def get_dividend_paras(doc_text):
start_trigger = "shall be entitled to receive dividends"
start_trigger = preprocess_text(start_trigger, stem=False).split(" ")
loc = find_loc(doc_text, [start_trigger], allow_contains=True)
if loc is None:
return "Failed to find dividend info text"
loc = loc[0]
numbers = get_nums_from_text(doc_text, min=0, max=1, decimal=True)
start_found = False
start = -1
for i in range(len(numbers)):
cur_num = numbers[i]
if not start_found and cur_num[2] > loc:
start = cur_num[2] - 10
start_found = True
continue
if start_found:
if (cur_num[2] - numbers[i - 1][2]) > 140: # value is too far aways
end = numbers[i-2][2] + 20
return " ".join(doc_text[start: end])
if not start_found:
return "Failed to find dividend info text"
return " ".join(doc_text[start: numbers[-1][2] + 20])
def get_IV_intro_text(filename,
beginning_intro_triggers_filename="Beginning_IV",
end_intro_triggers_filename="End_IV_intro"):
beginning_intro_triggers = get_proccessed_triggers(
beginning_intro_triggers_filename)
end_intro_triggers = get_proccessed_triggers(end_intro_triggers_filename)
text = read_contract(filename)
text = text.split(" ")
# print("text", text)
# text = text.split(" ")
# text = remove_periods(text)
if filename in Failures:
return "Failure expected"
try:
beginning_of_IV_intro = find_loc(text,
beginning_intro_triggers,
allow_contains=False)
end_of_IV_intro = find_loc(text, end_intro_triggers)[0]
except TypeError:
print("failed", filename)
print(text)
return None
# In case triggers didn't work, search the entire document
if beginning_of_IV_intro is None:
print("Beginning of IV intro is None")
beginning_of_IV_intro = 0
if beginning_of_IV_intro[0] > end_of_IV_intro:
print("End is less than beginning index of IV intro")
beginning_of_IV_intro = [0]
end_of_IV_intro = len(text)
# assert (beginning_of_IV_intro is not None), "Invalid info file " + " ".join(text) + filename
beginning_of_IV = get_closest_string(values=beginning_of_IV_intro,
target=end_of_IV_intro,
less=True)
if beginning_of_IV == -1:
print("no beginning found")
beginning_of_IV = end_of_IV_intro - 300
IV_intro_text = text[beginning_of_IV:end_of_IV_intro]
return IV_intro_text
def get_board_of_directors_paras(doc_text):
# start_trigger = "election of directors"
start_trigger = "director elect"
start_trigger = preprocess_text(start_trigger, stem=False).split(" ")
print(start_trigger)
loc = find_loc(doc_text, [start_trigger], allow_contains=False)
if loc is None:
loc = find_loc(doc_text, [start_trigger], allow_contains=True)
if loc is None:
return "Failed to find directors info text"
for i in loc:
if i < 200:
continue
print(i)
return " ".join(doc_text[i: i + 200])
# TODO: review find_loc function given sample results
return "Failed to find board of directors info text"
def get_securities_info_paras(doc_text, beginning_intro_triggers_filename="Beginning_IV",
end_intro_triggers_filename="End_IV_intro"):
beginning_intro_triggers = get_proccessed_triggers(beginning_intro_triggers_filename, preproccess=False)
end_intro_triggers = get_proccessed_triggers(end_intro_triggers_filename, preproccess=False)
try:
beginning_of_IV_intro = find_loc(doc_text, beginning_intro_triggers, allow_contains=True)
end_of_IV_intro = find_loc(doc_text, end_intro_triggers)
except TypeError:
return "Failed to find securities info text"
# In case triggers didn't work, search the entire document
diffs = sorted(product(beginning_of_IV_intro, end_of_IV_intro), key=lambda t: abs(t[0] - t[1]))
_, names = get_names(doc_text, stem=False)
names_used = get_names_from_text(doc_text, names)
names_used = " ".join([x[0] + ", " for x in names_used])[:-2]
for dif in diffs:
if dif[0] < dif[1]:
return "Types of shares found: " + names_used + "\n" + " ".join(doc_text[dif[0]:dif[1]])
return "Failed to find securities info text"
def get_liquidation_paras(doc_text):
start_triggers = ["event of any liquidation", "upon any such liquidation"]
start_triggers = [preprocess_text(start_trigger, stem=False).split(" ") for start_trigger in start_triggers]
start_locs = find_loc(doc_text, start_triggers, allow_contains=True)
print(start_locs)
print(len(doc_text))
if start_locs is None or start_locs == []:
return "Failed to find liquidation info text"
for i in start_locs:
if i > 200:
out = " ".join(doc_text[i: i + 200])
return out
return "Failed to find liquidation info text"
def get_original_issue_price(text, names, buffer=30):
loc = find_loc(text, ["riginal issue pric"], allow_contains=True)
print("LOC", loc)
text = text[loc:]
used_names = get_names_from_text(text, names)
numbers = get_nums_from_text(text, min=0, max=5, decimal=True)
pairs = match_nums_with_targets(numbers, used_names)
names_out = []
nums_out = []
if pairs is None:
print("NA")
return None, None
for name, num in pairs:
names_out.append(name)
nums_out.append(num)
print("nums_out", nums_out)
print("names_out", names_out)
return nums_out, names_out