电子信息与电气工程学院
HTML;
$adapter = new FacultyListHtmlAdapter;
$method = new \ReflectionMethod($adapter, 'extractFromHtml');
$method->setAccessible(true);
$items = $method->invoke(
$adapter,
$html,
[],
'https://faculty.sjtu.edu.cn/xyjs_list.jsp?urltype=tsites.CollegeTeacherList&wbtreeid=1001&st=0&id=1701&lang=zh_CN',
);
$this->assertCount(2, $items);
$this->assertSame('沈备军', $items[0]->title);
$this->assertSame('http://faculty.sjtu.edu.cn/bjshen/zh_CN/index.htm', $items[0]->canonicalUrl);
$this->assertSame('上海交通大学', $items[0]->schoolName);
$this->assertSame('faculty_html_tsites', $items[0]->extra['platform']);
$this->assertSame('副教授', $items[0]->extra['academic_title']);
$this->assertSame('电子信息与电气工程学院', $items[0]->extra['college_name']);
$this->assertSame('副教授', $items[0]->extra['lead_author']['academic_title']);
$this->assertSame('夏斌', $items[1]->title);
}
public function test_extracts_teacher_when_profile_href_is_empty(): void
{
$html = <<<'HTML'
HTML;
$adapter = new FacultyListHtmlAdapter;
$method = new \ReflectionMethod($adapter, 'extractFromHtml');
$method->setAccessible(true);
$items = $method->invoke(
$adapter,
$html,
[],
'https://faculty.sjtu.edu.cn/xyjs_list.jsp?id=1701',
);
$this->assertCount(1, $items);
$this->assertSame('邵海滨', $items[0]->title);
$this->assertStringContainsString('shaohaibin', (string) $items[0]->canonicalUrl);
$this->assertSame('副研究员', $items[0]->extra['academic_title']);
$this->assertSame('电子信息与电气工程学院', $items[0]->extra['college_name']);
}
public function test_extracts_email_from_teacher_profile_html(): void
{
$html = '电子邮箱：bjshen@sjtu.edu.cn';
$adapter = new FacultyListHtmlAdapter;
$method = new \ReflectionMethod($adapter, 'extractEmailFromProfileHtml');
$method->setAccessible(true);
$email = $method->invoke($adapter, $html);
$this->assertSame('bjshen@sjtu.edu.cn', $email);
}
public function test_apply_email_to_item_updates_lead_author(): void
{
$adapter = new FacultyListHtmlAdapter;
$method = new \ReflectionMethod($adapter, 'applyEmailToItem');
$method->setAccessible(true);
$item = $method->invoke(
$adapter,
new \App\Services\Crawl\CrawlItemDto(
externalId: 'faculty:test',
title: '沈备军',
canonicalUrl: 'http://faculty.sjtu.edu.cn/bjshen/zh_CN/index.htm',
extra: [
'lead_author' => [
'name' => '沈备军',
'email' => null,
'university_name' => '上海交通大学',
],
],
),
'bjshen@sjtu.edu.cn',
);
$this->assertSame('bjshen@sjtu.edu.cn', $item->extra['lead_author']['email']);
}
public function test_detects_total_pages_and_builds_pagenum_url(): void
{
$html = '下页';
$adapter = new FacultyListHtmlAdapter;
$detect = new \ReflectionMethod($adapter, 'detectTotalPages');
$detect->setAccessible(true);
$this->assertSame(20, $detect->invoke($adapter, $html));
$build = new \ReflectionMethod($adapter, 'buildPageUrl');
$build->setAccessible(true);
$url = $build->invoke(
$adapter,
'https://faculty.sjtu.edu.cn/xyjs_list.jsp?urltype=tsites.CollegeTeacherList&wbtreeid=1001&st=0&id=1701&lang=zh_CN',
3,
$html,
);
$this->assertStringContainsString('PAGENUM=3', $url);
$this->assertStringContainsString('totalpage=20', $url);
}
public function test_extracts_smse_staff_panel_list(): void
{
$html = <<<'HTML'
教师名录 - 上海交通大学材料科学与工程学院
HTML;
$adapter = new FacultyListHtmlAdapter;
$method = new \ReflectionMethod($adapter, 'extractFromHtml');
$method->setAccessible(true);
$items = $method->invoke(
$adapter,
$html,
[],
'https://smse.sjtu.edu.cn/people/staff_new/department',
);
$this->assertCount(2, $items);
$this->assertSame('陈军', $items[0]->title);
$this->assertSame('https://smse.sjtu.edu.cn/people/detail_new/20092', $items[0]->canonicalUrl);
$this->assertSame('上海交通大学', $items[0]->schoolName);
$this->assertSame('faculty_html_smse', $items[0]->extra['platform']);
$this->assertSame('塑性成形技术与装备研究院', $items[0]->extra['college_name']);
}
public function test_apply_profile_metadata_from_smse_detail_page(): void
{
$html = <<<'HTML'
所属二级机构：塑性成形技术与装备研究院
HTML;
