def extractnumber_fr(text): """Takes in a string and extracts a number. Args: text (str): the string to extract a number from Returns: (str): The number extracted or the original text. """ # normalize text, keep articles for ordinals versus fractionals text = normalize_fr(text, False) # split words by whitespace aWords = text.split() count = 0 result = None add = False while count < len(aWords): val = None word = aWords[count] wordNext = "" wordPrev = "" if count < (len(aWords) - 1): wordNext = aWords[count + 1] if count > 0: wordPrev = aWords[count - 1] if word in articles_fr: count += 1 continue if word in ["et", "plus", "+"]: count += 1 add = True continue # is current word a numeric number? if word.isdigit(): val = int(word) count += 1 elif is_numeric(word): val = float(word) count += 1 elif wordPrev in articles_fr and getOrdinal_fr(word): val = getOrdinal_fr(word) count += 1 # is current word the denominator of a fraction? elif isFractional_fr(word): val = isFractional_fr(word) count += 1 # is current word the numerator of a fraction? if val and wordNext: valNext = isFractional_fr(wordNext) if valNext: val = float(val) * valNext count += 1 if not val: count += 1 # is current word a numeric fraction like "2/3"? aPieces = word.split('/') # if (len(aPieces) == 2 and is_numeric(aPieces[0]) # and is_numeric(aPieces[1])): if look_for_fractions(aPieces): val = float(aPieces[0]) / float(aPieces[1]) # is current word followed by a decimal value? if wordNext == "virgule": zeros = 0 newWords = aWords[count + 1:] # count the number of zeros after the decimal sign for word in newWords: if word == "zéro" or word == "0": zeros += 1 else: break afterDotVal = None # extract the number after the zeros if newWords[zeros].isdigit(): afterDotVal = newWords[zeros] countDot = count + zeros + 2 # if a number was extracted (since comma is also a # punctuation sign) if afterDotVal: count = countDot if not val: val = 0 # add the zeros afterDotString = zeros * "0" + afterDotVal val = float(str(val) + "." + afterDotString) if val: if add: result += val add = False else: result = val # if result == False: if not result: return normalize_fr(text, True) return result
def extractnumber_de(text, short_scale=True, ordinals=False): """ This function prepares the given text for parsing by making numbers consistent, getting rid of contractions, etc. Args: text (str): the string to normalize Returns: (int) or (float): The value of extracted number undefined articles cannot be suppressed in German: 'ein Pferd' means 'one horse' and 'a horse' """ aWords = text.split() aWords = [word for word in aWords if word not in ["der", "die", "das", "des", "den", "dem"]] and_pass = False valPreAnd = False val = False count = 0 while count < len(aWords): word = aWords[count] if is_numeric(word): # if word.isdigit(): # doesn't work with decimals val = float(word) elif isFractional_de(word): val = isFractional_de(word) elif isOrdinal_de(word): val = isOrdinal_de(word) else: if word in de_numbers: val = de_numbers[word] if count < (len(aWords) - 1): wordNext = aWords[count + 1] else: wordNext = "" valNext = isFractional_de(wordNext) if valNext: val = val * valNext aWords[count + 1] = "" if not val: # look for fractions like "2/3" aPieces = word.split('/') # if (len(aPieces) == 2 and is_numeric(aPieces[0]) # and is_numeric(aPieces[1])): if look_for_fractions(aPieces): val = float(aPieces[0]) / float(aPieces[1]) elif and_pass: # added to value, quit here val = valPreAnd break else: count += 1 continue aWords[count] = "" if and_pass: aWords[count - 1] = '' # remove "and" val += valPreAnd elif count + 1 < len(aWords) and aWords[count + 1] == 'und': and_pass = True valPreAnd = val val = False count += 2 continue elif count + 2 < len(aWords) and aWords[count + 2] == 'und': and_pass = True valPreAnd = val val = False count += 3 continue break return val or False
