Exemplos de get_html_from_raw_text em Python

Exemplo n.º 1

Arquivo: generate_judge_stats.py Projeto: Andr3iC/courtlistener

def get_judge_and_court(case_path):
    raw_text = open(case_path).read()
    clean_html_tree, complete_html_tree, clean_html_str, body_text = get_html_from_raw_text(raw_text)
    judge = get_judge(clean_html_tree, case_path)
    court = get_court_object(clean_html_tree, case_path=case_path)
    if judge in judge_stats:
        if court in judge_stats[judge]:
            judge_stats[judge][court] += 1
        else:
            judge_stats[judge][court] = 1
    else:
        judge_stats[judge] = {court: 1}

Exemplo n.º 2

Arquivo: generate_judge_stats.py Projeto: zayedmohamed/courtlistener

def get_judge_and_court(case_path):
    raw_text = open(case_path).read()
    clean_html_tree, complete_html_tree, clean_html_str, body_text = get_html_from_raw_text(
        raw_text)
    judge = get_judge(clean_html_tree, case_path)
    court = get_court_object(clean_html_tree, case_path=case_path)
    if judge in judge_stats:
        if court in judge_stats[judge]:
            judge_stats[judge][court] += 1
        else:
            judge_stats[judge][court] = 1
    else:
        judge_stats[judge] = {court: 1}

Exemplo n.º 3

Arquivo: extract_courts_only.py Projeto: Andr3iC/courtlistener

def import_law_box_case(case_path):
    raw_text = open(case_path).read()
    clean_html_tree, complete_html_tree, clean_html_str, body_text = get_html_from_raw_text(raw_text)
    citations = get_citations_from_tree(complete_html_tree, case_path)
    court = get_court_object(clean_html_tree, citations, case_path)

    if not court and 'review_court_issues' in DEBUG:
        if 'review_court_issues' in DEBUG:
            subprocess.Popen(['firefox', 'file://%s' % case_path], shell=False).communicate()
            raw_input("No court identified! Can we fix this and restart, or just press enter to log it? ")
        if 'log_bad_courts' in DEBUG:
            # Write the failed case out to file.
            with open('missing_courts_post_focus.txt', 'a') as out:
                out.write('%s\n' % case_path)

Exemplo n.º 4

Arquivo: extract_courts_only.py Projeto: zayedmohamed/courtlistener

def import_law_box_case(case_path):
    raw_text = open(case_path).read()
    clean_html_tree, complete_html_tree, clean_html_str, body_text = get_html_from_raw_text(
        raw_text)
    citations = get_citations_from_tree(complete_html_tree, case_path)
    court = get_court_object(clean_html_tree, citations, case_path)

    if not court and 'review_court_issues' in DEBUG:
        if 'review_court_issues' in DEBUG:
            subprocess.Popen(['firefox', 'file://%s' % case_path],
                             shell=False).communicate()
            raw_input(
                "No court identified! Can we fix this and restart, or just press enter to log it? "
            )
        if 'log_bad_courts' in DEBUG:
            # Write the failed case out to file.
            with open('missing_courts_post_focus.txt', 'a') as out:
                out.write('%s\n' % case_path)