Merge pull request #491 from oicr-gsi/release-1.7.7

Release 1.7.7

CHANGELOG.md

@@ -1,11 +1,20 @@
 # CHANGELOG
 
+## 1.7.7: 2024-11-11
+- GCGI-1424: Fix rendering of non-Latin characters
+- GCGI-1437: Remove unnecessary checks in OncoKB cache
+- GCGI-981: Create virus (research) plugin
+- GCGI-1460: Make study a required parameter for PWGS
+- GCGI-1459: Update report date for PWGS assay from yyyy/mm/dd to date report was generated
+- GCGI-1458: Change geneticist sign-off date from date report was generated to yyyy-mm-dd
+- GCGI-1465: Fix for overzealous date format check from implementation of GCGI-1458
+
 ## 1.7.6: 2024-10-22
 - GCGI-1156: Merge long-running documentation branch to main
 - GCGI-1452: Update disclaimer with somatic mutation text and laboratory disclaimer
 - GCGI-1453: Update pWGS version number
 - GCGI-1456: Fix rounding of estimated tumour fraction in tar.sample
-- Increase csv field limit in plugins.fusions.tools.py to handle large fields in mavis_summary.txt 
+- Increase csv field limit in plugins.fusions.tools.py to handle large fields in mavis_summary.txt
 
 ## 1.7.5: 2024-10-02
 - GCGI-1378: Improve benchmarking to write an HTML summary page; update tests

src/lib/djerba/core/html_cache.py

@@ -90,7 +90,7 @@ def update_cached_html(self, new_html, old_cache):
         if replace_name:
             msg = "No end tag found for component name '{0}'".format(replace_name)
             self.logger.error(msg)
-            raise DjerbaHatmlCacheError(msg)
+            raise DjerbaHtmlCacheError(msg)
         self.logger.debug("HTML update done")
         return "\n".join(new_lines)
 

src/lib/djerba/core/main.py

@@ -353,7 +353,7 @@ def update_report_data(self, new_data, data, force):
 
     def update_data_from_file(self, new_data, json_path, force):
         """Read old JSON from a file, and return the updated data structure"""
-        with open(json_path) as in_file:
+        with open(json_path, encoding=cc.TEXT_ENCODING) as in_file:
             data = json.loads(in_file.read())
         return self.update_report_data(new_data, data, force)
 
@@ -579,7 +579,7 @@ def update(self, config_path, json_path, out_dir, archive, pdf, summary_only, fo
             config = self.configure_from_parser(config_in)
         else:
             config = self.configure(config_path)
-        with open(json_path) as in_file:
+        with open(json_path, encoding=cc.TEXT_ENCODING) as in_file:
             data = json.loads(in_file.read())
         data_new = self.base_extract(config)
         data = self.update_data_from_file(data_new, json_path, force)

src/lib/djerba/core/workspace.py

@@ -10,6 +10,7 @@
 import json
 import logging
 import os
+import djerba.core.constants as cc
 from djerba.util.logger import logger
 from djerba.util.validator import path_validator
 
@@ -51,15 +52,15 @@ def open_file(self, rel_path, mode='r'):
             self.validator.validate_input_file(file_path)
         else:
             self.validator.validate_output_file(file_path)
-        return open(file_path, mode)
+        return open(file_path, mode, encoding=cc.TEXT_ENCODING)
 
     def print_location(self):
         return self.dir_path
 
     def read_json(self, rel_path):
         in_path = os.path.join(self.dir_path, rel_path)
         self.validator.validate_input_file(in_path)
-        with open(in_path) as in_file:
+        with open(in_path, encoding=cc.TEXT_ENCODING) as in_file:
             data = json.loads(in_file.read())
         return data
 
@@ -82,7 +83,7 @@ def read_maybe_json(self, rel_path):
     def read_string(self, rel_path):
         in_path = os.path.join(self.dir_path, rel_path)
         self.validator.validate_input_file(in_path)
-        with open(in_path) as in_file:
+        with open(in_path, encoding=cc.TEXT_ENCODING) as in_file:
             content = in_file.read()
         return content
 
