Code is now functional but untidy and in need of refactoring and commenting

.gitignore

@@ -3,6 +3,7 @@ auto_matched.txt
 organised_for_processing.txt
 validated.txt
 output.txt
+output_file.txt
 output
 *.sh
 

src/gitea.py

@@ -1,6 +1,7 @@
 import requests
 import base64
-from urllib.parse import urljoin
+import json
+from urllib.parse import urljoin, urlencode
 
 class GiteaConnector:
     def __init__(self, repo_url, user, key):
@@ -14,22 +15,52 @@ class GiteaConnector:
             "Content-Type": "application/json",
         }
 
+    ## Get open issues in repo
+    def get_open_issue_titles(self):
+        url_suffix = "issues"
+        url = urljoin(self.repo_url, url_suffix)
+        queries = {
+            "state": "open",
+            "type": "issues",
+        }
+
+        full_url = f"{url}?{urlencode(queries)}"
+        response = requests.get(full_url, headers=self.header)
+        if response.status_code == 200:
+            try:
+                data = response.json()  # Parse JSON response
+                if isinstance(data, list):  # Confirm it's a list
+                    # Extract 'title' from each item in the list
+                    existing_titles = [item['title'] for item in data if 'title' in item]
+                    return existing_titles
+                else:
+                    print("Unexpected response format: Expected a list.")
+            except ValueError:
+                print("Error: Response is not valid JSON.")
+        else:
+            print("Error fetching existing issues")
+        return []
+
+
     ## Create an issue in the repo
     def create_issue(self,title,content):
         url_suffix = "issues"
         url = urljoin(self.repo_url, url_suffix)
-
+        existing_issue_titles = self.get_open_issue_titles()
+        if title in existing_issue_titles:
+            print("Issue regarding this code exists already")
+            return True
         issue_data = {
             "title": title,
             "body": content,
             "assignees": ["fred.boniface"],
-            "labels": [261],
+            "labels": [261,269],
         }
 
         print(f"URL: {url}")
         response = requests.post(url, headers=self.header, json=issue_data)
         if response.status_code == 201:
-            print("Succesfully creates issue: ", response.json())
+            print("Succesfully created issue")
             return True
         else:
             print("Failed to create issue: ", response.status_code, response.text)
@@ -50,36 +81,27 @@ class GiteaConnector:
     ## Create new branch in the repo
     def create_branch(self, branch_name):
         BASE_BRANCH = "main"
-        base_branch_url = urljoin(self.repo_url, f"branch/{BASE_BRANCH}")
-        branch_response = requests.get(base_branch_url, headers=self.header)
 
-        if branch_response.status_code == 200:
-            base_sha = branch_response.json()['commit']['id']
-
-            create_branch_url = urljoin(self.repo_url, "git/refs")
-            branch_data = {
-                "ref": f"refs/heads/{branch_name}",
-                "sha": base_sha,
-            }
-
-            create_branch_response = requests.post(create_branch_url, headers=self.header, data=json.dumps(branch_data))
-            if create_branch_response.status_code == 201:
-                print(f"Branch {branch_name} created successfully")
-                return True
-            else:
-                print(f"Failed to create branch: ", branch_name, create_branch_response.status_code, create_branch_response.text)
-                return False
+        create_branch_url = urljoin(self.repo_url, "branches")
+        branch_data = {
+            "new_branch_name": branch_name,
+            "old_branch_name": BASE_BRANCH,
+        }
 
+        create_branch_response = requests.post(create_branch_url, headers=self.header, data=json.dumps(branch_data))
+        if create_branch_response.status_code == 201:
+            print(f"Branch {branch_name} created successfully")
+            return True
         else:
-            print("Failed to retreive branch SHA: ", branch_response.status_code, branch_response.text)
+            print(f"Failed to create branch: ", branch_name, create_branch_response.status_code, create_branch_response.text)
             return False
 
     ## Create a new file in the repo
     def create_pis_file(self, branch, filename, file_content):
         file_path = f"pis/{filename}"
-        encoded_content = base64.encode(file_content.encode('utf-8'))
+        encoded_content = base64.b64encode(file_content.encode('utf-8')).decode('utf-8')
 
-        create_file_url = url_multijoin(self.repo_url, f'contents/{file_path}')
+        create_file_url = urljoin(self.repo_url, f'contents/{file_path}.yaml')
 
         file_data = {
             "content": encoded_content,

src/main.py

@@ -7,6 +7,8 @@ import json
 import parse_pdf
 import train_detail
 import validate
+import pis_file
+from gitea import GiteaConnector
 