def extractnumber_it(text, short_scale=False, ordinals=False): """ This function extracts a number from a text string, handles pronunciations in long scale and short scale https://en.wikipedia.org/wiki/Names_of_large_numbers Args: text (str): the string to normalize short_scale (bool): use short scale if True, long scale if False ordinals (bool): consider ordinal numbers, third=3 instead of 1/3 Returns: (int) or (float) or False: The extracted number or False if no number was found """ string_num_ordinal_it = {} # first, second... if ordinals: if short_scale: for num in SHORT_ORDINAL_STRING_IT: num_string = SHORT_ORDINAL_STRING_IT[num] string_num_ordinal_it[num_string] = num STRING_NUM_ITA[num_string] = num else: for num in LONG_ORDINAL_STRING_IT: num_string = LONG_ORDINAL_STRING_IT[num] string_num_ordinal_it[num_string] = num STRING_NUM_ITA[num_string] = num # negate next number (-2 = 0 - 2) negatives = ['meno'] # 'negativo' non è usuale in italiano # multiply the previous number (one hundred = 1 * 100) multiplies = ['decina', 'decine', 'dozzina', 'dozzine', 'centinaia', 'centinaio', 'migliaia', 'migliaio', 'mila'] # split sentence parse separately and sum ( 2 and a half = 2 + 0.5 ) fraction_marker = [' e '] # decimal marker ( 1 point 5 = 1 + 0.5) decimal_marker = [' punto ', ' virgola '] if short_scale: for num in SHORT_SCALE_IT: num_string = SHORT_SCALE_IT[num] STRING_NUM_ITA[num_string] = num multiplies.append(num_string) else: for num in LONG_SCALE_IT: num_string = LONG_SCALE_IT[num] STRING_NUM_ITA[num_string] = num multiplies.append(num_string) # 2 e 3/4 ed altri casi for separator in fraction_marker: components = text.split(separator) zeros = 0 if len(components) == 2: # count zeros in fraction part sub_components = components[1].split(' ') for element in sub_components: if element == 'zero' or element == '0': zeros += 1 else: break # ensure first is not a fraction and second is a fraction num1 = extractnumber_it(components[0]) num2 = extractnumber_it(components[1]) if num1 is not None and num2 is not None \ and num1 >= 1 and 0 < num2 < 1: return num1 + num2 # sette e quaranta sette e zero zero due elif num1 is not None and num2 is not None \ and num1 >= 1 and num2 > 1: return num1 + num2 / pow(10, len(str(num2)) + zeros) # 2 punto 5 for separator in decimal_marker: zeros = 0 # count zeros in fraction part components = text.split(separator) if len(components) == 2: sub_components = components[1].split(' ') for element in sub_components: if element == 'zero' or element == '0': zeros += 1 else: break number = int(extractnumber_it(components[0])) decimal = int(extractnumber_it(components[1])) if number is not None and decimal is not None: if '.' not in str(decimal): return number + decimal / pow(10, len(str(decimal)) + zeros) all_words = text.split() val = False prev_val = None to_sum = [] for idx, word in enumerate(all_words): if not word: continue prev_word = all_words[idx - 1] if idx > 0 else '' next_word = all_words[idx + 1] if idx + 1 < len(all_words) else '' # is this word already a number ? if is_numeric(word): val = float(word) # is this word the name of a number ? if word in STRING_NUM_ITA: val = STRING_NUM_ITA[word] # tre quarti un quarto trenta secondi if isFractional_it(word) and prev_val: if word[:-1] == 'second' and not ordinals: val = prev_val * 2 else: val = prev_val # is the prev word a number and should we multiply it? # twenty hundred, six hundred if word in multiplies: if not prev_val: prev_val = 1 val = prev_val * val # is this a spoken fraction? # mezza tazza if val is False: val = isFractional_it(word, short_scale=short_scale) # 2 quinti if not ordinals: next_value = isFractional_it(next_word, short_scale=short_scale) if next_value: if not val: val = 1 val = val * next_value # is this a negative number? if val and prev_word and prev_word in negatives: val = 0 - val if not val: val = extractnumber_long_it(word) # let's make sure it isn't a fraction if not val: # look for fractions like '2/3' all_pieces = word.split('/') if look_for_fractions(all_pieces): val = float(all_pieces[0]) / float(all_pieces[1]) else: prev_val = val # handle long numbers # six hundred sixty six # two million five hundred thousand if word in multiplies and next_word not in multiplies: to_sum.append(val) val = 0 prev_val = 0 elif extractnumber_long_it(word) > 100 and \ extractnumber_long_it(next_word) and \ next_word not in multiplies: to_sum.append(val) val = 0 prev_val = 0 if val is not None: for addend in to_sum: val = val + addend return val
