6 files changed, 73 insertions, 7 deletions
diff --git a/Smarker/assessments.py b/Smarker/assessments.py
index d5c8daf..a6c4df8 100644
--- a/Smarker/assessments.py
+++ b/Smarker/assessments.py
@@ -1,11 +1,29 @@
 import misc_classes
 import configparser
 import jinja_helpers
+import pycode_similar
+import operator
 import database
 import argparse
+import tempfile
 import yaml
 import os
 
+def generate_plagarism_report(codes):
+    for file_name, codes in codes.items():
+        with tempfile.TemporaryDirectory() as td:
+            un_added_student_nos = {i[0] for i in codes.keys()}
+            # print(un_added_student_nos)
+            for k, v in sorted(codes.keys(), key=operator.itemgetter(0, 1), reverse=True):
+                if k in un_added_student_nos:
+                    with open(os.path.join(td, "%i.py" % k), "w") as f:
+                        f.write(codes[(k, v)])
+                    
+                    # print("Written %s at %s" % (k, v))
+                    un_added_student_nos.remove(k)
+            input("%s..." % td)
+            print(pycode_similar.detect(os.listdir(td)))
+
 if __name__ == "__main__":
     config = configparser.ConfigParser()
     config.read(os.path.join(os.path.split(__file__)[0], "smarker.conf"))
@@ -47,6 +65,13 @@ if __name__ == "__main__":
         help = "Add a student in the form e.g. 123456789,Eden,Attenborough,E.Attenborough@uea.ac.uk",
         required = False
     )
+    parser.add_argument(
+        "-p", "--plagarism_report",
+        action = misc_classes.EnvDefault,
+        envvar = "plagarism_report",
+        help = "Generate a plagarism report for the given assessment",
+        required = False
+    )
     
     for option in config.options("mysql"):
         parser.add_argument(
@@ -91,6 +116,9 @@ if __name__ == "__main__":
 
             print("Added student %s" % name)
 
+        if args["plagarism_report"] is not None:
+            generate_plagarism_report(db.get_submission_codes(args["plagarism_report"]))
+
         
         # print(db.get_assessment_yaml("CMP-4009B-2020-A2"))
 
diff --git a/Smarker/database.py b/Smarker/database.py
index 38af6a7..c1b1df0 100644
--- a/Smarker/database.py
+++ b/Smarker/database.py
@@ -111,6 +111,8 @@ class SmarkerDatabase:
 
     def remove_assessment(self, name):
         with self.__connection.cursor() as cursor:
+            cursor.execute("DELETE FROM submitted_files WHERE submission_id IN (SELECT submission_id FROM submissions WHERE assessment_name = %s);", (name, ))
+            cursor.execute("DELETE FROM submissions WHERE assessment_name = %s;", (name, ))
             cursor.execute("DELETE FROM assessment_file WHERE assessment_name = %s;", (name, ))
             cursor.execute("DELETE FROM assessment WHERE assessment_name = %s;", (name, ))
         self.__connection.commit()
@@ -153,3 +155,35 @@ class SmarkerDatabase:
                     submission_id, file_name, file_contents
                 ))
         self.__connection.commit()
+
+    def get_submission_codes(self, assessment_name):
+        out = {}
+        with self.__connection.cursor() as cursor:
+            cursor.execute("SELECT file_id, file_name FROM assessment_file WHERE assessment_name = %s;", (assessment_name, ))
+            for file_id, file_name in cursor.fetchall():
+                out[file_name] = {}
+
+                cursor.execute("""
+                SELECT 
+                    submitted_files.file_text, 
+                    submissions.student_no, 
+                    submissions.submission_dt 
+                FROM submitted_files 
+                INNER JOIN submissions 
+                ON submissions.submission_id = submitted_files.submission_id 
+                WHERE submitted_files.file_id = %s;
+                """, (file_id, ))
+
+                for code, student_no, dt in cursor.fetchall():
+                    out[file_name][(int(student_no), dt)] = code
+        return out
+
+    def get_most_recent_submission_report(self, assessment_name):
+        with self.__connection.cursor() as cursor:
+            cursor.execute("SELECT MAX(submission_id), student_no FROM submissions WHERE assessment_name = %s GROUP BY student_no;", (assessment_name, ))
+            return [(int(i[0]), int(i[1]), yaml.safe_load(i[2])) for i in cursor.fetchall()]
+                
+
+if __name__ == "__main__":
+    with SmarkerDatabase(host = "vps.eda.gay", user="root", passwd=input("Input password: "), db="Smarker", port=3307) as db:
+        print(db.get_most_recent_submission_report("simple_assessment"))
diff --git a/Smarker/reflect.py b/Smarker/reflect.py
index 2a1f552..44d69de 100644
--- a/Smarker/reflect.py
+++ b/Smarker/reflect.py
@@ -376,7 +376,7 @@ def gen_reflection_report(client_code_path, assessment_struct, student_no, confi
                                 lines = lines.replace("\r", "")
                                 matches = {}
                                 for regex_ in contents["regexes"]:
-                                    matches[regex_] = re.findall(regex_, lines)
+                                    matches[regex_] = re.findall(str(regex_), lines)
                                 required_files_features["run"][j][cmd]["regexes"] = matches
                                 required_files_features["run"][j][cmd]["full_output"] = lines
 
