Add a PDF Parser which scans for missng codes

Signed-off-by: Fred Boniface <fred@fjla.uk>

src/find-missing-from-pdf-dir.py

@@ -0,0 +1,87 @@
+import os, sys, json, subprocess, re, yaml, requests
+
+report_file_path = "./report.txt"
+code_store_list_url = "https://git.fjla.uk/OwlBoard/data/raw/branch/main/pis/gw.yaml"
+
+def is_pdfgrep_installed():
+    try:
+        subprocess.check_call(["pdfgrep", "--version"], stdout=subprocess.PIPE, stderr=subprocess.PIPE)
+        return True
+    except subprocess.CalledProcessError:
+        return False
+
+def fetch_and_parse_yaml(url):
+    try:
+        response = requests.get(url)
+        response.raise_for_status()
+        existing_codes = yaml.safe_load(response.text)
+        return existing_codes
+    except Exception as e:
+        print(f"Error downloading and parsing codes: {e}")
+        sys.exit(1)
+
+def main():
+
+    if len(sys.argv) != 2:
+        print("Usage: python pdf_code_extraction.py <directory_path>")
+        sys.exit(1)
+
+    pdf_directory = sys.argv[1]
+
+    if not os.path.isdir(pdf_directory):
+        print(f"'{pdf_directory}' is not a valid directory.")
+        sys.exit(1)
+
+    if not is_pdfgrep_installed():
+        print("pdfgrep is not installed on your system.")
+        sys.exit(1)
+
+    code_list = []
+
+    pdfgrep_cmd = f"pdfgrep -Prno 'code\\s*:\\s*\\d{{4}}' {pdf_directory}"
+    pdfgrep_output = subprocess.getoutput(pdfgrep_cmd)
+    
+
+    for line in pdfgrep_output.splitlines():
+        match = re.search(r"^(.*?):\s*code\s*:\s*(\d{4})", line)
+        if match:
+            filename, code = match.groups()
+            code_list.append({"file":filename, "code":str(code)})
+    
+    existing_codes = fetch_and_parse_yaml(code_store_list_url)['pis']
+    existing_set = set()
+    for item in existing_codes:
+        code = item['code']
+        existing_set.add(str(code))
+
+    unique_codes = set()
+    unique_code_list = []
+    missing_codes = []
+    for item in code_list:
+        code = item['code']
+        if code not in unique_codes:
+            unique_codes.add(code)
+            unique_code_list.append(item)
+            if code not in existing_set:
+                missing_codes.append(item)
+
+    #print(missing_codes)
+
+    report = f"""
+    Number of missing codes found: {len(missing_codes)}
+
+Missing Codes:
+    """
+
+    for item in missing_codes:
+        report += f"\n  - code: {item['code']}\n    stops: (File: {item['file']})"
+
+    
+    print(f"Saving report to {report_file_path}")
+    with open(report_file_path, 'w') as report_file:
+        report_file.write(report)
+
+    print(report)
+
+if __name__ == "__main__":
+    main()