Compare commits

..

2 Commits

Author SHA1 Message Date
a813a3805e добавил закрытие бд для пресечения утечки данных
All checks were successful
continuous-integration/drone/push Build is passing
2026-04-11 12:14:01 +10:00
8c790b018c сохраняю перед добавлением во все бд, закрытие соединения 2026-04-11 12:12:28 +10:00
2 changed files with 202 additions and 119 deletions

View File

@@ -459,6 +459,10 @@ class Parserall(BaseModel):
url: HttpUrl
promt: str
@app.post("/add_sources" , summary="Добавление парсинга любого источника")
async def add_sources_all_ist(sources: Parserall):
return wp.add_sources(sources.url, sources.promt)
@app.post("/parser_all" , summary="Запуск процеса парсинга любого источника")
async def process_parser_all_ist(url: Parserall, background_tasks: BackgroundTasks):
background_tasks.add_task(start_pars_all_istochnik(str(url.url), url.promt))

View File

@@ -13,150 +13,228 @@ conn = psycopg2.connect(
)
conn.autocommit = True
def close_connection():
"""Закрывает подключение к БД"""
global conn
if conn:
conn.close()
conn = None
def create_table():
with conn.cursor() as cur:
cur.execute("""
CREATE TABLE IF NOT EXISTS work_parser (
id SERIAL PRIMARY KEY,
status VARCHAR(20) NOT NULL CHECK (status IN ('queued', 'in_progress', 'completed', 'failed')),
created_at TIMESTAMPTZ NOT NULL DEFAULT NOW(),
started_at TIMESTAMPTZ,
finished_at TIMESTAMPTZ,
source_url TEXT,
error_message TEXT,
attempts INTEGER DEFAULT 0,
priority INTEGER DEFAULT 0
);
""")
print("Таблица work_parser создана или уже существует")
try:
with conn.cursor() as cur:
cur.execute("""
CREATE TABLE IF NOT EXISTS work_parser (
id SERIAL PRIMARY KEY,
status VARCHAR(20) NOT NULL CHECK (status IN ('queued', 'in_progress', 'completed', 'failed')),
created_at TIMESTAMPTZ NOT NULL DEFAULT NOW(),
started_at TIMESTAMPTZ,
finished_at TIMESTAMPTZ,
source_url TEXT,
error_message TEXT,
attempts INTEGER DEFAULT 0,
priority INTEGER DEFAULT 0
);
""")
print("Таблица work_parser создана или уже существует")
finally:
if conn:
conn.close()
def insert_task(status, source_url=None, source_id=None, priority=0):
with conn.cursor() as cur:
cur.execute("""
INSERT INTO work_parser (status, source_url, priority)
VALUES (%s, %s, %s)
RETURNING id;
""", (status, source_url, priority))
task_id = cur.fetchone()[0]
return task_id
try:
with conn.cursor() as cur:
cur.execute("""
INSERT INTO work_parser (status, source_url, priority)
VALUES (%s, %s, %s)
RETURNING id;
""", (status, source_url, priority))
task_id = cur.fetchone()[0]
return task_id
finally:
if conn:
conn.close()
def get_tasks_offset(limit, offset):
with conn.cursor(cursor_factory=RealDictCursor) as cur:
cur.execute("""
SELECT * FROM work_parser
ORDER BY created_at DESC
LIMIT %s OFFSET %s
""", (limit, offset))
tasks = cur.fetchall()
return tasks
try:
with conn.cursor(cursor_factory=RealDictCursor) as cur:
cur.execute("""
SELECT * FROM work_parser
ORDER BY created_at DESC
LIMIT %s OFFSET %s
""", (limit, offset))
tasks = cur.fetchall()
return tasks
finally:
if conn:
conn.close()
def delete_task(task_id: int):
with conn.cursor(cursor_factory=RealDictCursor) as cur:
cur.execute("DELETE FROM work_parser WHERE id = %s RETURNING *;", (task_id,))
deleted_task = cur.fetchone()
if deleted_task:
return {"message": f"Задача {task_id} удалена", "deleted_task": dict(deleted_task)}
else:
return {"message": f"Задача с id {task_id} не найдена"}
try:
with conn.cursor(cursor_factory=RealDictCursor) as cur:
cur.execute("DELETE FROM work_parser WHERE id = %s RETURNING *;", (task_id,))
deleted_task = cur.fetchone()
if deleted_task:
return {"message": f"Задача {task_id} удалена", "deleted_task": dict(deleted_task)}
else:
return {"message": f"Задача с id {task_id} не найдена"}
finally:
if conn:
conn.close()
def update_task(task_id, **fields):
# dynamic update query generator
allowed_fields = ['status', 'started_at', 'finished_at', 'source_url', 'error_message', 'attempts', 'priority']
set_parts = []
values = []
for key, value in fields.items():
if key in allowed_fields:
set_parts.append(f"{key} = %s")
values.append(value)
if not set_parts:
return False
values.append(task_id)
set_sql = ", ".join(set_parts)
with conn.cursor() as cur:
cur.execute(f"UPDATE work_parser SET {set_sql} WHERE id = %s;", values)
return True
try:
allowed_fields = ['status', 'started_at', 'finished_at', 'source_url', 'error_message', 'attempts', 'priority']
set_parts = []
values = []
for key, value in fields.items():
if key in allowed_fields:
set_parts.append(f"{key} = %s")
values.append(value)
if not set_parts:
return False
values.append(task_id)
set_sql = ", ".join(set_parts)
with conn.cursor() as cur:
cur.execute(f"UPDATE work_parser SET {set_sql} WHERE id = %s;", values)
return True
finally:
