anti_plagiarism_check

app/main/check_packs/pack_config.py

@@ -52,6 +52,7 @@
     ["empty_task_page_check"],
     ["water_in_the_text_check"],
     ["report_task_tracker"],
+    ["anti_plagiarism_check"],
     ["report_was_were_check"],
 ]
 

app/main/checks/report_checks/anti_plagiarism_check.py

@@ -0,0 +1,75 @@
+import re
+from ..base_check import BaseReportCriterion, answer
+
+
+class AntiPlagiarismCheck(BaseReportCriterion):
+    label = "Проверка на заимствования"
+    _description = ''
+    id = 'anti_plagiarism_check'
+
+    def __init__(self, file_info, originality_threshold=70):
+        super().__init__(file_info)
+        self.chapters = []
+        self.originality_threshold = originality_threshold
+
+    def late_init(self):
+        self.chapters = self.file.make_chapters(self.file_type['report_type'])
+
+    def check(self):
+        if self.file.page_counter() < 4:
+            return answer(False, "В отчете недостаточно страниц. Нечего проверять.")
+        self.late_init()
+        text = ""
+        user_result_str = ""
+        for chapter in self.chapters:
+            if 'список использованных источников' in chapter['text'].lower():
+                break
+            for child in chapter['child']:
+                text += " " + child['text'] # заменить на нужную для метода шинглов структуру
+        result = self.run_antiplagiarism(text)
+        originality_percent = result["originality"]
+        borrowed_fragments = result["fragments"]
+        if originality_percent < self.originality_threshold:
+            user_result_str += f"Обнаружены заимствования в тексте отчета. Уникальность работы составляет {originality_percent}%.<br>"
+            admin_result_str = user_result_str
+            admin_result_str += f"Ниже приведены фраменты, содержащие заимстования.<br><br>"
+            for i, fragment in enumerate(borrowed_fragments, start=1):
+                source_check_id = str(fragment.get('source', ''))
+                admin_result_str += (f"Заимствованный фрагмент №{i} находится на странице {fragment['page_in_doc']}. "
+                    f"Источник: {fragment['source']}, страница {fragment['page_in_source']}."
+                    f"Совпадение {fragment['percent']}%<br>"
+                    f"Текст фрагмента:<br>"
+                    f"{fragment['text']}<br>"
+                    f"Подробнее: <a href=\"#\" target=\"_blank\" "
+                    f"onclick=\"const antiPath = window.location.pathname.replace('/results/', '/anti_plagiarism/');"
+                    f"window.open(`${{antiPath}}/{source_check_id}`, '_blank'); return false;\">"
+                    f"перейти к сравнению</a><br><br>"
+                )
+            return answer(False, user_result_str, admin_result_str)
+        return answer(True, f"Уникальность текста {originality_percent}%. Проверка пройдена.")
+
+    def run_antiplagiarism(self, text):
+        # вызов реального алгоритма антиплагиата
+        originality = 65
+        fragments = [
+            {
+                "text": "пример заимствованного текста",
+                "source": "doc_id",
+                "percent": 90,
+                "page_in_doc": 5,
+                "page_in_source": 8
+
+            },
+            {
+                "text": "еще один заимствованный фрагмент",
+                "source": "doc_id2",
+                "percent": 70,
+                "page_in_doc": 7,
+                "page_in_source": 9
+            }
+        ]
+
+        return {
+            "originality": originality,
+            "fragments": fragments
+        }

app/routes/anti_plagiarism.py

@@ -0,0 +1,50 @@
+import bson
+from bson import ObjectId
+
+from flask import Blueprint, render_template, url_for
+from flask_login import current_user, login_required
+
+from app.db import db_methods
+from app.root_logger import get_root_logger
+
+anti_plagiarism = Blueprint('anti_plagiarism', __name__, template_folder='templates', static_folder='static')
+logger = get_root_logger('web')
+
+
+@anti_plagiarism.route("/<string:_id>/<string:source_check_id>", methods=["GET"])
+@login_required
+def anti_plagiarism_page(_id, source_check_id):
+    try:
+        oid = ObjectId(_id)
+    except bson.errors.InvalidId:
+        logger.error('_id exception:', exc_info=True)
+        return render_template("./404.html")
+
+    check = db_methods.get_check(oid)
+    if check is None:
+        logger.info("Запрошенная проверка не найдена: " + _id)
+        return render_template("./404.html")
+
+    try:
+        source_oid = ObjectId(_id)
+    except bson.errors.InvalidId:
+        logger.error('source_check_id exception:', exc_info=True)
+        return render_template("./404.html")
+    source_check = db_methods.get_check(source_oid)
+    if source_check is None:
+        logger.info("Запрошенная проверка не найдена: " + source_check_id)
+        return render_template("./404.html")
+    if not (current_user.is_admin):
+        return "У вас нет прав на просмотр", 403
+
+    fragments = []
+
+    student_pdf_url = url_for("get_pdf.get_pdf_main", _id=check.conv_pdf_fs_id)
+    source_pdf_url = url_for("get_pdf.get_pdf_main", _id=source_check.conv_pdf_fs_id)
+
+    return render_template(
+        "anti_plagiarism.html",
+        fragments=fragments,
+        student_pdf_url=student_pdf_url,
+        source_pdf_url=source_pdf_url,
+    )

app/routes/results.py

@@ -16,6 +16,17 @@
 logger = get_root_logger('web')
 
