def get_court_by_paren(paren_string: str, citation: Citation) -> Optional[str]: """Takes the citation string, usually something like "2d Cir", and maps that back to the court code. Does not work on SCOTUS, since that court lacks parentheticals, and needs to be handled after disambiguation has been completed. """ if citation.year is None: court_str = strip_punct(paren_string) else: year_index = paren_string.find(str(citation.year)) court_str = strip_punct(paren_string[:year_index]) court_code = None if court_str == "": court_code = None else: # Map the string to a court, if possible. for court in courts: # Use startswith because citations are often missing final period, # e.g. "2d Cir" if court["citation_string"].startswith(court_str): court_code = court["id"] break return court_code
def extract_full_citation( words: List[str], reporter_index: int, ) -> Optional[FullCitation]: """Given a list of words and the index of a federal reporter, look before and after for volume and page. If found, construct and return a FullCitation object. Example: Adarand Constructors, Inc. v. Peña, 515 U.S. 200, 240 If we are given neutral, tax court opinions we treat them differently. The formats often follow {REPORTER} {YEAR}-{ITERATIVE_NUMBER} ex. T.C. Memo. 2019-13 """ # Get reporter reporter = words[reporter_index] # Variables to extact volume: Union[int, str, None] page: Union[int, str, None] # Handle tax citations is_tax_citation = is_neutral_tc_reporter(reporter) if is_tax_citation: volume, page = words[reporter_index + 1].replace("–", "-").split("-") # Handle "normal" citations, e.g., XX F.2d YY else: # Don't check if we are at the beginning of a string if reporter_index == 0: return None volume = strip_punct(words[reporter_index - 1]) page = strip_punct(words[reporter_index + 1]) # Get volume if volume.isdigit(): volume = int(volume) else: # No volume, therefore not a valid citation return None # Get page page = parse_page(page) if not page: return None # Return FullCitation return FullCitation( reporter, page, volume, reporter_found=reporter, reporter_index=reporter_index, )
def parse_page(page: Union[str, int]) -> Optional[str]: """Test whether something is a valid page number.""" page = strip_punct(str(page)) if page.isdigit(): # First, check whether the page is a simple digit. Most will be. return page # Otherwise, check whether the "page" is really one of the following: # (ordered in descending order of likelihood) # 1) A numerical page range. E.g., "123-124" # 2) A roman numeral. E.g., "250 Neb. xxiv (1996)" # 3) A special Connecticut or Illinois number. E.g., "13301-M" # 4) A page with a weird suffix. E.g., "559 N.W.2d 826|N.D." # 5) A page with a ¶ symbol, star, and/or colon. E.g., "¶ 119:12-14" match = ( re.match(r"\d{1,6}-\d{1,6}", page) # Simple page range or is_roman(page) # Roman numeral or re.match(r"\d{1,6}[-]?[a-zA-Z]{1,6}", page) # CT/IL page or re.match(r"\d{1,6}", page) # Weird suffix or re.match(r"[*\u00b6\ ]*[0-9:\-]+", page) # ¶, star, colon ) if match: return str(match.group(0)) return None
def extract_shortform_citation( words: List[str], reporter_index: int, ) -> Optional[ShortformCitation]: """Given a list of words and the index of a federal reporter, look before and after to see if this is a short form citation. If found, construct and return a ShortformCitation object. Shortform 1: Adarand, 515 U.S., at 241 Shortform 2: 515 U.S., at 241 """ # Don't check if we are at the beginning of a string if reporter_index <= 2: return None # Variables to extact volume: Union[int, str, None] page: Union[int, str, None] antecedent_guess: str # Get volume volume = strip_punct(words[reporter_index - 1]) if volume.isdigit(): volume = int(volume) else: # No volume, therefore not a valid citation return None # Get page try: page = parse_page(words[reporter_index + 2]) if not page: # There might be a comma in the way, so try one more index page = parse_page(words[reporter_index + 3]) if not page: # No page, therefore not a valid citation return None except IndexError: return None # Get antecedent antecedent_guess = words[reporter_index - 2] if antecedent_guess == ",": antecedent_guess = words[reporter_index - 3] + "," # Get reporter reporter = words[reporter_index] # Return ShortformCitation return ShortformCitation( reporter, page, volume, antecedent_guess, reporter_found=reporter, reporter_index=reporter_index, )