if conn:
conn.close()
def create_table_config_gpt():
with conn.cursor() as cur:
cur.execute("""
CREATE TABLE IF NOT EXISTS config_gpt (
url TEXT PRIMARY KEY,
name VARCHAR(20),
promt TEXT
);
""")
print("Таблица config_gpt создана или уже существует")
try:
with conn.cursor() as cur:
cur.execute("""
CREATE TABLE IF NOT EXISTS config_gpt (
url TEXT PRIMARY KEY,
name VARCHAR(20),
promt TEXT
);
""")
print("Таблица config_gpt создана или уже существует")
finally:
if conn:
conn.close()
def update_promt(url: str, name: str, promt: str):
with conn.cursor() as cur:
cur.execute("""
INSERT INTO config_gpt (url, name, promt)
VALUES (%s, %s, %s)
ON CONFLICT (url) DO UPDATE SET
name = EXCLUDED.name,
promt = EXCLUDED.promt
""", (url, name, promt))
conn.commit()
try:
with conn.cursor() as cur:
cur.execute("""
INSERT INTO config_gpt (url, name, promt)
VALUES (%s, %s, %s)
ON CONFLICT (url) DO UPDATE SET
name = EXCLUDED.name,
promt = EXCLUDED.promt
""", (url, name, promt))
conn.commit()
finally:
if conn:
conn.close()
def get_promt(promt_name_url):
with conn.cursor(cursor_factory=RealDictCursor) as cur:
cur.execute("SELECT promt FROM config_gpt WHERE url = %s", (promt_name_url,))
promt = cur.fetchone()
return promt['promt']
try:
with conn.cursor(cursor_factory=RealDictCursor) as cur:
cur.execute("SELECT promt FROM config_gpt WHERE url = %s", (promt_name_url,))
promt = cur.fetchone()
return promt['promt']
finally:
if conn:
conn.close()
def get_all_promt():
with conn.cursor(cursor_factory=RealDictCursor) as cur:
cur.execute("SELECT * FROM config_gpt")
rows = cur.fetchall()
try:
with conn.cursor(cursor_factory=RealDictCursor) as cur:
cur.execute("SELECT * FROM config_gpt")
rows = cur.fetchall()
sources = [{"url": row["url"], "name": row["name"], "promt": row["promt"]} for row in rows]
return {"sources": sources}
sources = [{"url": row["url"], "name": row["name"], "promt": row["promt"]} for row in rows]
return {"sources": sources}
finally:
if conn:
conn.close()
def get_all_categories_promt():
"""Возвращает список всех значений поля name из таблицы config_gpt"""
with conn.cursor(cursor_factory=RealDictCursor) as cur:
cur.execute("SELECT name FROM config_gpt")
rows = cur.fetchall()
try:
with conn.cursor(cursor_factory=RealDictCursor) as cur:
cur.execute("SELECT name FROM config_gpt")
rows = cur.fetchall()
return [row["name"] for row in rows]
return [row["name"] for row in rows]
finally:
if conn:
conn.close()
def create_table_error_url():
with conn.cursor() as cur:
cur.execute("""
CREATE TABLE IF NOT EXISTS error_url (
id SERIAL PRIMARY KEY,
source_url TEXT NOT NULL,
error_sources_url TEXT NOT NULL
);
""")
print("Таблица error_url создана или уже существует")
try:
with conn.cursor() as cur:
cur.execute("""
CREATE TABLE IF NOT EXISTS error_url (
id SERIAL PRIMARY KEY,
source_url TEXT NOT NULL,
error_sources_url TEXT NOT NULL
);
""")
print("Таблица error_url создана или уже существует")
finally:
if conn:
conn.close()
def add_error_url(source_url: str, error_sources_url: str):
"""Добавляет запись об ошибке URL"""
with conn.cursor() as cur:
cur.execute("""
INSERT INTO error_url (source_url, error_sources_url)
VALUES (%s, %s)
RETURNING id;
""", (source_url, error_sources_url))
return cur.fetchone()[0]
try:
with conn.cursor() as cur:
cur.execute("""
INSERT INTO error_url (source_url, error_sources_url)
VALUES (%s, %s)
RETURNING id;
""", (source_url, error_sources_url))
return cur.fetchone()[0]
finally:
if conn:
conn.close()
def check_error_url(error_sources_url: str) -> bool:
"""Проверяет, есть ли запись в таблице error_url с таким URL"""
with conn.cursor(cursor_factory=RealDictCursor) as cur:
cur.execute("""
SELECT 1 FROM error_url
WHERE error_sources_url = %s
LIMIT 1;
""", (error_sources_url,))
row = cur.fetchone()
try:
with conn.cursor(cursor_factory=RealDictCursor) as cur:
cur.execute("""
SELECT 1 FROM error_url
WHERE error_sources_url = %s
LIMIT 1;
""", (error_sources_url,))
row = cur.fetchone()
return row is None
return row is None
finally:
if conn:
conn.close()
def create_table_add_sourse():
try:
with conn.cursor() as cur:
cur.execute("""
CREATE TABLE IF NOT EXISTS sourse (
url TEXT PRIMARY KEY,
promt TEXT
);
""")
print("Таблица sourse создана или уже существует")
finally:
if conn:
conn.close()
def add_sources(url: str, promt: str):
try:
with conn.cursor() as cur:
cur.execute("""
INSERT INTO config_gpt (url, promt)
VALUES (%s, %s)
ON CONFLICT (url) DO UPDATE SET
promt = EXCLUDED.promt
""", (url, promt))
conn.commit()
finally:
if conn:
conn.close()
# Пример использования
# if __name__ == "__main__":
@@ -168,3 +246,4 @@ def check_error_url(error_sources_url: str) -> bool:
# })
# # print(get_promt("japan"))
# # create_table_error_url()
# create_table_add_sourse()