 
+def _prepare_verdicts_for_view(check, is_admin):
+    if not isinstance(check.enabled_checks, list):
+        return check
+
+    for criterion_info in check.enabled_checks:
+        verdict = criterion_info.get('verdict')
+        if isinstance(verdict, (list, tuple)) and len(verdict) > 1:
+            criterion_info['verdict'] = [verdict[1] if is_admin else verdict[0]]
+    return check
+
+
 @results_bp.route("/<string:_id>", methods=["GET"])
 @login_required
 def results_main(_id):
@@ -30,6 +41,7 @@ def results_main(_id):
         if current_user.is_admin or current_user.username == check.user or check.user == "api_access_token":
             # show processing time for user
             avg_process_time = None if check.is_ended else db_methods.get_average_processing_time()
+            check = _prepare_verdicts_for_view(check, current_user.is_admin)
             return render_template("./results.html", navi_upload=True, results=check,
                                 columns=TABLE_COLUMNS, avg_process_time=avg_process_time,
                                 stats=format_check(check.pack()))

app/server.py

@@ -51,6 +51,7 @@
 from routes.version import version
 from routes.capacity import capacity
 from routes.profile import profile
+from routes.anti_plagiarism import anti_plagiarism
 
 from server_consts import UPLOAD_FOLDER
 
@@ -87,6 +88,7 @@
 app.register_blueprint(version, url_prefix='/version')
 app.register_blueprint(capacity, url_prefix='/capacity')
 app.register_blueprint(profile, url_prefix='/profile')
+app.register_blueprint(anti_plagiarism, url_prefix='/anti_plagiarism')
 
 app.logger.addHandler(get_logging_stdout_handler())
 app.logger.propagate = False

app/templates/anti_plagiarism.html

@@ -0,0 +1,49 @@
+{% extends "root.html" %}
+
+{% block title %}Сравнение документов{% endblock %}
+
+{% block main %}
+    <div class="holder" id="anti_plagiarism_holder">
+        <h3 id="results_title" class="texteous ins" style="text-align: center;">
+            Проверка на заимствования
+        </h3>
+
+        <div class="pdf_compare_container">
+            <div class="pdf_block">
+                <h5 class="title">Документ студента</h5>
+                <div class="pdf_controls">
+                    <button class="btn" id="student-prev-page"><i class="bi bi-arrow-left"></i></button>
+                    <button class="btn" id="student-next-page"><i class="bi bi-arrow-right"></i></button>
+                    <span class="page-info">
+                        Страница <span id="student-page-num"></span> из <span id="student-page-count"></span>
+                    </span>
+                    <a href="{{ student_pdf_url }}" id="student-pdf-download" class="hidden"></a>
+                </div>
+                <div class="pdf_canvas_frame" id="student-canvas-frame">
+                    <canvas id="student-canvas" style="direction:ltr;"></canvas>
+                </div>
+            </div>
+
+            <div class="pdf_block">
+                <h5 class="title">Документ для сравнения</h5>
+                <div class="pdf_controls">
+                    <button class="btn" id="source-prev-page"><i class="bi bi-arrow-left"></i></button>
+                    <button class="btn" id="source-next-page"><i class="bi bi-arrow-right"></i></button>
+                    <span class="page-info">
+                        Страница <span id="source-page-num"></span> из <span id="source-page-count"></span>
+                    </span>
+                    <a href="{{ source_pdf_url }}" id="source-pdf-download" class="hidden"></a>
+                </div>
+                <div class="pdf_canvas_frame" id="source-canvas-frame">
+                    <canvas id="source-canvas" style="direction:ltr;"></canvas>
+                </div>
+            </div>
+        </div>
+    </div>
+{% endblock %}
+{% block script %}
+<script>
+    const fragments = {{ fragments|tojson|safe }};
+</script>
+<script src="{{ url_for('static', filename='anti_plagiarism.js') }}"></script>
+{% endblock script %}