def extractnumber_pt(text, short_scale=True, ordinals=False): """ This function prepares the given text for parsing by making numbers consistent, getting rid of contractions, etc. Args: text (str): the string to normalize Returns: (int) or (float): The value of extracted number """ text = text.lower() aWords = text.split() count = 0 result = None while count < len(aWords): val = 0 word = aWords[count] next_next_word = None if count + 1 < len(aWords): next_word = aWords[count + 1] if count + 2 < len(aWords): next_next_word = aWords[count + 2] else: next_word = None # is current word a number? if word in _NUMBERS_PT: val = _NUMBERS_PT[word] elif word.isdigit(): # doesn't work with decimals val = int(word) elif is_numeric(word): val = float(word) elif isFractional_pt(word): if not result: result = 1 result = result * isFractional_pt(word) count += 1 continue if not val: # look for fractions like "2/3" aPieces = word.split('/') # if (len(aPieces) == 2 and is_numeric(aPieces[0]) # and is_numeric(aPieces[1])): if look_for_fractions(aPieces): val = float(aPieces[0]) / float(aPieces[1]) if val: if result is None: result = 0 # handle fractions if next_word != "avos": result += val else: result = float(result) / float(val) if next_word is None: break # number word and fraction ands = ["e"] if next_word in ands: zeros = 0 if result is None: count += 1 continue newWords = aWords[count + 2:] newText = "" for word in newWords: newText += word + " " afterAndVal = extractnumber_pt(newText[:-1]) if afterAndVal: if result < afterAndVal or result < 20: while afterAndVal > 1: afterAndVal = afterAndVal / 10.0 for word in newWords: if word == "zero" or word == "0": zeros += 1 else: break for _ in range(0, zeros): afterAndVal = afterAndVal / 10.0 result += afterAndVal break elif next_next_word is not None: if next_next_word in ands: newWords = aWords[count + 3:] newText = "" for word in newWords: newText += word + " " afterAndVal = extractnumber_pt(newText[:-1]) if afterAndVal: if result is None: result = 0 result += afterAndVal break decimals = ["ponto", "virgula", "vírgula", ".", ","] if next_word in decimals: zeros = 0 newWords = aWords[count + 2:] newText = "" for word in newWords: newText += word + " " for word in newWords: if word == "zero" or word == "0": zeros += 1 else: break afterDotVal = str(extractnumber_pt(newText[:-1])) afterDotVal = zeros * "0" + afterDotVal result = float(str(result) + "." + afterDotVal) break count += 1 # Return the $str with the number related words removed # (now empty strings, so strlen == 0) # aWords = [word for word in aWords if len(word) > 0] # text = ' '.join(aWords) if "." in str(result): integer, dec = str(result).split(".") # cast float to int if dec == "0": result = int(integer) return result or False
def extractnumber_sv(text): """ This function prepares the given text for parsing by making numbers consistent, getting rid of contractions, etc. Args: text (str): the string to normalize Returns: (int) or (float): The value of extracted number """ aWords = text.split() and_pass = False valPreAnd = False val = False count = 0 while count < len(aWords): word = aWords[count] if is_numeric(word): val = float(word) elif word == "första": val = 1 elif word == "andra": val = 2 elif word == "tredje": val = 3 elif word == "fjärde": val = 4 elif word == "femte": val = 5 elif word == "sjätte": val = 6 elif is_fractional_sv(word): val = is_fractional_sv(word) else: if word == "en": val = 1 if word == "ett": val = 1 elif word == "två": val = 2 elif word == "tre": val = 3 elif word == "fyra": val = 4 elif word == "fem": val = 5 elif word == "sex": val = 6 elif word == "sju": val = 7 elif word == "åtta": val = 8 elif word == "nio": val = 9 elif word == "tio": val = 10 if val: if count < (len(aWords) - 1): wordNext = aWords[count + 1] else: wordNext = "" valNext = is_fractional_sv(wordNext) if valNext: val = val * valNext aWords[count + 1] = "" if not val: # look for fractions like "2/3" aPieces = word.split('/') if look_for_fractions(aPieces): val = float(aPieces[0]) / float(aPieces[1]) elif and_pass: # added to value, quit here val = valPreAnd break else: count += 1 continue aWords[count] = "" if and_pass: aWords[count - 1] = '' # remove "och" val += valPreAnd elif count + 1 < len(aWords) and aWords[count + 1] == 'och': and_pass = True valPreAnd = val val = False count += 2 continue elif count + 2 < len(aWords) and aWords[count + 2] == 'och': and_pass = True valPreAnd = val val = False count += 3 continue break if not val: return False return val