$adapter = new FacultyListHtmlAdapter;
$method = new \ReflectionMethod($adapter, 'applyProfileMetadataToItem');
$method->setAccessible(true);
$item = $method->invoke(
$adapter,
new \App\Services\Crawl\CrawlItemDto(
externalId: 'faculty:test',
title: '陈军',
canonicalUrl: 'https://smse.sjtu.edu.cn/people/detail_new/20092',
extra: [
'lead_author' => [
'name' => '陈军',
'email' => null,
'university_name' => '上海交通大学',
],
],
),
$html,
);
$this->assertSame('教授', $item->extra['lead_author']['academic_title']);
$this->assertSame('塑性成形技术与装备研究院', $item->extra['college_name']);
}
public function test_extracts_sais_js_list_from_ajax_content(): void
{
$html = <<<'HTML'
教师名录-上海交通大学自动化与感知学院
HTML;
$adapter = new FacultyListHtmlAdapter;
$method = new \ReflectionMethod($adapter, 'extractFromAjaxTeacherContent');
$method->setAccessible(true);
$items = $method->invoke(
$adapter,
$html,
[],
'https://sais.sjtu.edu.cn/faculty.html',
'faculty',
);
$this->assertCount(2, $items);
$this->assertSame('白洋', $items[0]->title);
$this->assertSame('https://sais.sjtu.edu.cn/faculty/baiyang.html', $items[0]->canonicalUrl);
$this->assertSame('faculty_html_ajax', $items[0]->extra['platform']);
$this->assertSame('上海交通大学', $items[0]->schoolName);
}
public function test_extracts_cs_rc_item_teacher_list(): void
{
$html = <<<'HTML'
教师名录-上海交通大学计算机学院（网络空间安全学院、密码学院）
HTML;
$adapter = new FacultyListHtmlAdapter;
$method = new \ReflectionMethod($adapter, 'extractFromAjaxTeacherContent');
$method->setAccessible(true);
$items = $method->invoke(
$adapter,
$html,
[],
'https://www.cs.sjtu.edu.cn/jiaoshiml.html',
'jiaoshiml',
);
$this->assertCount(2, $items);
$this->assertSame('臧斌宇', $items[0]->title);
$this->assertSame('并行与分布式系统研究所', $items[0]->extra['college_name']);
$this->assertSame('https://www.cs.sjtu.edu.cn/jiaoshiml/chenhaibo.html', $items[1]->canonicalUrl);
}
public function test_resolve_profile_enrich_max_caps_large_batches(): void
{
$adapter = new FacultyListHtmlAdapter;
$method = new \ReflectionMethod($adapter, 'resolveProfileEnrichMax');
$method->setAccessible(true);
$this->assertSame(32, $method->invoke($adapter, [], 500));
$this->assertSame(10, $method->invoke($adapter, ['profile_enrich_max' => 10], 500));
$this->assertSame(0, $method->invoke($adapter, ['skip_profile_enrich' => true], 500));
}
public function test_response_body_from_pool_result_ignores_connection_exception(): void
{
$adapter = new FacultyListHtmlAdapter;
$method = new \ReflectionMethod($adapter, 'responseBodyFromPoolResult');
$method->setAccessible(true);
$this->assertNull($method->invoke($adapter, new \GuzzleHttp\Exception\ConnectException(
'Connection timed out',
new \GuzzleHttp\Psr7\Request('GET', 'https://faculty.sjtu.edu.cn/test'),
)));
$this->assertNull($method->invoke($adapter, null));
}
public function test_parses_icisee_ajax_teacher_config_without_cat_id(): void
{
$html = <<<'HTML'
HTML;
$adapter = new FacultyListHtmlAdapter;
$method = new \ReflectionMethod($adapter, 'parseAjaxTeacherConfig');
$method->setAccessible(true);
$config = $method->invoke($adapter, $html, 'https://icisee.sjtu.edu.cn/jiaoshiml.html');
$this->assertSame('simple', $config['variant']);
$this->assertNull($config['cat_id']);
$this->assertSame('jiaoshiml', $config['cat_code']);
$this->assertSame('https://icisee.sjtu.edu.cn/active/ajax_teacher_list.html', $config['api_url']);
$this->assertTrue($config['uses_page']);
}
public function test_extracts_icisee_card_style_teacher_list(): void
{
$html = <<<'HTML'
教师名录-上海交通大学集成电路学院（信息与电子工程学院）
张三副教授
HTML;
$adapter = new FacultyListHtmlAdapter;
$method = new \ReflectionMethod($adapter, 'extractFromAjaxTeacherContent');
$method->setAccessible(true);
$items = $method->invoke(
$adapter,
$html,
[],
'https://icisee.sjtu.edu.cn/jiaoshiml.html',
'jiaoshiml',
);
$this->assertCount(2, $items);
$this->assertSame('蔡星汉', $items[0]->title);
$this->assertSame('教授', $items[0]->extra['academic_title']);