assets/scripts/anti_plagiarism.js

@@ -0,0 +1,123 @@
+import '../styles/anti_plagiarism.css';
+import * as pdfjsLib from 'pdfjs-dist';
+import pdfjsWorker from "pdfjs-dist/build/pdf.worker.entry";
+
+$(function(){
+    if ($("#student-pdf-download").length === 0 || $("#source-pdf-download").length === 0) {
+        return;
+    }
+
+    pdfjsLib.GlobalWorkerOptions.workerSrc = pdfjsWorker;
+
+    function initPdfViewer(canvasId, canvasFrameId, prevBtnId, nextBtnId, pageNumId, pageCountId, pdfUrl) {
+        let pdfDoc = null;
+        let pageNum = 1;
+        let pageIsRendering = false;
+        let pageNumIsPending = null;
+
+        const canvas = document.getElementById(canvasId);
+        const canvasFrame = document.getElementById(canvasFrameId);
+        const ctx = canvas.getContext('2d');
+
+        const renderPage = num => {
+            pageIsRendering = true;
+
+            pdfDoc.getPage(num).then(page => {
+                const baseViewport = page.getViewport({scale: 1});
+                const targetWidth = canvasFrame.clientWidth || baseViewport.width;
+                const scale = targetWidth / baseViewport.width;
+                const viewport = page.getViewport({scale: scale});
+
+                canvas.height = viewport.height;
+                canvas.width = viewport.width;
+
+                const renderCtx = {
+                    canvasContext: ctx,
+                    viewport
+                };
+
+                page.render(renderCtx).promise.then(() => {
+                    pageIsRendering = false;
+
+                    if (pageNumIsPending !== null) {
+                        renderPage(pageNumIsPending);
+                        pageNumIsPending = null;
+                    }
+                });
+
+                document.getElementById(pageNumId).textContent = num;
+            });
+        };
+
+        const queueRenderPage = num => {
+            if (pageIsRendering) {
+                pageNumIsPending = num;
+            } else {
+                renderPage(num);
+            }
+        };
+
+        const showPrevPage = () => {
+            if (pageNum <= 1) {
+                return;
+            }
+            pageNum--;
+            queueRenderPage(pageNum);
+        };
+
+        const showNextPage = () => {
+            if (pageNum >= pdfDoc.numPages) {
+                return;
+            }
+            pageNum++;
+            queueRenderPage(pageNum);
+        };
+
+        const handleResize = () => {
+            if (!pdfDoc) {
+                return;
+            }
+            queueRenderPage(pageNum);
+        };
+
+        pdfjsLib
+            .getDocument(pdfUrl)
+            .promise.then(pdfDoc_ => {
+                pdfDoc = pdfDoc_;
+                document.getElementById(pageCountId).textContent = pdfDoc.numPages;
+                renderPage(pageNum);
+            });
+
+        $('#' + prevBtnId).click(showPrevPage);
+        $('#' + nextBtnId).click(showNextPage);
+        window.addEventListener('resize', handleResize);
+
+        if (window.ResizeObserver) {
+            const resizeObserver = new ResizeObserver(handleResize);
+            resizeObserver.observe(canvasFrame);
+        }
+    }
+
+    const studentPdfUrl = $("#student-pdf-download").attr('href');
+    const sourcePdfUrl = $("#source-pdf-download").attr('href');
+
+    initPdfViewer(
+        'student-canvas',
+        'student-canvas-frame',
+        'student-prev-page',
+        'student-next-page',
+        'student-page-num',
+        'student-page-count',
+        studentPdfUrl
+    );
+
+    initPdfViewer(
+        'source-canvas',
+        'source-canvas-frame',
+        'source-prev-page',
+        'source-next-page',
+        'source-page-num',
+        'source-page-count',
+        sourcePdfUrl
+    );
+});

assets/styles/anti_plagiarism.css

@@ -0,0 +1,48 @@
+#anti_plagiarism_holder {
+    overflow: auto;
+}
+
+#results_title {
+    margin-top: 1rem;
+}
+
+.pdf_compare_container {
+    display: flex;
+    flex-wrap: wrap;
+    justify-content: center;
+    gap: 2rem;
+    margin-bottom: 1rem;
+}
+
+.pdf_block {
+    width: 600px;
+    max-width: 100%;
+}
+
+.pdf_block .title {
+    margin-bottom: 1rem;
+    text-align: center;
+}
+
+.pdf_controls {
+    display: flex;
+    align-items: center;
+    justify-content: center;
+    gap: 0.5rem;
+    margin-bottom: 0.5rem;
+}
+
+.pdf_canvas_frame {
+    width: 100%;
+    aspect-ratio: 210 / 297;
+    overflow: hidden;
+    background: var(--bright-color);
+    border: 1px solid var(--border-color);
+}
+
+canvas {
+    display: block;
+    width: 100%;
+    height: auto;
+    background: var(--bright-color);
+}