Release_2.0 next

General

• Move constants to file constants.py
• Set default config file in scripts.
• Update description of scripts in main.
• Update README.md

Input config file

• Change config root to dict and update parse_config().
• Change some config parameters name (galaxy_persist_data, tripal_banner_path, jbrowse_menu_url), update example config file and compose templates.
• Remove apollo config variables (webapollo_user, webapollo_password) in config file and in gspecies compose template, because apollo service not deployed here.

Utilities.py

• Factorize parse_input() et parse_config().
• Get input yml keys from constants (update organisms.yml.j2 and add create_org_param_dict_from_constants())

gspecies_compose.yml

• set ENABLE_WIKI and ENABLE_APOLLO to 0 in gspecies compose template
• fix ENABLE_BLAST value in gspecies compose template

gga_get_data.py

• Remove useless variable datasets_to_get

gga_init.py

• replace subprocess.call() with subprocess.run(). Factorize run_command() and run_docker_stack_deploy()
• fix banner copy

utilities.py

@@ -7,7 +7,25 @@ import sys
 import os
 import subprocess
 import bioblend
+import constants
 
+def load_yaml(yaml_file):
+
+    try:
+        with open(yaml_file, 'r') as stream:
+            try:
+                data = yaml.safe_load(stream)
+            except yaml.YAMLError as err:
+                logging.critical("Input file %s is not in YAML format" % yaml_file)
+                sys.exit(err)
+    except FileNotFoundError:
+        logging.critical("Input file doesn't exist (%s)" % yaml_file)
+        sys.exit()
+    except OSError:
+        logging.critical("Input file cannot be read (%s)" % yaml_file)
+        sys.exit()
+
+    return data
 
 def parse_config(config_file):
     """
@@ -17,25 +35,14 @@ def parse_config(config_file):
     :return:
     """
 
-    config_variables = {}
-    logging.debug("Using config: %s" % os.path.abspath(config_file))
-    try:
-        with open(config_file, 'r') as stream:
-            yaml_dict = yaml.safe_load(stream)
-            for k, v in yaml_dict.items():
-                for k2, v2 in v.items():
-                    config_variables[k2] = v2  # Add a key:value pair to variables for replacement in the compose template file
-
-    except FileNotFoundError:
-        logging.critical("The config file specified doesn't exist (%s)" % config_file)
-        sys.exit()
-    except OSError:
-        logging.critical("The config file specified cannot be read (%s)" % config_file)
+    config_dict = load_yaml(config_file)
+    if isinstance(config_dict, dict):
+        #logging.debug("Config dictionary: {0}".format(config_dict))
+        return config_dict
+    else:
+        logging.critical("Config yaml file is not a dictionary" % config_file)
         sys.exit()
 
-    return config_variables
-
-
 def parse_input(input_file):
     """
     Parse the yml input file to extract data to create the SpeciesData objects
@@ -45,25 +52,13 @@ def parse_input(input_file):
     :return:
     """
 
-    sp_dict_list = []
-
-    try:
-        with open(input_file, 'r') as stream:
-            try:
-                sp_dict_list = yaml.safe_load(stream)
-            except yaml.YAMLError as err:
-                logging.critical("Input file is not in YAML format")
-                sys.exit(err)
-    except FileNotFoundError:
-        logging.critical("The specified input file doesn't exist (%s)" % input_file)
-        sys.exit()
-    except OSError:
-        logging.critical("The specified input file cannot be read (%s)" % input_file)
+    sp_dict_list = load_yaml(input_file)
+    if isinstance(sp_dict_list, list):
+        return sp_dict_list
+    else:
+        logging.critical("Input organisms yaml file is not a list" % input_file)
         sys.exit()
 
-    return sp_dict_list
-
-
 def filter_empty_not_empty_items(li):
     """
     Separate a list between empty items and non empty items.
@@ -124,8 +119,8 @@ def get_species_history_id(instance, full_name):
 
 def get_gspecies_string_from_sp_dict(sp_dict):
 