 # List all PDF files in the given directory
 def list_pdf_files(directory):
@@ -120,6 +122,18 @@ def main():
 
     validated = validate.check_and_validate_against_owlboard(auto_matched)
 
+    output_file = pis_file.create_new_pis_file(validated)
+
+    ## Create new branch and post output_file
+    gitea_client = GiteaConnector("https://git.fjla.uk/api/v1/repos/owlboard/data/", "fred.boniface", os.getenv("GITEA_PASS"))
+    BRANCH_NAME = f"auto-dgp2-{datetime.datetime.now().strftime('%Y-%m-%d_%H-%M-%S')}"
+    branch = gitea_client.create_branch(BRANCH_NAME)
+    if not branch:
+        print("Branch creation failed.  Unable to continue")
+        os.exit()
+    gitea_client.create_pis_file(BRANCH_NAME, BRANCH_NAME, output_file)
+
+
 
     # print(trains)
     out = open("organised_for_processing.txt", "w")
@@ -131,6 +145,9 @@ def main():
     out = open("validated.txt", "w")
     out.write(json.dumps(validated, indent=4, default=str))
     out.close()
+    out = open("output_file.txt", "w")
+    out.write(output_file)
+    out.close()
     
 if __name__ == "__main__":
     main()

src/pis_file.py

@@ -0,0 +1,25 @@
+## Produces a PIS YAML file in the OwlBoard/data format
+## and pushes it to a new branch for review
+
+def create_new_pis_file(input):
+    file_content = "# File produced by DGP2\n\nFor Review before merging\nXXXX\n\n"
+    for train in input:
+        file_content += f"""\n
+- code: {train.get('diagram_pis_code', 'N/A')}
+    # diagramDate: {train.get('diagram_date', 'N/A')}
+    # diagramHeadcode: {train.get('train_headcode', 'N/A')}
+    # diagram_time1: {train.get('diagram_time1', 'N/A')}
+    # diagram_time2: {train.get('diagram_time2', 'N/A')}
+"""
+        for entry in train.get('timetable_entries', []):
+            file_content += f"""
+  stops: {",".join(stop.lower() for stop in entry.get('stops', []))}
+  # stpIndicator: {entry.get('stpIndicator', 'N/A')}
+  # trainUid: {entry.get('trainUid', 'N/A')}
+  # scheduleStart: {entry.get('scheduleStart', 'N/A')}
+  # scheduleEnd: {entry.get('scheduleEnd', 'N/A')}
+  # daysRun: {",".join(day.upper() for day in entry.get('daysRun', []))}
+  # trainStartTime: {entry.get('trainStartTime', 'N/A')}
+"""
+    return file_content
+

src/validate.py

@@ -58,14 +58,24 @@ def check_and_validate_against_owlboard(train_entries):
         elif len(train_entry['timetable_entries']) == 1:
             issue_title = f"PIS Error | Code: {train_entry['diagram_pis_code']}"
             issue_content = f"""
-            PIS Code {train_entry['diagram_pis_code']}.
-            Diagram dated {train_entry['diagram_date']} has been parsed and produced a mismatch with the OwlBoard database.
+PIS Code {train_entry['diagram_pis_code']}.
+Diagram dated {train_entry['diagram_date']} has been parsed and produced a mismatch with the OwlBoard database.
 
-            Diagram  Stops: {",".join(train_entry['timetable_entries'][0]["stops"])}
+Diagram  Stops: {",".join(train_entry['timetable_entries'][0]["stops"])}
 
-            Database Stops: {",".join(database_stops)}
+Database Stops: {",".join(database_stops)}
 
-            This requires a manual check of the PIS file to identify mistakes or changes.
+The train which was identified in the database is:
+stpIndicator: {train_entry['timetable_entries'][0]['stpIndicator']}
+operator: {train_entry['timetable_entries'][0]['operator']}
+trainUID: {train_entry['timetable_entries'][0]['trainUid']}
+scheduleStart: {train_entry['timetable_entries'][0]['scheduleStart']}
+scheduleEnd: {train_entry['timetable_entries'][0]['scheduleEnd']}
+daysRun: {",".join(train_entry['timetable_entries'][0]['daysRun'])}
+
+This requires a manual check of the PIS file to identify mistakes or changes.
+Make sure that you validate that the service runs on the correct day.
+If the stpIndicator is 'O' or 'N', it is possible that the train schedule is newer than the parsed diagram.
             """
             print(issue_content)
             gitea_client.create_issue(issue_title, issue_content)