@@ -92,9 +93,11 @@ def remove_file(self, rel_path):
     # no need to validate paths for write_* methods; output dir already validated as writable
 
     def write_json(self, rel_path, data):
-        with open(os.path.join(self.dir_path, rel_path), 'w') as out_file:
+        out_path = os.path.join(self.dir_path, rel_path)
+        with open(out_path, 'w', encoding=cc.TEXT_ENCODING) as out_file:
             out_file.write(json.dumps(data))
 
     def write_string(self, rel_path, output_string):
-        with open(os.path.join(self.dir_path, rel_path), 'w') as out_file:
+        out_path = os.path.join(self.dir_path, rel_path)
+        with open(out_path, 'w', encoding=cc.TEXT_ENCODING) as out_file:
             out_file.write(output_string)

src/lib/djerba/plugins/pwgs/analysis/plugin.py

@@ -67,7 +67,7 @@ def extract(self, config):
                 json_data = json.load(json_file)
                 assay = json_data.get("results", {}).get("assay", "Assay name not found")
                 primary_cancer = json_data.get("results", {}).get("primary_cancer", "Primary cancer not found")
-                study_title = json_data.get("results", {}).get("study_title", "Study title not found")
+                study_title = json_data.get("results", {}).get("study", "Study title not found")
         else:
             assay = "Assay name not found"
             primary_cancer = "Primary cancer not found"

src/lib/djerba/plugins/pwgs/case_overview/case_template.html

@@ -1,6 +1,7 @@
 <%
 from djerba.util.html import html_builder
 import djerba.plugins.pwgs.constants as constants
+from djerba.util.date import get_todays_date
 %>
 
 ${html_builder.section_cells_begin("Case Overview", True)}
@@ -37,11 +38,11 @@
       <td>Plasma Sample ID:</td><td>${results.get(constants.GROUP_ID)}</td>
     </tr>
     <tr>
-      <td>Date of Report:</td> <td>yyyy/mm/dd</td>
+      <td>Date of Report:</td> <td>${get_todays_date()}</td>
       <td>pWGS Report ID:</td><td>${results.get(constants.PWGS_REPORT)}</td>
     </tr>
 
 
   </table>
 
-${html_builder.section_cells_end()}
+${html_builder.section_cells_end()}

src/lib/djerba/plugins/pwgs/case_overview/plugin.py

@@ -30,9 +30,6 @@ def configure(self, config):
                 wrapper.set_my_param(pc.GROUP_ID, sample_info[core_constants.TUMOUR_ID])
             if wrapper.my_param_is_null(pc.PATIENT_ID_LOWER):
                 wrapper.set_my_param(pc.PATIENT_ID_LOWER, sample_info[pc.PATIENT_ID_LOWER])
-            if wrapper.my_param_is_null(pc.STUDY):
-                # if study id is unspecifided, default to project id
-                wrapper.set_my_param(pc.STUDY, sample_info[pc.PROJECT])
         else:
             msg = 'sample info file not found, make sure case overview parameters are in INI'
             self.logger.warning(msg)
@@ -66,15 +63,15 @@ def specify_params(self):
         required = [
             pc.REQ_APPROVED,
             pc.PRIMARY_CANCER,
-            pc.WGS_REPORT
+            pc.WGS_REPORT, 
+            pc.STUDY
         ]
         for key in required:
             self.add_ini_required(key)
         discovered = [
             pc.DONOR,
             pc.GROUP_ID,
             pc.PATIENT_ID_LOWER,
-            pc.STUDY
         ]
         for key in discovered:
             self.add_ini_discovered(key)

src/lib/djerba/plugins/pwgs/case_overview/test/plugin_test.py

@@ -40,7 +40,7 @@ def testPwgsCase(self):
         params = {
             self.INI: self.INI_NAME,
             self.JSON: json_location,
-            self.MD5: '9bf12ff6b6f8af8382480cb870f3f666'
+            self.MD5: '3d7ff8107b879dd01db629482971969d'
         }
         self.run_basic_test(input_dir, params)
 

src/lib/djerba/plugins/pwgs/case_overview/test/pwgs.case.ini

@@ -1,10 +1,10 @@
 [core]
 
 [pwgs.case_overview]
-requisition_approved = 2023/01/01
-study_title = Plasma Val
+requisition_approved = 2023-01-01
+study = Plasma Val
 primary_cancer = Uterine Clear Cell Carcinoma
 donor = OCT_011328
 group_id = OCT_010384_Ct_T_nn_1-11_LB01
 patient_study_id= OCT-01-1328
-wgs_report_id=OCT-01-1328_Ut_P-v2
+wgs_report_id=OCT-01-1328_Ut_P-v2

src/lib/djerba/plugins/pwgs/constants.py

@@ -32,7 +32,7 @@
 PWGS_REPORT = 'pwgs_report_id'
 REQ_APPROVED = 'requisition_approved'
 REQ_ID = 'requisition_id'
-STUDY = 'study_title'
+STUDY = 'study'
 WGS_REPORT = 'wgs_report_id'
 
