1010
1111from utils import read_gitignore , is_ignored , find_projects
1212
13+ IGNORED_DIRS = {'.git' , '__pycache__' , '.idea' , '.vscode' , '.venv' , '.eggs' }
14+
1315
1416def analyze_project_structure (directory , task_queue ):
1517 ignored_paths = read_gitignore (directory )
@@ -50,14 +52,30 @@ def analyze_project_structure(directory, task_queue):
5052
5153
5254
53- def parse_python_files (projects_dir ):
55+ def parse_python_files (projects_dir , export = True , max_files = 5000 , max_depth = 6 ):
56+ import os , ast , datetime
57+ import pandas as pd
58+
59+ IGNORED_DIRS = {'.git' , '__pycache__' , '.idea' , '.vscode' , 'venv' , '.venv' , 'env' , '.env' , '.mypy_cache' }
60+
5461 project_stats = {}
62+ scanned_files = 0
63+
64+ for root , dirs , files in os .walk (projects_dir ):
65+ # Удаление игнорируемых директорий
66+ dirs [:] = [d for d in dirs if d not in IGNORED_DIRS ]
67+
68+ # Ограничение глубины
69+ rel_root = os .path .relpath (root , projects_dir )
70+ depth = rel_root .count (os .sep )
71+ if depth > max_depth :
72+ continue
5573
56- # Используем find_projects для поиска папок с проектами
57- projects = find_projects (projects_dir )
74+ py_files = [f for f in files if f .endswith (".py" )]
75+ if not py_files :
76+ continue
5877
59- for project_dir in projects :
60- project_name = os .path .relpath (project_dir , projects_dir ).split (os .sep )[0 ]
78+ project_name = rel_root .replace (os .sep , " / " ) if rel_root != "." else "ROOT"
6179
6280 if project_name not in project_stats :
6381 project_stats [project_name ] = {
@@ -67,98 +85,151 @@ def parse_python_files(projects_dir):
6785 "dirs" : set ()
6886 }
6987
70- # Склонение директорий с проектами
71- for root , dirs , files in os .walk (project_dir ):
72- for file in files :
73- if file .endswith (".py" ):
74- file_path = os .path .join (root , file )
75-
76- # Обновляем счётчик Python файлов
77- project_stats [project_name ]["py_count" ] += 1
78-
79- # Обновляем дату создания проекта
80- creation_time = os .path .getctime (file_path )
81- creation_date = datetime .datetime .fromtimestamp (creation_time )
82-
83- current_created = project_stats [project_name ]["created" ]
84- if current_created is None or creation_date < current_created :
85- project_stats [project_name ]["created" ] = creation_date
86-
87- # Сбор директорий
88- rel_dir = os .path .relpath (root , os .path .join (projects_dir , project_name ))
89- if rel_dir != "." :
90- project_stats [project_name ]["dirs" ].add (rel_dir )
91-
92- # Парсим файл для библиотек
93- try :
94- with open (file_path , "r" , encoding = "utf-8" ) as f :
95- node = ast .parse (f .read (), filename = file_path )
96-
97- for sub_node in ast .walk (node ):
98- if isinstance (sub_node , ast .Import ):
99- for alias in sub_node .names :
100- project_stats [project_name ]["libs" ].add (alias .name .split ('.' )[0 ])
101- elif isinstance (sub_node , ast .ImportFrom ):
102- if sub_node .module :
103- project_stats [project_name ]["libs" ].add (sub_node .module .split ('.' )[0 ])
104- except Exception :
105- continue
106-
107- # Приведение к нужному формату
108- for proj in project_stats :
109- project_stats [proj ]["libs" ] = sorted (project_stats [proj ]["libs" ])
110- project_stats [proj ]["dirs" ] = sorted (project_stats [proj ]["dirs" ])
111- if project_stats [proj ]["created" ]:
112- project_stats [proj ]["created" ] = project_stats [proj ]["created" ].strftime ("%Y-%m-%d %H:%M:%S" )
113-
114- return project_stats
88+ for file in py_files :
89+ """if scanned_files >= max_files:
90+ print(f"⚠ Превышен лимит {max_files} файлов. Анализ остановлен.")
