Merge pull request #252 from AgriculturalModelExchangeInitiative/issue251

Remove duplicated functions in pyx

Author: pradal

src/pycropml/render_cyml.py

@@ -14,6 +14,7 @@
 from datetime import datetime
 from pycropml.modelunit import ModelUnit
 from . import error
+from . import split_function
 
 DATATYPE = {}
 DATATYPE['INT'] = "int"
@@ -70,6 +71,7 @@ def generate_package(self):
             self.dir = directory.mkdir()
 
         files = []
+    
         count = 0
         for model in self.models:          
             self.generate_component(model) 
@@ -84,10 +86,13 @@ def generate_package(self):
 
     def generate_component(self, model_unit):
         """Todo"""
+
+        functions = []
         if model_unit.modelid.split(".")[0] != "function":
             func_name = f"model_{signature(model_unit)}"
         else:
             func_name = signature(model_unit)
+        
         types = [inp.datatype for inp in model_unit.inputs] + [out.datatype for out in model_unit.outputs]
         self.code = "import numpy\n"
         self.code += "from math import *\n"
@@ -99,19 +104,15 @@ def generate_component(self, model_unit):
 
         self.code += self.generate_function_signature(func_name, model_unit.inputs) + "\n"
         self.code += self.generate_function_doc(model_unit) + "\n"
-        if sys.version_info[0] >= 3:
-            self.code += self.generate_algorithm(model_unit) + "\n"
-        else:
-            self.code += self.generate_algorithm(model_unit).decode("utf-8") + "\n"
-
-        if model_unit.function:
-            for function in model_unit.function:
-                if function.language in ("Cyml", "cyml"):
-                    filefunc = Path(os.path.join(model_unit.path, "crop2ml", function.filename))
-                    with open(filefunc.encode('utf-8'), 'r') as f:
-                        source = f.read()
-                        self.code += source 
-                        self.code += "\n\n\n"
+        self.code += self.generate_algorithm(model_unit) + "\n"
+
+        if model_unit.function: 
+            function_files = [Path(os.path.join(model_unit.path, "crop2ml", function.filename))
+                              for function in model_unit.function if function.language.lower() == "cyml"]
+
+            content = split_function.unique_functions(function_files)
+            self.code += content
+            self.code += "\n"
 
         return self.code
 

src/pycropml/split_function.py

@@ -0,0 +1,73 @@
+""" Merge files into one file with unique functions.
+
+Render a set of pyx files containing the same function several into one file containing one and only one function of each name.
+
+"""
+import re
+
+def function_names(code):
+    # Regular expression pattern to match function names
+    # It assumes functions start with 'def' or 'cpdef'
+    pattern = r'\b(?:def|cpdef)\s+([a-zA-Z_]\w*)\s*\('
+
+    # Find all function names using regex
+    function_names = re.findall(pattern, code)
+
+    # Remove duplicates by converting to a set
+    unique_function_names = list(function_names)
+
+    return unique_function_names
+
+# Define a function to retrieve the code of a specific function
+def get_function_code(content, function_name):
+    # Regex to capture function definition
+    function_pattern = rf'(def|cpdef)\s+{function_name}\s*\((?:[^()]*(?:\([^()]*\))?)*\)\s*:[\s\S]*?'
+    
+    # Position of the function
+    match = re.search(function_pattern, content, re.MULTILINE)
+    
+    if not match:
+        return ''
+
+    start = match.start()
+    lines = content[start:].splitlines()
+    function_lines = []
+    
+    # Iter on the next lines from the function
+    # First line is the function definition
+    count = 0
+    for line in lines:
+        if (line.startswith('def ') or line.startswith('cpdef ')) and count != 0:
+            # Stop after the first line begining with def without indent
+            break  
+        function_lines.append(line)
+        count +=1
+    
+    function_lines.append('')
+    return '\n'.join(function_lines)
+
+    
+
+def unique_functions(files):
+    unique_functions = []
+
+    codes = []
+    for fn in files:
+        with open(fn.encode('utf-8'), 'r') as file:
+            content = file.read()
+
+        funs = function_names(content)
+        for fun in funs:
+            if fun in unique_functions:
+                print(f'{fun} is duplicated')
+            else:
+                unique_functions.append(fun)
+                function_code = get_function_code(content, fun)
+                #print(function_code)
+                codes.append(function_code)
+
+    content = '\n'.join(codes)
+
+    return content
+                
+