writing data

This commit is contained in:
2023-04-12 01:00:18 +03:00
parent e25ad40379
commit e31ca85485
3 changed files with 2624 additions and 6 deletions

View File

@@ -10,7 +10,7 @@
},
{
"cell_type": "code",
"execution_count": 8,
"execution_count": 16,
"metadata": {},
"outputs": [],
"source": [
@@ -18,12 +18,18 @@
"import scrapping\n",
"import psycopg2\n",
"from psycopg2.extras import execute_batch\n",
"# from psycopg2.extensions import register_adapter"
"import re\n",
"import logging\n",
"import datetime\n",
"logging.basicConfig(\n",
" level=logging.INFO, filename=f'{datetime.datetime.now().strftime(\"%Y-%m-%d_%H-%M-%S\")}.log', filemode='w',\n",
" format='%(name)s - %(levelname)s - %(message)s', datefmt='%d-%b-%y %H:%M:%S')\n",
"# from psycopg2.extensions import register_adapter\n"
]
},
{
"cell_type": "code",
"execution_count": 12,
"execution_count": 17,
"metadata": {},
"outputs": [],
"source": [
@@ -32,9 +38,14 @@
"with open('project_list') as f:\n",
" for project in f.read().split('\\n'):\n",
" scrap = scrapping.get_raw_english_texts_of_project(project)\n",
" connection = connection.execute(\"insert into raw_texts values (%(id_text)s, %(project_name)s, %(raw_text)s)\", scrap)\n",
" execute_batch(connection, \"insert into raw_texts values (%(id_text)s, %(project_name)s, %(raw_text)s)\", scrap)\n",
" conn.commit()"
" # connection = connection.execute(\"insert into raw_texts values (%(id_text)s, %(project_name)s, %(raw_text)s)\", scrap)\n",
" try:\n",
" execute_batch(connection, \"insert into raw_texts values (%(id_text)s, %(project_name)s, %(raw_text)s)\", scrap)\n",
" conn.commit()\n",
" except Exception:\n",
" with open(project, 'w') as f:\n",
" f.write(json.dumps(scrap))\n",
" logging.error(f\"Error in {project}:{Exception}\")"
]
},
{