FIX: change name weekly-check to db-check

pmodules_tools/db_diff/check.py

@@ -0,0 +1,157 @@
+import difflib
+import hashlib
+import os
+import re
+import shutil
+
+
+def db_check_diff(current_pmodule_state, Pmodules_db_path, Pmodules_states):
+    current_sha = hashlib.sha256(current_pmodule_state.encode()).hexdigest()
+
+    no_current_db = False
+
+    # Retrieve old state for comparison
+    try:
+        old_pmodule_file = sorted(
+            [Pmodules_db_path + f for f in os.listdir(Pmodules_db_path)],
+            key=os.path.getctime,
+        )[0]
+        old_pmodule_state = open(old_pmodule_file, "r").read()
+    except:
+        print(
+            "There is no old Pmodule database available on path "
+            + Pmodules_db_path
+            + "... Writing current one"
+        )
+        no_current_db = True
+
+    # There is a database and we have to check the differences between the pmodule states.
+    if not no_current_db and compare_states_sha(
+        current_pmodule_state, old_pmodule_state, current_sha
+    ):
+        print_pmodules_differences(
+            current_pmodule_state, old_pmodule_state, Pmodules_states
+        )
+
+    # There is no database available or there are differences with the old pmodule state, writing current state.
+    if no_current_db or compare_states_sha(
+        current_pmodule_state, old_pmodule_state, current_sha
+    ):
+        write_curent_state(current_pmodule_state, current_sha, Pmodules_db_path)
+
+
+def compare_states_sha(current_pmodule_state, old_pmodule_state, current_sha):
+    if (len(old_pmodule_state)) != 0:
+        old_sha = hashlib.sha256(old_pmodule_state.encode()).hexdigest()
+        if current_sha != old_sha:
+            return True
+        else:
+            return False
+
+
+def print_to_markdown(new_module_list, deleted_module_list, changed_module_list):
+    with open("pmodules_changes.md", "w") as md_file:
+        standard_print(md_file, new_module_list, "New modules:")
+
+        md_file.write("\n# Changed modules: \n")
+        if len(changed_module_list) > 1:
+            for index in range(0, len(changed_module_list) - 1, 2):
+                md_file.write(
+                    changed_module_list[index]
+                    + "<br>Changed state from "
+                    + changed_module_list[index + 1].split()[1]
+                    + " to ---> "
+                    + changed_module_list[index].split()[1]
+                    + "<br>"
+                )
+
+        standard_print(md_file, deleted_module_list, "Deleted modules:")
+
+
+def print_pmodules_differences(
+    current_pmodule_state, old_pmodule_state, Pmodules_states
+):
+    # Make sure the whitespaces in between the module descriptions are always one space only.
+    current_db = set(" ".join(i.split()) for i in current_pmodule_state.splitlines())
+    old_db = set(" ".join(i.split()) for i in old_pmodule_state.splitlines())
+
+    new_module_list = list(current_db - old_db)
+    deleted_module_list = list(old_db - current_db)
+    changed_module_list = []
+
+    # Replace the state of the modules with * for Regex comparison
+    new_module_list_no_state = new_module_list.copy()
+    deleted_module_list_no_state = deleted_module_list.copy()
+    for state in Pmodules_states:
+        new_module_list_no_state = [
+            diff.replace(state, ".*") for diff in new_module_list_no_state
+        ]
+        deleted_module_list_no_state = [
+            diff.replace(state, ".*") for diff in deleted_module_list_no_state
+        ]
+
+    # Check if the state of the module is the only thing that changed
+    # Append changed_module_list if yes
+    if len(deleted_module_list_no_state) >= len(new_module_list_no_state):
+        set_changed_module_list(
+            new_module_list_no_state,
+            deleted_module_list_no_state,
+            new_module_list,
+            deleted_module_list,
+            changed_module_list,
+        )
+    else:
+        set_changed_module_list(
+            deleted_module_list_no_state,
+            new_module_list_no_state,
+            new_module_list,
+            deleted_module_list,
+            changed_module_list,
+        )
+
+    print_to_markdown(new_module_list, deleted_module_list, changed_module_list)
+
+
+def set_changed_module_list(
+    small_module_list_no_state,
+    big_module_list_no_state,
+    new_module_list,
+    deleted_module_list,
+    changed_module_list,
+):
+    for diff in small_module_list_no_state:
+        if diff in big_module_list_no_state:
+            new_module_index = [
+                idx
+                for idx, string in enumerate(new_module_list)
+                if re.search(diff, string)
+            ][0]
+            deleted_module_index = [
+                idx
+                for idx, string in enumerate(deleted_module_list)
+                if re.search(diff, string)
+            ][0]
+            changed_module_list += [
+                new_module_list[new_module_index],
+                deleted_module_list[deleted_module_index],
+            ]
+            new_module_list[new_module_index] = ""
+            deleted_module_list[deleted_module_index] = ""
+
+
+def standard_print(file, module_list, string):
+    file.write("\n# " + string + " \n")
+    for diff in module_list:
+        if diff != "":
+            file.write(diff + "<br>")
+
+
+def write_curent_state(current_pmodule_state, current_sha, Pmodules_db_path):
+    # Emptying Pmodules database first
+    if os.path.exists(Pmodules_db_path):
+        shutil.rmtree(Pmodules_db_path)
+
+    # Recreating dir and writing new database
+    os.makedirs(Pmodules_db_path)
+    with open(Pmodules_db_path + current_sha, "w") as current_pmodule_file:
+        current_pmodule_file.write(current_pmodule_state)