Mercurial > repos > tduigou > seq_to_db

comparison save_to_db.py @ 2:dad2c4c3450a draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/brsynth/galaxytools/tree/main/tools commit db4ac861e1d03fcdfe94321d858839124e493930-dirty
author tduigou
date Wed, 23 Jul 2025 09:44:33 +0000
parents 3daf04425ea1
children
comparison
equal deleted inserted replaced
1:3daf04425ea1 2:dad2c4c3450a
7 import json 7 import json
8 from sqlalchemy import create_engine, inspect 8 from sqlalchemy import create_engine, inspect
9 from sqlalchemy.engine.url import make_url 9 from sqlalchemy.engine.url import make_url
10 from sqlalchemy.sql import text 10 from sqlalchemy.sql import text
11 from sqlalchemy.exc import OperationalError 11 from sqlalchemy.exc import OperationalError
12
13
14 def resolve_parameters(user_params: dict, json_params: dict, keys: list):
15 resolved = {}
16 for key in keys:
17 # Prefer user parameter if it's provided (not None or empty string)
18 if key in user_params and user_params[key]:
19 resolved[key] = user_params[key]
20 else:
21 resolved[key] = json_params.get(f"JSON_{key}")
22 return resolved
23 12
24 13
25 def fix_db_uri(uri): 14 def fix_db_uri(uri):
26 """Replace __at__ with @ in the URI if needed.""" 15 """Replace __at__ with @ in the URI if needed."""
27 return uri.replace("__at__", "@") 16 return uri.replace("__at__", "@")
191 180
192 181
193 def main(): 182 def main():
194 parser = argparse.ArgumentParser(description="Fetch annotations from PostgreSQL database and save as JSON.") 183 parser = argparse.ArgumentParser(description="Fetch annotations from PostgreSQL database and save as JSON.")
195 parser.add_argument("--input", required=True, help="Input gb files") 184 parser.add_argument("--input", required=True, help="Input gb files")
196 parser.add_argument("--sequence_column", required=True, help="DB column contains sequence for ganbank file") 185 parser.add_argument("--use_json_paramers", required=False, help="Use parameters from JSON: true/false")
197 parser.add_argument("--annotation_column", required=True, help="DB column contains head for ganbank file") 186 parser.add_argument("--sequence_column", required=False, help="DB column contains sequence for GenBank file")
198 parser.add_argument("--db_uri", required=True, help="Database URI connection string") 187 parser.add_argument("--annotation_column", required=False, help="DB column contains head for GenBank file")
199 parser.add_argument("--table", required=True, help="Table name in the database") 188 parser.add_argument("--db_uri", required=False, help="Database URI connection string")
200 parser.add_argument("--fragment_column", required=True, help="Fragment column name in the database") 189 parser.add_argument("--table", required=False, help="Table name in the database")
201 parser.add_argument("--output", required=True, help="Text report") 190 parser.add_argument("--fragment_column", required=False, help="Fragment column name in the database")
202 parser.add_argument("--file_name_mapping", required=True, help="real fragments names") 191 parser.add_argument("--output", required=False, help="Text report")
192 parser.add_argument("--file_name_mapping", required=False, help="Real fragment names")
203 parser.add_argument("--json_conf", required=False, help="JSON config file with DB parameters") 193 parser.add_argument("--json_conf", required=False, help="JSON config file with DB parameters")
204 parser.add_argument("--execution_enable", required=True, help="enabbe or desable execution directly from the tool option") 194 parser.add_argument("--execution_enable", required=False, help="Enable or disable execution")
195 parser.add_argument("--json_generating", required=False, help="Generate JSON: true/false")
196 parser.add_argument("--json_output", required=False, help="Output path for generated JSON")
197
