Merge pull request #81 from casparvl/find_modules_regex

Improve find_modules with regex functionality

eessi/testsuite/utils.py

@@ -57,15 +57,60 @@ def is_cuda_required_module(module_name: str) -> bool:
     return requires_cuda
 
 
-def find_modules(substr: str) -> Iterator[str]:
-    """Return all modules in the current system that contain ``substr`` in their name."""
-    if not isinstance(substr, str):
+def find_modules(regex: str, name_only=True) -> Iterator[str]:
+    """
+    Return all modules matching the regular expression regex. Note that since we use re.search,
+    a module matches if the regex matches the module name at any place. I.e. the match does
+    not have to be at the start of the smodule name
+    
+    Arguments:
+    - regex: a regular expression
+    - name_only: regular expressions will only be matched on the module name, not the version (default: True).
+    
+    Note: the name_only feature assumes anything after the last forward '/' is the version,
+    and strips that before doing a match.
+
+    Example
+
+    Suppose we have the following modules on a system:
+
+    gompic/2022a
+    gompi/2022a
+    CGAL/4.14.3-gompi-2022a
+
+    The following calls would return the following respective modules
+
+    find_modules('gompi') => [gompic/2022a, gompi/2022a]
+    find_modules('gompi$') => [gompi/2022a]
+    find_modules('gompi', name_only = False) => [gompic/2022a, gompi/2022a, CGAL/4.14.3-gompi-2022a]
+    find_modules('^gompi', name_only = False) => [gompic/2022a, gompi/2022a]
+    find_modules('^gompi/', name_only = False) => [gompi/2022a]
+    find_modules('-gompi-2022a', name_only = False) => [CGAL/4.14.3-gompi-2022a]
+
+    """
+
+    if not isinstance(regex, str):
         raise TypeError("'substr' argument must be a string")
 
     ms = rt.runtime().modules_system
-    modules = OrderedSet(ms.available_modules(substr))
-    for m in modules:
-        yield m
+    # Returns e.g. ['Bison/', 'Bison/3.7.6-GCCcore-10.3.0', 'BLIS/', 'BLIS/0.8.1-GCC-10.3.0']
+    modules = ms.available_modules('')
+    for mod in modules:
+        # Exclude anything without version, i.e. ending with / (e.g. Bison/)
+        if re.search('.*/$', mod):
+            continue
+        # The thing we yield should always be the original module name (orig_mod), including version
+        orig_mod = mod
+        if name_only:
+            # Remove trailing slashes from the regex (in case the callee forgot)
+            regex = regex.rstrip('/')
+            # Remove part after the last forward slash, as we assume this is the version
+            mod = re.sub('/[^/]*$', '', mod)
+        # Match the actual regular expression
+        log(f"Matching module {mod} with regex {regex}")
+        if re.search(regex, mod):
+            log(f"Match!")
+            yield orig_mod
 
 def check_proc_attribute_defined(test: rfm.RegressionTest, attribute) -> bool:
     """
@@ -104,4 +149,4 @@ def check_proc_attribute_defined(test: rfm.RegressionTest, attribute) -> bool:
             "This is a programming error, please report this issue."
         )
         raise AttributeError(msg)
-
+