOT summarize or condense. Write extensively — include every detail you find. This data will be stored as a reference database and used later for targeted content generation. More detail is always better. |
| 37 | |
| 38 | Cover ALL of the following sections in depth: |
| 39 | |
| 40 | ### 1. Company Overview |
| 41 | - Full legal/operating name, any DBA names or brand names |
| 42 | - Detailed description of what the company does (multiple paragraphs) |
| 43 | - Mission statement, vision, and core values (quote verbatim if found) |
| 44 | - Founding year, founders' names and backgrounds |
| 45 | - Headquarters location (full address if available) |
| 46 | - Other office locations |
| 47 | - Company size: number of employees, revenue (if public) |
| 48 | - Ownership structure (private, public, employee-owned, PE-backed, etc.) |
| 49 | - Brief company history and major milestones |
| 50 | |
| 51 | ### 2. Products & Services |
| 52 | - List EVERY product and service offering individually |
| 53 | - For each product: name, detailed description of what it does, who it's for, key capabilities |
| 54 | - Product categories or product lines |
| 55 | - Any free trials, freemium tiers, or demo offerings |
| 56 | - Platform details (web app, mobile app, browser extension, API, etc.) |
| 57 | - Any white-label or reseller programs |
| 58 | |
| 59 | ### 3. Target Market |
| 60 | - ALL industries and verticals they serve — list every one mentioned on their site |
| 61 | - Specific market segments or niches they specialize in |
| 62 | - Company sizes they target (startups, SMB, mid-market, enterprise) |
| 63 | - Geographic markets served (local, national, international — which countries/regions) |
| 64 | - Target buyer personas: job titles, departments, seniority levels |
| 65 | - Any specific use cases or scenarios they highlight |
| 66 | |
| 67 | ### 4. Value Propositions & Differentiators |
| 68 | - Every competitive advantage and unique selling point |
| 69 | - Specific claims they make (e.g., "saves X hours", "increases Y by Z%") |
| 70 | - Customer testimonials or success metrics mentioned on their site |
| 71 | - Awards, certifications, recognitions |
| 72 | - How they position themselves against competitors |
| 73 | |
| 74 | ### 5. Pricing Model |
| 75 | - Every pricing tier, plan name, and price point (if publicly available) |
| 76 | - What's included in each tier |
| 77 | - Free vs. paid feature differences |
| 78 | - Enterprise/custom pricing indicators |
| 79 | - Any volume discounts, annual vs. monthly pricing |
| 80 | |
| 81 | ### 6. Clients, Partners & Case Studies |
| 82 | - Every named client, customer, or logo featured on their site |
| 83 | - Partner ecosystem: technology partners, resellers, integrations partners |
| 84 | - Case studies: company name, industry, problem solved, results achieved |
| 85 | - Testimonials: quote, person's name and title, company |
| 86 | |
| 87 | ### 7. Leadership & Team |
| 88 | - Founders and C-suite executives: names, titles, backgrounds |
| 89 | - Board members or advisors if listed |
| 90 | - Total team size and any notable team information |
| 91 | - DEI initiatives, culture programs |
| 92 | |
| 93 | ### 8. Company Culture & Employer Brand |
| 94 | - Core values and cultural principles |
| 95 | - Employee benefits and perks highlighted |
| 96 | - Glassdoor ratings and notable employee feedback themes |
| 97 | - Charitable initiatives, CSR programs |
| 98 | |
| 99 | ### 9. Recent News & Developments |
| 100 | - Product launches, feature releases (last 2 years) |
| 101 | - Funding rounds, acquisitions, mergers |
| 102 | - Strategic partnerships announced |
| 103 | - Media coverage, press mentions |
| 104 | - Blog topics and thought leadership themes |
| 105 | |
| 106 | ### 10. Technology & Integrations |
| 107 | - Technology stack and platforms |
| 108 | - Every integration listed (CRM, email, social, productivity tools) |
| 109 | - API availability and developer resources |
| 110 | - Security certifications, compliance (SOC2, GDPR, HIPAA, etc.) |
| 111 | - Mobile/desktop/browser support |
| 112 | |
| 113 | ### 11. Pain Points They Address |
| 114 | - Every problem or challenge they claim to solve |
| 115 | - Before/after scenarios described on their site |
| 116 | - Industry-specific pain points they target |
| 117 | - Objections they preemptively address in their marketing |
| 118 | |
| 119 | ### 12. Competitive Landscape |
| 120 | - Named competitors or "alternatives to" mentioned anywhere |
| 121 | - How they differentiate from competitors |
| 122 | - Market category or segment they compete in |
| 123 | |
| 124 | IMPORTANT RULES: |
| 125 | - Write extensively. Include ALL details found — do not leave out information to save space. |
| 126 | - Quote specific numbers, metrics, and claims verbatim when found. |
| 127 | - If a section has limited information available, state what you found and note the limitation. |
| 128 | - Use bullet points and sub-bullets for maximum detail density. |