 # file constants
@@ -58,4 +58,4 @@
 ANALYSIS_TEMPLATE_NAME = 'analysis_template.html'
 SAMPLE_TEMPLATE_NAME = 'sample_template.html'
 SUMMARY_TEMPLATE_NAME = 'summary_template.html'
-CASE_OVERVIEW_TEMPLATE_NAME = 'case_template.html'
+CASE_OVERVIEW_TEMPLATE_NAME = 'case_template.html'

src/lib/djerba/plugins/summary/plugin.py

@@ -32,7 +32,7 @@ def configure(self, config):
     def extract(self, config):
         wrapper = self.get_config_wrapper(config)
         summary_path = wrapper.get_my_string(self.SUMMARY_FILE)
-        with open(summary_path) as in_file:
+        with open(summary_path, encoding=core_constants.TEXT_ENCODING) as in_file:
             summary_text = in_file.read()
         self.logger.debug('Read summary from {0}: "{1}"'.format(summary_path, summary_text))
         data = self.get_starting_plugin_data(wrapper, self.PLUGIN_VERSION)

src/lib/djerba/plugins/summary/test/custom_summary.txt

@@ -0,0 +1 @@
+Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. \u0391\u0392\u0393\u0394\u03B1\u03B2\u03B3\u03B4

src/lib/djerba/plugins/summary/test/plugin_test.py

@@ -5,6 +5,7 @@
 import os
 import unittest
 import tempfile
+from configparser import ConfigParser
 
 from djerba.util.validator import path_validator
 from djerba.plugins.plugin_tester import PluginTester
@@ -20,13 +21,32 @@ def setUp(self):
 
     def testSummary(self):
         test_source_dir = os.path.realpath(os.path.dirname(__file__))
-        json_location = os.path.join(self.data_dir_root ,"plugins/summary/report_json/summary.json")
+        json_location = os.path.join(self.data_dir_root, "plugins", "summary", "report_json", "summary.json")
         params = {
             self.INI: 'summary.ini',
             self.JSON: json_location,
             self.MD5: '155e22cc02a45e04dc9058112354367c'
         }
         self.run_basic_test(test_source_dir, params)
 
+    def testSummaryWithCustomText(self):
+        test_source_dir = os.path.realpath(os.path.dirname(__file__))
+        summary_path = os.path.join(test_source_dir, 'custom_summary.txt')
+        config = ConfigParser()
+        config.add_section('core')
+        config.add_section('summary')
+        config.set('summary', 'summary_file', summary_path)
+        ini_path = os.path.join(self.tmp_dir, 'custom_summary.ini')
+        with open(ini_path, 'w') as ini_file:
+            config.write(ini_file)
+        json_location = os.path.join(self.data_dir_root, "plugins", "summary", "report_json", "custom_summary.json")
+        params = {
+            self.INI: ini_path,
+            self.JSON: json_location,
+            self.MD5: 'cebbb53b9b074131e309dca71704a896'
+        }
+        self.run_basic_test(test_source_dir, params)
+
+
 if __name__ == '__main__':
     unittest.main()

src/lib/djerba/plugins/supplement/body/plugin.py

@@ -23,6 +23,7 @@ class main(plugin_base):
     INCLUDE_SIGNOFFS = "include_signoffs"
     GENETICIST_DEFAULT = 'PLACEHOLDER'
     GENETICIST_ID_DEFAULT = 'XXXXXXX'
+    REPORT_SIGNOUT_DEFAULT = 'yyyy-mm-dd'
 
     def check_assay_name(self, wrapper):
         [ok, msg] = assays.name_status(wrapper.get_my_string(self.ASSAY))
@@ -54,13 +55,15 @@ def configure(self, config):
         # Check if dates are valid
         user_supplied_date = wrapper.get_my_string(self.USER_SUPPLIED_DRAFT_DATE)
         report_signoff_date = wrapper.get_my_string(self.REPORT_SIGNOFF_DATE)
-        for date in [user_supplied_date, report_signoff_date]:
+        dates_to_check = [user_supplied_date, ]
+        if report_signoff_date != self.REPORT_SIGNOUT_DEFAULT:
+            dates_to_check.append(report_signoff_date)
+        for date in dates_to_check:
             if not is_valid_date(date):
                 msg = "Invalid requisition approved date '{0}': ".format(date)+\
                       "Must be in yyyy-mm-dd format"
                 self.logger.error(msg)
                 raise ValueError(msg)
-
         return wrapper.get_config()
 
     def extract(self, config):
@@ -113,7 +116,7 @@ def specify_params(self):
         ]
         for key in discovered:
             self.add_ini_discovered(key)
-        self.set_ini_default(self.REPORT_SIGNOFF_DATE, get_todays_date())
+        self.set_ini_default(self.REPORT_SIGNOFF_DATE, self.REPORT_SIGNOUT_DEFAULT)
         self.set_ini_default(self.USER_SUPPLIED_DRAFT_DATE, get_todays_date())
         self.set_ini_default(self.GENETICIST, self.GENETICIST_DEFAULT)
         self.set_ini_default(self.GENETICIST_ID, self.GENETICIST_ID_DEFAULT)