91+ break"""
92+
93+ file_path = os .path .join (root , file )
94+ scanned_files += 1
95+ project_stats [project_name ]["py_count" ] += 1
96+
97+ # Обработка даты
98+ try :
99+ creation_time = os .path .getctime (file_path )
100+ creation_date = datetime .datetime .fromtimestamp (creation_time )
101+ current_created = project_stats [project_name ]["created" ]
102+ if current_created is None or creation_date < current_created :
103+ project_stats [project_name ]["created" ] = creation_date
104+ except Exception :
105+ pass
106+
107+ # Добавление относительной директории
108+ rel_dir = os .path .relpath (root , projects_dir )
109+ if rel_dir != "." :
110+ project_stats [project_name ]["dirs" ].add (rel_dir )
111+
112+ # Парсинг импортов
113+ try :
114+ with open (file_path , "r" , encoding = "utf-8" ) as f :
115+ content = f .read ()
116+ node = ast .parse (content , filename = file_path )
117+ for sub_node in ast .walk (node ):
118+ if isinstance (sub_node , ast .Import ):
119+ for alias in sub_node .names :
120+ project_stats [project_name ]["libs" ].add (alias .name .split ('.' )[0 ])
121+ elif isinstance (sub_node , ast .ImportFrom ) and sub_node .module :
122+ project_stats [project_name ]["libs" ].add (sub_node .module .split ('.' )[0 ])
123+ except Exception :
124+ continue
125+
126+ print (f"[✓] { project_name } { len (py_files )} )
127+
128+ # Финальная сборка
129+ result = []
130+ for proj , data in project_stats .items ():
131+ date_str = data ["created" ].strftime ("%Y-%m-%d %H:%M:%S" ) if data ["created" ] else None
132+ result .append ({
133+ "name" : proj ,
134+ "stack" : sorted (data ["libs" ]),
135+ "dirs" : sorted (data ["dirs" ]),
136+ "date" : date_str ,
137+ "py_count" : data ["py_count" ]
138+ })
139+
140+ if not result :
141+ print ("⚠ Не найдено проектов с .py файлами." )
142+ return []
143+
144+ df = pd .DataFrame (result )
145+ df ["date" ] = pd .to_datetime (df ["date" ], errors = "coerce" )
146+
147+ if export :
148+ df .to_csv ("project_stats.csv" , index = False , encoding = "utf-8-sig" )
149+ df .to_html ("project_stats.html" , index = False )
150+
151+ print ("==== Итог ====" )
152+ print (df [["name" , "date" ]])
153+ return df .to_dict ("records" )
154+
155+
156+
115157
116158
117159def open_stats_window (root , project_data : list [dict ]):
118160 stats_win = tk .Toplevel (root )
119- stats_win .title ("Статистика проектов " )
120- stats_win .geometry ("800x600 " )
161+ stats_win .title ("📊 Статистика по проектам " )
162+ stats_win .geometry ("1000x700 " )
121163
122- # Проверка наличия данных в project_data
123164 if not project_data :
124- tk .Label (stats_win , text = "Нет данных для отображения статистики." ).pack (pady = 20 )
165+ tk .Label (stats_win , text = "❌ Нет данных для отображения." , font = ( "Arial" , 12 ) ).pack (pady = 20 )
125166 return
126167
127168 df = pd .DataFrame (project_data )
128169
129- # Если данных в столбце 'date' нет, выводим сообщение
130170 if df .empty or 'date' not in df .columns :
131- tk .Label (stats_win , text = "Недостаточно данных для отображения статистики." ).pack (pady = 20 )
171+ tk .Label (stats_win , text = "❌ Недостаточно данных для статистики." , font = ( "Arial" , 12 ) ).pack (pady = 20 )
132172 return
133173
134- # Конвертируем 'date' в формат datetime для графиков
135174 df ['date' ] = pd .to_datetime (df ['date' ], errors = 'coerce' )
136-
137- # Убираем записи с некорректными датами
