from __future__ import division __copyright__ = "Copyright (C) 2017 Dong Zhuang" __license__ = """ Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions: The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software. THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. """ from django.test import TestCase import json try: from mock import patch except ImportError: from unittest.mock import patch from .test_sandbox import SingleCoursePageSandboxTestBaseMixin # {{{ Test Nbconvert for rendering ipynb notebook QUESTION_MARKUP_FULL = """ type: Page id: ipynb content: | # Ipython notebook Examples {{ render_notebook_cells("test.ipynb") }} """ QUESTION_MARKUP_SLICED1 = """ type: Page id: ipynb content: | # Ipython notebook Examples {{ render_notebook_cells("test.ipynb", indices=[0, 1, 2]) }} """ QUESTION_MARKUP_SLICED2 = """ type: Page id: ipynb content: | # Ipython notebook Examples {{ render_notebook_cells("test.ipynb", indices=[1, 2]) }} """ QUESTION_MARKUP_CLEAR_MARKDOWN = """ type: Page id: ipynb content: | # Ipython notebook Examples {{ render_notebook_cells("test.ipynb", clear_markdown=True) }} """ QUESTION_MARKUP_CLEAR_OUTPUT = """ type: Page id: ipynb content: | # Ipython notebook Examples {{ render_notebook_cells("test.ipynb", clear_output=True) }} """ QUESTION_MARKUP_CLEAR_ALL = """ type: Page id: ipynb content: | # Ipython notebook Examples {{ render_notebook_cells("test.ipynb", clear_markdown=True, clear_output=True) }} """ MARKDOWN_PLACEHOLDER = "wzxhzdk" TEST_IPYNB_BYTES = json.dumps({ "cells": [ { "cell_type": "markdown", "metadata": {}, "source": [ "# First Title of Test NoteBook" ] }, { "cell_type": "code", "execution_count": 1, "metadata": { "scrolled": True }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "This is function1\n" ] } ], "source": [ "def function1():\n", " print(\"This is function1\")\n", "\n", "function1()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "# Second Title of Test NoteBook" ] }, { "cell_type": "code", "execution_count": 2, "metadata": { "collapsed": True }, "outputs": [], "source": [ "def function2():\n", " print(\"This is function2\")" ] }, { "cell_type": "code", "execution_count": 3, "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "This is function2\n" ] } ], "source": [ "function2()" ] }, { "cell_type": "code", "execution_count": None, "metadata": { "collapsed": True }, "outputs": [], "source": [ "print(`5**18`)" ] } ], "metadata": { "kernelspec": { "display_name": "Python 3", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.5.0" } }, "nbformat": 4, "nbformat_minor": 2 }).encode() FIRST_TITLE_TEXT = "First Title of Test NoteBook" SECOND_TITLE_TEXT = "Second Title of Test NoteBook" TEXT_CELL_HTML_CLASS = "text_cell_render" CODE_CELL_HTML_CLASS = "code_cell" CODE_CELL_IN_STR_PATTERN = '
In[%s]:
' CODE_CELL_PRINT_STR1 = "This is function1" CODE_CELL_PRINT_STR2 = "This is function2" def strip_nbsp(s): """ Returns the given HTML with ' ' (introduced by nbconvert) stripped """ from django.utils.encoding import force_text return force_text(s).replace(' ', '').replace(u'\xa0', '') def get_nb_html_from_response(response): from django.utils.safestring import mark_safe return strip_nbsp(mark_safe(response.context["body"])) class NbconvertRenderTestMixin(SingleCoursePageSandboxTestBaseMixin): def assertIsValidNbConversion(self, response): # noqa self.assertNotContains(response, MARKDOWN_PLACEHOLDER) self.assertNotContains(response, "```") self.assertNotContains(response, "# First Title of Test NoteBook") self.assertNotContains(response, "# Second Title of Test NoteBook") def setUp(self): super(NbconvertRenderTestMixin, self).setUp() patcher = patch("course.content.get_repo_blob_data_cached") self.mock_func = patcher.start() self.mock_func.return_value = TEST_IPYNB_BYTES self.addCleanup(patcher.stop) class NbconvertRenderTest(NbconvertRenderTestMixin, TestCase): @classmethod def setUpTestData(cls): # noqa super(NbconvertRenderTest, cls).setUpTestData() cls.c.force_login(cls.instructor_participation.user) def test_full_notebook_render(self): resp = self.get_page_sandbox_preview_response(QUESTION_MARKUP_FULL) self.assertIsValidNbConversion(resp) self.assertContains(resp, TEXT_CELL_HTML_CLASS, count=2) self.assertContains(resp, CODE_CELL_HTML_CLASS, count=4) self.assertContains(resp, FIRST_TITLE_TEXT, count=1) self.assertContains(resp, SECOND_TITLE_TEXT, count=1) self.assertContains(resp, CODE_CELL_PRINT_STR1, count=2) self.assertContains(resp, CODE_CELL_PRINT_STR2, count=2) # backtick is properly rendered with highlight # for "`5**18`". though this syntax is not allowed in PY3 self.assertContains( resp, '`5') nb_html = get_nb_html_from_response(resp) for i in range(1, 4): self.assertInHTML(CODE_CELL_IN_STR_PATTERN % i, nb_html) def test_notebook_sliced1(self): resp = self.get_page_sandbox_preview_response(QUESTION_MARKUP_SLICED1) self.assertIsValidNbConversion(resp) self.assertContains(resp, TEXT_CELL_HTML_CLASS, count=2) self.assertContains(resp, CODE_CELL_HTML_CLASS, count=1) self.assertContains(resp, FIRST_TITLE_TEXT, count=1) self.assertContains(resp, SECOND_TITLE_TEXT, count=1) self.assertContains(resp, CODE_CELL_PRINT_STR1, count=2) self.assertNotContains(resp, CODE_CELL_PRINT_STR2) nb_html = get_nb_html_from_response(resp) self.assertInHTML(CODE_CELL_IN_STR_PATTERN % 1, nb_html, count=1) self.assertInHTML(CODE_CELL_IN_STR_PATTERN % 2, nb_html, count=0) self.assertInHTML(CODE_CELL_IN_STR_PATTERN % 3, nb_html, count=0) def test_notebook_sliced2(self): resp = self.get_page_sandbox_preview_response(QUESTION_MARKUP_SLICED2) self.assertIsValidNbConversion(resp) self.assertContains(resp, TEXT_CELL_HTML_CLASS, count=1) self.assertContains(resp, CODE_CELL_HTML_CLASS, count=1) self.assertNotContains(resp, FIRST_TITLE_TEXT) self.assertContains(resp, SECOND_TITLE_TEXT, count=1) self.assertContains(resp, CODE_CELL_PRINT_STR1, count=2) self.assertNotContains(resp, CODE_CELL_PRINT_STR2) # code highlight functions (in terms of rendered ipynb notebook cells only) import six if six.PY3: self.assertRegex(resp.context["body"], 'class="\w*\s*highlight[^\w]') self.assertContains(resp, " highlight hl-ipython3") self.assertContains(resp, 'print' '(', count=1) nb_html = get_nb_html_from_response(resp) self.assertInHTML(CODE_CELL_IN_STR_PATTERN % 1, nb_html, count=1) self.assertInHTML(CODE_CELL_IN_STR_PATTERN % 2, nb_html, count=0) self.assertInHTML(CODE_CELL_IN_STR_PATTERN % 3, nb_html, count=0) def test_notebook_clear_markdown(self): resp = self.get_page_sandbox_preview_response(QUESTION_MARKUP_CLEAR_MARKDOWN) self.assertIsValidNbConversion(resp) self.assertNotContains(resp, TEXT_CELL_HTML_CLASS) self.assertContains(resp, CODE_CELL_HTML_CLASS, count=4) self.assertNotContains(resp, FIRST_TITLE_TEXT) self.assertNotContains(resp, SECOND_TITLE_TEXT) nb_html = get_nb_html_from_response(resp) for i in range(1, 4): self.assertInHTML(CODE_CELL_IN_STR_PATTERN % i, nb_html, count=1) def test_notebook_clear_output(self): resp = self.get_page_sandbox_preview_response(QUESTION_MARKUP_CLEAR_OUTPUT) self.assertIsValidNbConversion(resp) self.assertContains(resp, TEXT_CELL_HTML_CLASS, count=2) self.assertContains(resp, CODE_CELL_HTML_CLASS, count=4) self.assertContains(resp, FIRST_TITLE_TEXT, count=1) self.assertContains(resp, SECOND_TITLE_TEXT, count=1) self.assertContains(resp, CODE_CELL_PRINT_STR1, count=1) self.assertContains(resp, CODE_CELL_PRINT_STR2, count=1) nb_html = get_nb_html_from_response(resp) for i in range(1, 4): self.assertInHTML(CODE_CELL_IN_STR_PATTERN % i, nb_html, count=0) self.assertInHTML(CODE_CELL_IN_STR_PATTERN % "", nb_html, count=4) def test_notebook_clear_markdown_and_output(self): resp = self.get_page_sandbox_preview_response(QUESTION_MARKUP_CLEAR_ALL) self.assertIsValidNbConversion(resp) self.assertNotContains(resp, TEXT_CELL_HTML_CLASS) self.assertContains(resp, CODE_CELL_HTML_CLASS, count=4) self.assertNotContains(resp, FIRST_TITLE_TEXT) self.assertNotContains(resp, SECOND_TITLE_TEXT) self.assertContains(resp, CODE_CELL_PRINT_STR1, count=1) self.assertContains(resp, CODE_CELL_PRINT_STR2, count=1) nb_html = get_nb_html_from_response(resp) for i in range(1, 4): self.assertInHTML(CODE_CELL_IN_STR_PATTERN % i, nb_html, count=0) self.assertInHTML(CODE_CELL_IN_STR_PATTERN % "", nb_html, count=4) # }}}