def get_year(token: str) -> Optional[int]: """Given a string token, look for a valid 4-digit number at the start and return its value. """ token = strip_punct(token) if not token.isdigit(): # Sometimes funny stuff happens? token = re.sub(r"(\d{4}).*", r"\1", token) if not token.isdigit(): return None if len(token) != 4: return None year = int(token) if year < 1754: # Earliest case in the database return None return year
def add_defendant(citation: Citation, words: List[str]) -> None: """Scan backwards from 2 tokens before reporter until you find v., in re, etc. If no known stop-token is found, no defendant name is stored. In the future, this could be improved. """ start_index = None back_seek = citation.reporter_index - BACKWARD_SEEK for index in range(citation.reporter_index - 1, max(back_seek, 0), -1): word = words[index] if word == ",": # Skip it continue if strip_punct(word).lower() in STOP_TOKENS: if word == "v.": citation.plaintiff = words[index - 1] start_index = index + 1 break if word.endswith(";"): # String citation break if start_index: citation.defendant = " ".join( words[start_index:citation.reporter_index - 1])
def get_citations( text: str, html: bool = True, do_post_citation: bool = True, do_defendant: bool = True, disambiguate: bool = True, ) -> List[Union[NonopinionCitation, Citation]]: """Main function""" if html: text = get_visible_text(text) words = tokenize(text) citations: List[Union[Citation, NonopinionCitation]] = [] for i in range(0, len(words) - 1): citation_token = words[i] citation: Union[Citation, NonopinionCitation, None] = None # CASE 1: Citation token is a reporter (e.g., "U. S."). # In this case, first try extracting it as a standard, full citation, # and if that fails try extracting it as a short form citation. if citation_token in list(EDITIONS.keys()) + list( VARIATIONS_ONLY.keys()): citation = extract_full_citation(words, i) if citation: # CASE 1A: Standard citation found, try to add additional data if do_post_citation: add_post_citation(citation, words) if do_defendant: add_defendant(citation, words) else: # CASE 1B: Standard citation not found, so see if this # reference to a reporter is a short form citation instead citation = extract_shortform_citation(words, i) if not citation: # Neither a full nor short form citation continue # CASE 2: Citation token is an "Id." or "Ibid." reference. # In this case, the citation should simply be to the item cited # immediately prior, but for safety we will leave that resolution up # to the user. elif citation_token.lower() in {"id.", "id.,", "ibid."}: citation = extract_id_citation(words, i) # CASE 3: Citation token is a "supra" reference. # In this case, we're not sure yet what the citation's antecedent is. # It could be any of the previous citations above. Thus, like an Id. # citation, for safety we won't resolve this reference yet. elif strip_punct(citation_token.lower()) == "supra": citation = extract_supra_citation(words, i) # CASE 4: Citation token is a section marker. # In this case, it's likely that this is a reference to a non- # opinion document. So we record this marker in order to keep # an accurate list of the possible antecedents for id citations. elif "§" in citation_token: citation = NonopinionCitation(match_token=citation_token) # CASE 5: The token is not a citation. else: continue if citation is not None: citations.append(citation) # Disambiguate each citation's reporter if disambiguate: citations = disambiguate_reporters(citations) citations = remove_address_citations(citations) # Set each citation's court property to "scotus" by default for citation in citations: if (isinstance(citation, Citation) and not citation.court and is_scotus_reporter(citation)): citation.court = "scotus" # Returns a list of citations ordered in the sequence that they appear in # the document. The ordering of this list is important for reconstructing # the references of the ShortformCitation, SupraCitation, and # IdCitation objects. return citations