def create_syllable_guide(self):
syllable_count = SortedStringTrie()
syllable_pronounciation = SortedStringTrie()
for line in self.dictionary:
index_space = line.find(" ")
word = line[:index_space]
pronounciation = line[index_space + 2:]
syllables = self.count_syllables(pronounciation)
syllable_count.__setitem__(word, syllables)
syllable_pronounciation.__setitem__(word, pronounciation)
return syllable_count, syllable_pronounciation
def setUp(self):
"""
测试前准备环境的搭建(setUp)
:return:
"""
self.words = 'an ant all allot alloy aloe are ate be'.split()
self.trie = SortedStringTrie(zip(self.words, range(len(self.words))))
def setUp(self):
self.words = 'an ant all allot alloy aloe are ate be'.split()
self.trie = SortedStringTrie(zip(self.words, range(len(self.words))))
class application(tkinter.Frame):
top_frame = None
homepage_frame = None
results_frame = None
results_text = None
__s_engine = search_engine.crawler()
bg_image = os.path.join('resources', 'background_1.jpg')
search_history = os.path.join('data', 'search_history.txt')
search_freq = os.path.join('data', 'search_freq.txt')
history = SortedStringTrie()
def __init__(self, master):
#for GUI
tkinter.Frame.__init__(self, master)
self.master = master
self.search_query = tkinter.StringVar()
self.search_query.trace("w", self.update_listbox)
self.initUI()
#for loading search history
self.__restore_history()
#for loading data into crawler
self.__s_engine.restore()
self.__s_engine.calc_pagerank(30)
def initUI(self):
self.master.title("SPIDER")
self.master.state('zoomed')
self.grid()
#set icon
icon = tkinter.PhotoImage(file=os.path.join('resources', "logo_t.png"))
self.master.tk.call('wm', 'iconphoto', self.master._w, icon)
application.top_frame = tkinter.Frame(self.master,
bg='black',
bd=3,
highlightcolor='blue')
application.top_frame.grid(row=0)
#tool bar on top of master
self.tabs = tkinter.Label(application.top_frame,
text="reserved for tabs")
self.tabs.grid(row=0)
self.options = tkinter.PhotoImage(
file=os.path.join('resources', "options_3.png"))
self.options = self.options.subsample(20, 20)
self.options_button = tkinter.Button(application.top_frame,
relief='flat',
image=self.options,
command=self.hide)
self.options_button.grid(row=1, column=8)
self.bookmark = tkinter.PhotoImage(
file=os.path.join('resources', "bookmark_1.png"))
self.bookmark = self.bookmark.subsample(40, 40)
self.bookmark_button = tkinter.Button(application.top_frame,
relief='flat',
bg='white',
image=self.bookmark,
command=self.hide)
self.bookmark_button.grid(row=1, column=7)
self.search_box1 = tkinter.Entry(application.top_frame,
textvariable=self.search_query,
bg='white',
fg='black',
selectborderwidth=2,
width=200)
self.search_box1.grid(row=1, column=0)
self.search_box1.bind('<Return>', self.__get)
#self.search_box.bind('<Configure>',self._resize_search_box)
#homepage
self.homepage_frame = tkinter.Frame()
self.homepage = self.create_homepage()
self.homepage_frame.grid(row=2)
#results page
self.results_frame = tkinter.Frame()
def hide(self):
self.frame.pack_forget()
'''def _resize_search_box(self,event):
new_width = event.width
self.search_box.destroy()
#if new_width > 70:
self.search_box = tkinter.Entry(application.top_frame,bg = 'white',fg = 'black',selectborderwidth = 1,bd = 10,font = ("COPRGTL.TTF"),relief = 'flat',width = new_width)
#else:
# self.search_box = tkinter.Entry(application.top_frame,bg = 'white',fg = 'black',selectborderwidth = 1,bd = 4,font = ("COPRGTL.TTF"),relief = 'flat',width = 77)'''
def create_homepage(self):