diff --git a/Smarker/requirements.txt b/Smarker/requirements.txt
index 944b77a..a8fef17 100644
--- a/Smarker/requirements.txt
+++ b/Smarker/requirements.txt
@@ -9,3 +9,4 @@ junit2html
 pdfkit
 lxml
 pymysql
+pycode_similar
diff --git a/Smarker/templates/tex.jinja2 b/Smarker/templates/tex.jinja2
index eaa7db7..5985875 100644
--- a/Smarker/templates/tex.jinja2
+++ b/Smarker/templates/tex.jinja2
@@ -57,7 +57,7 @@ breaklines=true
 \newcommand{\errortext}[1]{\textcolor{red}{\textbf{#1}}}
 
 \author{((( student_no )))}
-\title{((( name ))) - Automatic marking report}
+\title{((( tex_escape(name) ))) - Automatic marking report}
 
 \begin{document}
 
@@ -70,16 +70,19 @@ breaklines=true
 
 \begin{figure}[H]
     \centering
-    \begin{forest}
+    ((# \begin{forest}
         ((( recurse_class_tree_forest(class_tree)|indent(8, False) )))
-    \end{forest}
+    \end{forest} #))
+    \begin{lstlisting}
+((( recurse_class_tree_text(class_tree) )))
+    \end{lstlisting}
     \caption{Class inheritance tree}
 \end{figure}
 
 \section{File Analysis}
 ((* set flat_files = flatten_struct(files) *))
 ((* for filename, files_contents in flat_files.items() *))
-    \subsection{\texttt{((( filename )))}}
+    \subsection{\texttt{((( tex_escape(filename) )))}}
     ((* if files_contents["present"] *))
         ((* if files_contents["has_exception"] *))
             \errortext{File cannot be run - has compile time exception.}
@@ -87,9 +90,9 @@ breaklines=true
             Please note that this file cannot be analysed or have tests preformed upon it-
             this can lead to the whole test suite failing if another module imports this.
                     
-            \textbf{Exception Type:} \texttt{((( files_contents["exception"]["type"] )))}
+            \textbf{Exception Type:} \texttt{((( tex_escape(files_contents["exception"]["type"]) )))}
             
-            \textbf{Exception String:} \texttt{((( files_contents["exception"]["str"] )))}
+            \textbf{Exception String:} \texttt{((( tex_escape(files_contents["exception"]["str"]) )))}
 
             \textbf{Full Traceback:}
 
diff --git a/Smarker/templates/text.jinja2 b/Smarker/templates/text.jinja2
deleted file mode 100644
index eca6ebd..0000000
--- a/Smarker/templates/text.jinja2
+++ /dev/null