205 args = parser.parse_args() 198 args = parser.parse_args()
206 199
207 # enabbe or desable execution based on galaxy param (execution_enable)
208
209 if args.execution_enable == 'false': 200 if args.execution_enable == 'false':
210 print("Execution disabled. 'Send Requenst to DB' is set to 'false'") 201 print("Execution disabled. 'Send Request to DB' is set to 'false'")
211 return 202 return
212 203
213 # Load JSON config if provided 204 config_params = {}
214 json_config = {} 205 use_json = args.use_json_paramers == 'true'
215 if args.json_conf != 'None' or '': 206 generate_json = args.json_generating == 'true'
207
208 if use_json:
209 if not args.json_conf:
210 raise ValueError("You must provide --json_conf when --use_json_paramers is 'true'")
216 with open(args.json_conf, "r") as f: 211 with open(args.json_conf, "r") as f:
217 json_config = json.load(f) 212 config_params = json.load(f)
218 if "execution" in json_config and json_config["execution"] == "false": 213 if config_params.get("execution", "") == "false":
219 print("Execution was blocked by config (execution = false)") 214 print("Execution blocked by config (execution = false)")
220 return 215 return
221 216 else:
222 # Prefer user input; fallback to JSON_ values if not provided 217 config_params = {
223 user_params = { 218 "table": args.table,
224 "table": args.table, 219 "sequence_column": args.sequence_column,
225 "sequence_column": args.sequence_column, 220 "annotation_column": args.annotation_column,
226 "annotation_column": args.annotation_column, 221 "fragment_column": args.fragment_column,
227 "fragment_column": args.fragment_column, 222 "db_uri": args.db_uri,
228 "db_uri": args.db_uri 223 "execution": args.execution_enable
229 } 224 }
230 225
231 keys = ["table", "sequence_column", "annotation_column", "fragment_column", "db_uri"] 226 if generate_json:
232 resolved = resolve_parameters(user_params, json_config, keys) 227 if not args.json_output:
233 228 raise ValueError("You must provide --json_output when --json_generating is 'true'")
234 # Unpack resolved parameters 229 with open(args.json_output, "w") as f:
235 table = resolved["table"] 230 json.dump(config_params, f, indent=2)
236 sequence_column = resolved["sequence_column"] 231 print(f"JSON configuration written to: {args.json_output}")
237 annotation_column = resolved["annotation_column"] 232
238 fragment_column = resolved["fragment_column"] 233 # Extract final resolved parameters
239 db_uri = fix_db_uri(resolved["db_uri"]) 234 table = config_params["table"]
240 235 sequence_column = config_params["sequence_column"]
241 # Prepare gb files 236 annotation_column = config_params["annotation_column"]
237 fragment_column = config_params["fragment_column"]
238 db_uri = fix_db_uri(config_params["db_uri"])
239
242 gb_file_list = [f.strip() for f in args.input.split(",") if f.strip()] 240 gb_file_list = [f.strip() for f in args.input.split(",") if f.strip()]
243 241
244 # Start and wait for DB 242 # Connect to DB
245 # db_name = extract_db_name(db_uri)
246 # start_postgres_container(db_name)
247 MAX_RETRIES = 3 243 MAX_RETRIES = 3
248 for attempt in range(1, MAX_RETRIES + 1): 244 for attempt in range(1, MAX_RETRIES + 1):
249 try: 245 try:
250 wait_for_db(db_uri) 246 wait_for_db(db_uri)
251 break # Success 247 break
252 except Exception as e: 248 except Exception as e:
253 if attempt == MAX_RETRIES: 249 if attempt == MAX_RETRIES:
254 print(f"Attempt {attempt} failed: Could not connect to database at {db_uri}.") 250 print(f"Attempt {attempt} failed: Could not connect to database at {db_uri}.")
255 raise e 251 raise e
256 else: 252 time.sleep(2)
257 time.sleep(2) 253
258 254 # Push annotations
259 # Push annotations
260 push_gb_annotations( 255 push_gb_annotations(
261 gb_file_list, 256 gb_file_list,
262 sequence_column, 257 sequence_column,
263 annotation_column, 258 annotation_column,
264 db_uri, 259 db_uri,