def extract_number_ca(text, short_scale=True, ordinals=False): """ This function prepares the given text for parsing by making numbers consistent, getting rid of contractions, etc. Args: text (str): the string to normalize Returns: (int) or (float): The value of extracted number """ # TODO: short_scale and ordinals don't do anything here. # The parameters are present in the function signature for API compatibility # reasons. text = text.lower() aWords = text.split() count = 0 result = None while count < len(aWords): val = 0 word = aWords[count] next_next_word = None if count + 1 < len(aWords): next_word = aWords[count + 1] if count + 2 < len(aWords): next_next_word = aWords[count + 2] else: next_word = None # is current word a number? if word in _NUMBERS_CA: val = _NUMBERS_CA[word] elif '-' in word: wordparts = word.split('-') # trenta-cinc > 35 if len(wordparts) == 2 and (wordparts[0] in _TENS_CA and wordparts[1] in _AFTER_TENS_CA): val = _TENS_CA[wordparts[0]] + _AFTER_TENS_CA[wordparts[1]] # vint-i-dues > 22 elif len(wordparts) == 3 and wordparts[1] == 'i' and ( wordparts[0] in _TENS_CA and wordparts[2] in _AFTER_TENS_CA): val = _TENS_CA[wordparts[0]] + _AFTER_TENS_CA[wordparts[2]] # quatre-centes > 400 elif len(wordparts) == 2 and (wordparts[0] in _BEFORE_HUNDREDS_CA and wordparts[1] in _HUNDREDS_CA): val = _BEFORE_HUNDREDS_CA[wordparts[0]] * 100 elif word.isdigit(): # doesn't work with decimals val = int(word) elif is_numeric(word): val = float(word) elif is_fractional_ca(word): if not result: result = 1 result = result * is_fractional_ca(word) count += 1 continue if not val: # look for fractions like "2/3" aPieces = word.split('/') # if (len(aPieces) == 2 and is_numeric(aPieces[0]) # and is_numeric(aPieces[1])): if look_for_fractions(aPieces): val = float(aPieces[0]) / float(aPieces[1]) if val: if result is None: result = 0 # handle fractions #TODO: caution, review use of "ens" word if next_word != "ens": result += val else: result = float(result) / float(val) if next_word is None: break # number word and fraction ands = ["i"] if next_word in ands: zeros = 0 if result is None: count += 1 continue newWords = aWords[count + 2:] newText = "" for word in newWords: newText += word + " " afterAndVal = extract_number_ca(newText[:-1]) if afterAndVal: if result < afterAndVal or result < 20: while afterAndVal > 1: afterAndVal = afterAndVal / 10.0 for word in newWords: if word == "zero" or word == "0": zeros += 1 else: break for _ in range(0, zeros): afterAndVal = afterAndVal / 10.0 result += afterAndVal break elif next_next_word is not None: if next_next_word in ands: newWords = aWords[count + 3:] newText = "" for word in newWords: newText += word + " " afterAndVal = extract_number_ca(newText[:-1]) if afterAndVal: if result is None: result = 0 result += afterAndVal break decimals = ["coma", "amb", "punt", ".", ","] if next_word in decimals: zeros = 0 newWords = aWords[count + 2:] newText = "" for word in newWords: newText += word + " " for word in newWords: if word == "zero" or word == "0": zeros += 1 else: break afterDotVal = str(extract_number_ca(newText[:-1])) afterDotVal = zeros * "0" + afterDotVal result = float(str(result) + "." + afterDotVal) break count += 1 # Return the $str with the number related words removed # (now empty strings, so strlen == 0) # aWords = [word for word in aWords if len(word) > 0] # text = ' '.join(aWords) if "." in str(result): integer, dec = str(result).split(".") # cast float to int if dec == "0": result = int(integer) return result or False