background_image = Image.open(self.bg_image)
background_image = background_image.resize((1536, 841))
self.background_photo = ImageTk.PhotoImage(background_image)
self.background_label = tkinter.Label(self.homepage_frame,
image=self.background_photo)
self.background_label.grid(row=1)
#name
self.name = tkinter.Label(self.homepage_frame,
text=" ADS ",
bg='black',
fg='white',
font=("SHOWG.TTF", 60))
self.name.grid()
self.name.place(x=620, y=275)
#search box
self.search_box = tkinter.Entry(self.homepage_frame,
textvariable=self.search_query,
bg='white',
fg='black',
selectborderwidth=1,
bd=6,
font=("COPRGTL.TTF", 16),
relief='flat',
width=40)
self.search_box.bind('<Return>', self.__get)
self.search_box.grid()
self.search_box.place(x=525, y=400) #centre 541,403.5
#search image
self.search_image = tkinter.PhotoImage(file=os.path.join(
'resources', "search_4.png")) #1-16 /4 - 8
self.search_image = self.search_image.subsample(8, 8)
self.search_button = tkinter.Button(self.homepage_frame,
image=self.search_image,
bg='white',
relief='flat')
self.search_button.grid(row=1)
self.search_button.place(x=1020, y=400)
#listbox
self.listbox = tkinter.Listbox(self.homepage_frame,
selectborderwidth=2,
width=0,
height=0)
#self.listbox.bind('<Configure>',self.resize_listbox)
self.listbox.grid(row=1)
self.listbox.place(x=525, y=439)
def __get(self, event):
input_str = self.search_box.get()
if input_str in self.history:
self.history[input_str] += 0.1
else:
self.history[input_str] = 0.1
results = self.__s_engine.query(input_str)
self.homepage_frame.grid_forget()
self.resultspage = self.create_resultspage(results)
self.results_frame.grid()
def update_listbox(self, *args):
text = self.search_query.get()
if text.isprintable():
relevent_words = self.__from_history(
text) #search the term from history ;return a list
self.listbox.delete(0, self.listbox.size())
for each_item in relevent_words:
self.listbox.insert(self.listbox.size(), each_item)
def __from_history(self, query):
if query in self.history:
return sorted(self.history, key=self.history.values)
return
def create_resultspage(self, results):
if self.results_text:
self.results_text.grid_forget()
self.results_text = tkinter.Label(self.results_frame, text="results\n")
self.results_text.grid()
self.results = tkinter.Label(self.results_frame, font=('', 20))
self.results.grid(row=1, column=0)
s = ''
for each_url in results:
s += each_url + '\n'
# arr = box(self.results_frame,each_url)
self.results.config(text=s)
def __restore_history(self):
words_file = open(search_history)
freq_file = open(search_freq)
words = words_file.read()
freq = freq_file.read()
words_list = []
freq_list = []
if words:
words_list = json.loads(words)
if freq:
freq_list = json.loads(freq)
words_file.close()
freq_file.close()
for loop_var in range(len(words_list)):
self.history[words_list[loop_var]] = freq_list[loop_var]
def store_history(self):
words_file = open(search_history, 'w')
freq_file = open(search_freq, 'w')
words_list = self.history.keys()
freq_list = self.history.values()
words_file.write(json.dumps(words_list))
freq_file.write(json.dumps(freq_list))
words_file.close()
freq_file.close()
def reset(cls):
cls._shared_state.clear()
cls._arg_trie = SortedStringTrie()
cls._registries.clear()
class _Repository:
"""Copied from https://stackoverflow.com/questions/6255050/python-thinking-of-a-module-and-its-variables-as-a-singleton-clean-approach."""