| 129 | - Do not fabricate information. Only include facts you can verify from public sources. |
| 130 | - Target 3000+ words of output. Longer is better. |
| 131 | PROMPT; |
| 132 | |
| 133 | public function __construct( |
| 134 | private readonly WebScraperService $scraper, |
| 135 | private readonly NodeJsAIBridgeService $aiService, |
| 136 | ) {} |
| 137 | |
| 138 | /** |
| 139 | * Return an existing (fresh) research record or create/refresh one. |
| 140 | * |
| 141 | * - Fresh + completed => return immediately. |
| 142 | * - Stale + completed => dispatch async refresh, return stale record. |
| 143 | * - Processing => return as-is (caller polls). |
| 144 | * - Not found => create + run synchronously, return record. |
| 145 | * |
| 146 | * @param string $url The website URL to research |
| 147 | * @param mixed $user The authenticated user (for logging / requested_by) |
| 148 | */ |
| 149 | public function getOrCreateForUrl(string $url, mixed $user): CompanyResearch |
| 150 | { |
| 151 | $domain = CompanyResearch::extractDomain($url); |
| 152 | $existing = CompanyResearch::where('domain', $domain)->first(); |
| 153 | |
| 154 | if ($existing) { |
| 155 | if ($existing->status === 'completed' && $this->isFresh($existing)) { |
| 156 | return $existing; |
| 157 | } |
| 158 | |
| 159 | if ($existing->status === 'completed' && ! $this->isFresh($existing)) { |
| 160 | RefreshCompanyResearchJob::dispatch($existing); |
| 161 | |
| 162 | return $existing; |
| 163 | } |
| 164 | |
| 165 | // processing, pending, or failed — return as-is |
| 166 | return $existing; |
| 167 | } |
| 168 | |
| 169 | // New domain — create and run synchronously |
| 170 | $research = CompanyResearch::create([ |
| 171 | 'domain' => $domain, |
| 172 | 'url' => $url, |
| 173 | 'status' => 'processing', |
| 174 | 'requested_by' => (string) ($user->id ?? ''), |
| 175 | 'research' => '', |
| 176 | 'scraped_content' => '', |
| 177 | ]); |
| 178 | |
| 179 | $this->runResearch($research); |
| 180 | |
| 181 | return $research->fresh(); |
| 182 | } |
| 183 | |
| 184 | /** |
| 185 | * Execute the full research pipeline: scrape + AI deep research. |
| 186 | * |
| 187 | * On success, updates the record to `completed` with the research output. |
| 188 | * On failure, marks the record as `failed` with the error message and |
| 189 | * rethrows the exception so callers/jobs can handle retries. |
| 190 | * |
| 191 | * @param CompanyResearch $research The research record to populate |
| 192 | * |
| 193 | * @throws \Exception On any scraping or AI failure |
| 194 | */ |
| 195 | public function runResearch(CompanyResearch $research): void |
| 196 | { |
| 197 | try { |
| 198 | $scrapedContent = $this->scraper->scrape($research->url); |
| 199 | |
| 200 | $prompt = str_replace('{domain}', $research->domain, self::RESEARCH_PROMPT); |
| 201 | |
| 202 | $result = $this->aiService->generate( |
| 203 | [ |
| 204 | 'provider' => 'vertex', |
| 205 | 'model' => 'gemini-3.1-flash-lite-preview', |
| 206 | 'prompt' => $prompt, |
| 207 | 'config' => [ |
| 208 | 'maxOutputTokens' => 16000, |
| 209 | 'temperature' => 0.7, |
| 210 | 'topP' => 0.95, |
| 211 | 'thinkingBudget' => 4096, |
| 212 | 'enableGoogleSearch' => true, |
| 213 | ], |
| 214 | ], |
| 215 | [ |
| 216 | 'feature' => 'company_research', |
| 217 | 'user_id' => $research->requested_by, |
| 218 | ] |
| 219 | ); |
| 220 | |
| 221 | $research->update([ |
| 222 | 'status' => 'completed', |
| 223 | 'research' => $result, |
| 224 | 'scraped_content' => $scrapedContent, |
| 225 | 'error' => null, |
| 226 | ]); |
| 227 | } catch (\Exception $e) { |
| 228 | Log::error('[CompanyResearchService] Research failed', [ |
| 229 | 'domain' => $research->domain, |
| 230 | 'error' => $e->getMessage(), |
| 231 | ]); |
| 232 | |
| 233 | $research->update([ |
| 234 | 'status' => 'failed', |
| 235 | 'error' => $e->getMessage(), |
| 236 | ]); |
| 237 | |
| 238 | throw $e; |
| 239 | } |
| 240 | } |
| 241 | |
| 242 | /** |
| 243 | * Read the TTL (in days) from the system parameter, with a 30-day default. |
| 244 | * |
| 245 | * @return int Number of days a research record is considered fresh |
| 246 | */ |
| 247 | public function getTtlDays(): int |
| 248 | { |
| 249 | $param = Parameter::where('name', 'company_research_ttl_days')->first(); |
| 250 | |
| 251 | return (int) ($param?->value ?? 30); |
| 252 | } |
| 253 | |
| 254 | /** |
| 255 | * Check whether a research record is still within the TTL window. |
| 256 | * |
| 257 | * @return bool True if the record's updated_at is within TTL days of now |
| 258 | */ |
| 259 | public function isFresh(CompanyResearch $research): bool |
| 260 | { |
| 261 | return $research->updated_at->diffInDays(now()) < $this->getTtlDays(); |
| 262 | } |
| 263 | |
| 264 | /** |
| 265 | * Look up a research record by normalized domain. |
| 266 | * |
| 267 | * @param string $domain The normalized domain to search for |
| 268 | */ |
| 269 | public function getByDomain(string $domain): ?CompanyResearch |
| 270 | { |
| 271 | return CompanyResearch::where('domain', $domain)->first(); |
| 272 | } |
| 273 | } |