$this->assertSame('https://icisee.sjtu.edu.cn/jiaoshiml/caixinghan.html', $items[0]->canonicalUrl);
$this->assertSame('张三', $items[1]->title);
$this->assertSame('副教授', $items[1]->extra['academic_title']);
$this->assertSame('faculty_html_ajax', $items[0]->extra['platform']);
}
public function test_detects_nju_teacher_home_page(): void
{
$html = '';
$adapter = new FacultyListHtmlAdapter;
$method = new \ReflectionMethod($adapter, 'isNjuTeacherHomePage');
$method->setAccessible(true);
$this->assertTrue($method->invoke($adapter, $html));
$this->assertFalse($method->invoke($adapter, ''));
}
public function test_builds_nju_teacher_home_conditions_for_all_faculty(): void
{
$adapter = new FacultyListHtmlAdapter;
$method = new \ReflectionMethod($adapter, 'buildNjuTeacherHomeConditions');
$method->setAccessible(true);
$conditions = $method->invoke($adapter, null, null);
$this->assertCount(1, $conditions);
$this->assertSame('published', $conditions[0]['field']);
}
public function test_builds_nju_teacher_home_conditions_for_professor_category(): void
{
$adapter = new FacultyListHtmlAdapter;
$method = new \ReflectionMethod($adapter, 'buildNjuTeacherHomeConditions');
$method->setAccessible(true);
$conditions = $method->invoke($adapter, '教授', null);
$this->assertCount(2, $conditions);
$this->assertSame('exField2', $conditions[1]['field']);
$this->assertSame('教授', $conditions[1]['value']);
}
public function test_parses_nju_site_id_from_html(): void
{
$html = '';
$adapter = new FacultyListHtmlAdapter;
$method = new \ReflectionMethod($adapter, 'parseNjuSiteId');
$method->setAccessible(true);
$this->assertSame(786, $method->invoke($adapter, $html));
}
public function test_infers_college_from_meta_description(): void
{
$html = '师资力量';
$adapter = new FacultyListHtmlAdapter;
$method = new \ReflectionMethod($adapter, 'inferCollegeFromPageTitle');
$method->setAccessible(true);
$this->assertSame('智能科学与技术学院', $method->invoke($adapter, $html));
}
public function test_extracts_sudy_news_faculty_list(): void
{
$html = <<<'HTML'
师资力量-南京大学前沿科学学院
功能材料与智能制造研究院
HTML;
$adapter = new FacultyListHtmlAdapter;
$method = new \ReflectionMethod($adapter, 'extractFromSudyNewsFacultyList');
$method->setAccessible(true);
$items = $method->invoke($adapter, $html, [], 'https://frontier.nju.edu.cn/zrjs/list.htm');
$this->assertCount(1, $items);
$this->assertSame('王保明', $items[0]->title);
$this->assertSame('功能材料与智能制造研究院', $items[0]->extra['college_name']);
$this->assertSame('faculty_html_sudy_news', $items[0]->extra['platform']);
}
public function test_extracts_ra_teacher_cards(): void
{
$html = <<<'HTML'
专职教师-南京大学机器人与自动化学院
HTML;
$adapter = new FacultyListHtmlAdapter;
$method = new \ReflectionMethod($adapter, 'extractFromRaTeacherList');
$method->setAccessible(true);
$items = $method->invoke($adapter, $html, [], 'https://ra.nju.edu.cn/szll/zzjs/index.html');
$this->assertCount(1, $items);
$this->assertSame('周克敏', $items[0]->title);
$this->assertSame('教授', $items[0]->extra['academic_title']);
$this->assertSame('faculty_html_ra', $items[0]->extra['platform']);
}
public function test_extracts_vsb_faculty_table(): void
{
$html = <<<'HTML'
专兼职教师-南京大学智能软件与工程学院
HTML;
$adapter = new FacultyListHtmlAdapter;
$method = new \ReflectionMethod($adapter, 'extractFromVsbFacultyTable');
$method->setAccessible(true);
$items = $method->invoke($adapter, $html, [], 'https://ise.nju.edu.cn/szll/zjzjs.htm');
$this->assertCount(3, $items);
$names = array_map(fn ($item) => $item->title, $items);
$this->assertContains('杨鲲', $names);
$this->assertContains('陶先平', $names);
$this->assertContains('邵栋', $names);
$titles = array_column(array_map(fn ($item) => $item->extra, $items), 'academic_title', null);
$this->assertSame('教授', $items[array_search('杨鲲', $names, true)]->extra['academic_title']);
$this->assertSame('副教授', $items[array_search('邵栋', $names, true)]->extra['academic_title']);
}
}