_shared_state = dict()
_arg_trie = SortedStringTrie()
_registries = dict()
@classmethod
def reset(cls):
cls._shared_state.clear()
cls._arg_trie = SortedStringTrie()
cls._registries.clear()
def __init__(self):
self.__dict__ = self._shared_state
def add_argument(self,
name,
*aliases,
scope=None,
dtype=str,
default=None,
nargs=1,
msg='',
choices=None):
if scope is None:
raise ArgumentScopeNotSupplied(
'You have to explicitly set scope to a value.')
arg = Argument(name,
*aliases,
scope=scope,
dtype=dtype,
default=default,
nargs=nargs,
msg=msg,
choices=choices)
if arg.name in self.__dict__:
raise DuplicateArgument(
f'An argument named "{arg.name}" has been declared.')
if arg.name in SUPPORTED_VIEW_ATTRS:
raise ReservedNameError(
f'Name "{arg.name}" is reserved for something else.')
self.__dict__[arg.name] = arg
self._arg_trie[
arg.
name] = arg # NOTE(j_luo) This is class attribute, therefore not part of __dict__.
if dtype == bool:
self._arg_trie[f'no_{arg.name}'] = arg
return arg
def set_argument(self, name, value, *, _force=False):
if not _force:
raise MustForceSetArgument(
f'You must explicitliy set _force to True in order to set an argument.'
)
arg = self._get_argument_by_string(name)
arg.value = value
def add_registry(self, registry, scope):
try:
arg = self.add_argument(registry.name, scope=scope, dtype=str)
except DuplicateArgument:
raise DuplicateRegistry(
f'A registry named "{registry.name}" already exists.')
self._registries[arg.name] = registry
@property
def configs(self) -> Dict[str, str]:
ret = dict()
for name, registry in self._registries.items():
arg = self._get_argument_by_string(name)
ret[name] = arg.value
return ret
def get_view(self):
return _RepositoryView(self._shared_state)
def _get_argument_by_string(self, name, source='CLI'):
args = self._arg_trie.values(prefix=name)
if len(args) > 1:
found_names = [f'"{arg.name}"' for arg in args]
raise MultipleMatches(
f'Found more than one match for name "{name}": {", ".join(found_names)}.'
)
elif len(args) == 0:
raise MatchNotFound(
f'Found no argument named "{name}" from "{source}".')
arg = args[0]
return arg
def parse_args(self, known_only=False):
arg_groups = list()
group = list()
for seg in sys.argv[1:]:
if seg.startswith('-'):
if group:
arg_groups.append(group)
group = seg.split('=')
else:
group.append(seg)
if group:
arg_groups.append(group)
# Parse the CLI string first.
parsed = list()
for group in arg_groups:
name, *values = group
name = name.strip('-')
# NOTE(j_luo) Help mode. Note that if known_only is True, then help is ignored.
if name == 'h' or name == 'help':
if not known_only:
_print_all_args(log_also=False, and_exit=True)
continue # NOTE(j_luo) Some other args might start with "h"!
try:
arg = self._get_argument_by_string(name, source='CLI')
if arg.dtype == bool:
new_value = [not name.startswith('no_')] + values
else:
new_value = values
parsed.append((arg, new_value))
except MatchNotFound as e:
if not known_only:
raise e
# Deal with config files and use their values to set the new default values.
cfg_names = set()
for arg, new_value in parsed:
if arg.name in self._registries:
arg.value = new_value
reg = self._registries[arg.name]
cfg_cls = reg[arg.value]
cfg = vars(cfg_cls())
for cfg_name, cfg_value in cfg.items():
cfg_arg = self._get_argument_by_string(
cfg_name, source=cfg_cls.__name__)
cfg_arg.value = cfg_value
if cfg_name in cfg_names:
raise OverlappingRegistries(
f'Argument named "{cfg_name}" has been found in multiple registries.'
)
cfg_names.add(cfg_name)
# Set the remaning CLI arguments.
for arg, new_value in parsed:
if arg.name not in self._registries:
arg.value = new_value
return g
