def scrape_from_nanapi_and_build_heading_tree():
query = request.form['query']
head = 'http://nanapi.jp/search/q:'
query_url = head + query
nanapi_search_result_page = WebPage(query_url)
nanapi_search_result_page.fetch_html()
urls = nanapi_search_result_page.find_urls_from_nanapi_search_result()
results = []
for url in urls:
# result_pageはnanapiの1記事
result_page = WebPage(url)
result_page.fetch_html()
result_page.set_title()
# task_steps => [task_step, task_step, ...]
result_page.build_heading_tree()
result = {'title': result_page.title, 'nodes': result_page.top_nodes, 'url': result_page.url}
results.append(result)
return render_template('headings_and_li_texts.tmpl', results=results)
def scrape_from_nanapi():
query = request.form['query']
head = 'http://nanapi.jp/search/q:'
query_url = head + query
nanapi_search_result_page = WebPage(query_url)
nanapi_search_result_page.fetch_html()
urls = nanapi_search_result_page.find_urls_from_nanapi_search_result()
tasks = []
for url in urls:
# result_pageはnanapiの1記事
result_page = WebPage(url)
result_page.fetch_html()
# task_steps => [task_step, task_step, ...]
task = result_page.find_task_from_nanapi_with_headings()
# task_steps[0].h2 => 'はじめに'
# task_steps[0].h3s[0] => 'はじめに'
tasks.append(task)
# tasks => [task, task, ...]
# tasks[0][0].h2 => 'はじめに'
return render_template('nanapi_tasks.tmpl', tasks=tasks)