138175 df = df .dropna (subset = ['date' ])
139176
140- stats_canvas = tk .Frame (stats_win )
141- stats_canvas .pack (fill = "both" , expand = True )
177+ if df .empty :
178+ tk .Label (stats_win , text = "❌ Все даты повреждены или отсутствуют." , font = ("Arial" , 12 )).pack (pady = 20 )
179+ return
180+
181+ df ['month' ] = df ['date' ].dt .to_period ('M' )
182+
183+ # === Верхняя сводка ===
184+ summary_frame = tk .Frame (stats_win )
185+ summary_frame .pack (pady = 10 )
186+
187+ total_projects = len (df )
188+ total_files = df ['py_count' ].sum ()
189+ earliest = df ['date' ].min ().strftime ("%Y-%m-%d" )
190+ latest = df ['date' ].max ().strftime ("%Y-%m-%d" )
191+ unique_months = df ['month' ].nunique ()
142192
143- # === График по дате создания ===
144- fig , ax = plt .subplots (figsize = (6 , 4 ))
145- df_by_month = df .groupby (df ['date' ].dt .to_period ('M' )).size ().sort_index ()
146- df_by_month .plot (kind = 'bar' , ax = ax , title = 'Проекты по месяцам' , rot = 45 )
193+ summary_text = (
194+ f"📦 Всего проектов: { total_projects } { total_files } \n "
195+ f"📆 Период: { earliest } { latest } { unique_months }
196+ )
197+ tk .Label (summary_frame , text = summary_text , font = ("Arial" , 11 ), justify = "left" ).pack ()
198+
199+ # === График по месяцам ===
200+ chart_frame = tk .Frame (stats_win )
201+ chart_frame .pack (fill = "both" , expand = False , pady = 5 )
202+
203+ fig , ax = plt .subplots (figsize = (8 , 4 ))
204+ df_by_month = df .groupby ('month' ).size ().sort_index ()
205+ df_by_month .plot (kind = 'bar' , ax = ax , color = 'skyblue' , edgecolor = 'black' )
206+
207+ ax .set_title ('📅 Количество проектов по месяцам' , fontsize = 14 )
208+ ax .set_ylabel ('Проекты' , fontsize = 12 )
209+ ax .set_xlabel ('Месяц' , fontsize = 12 )
210+ ax .grid (True , axis = 'y' , linestyle = '--' , alpha = 0.5 )
147211 fig .tight_layout ()
148212
149- canvas = FigureCanvasTkAgg (fig , master = stats_canvas )
213+ canvas = FigureCanvasTkAgg (fig , master = chart_frame )
150214 canvas .draw ()
151- canvas .get_tk_widget ().pack (pady = 10 )
215+ canvas .get_tk_widget ().pack ()
152216
153217 # === Анализ технологий ===
218+ tech_frame = tk .LabelFrame (stats_win , text = "📚 Используемые библиотеки" , font = ("Arial" , 12 ))
219+ tech_frame .pack (fill = "both" , expand = True , padx = 10 , pady = 10 )
220+
154221 if 'stack' in df .columns :
155222 all_stacks = sum (df ['stack' ].tolist (), [])
156223 stack_series = pd .Series (all_stacks ).value_counts ()
157- stack_text = "\n " .join (f"{ lang } { count } for lang , count in stack_series .items ())
224+ if not stack_series .empty :
225+ stack_text = "\n " .join (f"• { lib :<20} { count } for lib , count in stack_series .items ())
226+ else :
227+ stack_text = "Нет данных о библиотеках."
158228 else :
159229 stack_text = "Нет данных по стеку технологий."
160230
161- tk .Label ( stats_canvas , text = "Анализ технологий: \n " + stack_text , justify = "left "font = ( "Arial" , 12 )) .pack (pady = 5 )
231+ tk .Message ( tech_frame , text = stack_text , font = ( "Courier New" , 10 ), width = 900 , anchor = "w "justify = "left" ) .pack ()
162232
163233 # === Кнопка закрытия ===
164- tk .Button (stats_canvas , text = "Закрыть" , command = stats_win .destroy ).pack (pady = 10 )
234+ tk .Button (stats_win , text = "Закрыть" , command = stats_win .destroy ).pack (pady = 10 )
235+
0 commit comments