From c0937095e7214ec6b0c279cd51ab851a0205de3d Mon Sep 17 00:00:00 2001
From: Calvin <calvinpieters@gmail.com>
Date: Wed, 25 Dec 2024 12:24:09 +0200
Subject: [PATCH] Enhance GaussianAdapter to support integral algorithm in job
 type and update TRSH keyword handling. Refactor find_crest_executable for
 improved version parsing and comparison.

---
 arc/job/adapters/gaussian.py |  3 ++
 arc/job/trsh.py              |  3 +-
 arc/job/trsh_test.py         |  2 +-
 arc/settings/settings.py     | 59 +++++++++++++++++++++++++-----------
 4 files changed, 47 insertions(+), 20 deletions(-)

diff --git a/arc/job/adapters/gaussian.py b/arc/job/adapters/gaussian.py
index 476af80e67..f75af27ca3 100644
--- a/arc/job/adapters/gaussian.py
+++ b/arc/job/adapters/gaussian.py
@@ -303,6 +303,9 @@ def write_input_file(self) -> None:
                     keywords.extend(['tight', 'maxstep=5'])
                 else:
                     keywords.extend(['tight', 'maxstep=5', f'maxcycle={max_c}'])
+            else:
+                if integral_algorithm == 'Acc2E=14':
+                    input_dict['job_type_1'] = input_dict['job_type_1'] + (f' integral=({integral_algorithm})')
             input_dict['job_type_1'] = "opt" if self.level.method_type not in ['dft', 'composite', 'wavefunction']\
                 else f"opt=({', '.join(key for key in keywords)})"
 
diff --git a/arc/job/trsh.py b/arc/job/trsh.py
index 15b9b5c603..10c5d7d79c 100644
--- a/arc/job/trsh.py
+++ b/arc/job/trsh.py
@@ -1937,7 +1937,8 @@ def trsh_keyword_opt_maxcycles(job_status, ess_trsh_methods, trsh_keyword, could
 
             new_opt_keyword = 'opt=(' + ','.join(filtered_methods) + ')'
 
-            trsh_keyword = [kw if not kw.startswith('opt') else new_opt_keyword for kw in trsh_keyword]
+            trsh_keyword = [kw for kw in trsh_keyword if not kw.startswith('opt')]
+            trsh_keyword.append(new_opt_keyword)    
     
     return ess_trsh_methods, trsh_keyword, couldnt_trsh
 
diff --git a/arc/job/trsh_test.py b/arc/job/trsh_test.py
index 72513f495c..0a6e3c4fa0 100644
--- a/arc/job/trsh_test.py
+++ b/arc/job/trsh_test.py
@@ -398,7 +398,7 @@ def test_trsh_ess_job(self):
                                                                        num_heavy_atoms, cpu_cores, ess_trsh_methods)
 
         self.assertTrue(remove_checkfile)
-        self.assertEqual(ess_trsh_methods, ['change_node', 'int=(Acc2E=14)', 'checkfile=None', 'cartesian', 'NoSymm'])
+        self.assertEqual(ess_trsh_methods, ['change_node', 'int=(Acc2E=14)', 'checkfile=None', 'opt=(cartesian)', 'NoSymm'])
         self.assertEqual(trsh_keyword, ['opt=(cartesian)', 'int=(Acc2E=14)', 'nosymm'] )
 
         # Gaussian: test 3
diff --git a/arc/settings/settings.py b/arc/settings/settings.py
index 1514f16182..8924c1a431 100644
--- a/arc/settings/settings.py
+++ b/arc/settings/settings.py
@@ -817,28 +817,17 @@ def find_highest_version_in_directory(directory, name_contains):
             return None
 
         highest_version_path = None
-        highest_version = None
+        highest_version = ()
 
         for folder in os.listdir(directory):
             file_path = os.path.join(directory, folder)
             if name_contains.lower() in folder.lower() and os.path.isdir(file_path):
-                    # check if 'crest' is an executable file in the directory
-                    crest_path = os.path.join(file_path, 'crest')
-                    if os.path.isfile(crest_path) and os.access(crest_path, os.X_OK):
-                        match = re.search(r"(\d+\.\d+(\.\d+)*)", folder)  # Match version patterns like 1.0, 1.0.1
-                        if match:
-                            version_str = match.group(1)
-                            version = tuple(map(int, version_str.split('.')))
-                            if highest_version is None:
-                                highest_version = version
-                                highest_version_path = crest_path
-                            elif version[0] > highest_version[0]:
-                                highest_version = version
-                                highest_version_path = crest_path
-                            elif version[0] == highest_version[0] and version[1] > highest_version[1]:
-                                highest_version = version
-                                highest_version_path = crest_path
-
+                crest_path = os.path.join(file_path, 'crest')
+                if os.path.isfile(crest_path) and os.access(crest_path, os.X_OK):
+                    version = parse_version(folder)
+                    if highest_version == () or version > highest_version:
+                        highest_version = version
+                        highest_version_path = crest_path
         return highest_version_path
 
     # Priority 1: Search in /Local/ce_dana for the highest version of 'crest'
@@ -881,5 +870,39 @@ def find_highest_version_in_directory(directory, name_contains):
     # 'crest' not found
     return None, None
 
+def parse_version(folder_name):
+    """
+    Parses the version from the folder name and returns a tuple for comparison.
+
+    Supports versions like:
+    - 3.0.2
+    - v212
+    - 2.1
+    - 2
+
+    Args:
+        folder_name (str): The folder name containing the version.
+
+    Returns:
+        tuple: A version tuple (major, minor, patch)
+    """
+    version_regex = re.compile(r"(?:v?(\d+)(?:\.(\d+))?(?:\.(\d+))?)", re.IGNORECASE)
+    match = version_regex.search(folder_name)
+    if not match:
+        return (0, 0, 0)  # Default version if none found
+
+    major = int(match.group(1)) if match.group(1) else 0
+    minor = int(match.group(2)) if match.group(2) else 0
+    patch = int(match.group(3)) if match.group(3) else 0
+
+    # Handle specific cases
+    if major >= 100 and match.group(2) is None and match.group(3) is None:
+        # Example: v212 -> (2, 1, 2)
+        major = major // 100
+        minor = (major % 100) // 10
+        patch = major % 10
+
+    return (major, minor, patch)
+
 
 CREST_PATH, CREST_ENV_PATH = find_crest_executable()