-    genus = sp_dict["description"]["genus"]
-    species = sp_dict["description"]["species"]
+    genus = sp_dict[constants.ORG_PARAM_DESC][constants.ORG_PARAM_DESC_GENUS]
+    species = sp_dict[constants.ORG_PARAM_DESC][constants.ORG_PARAM_DESC_SPECIES]
     gspecies = genus.lower() + "_" + species.lower()
     return gspecies
 
@@ -164,7 +159,8 @@ def get_unique_species_dict_list(sp_dict_list):
 
     for sp in sp_dict_list:
         gspecies = get_gspecies_string_from_sp_dict(sp)
-        if gspecies not in unique_species_dict.keys() or sp["description"]["main_species"] == True :
+        if gspecies not in unique_species_dict.keys() or ( constants.ORG_PARAM_DESC_MAIN_SPECIES in sp[constants.ORG_PARAM_DESC].keys() and
+                                                           sp[constants.ORG_PARAM_DESC][constants.ORG_PARAM_DESC_MAIN_SPECIES] == True ) :
             unique_species_dict[gspecies] = sp
         else:
             continue
@@ -193,4 +189,37 @@ def run_tool_and_get_single_output_dataset_id(instance, tool_id, history_id, too
     output_dict = run_tool(instance, tool_id, history_id, tool_inputs)
     single_output_dataset_id = output_dict["outputs"][0]["id"]
 
-    return single_output_dataset_id
\ No newline at end of file
+    return single_output_dataset_id
+
+def create_org_param_dict_from_constants():
+    """
+    Create a dictionary of variables containing the keys needed to render the organisms.yml.j2 (NOT the values)
+    Created from the constants
+    """
+
+    org_param_dict={}
+    org_param_dict["org_param_name"] = constants.ORG_PARAM_NAME
+    org_param_dict["org_param_desc"] = constants.ORG_PARAM_DESC
+    org_param_dict["org_param_desc_genus"] = constants.ORG_PARAM_DESC_GENUS
+    org_param_dict["org_param_desc_species"] = constants.ORG_PARAM_DESC_SPECIES
+    org_param_dict["org_param_desc_sex"] = constants.ORG_PARAM_DESC_SEX
+    org_param_dict["org_param_desc_strain"] = constants.ORG_PARAM_DESC_STRAIN
+    org_param_dict["org_param_desc_common_name"] = constants.ORG_PARAM_DESC_COMMON_NAME
+    org_param_dict["org_param_desc_origin"] = constants.ORG_PARAM_DESC_ORIGIN
+    org_param_dict["org_param_desc_main_species"] = constants.ORG_PARAM_DESC_MAIN_SPECIES
+    org_param_dict["org_param_data"] = constants.ORG_PARAM_DATA
+    org_param_dict["org_param_data_genome_path"] = constants.ORG_PARAM_DATA_GENOME_PATH
+    org_param_dict["org_param_data_transcripts_path"] = constants.ORG_PARAM_DATA_TRANSCRIPTS_PATH
+    org_param_dict["org_param_data_proteins_path"] = constants.ORG_PARAM_DATA_PROTEINS_PATH
+    org_param_dict["org_param_data_gff_path"] = constants.ORG_PARAM_DATA_GFF_PATH
+    org_param_dict["org_param_data_interpro_path"] = constants.ORG_PARAM_DATA_INTERPRO_PATH
+    org_param_dict["org_param_data_orthofinder_path"] = constants.ORG_PARAM_DATA_ORTHOFINDER_PATH
+    org_param_dict["org_param_data_blastp_path"] = constants.ORG_PARAM_DATA_BLASTP_PATH
+    org_param_dict["org_param_data_blastx_path"] = constants.ORG_PARAM_DATA_BLASTX_PATH
+    org_param_dict["org_param_data_genome_version"] = constants.ORG_PARAM_DATA_GENOME_VERSION
+    org_param_dict["org_param_data_ogs_version"] = constants.ORG_PARAM_DATA_OGS_VERSION
+    org_param_dict["org_param_data_performed_by"] = constants.ORG_PARAM_DATA_PERFORMED_BY
+    org_param_dict["org_param_services"] = constants.ORG_PARAM_SERVICES
+    org_param_dict["org_param_services_blast"] = constants.ORG_PARAM_SERVICES_BLAST
+
+    return org_param_dict
\ No newline at end of file