diff options
Diffstat (limited to 'test/unit/specs/services/html_converter')
3 files changed, 340 insertions, 0 deletions
diff --git a/test/unit/specs/services/html_converter/html_line_converter.spec.js b/test/unit/specs/services/html_converter/html_line_converter.spec.js new file mode 100644 index 00000000..86bd7e8b --- /dev/null +++ b/test/unit/specs/services/html_converter/html_line_converter.spec.js @@ -0,0 +1,171 @@ +import { convertHtmlToLines } from 'src/services/html_converter/html_line_converter.service.js' + +const greentextHandle = new Set(['p', 'div']) +const mapOnlyText = (processor) => (input) => { + if (input.text && input.level.every(l => greentextHandle.has(l))) { + return processor(input.text) + } else if (input.text) { + return input.text + } else { + return input + } +} + +describe('html_line_converter', () => { + describe('with processor that keeps original line should not make any changes to HTML when', () => { + const processorKeep = (line) => line + it('fed with regular HTML with newlines', () => { + const inputOutput = '1<br/>2<p class="lol">3 4</p> 5 \n 6 <p > 7 <br> 8 </p> <br>\n<br/>' + const result = convertHtmlToLines(inputOutput) + const comparableResult = result.map(mapOnlyText(processorKeep)).join('') + expect(comparableResult).to.eql(inputOutput) + }) + + it('fed with possibly broken HTML with invalid tags/composition', () => { + const inputOutput = '<feeee dwdwddddddw> <i>ayy<b>lm</i>ao</b> </section>' + const result = convertHtmlToLines(inputOutput) + const comparableResult = result.map(mapOnlyText(processorKeep)).join('') + expect(comparableResult).to.eql(inputOutput) + }) + + it('fed with very broken HTML with broken composition', () => { + const inputOutput = '</p> lmao what </div> whats going on <div> wha <p>' + const result = convertHtmlToLines(inputOutput) + const comparableResult = result.map(mapOnlyText(processorKeep)).join('') + expect(comparableResult).to.eql(inputOutput) + }) + + it('fed with sorta valid HTML but tags aren\'t closed', () => { + const inputOutput = 'just leaving a <div> hanging' + const result = convertHtmlToLines(inputOutput) + const comparableResult = result.map(mapOnlyText(processorKeep)).join('') + expect(comparableResult).to.eql(inputOutput) + }) + + it('fed with not really HTML at this point... tags that aren\'t finished', () => { + const inputOutput = 'do you expect me to finish this <div class=' + const result = convertHtmlToLines(inputOutput) + const comparableResult = result.map(mapOnlyText(processorKeep)).join('') + expect(comparableResult).to.eql(inputOutput) + }) + + it('fed with dubiously valid HTML (p within p and also div inside p)', () => { + const inputOutput = 'look ma <p> p \nwithin <p> p! </p> and a <br/><div>div!</div></p>' + const result = convertHtmlToLines(inputOutput) + const comparableResult = result.map(mapOnlyText(processorKeep)).join('') + expect(comparableResult).to.eql(inputOutput) + }) + + it('fed with maybe valid HTML? self-closing divs and ps', () => { + const inputOutput = 'a <div class="what"/> what now <p aria-label="wtf"/> ?' + const result = convertHtmlToLines(inputOutput) + const comparableResult = result.map(mapOnlyText(processorKeep)).join('') + expect(comparableResult).to.eql(inputOutput) + }) + + it('fed with valid XHTML containing a CDATA', () => { + const inputOutput = 'Yes, it is me, <![CDATA[DIO]]>' + const result = convertHtmlToLines(inputOutput) + const comparableResult = result.map(mapOnlyText(processorKeep)).join('') + expect(comparableResult).to.eql(inputOutput) + }) + + it('fed with some recognized but not handled elements', () => { + const inputOutput = 'testing images\n\n<img src="benis.png">' + const result = convertHtmlToLines(inputOutput) + const comparableResult = result.map(mapOnlyText(processorKeep)).join('') + expect(comparableResult).to.eql(inputOutput) + }) + }) + describe('with processor that replaces lines with word "_" should match expected line when', () => { + const processorReplace = (line) => '_' + it('fed with regular HTML with newlines', () => { + const input = '1<br/>2<p class="lol">3 4</p> 5 \n 6 <p > 7 <br> 8 </p> <br>\n<br/>' + const output = '_<br/>_<p class="lol">_</p>_\n_<p >_<br>_</p> <br>\n<br/>' + const result = convertHtmlToLines(input) + const comparableResult = result.map(mapOnlyText(processorReplace)).join('') + expect(comparableResult).to.eql(output) + }) + + it('fed with possibly broken HTML with invalid tags/composition', () => { + const input = '<feeee dwdwddddddw> <i>ayy<b>lm</i>ao</b> </section>' + const output = '_' + const result = convertHtmlToLines(input) + const comparableResult = result.map(mapOnlyText(processorReplace)).join('') + expect(comparableResult).to.eql(output) + }) + + it('fed with very broken HTML with broken composition', () => { + const input = '</p> lmao what </div> whats going on <div> wha <p>' + const output = '_<div>_<p>' + const result = convertHtmlToLines(input) + const comparableResult = result.map(mapOnlyText(processorReplace)).join('') + expect(comparableResult).to.eql(output) + }) + + it('fed with sorta valid HTML but tags aren\'t closed', () => { + const input = 'just leaving a <div> hanging' + const output = '_<div>_' + const result = convertHtmlToLines(input) + const comparableResult = result.map(mapOnlyText(processorReplace)).join('') + expect(comparableResult).to.eql(output) + }) + + it('fed with not really HTML at this point... tags that aren\'t finished', () => { + const input = 'do you expect me to finish this <div class=' + const output = '_' + const result = convertHtmlToLines(input) + const comparableResult = result.map(mapOnlyText(processorReplace)).join('') + expect(comparableResult).to.eql(output) + }) + + it('fed with dubiously valid HTML (p within p and also div inside p)', () => { + const input = 'look ma <p> p \nwithin <p> p! </p> and a <br/><div>div!</div></p>' + const output = '_<p>_\n_<p>_</p>_<br/><div>_</div></p>' + const result = convertHtmlToLines(input) + const comparableResult = result.map(mapOnlyText(processorReplace)).join('') + expect(comparableResult).to.eql(output) + }) + + it('fed with maybe valid HTML? (XHTML) self-closing divs and ps', () => { + const input = 'a <div class="what"/> what now <p aria-label="wtf"/> ?' + const output = '_<div class="what"/>_<p aria-label="wtf"/>_' + const result = convertHtmlToLines(input) + const comparableResult = result.map(mapOnlyText(processorReplace)).join('') + expect(comparableResult).to.eql(output) + }) + + it('fed with valid XHTML containing a CDATA', () => { + const input = 'Yes, it is me, <![CDATA[DIO]]>' + const output = '_' + const result = convertHtmlToLines(input) + const comparableResult = result.map(mapOnlyText(processorReplace)).join('') + expect(comparableResult).to.eql(output) + }) + + it('Testing handling ignored blocks', () => { + const input = ` + <pre><code>> rei = "0" + '0' + > rei == 0 + true + > rei == null + false</code></pre><blockquote>That, christian-like JS diagram but it’s evangelion instead.</blockquote> + ` + const result = convertHtmlToLines(input) + const comparableResult = result.map(mapOnlyText(processorReplace)).join('') + expect(comparableResult).to.eql(input) + }) + it('Testing handling ignored blocks 2', () => { + const input = ` + <blockquote>An SSL error has happened.</blockquote><p>Shakespeare</p> + ` + const output = ` + <blockquote>An SSL error has happened.</blockquote><p>_</p> + ` + const result = convertHtmlToLines(input) + const comparableResult = result.map(mapOnlyText(processorReplace)).join('') + expect(comparableResult).to.eql(output) + }) + }) +}) diff --git a/test/unit/specs/services/html_converter/html_tree_converter.spec.js b/test/unit/specs/services/html_converter/html_tree_converter.spec.js new file mode 100644 index 00000000..7283021b --- /dev/null +++ b/test/unit/specs/services/html_converter/html_tree_converter.spec.js @@ -0,0 +1,132 @@ +import { convertHtmlToTree } from 'src/services/html_converter/html_tree_converter.service.js' + +describe('html_tree_converter', () => { + describe('convertHtmlToTree', () => { + it('converts html into a tree structure', () => { + const input = '1 <p>2</p> <b>3<img src="a">4</b>5' + expect(convertHtmlToTree(input)).to.eql([ + '1 ', + [ + '<p>', + ['2'], + '</p>' + ], + ' ', + [ + '<b>', + [ + '3', + ['<img src="a">'], + '4' + ], + '</b>' + ], + '5' + ]) + }) + it('converts html to tree while preserving tag formatting', () => { + const input = '1 <p >2</p><b >3<img src="a">4</b>5' + expect(convertHtmlToTree(input)).to.eql([ + '1 ', + [ + '<p >', + ['2'], + '</p>' + ], + [ + '<b >', + [ + '3', + ['<img src="a">'], + '4' + ], + '</b>' + ], + '5' + ]) + }) + it('converts semi-broken html', () => { + const input = '1 <br> 2 <p> 42' + expect(convertHtmlToTree(input)).to.eql([ + '1 ', + ['<br>'], + ' 2 ', + [ + '<p>', + [' 42'] + ] + ]) + }) + it('realistic case 1', () => { + const input = '<p><span class="h-card"><a class="u-url mention" data-user="9wRC6T2ZZiKWJ0vUi8" href="https://cawfee.club/users/benis" rel="ugc">@<span>benis</span></a></span> <span class="h-card"><a class="u-url mention" data-user="194" href="https://shigusegubu.club/users/hj" rel="ugc">@<span>hj</span></a></span> nice</p>' + expect(convertHtmlToTree(input)).to.eql([ + [ + '<p>', + [ + [ + '<span class="h-card">', + [ + [ + '<a class="u-url mention" data-user="9wRC6T2ZZiKWJ0vUi8" href="https://cawfee.club/users/benis" rel="ugc">', + [ + '@', + [ + '<span>', + [ + 'benis' + ], + '</span>' + ] + ], + '</a>' + ] + ], + '</span>' + ], + ' ', + [ + '<span class="h-card">', + [ + [ + '<a class="u-url mention" data-user="194" href="https://shigusegubu.club/users/hj" rel="ugc">', + [ + '@', + [ + '<span>', + [ + 'hj' + ], + '</span>' + ] + ], + '</a>' + ] + ], + '</span>' + ], + ' nice' + ], + '</p>' + ] + ]) + }) + it('realistic case 2', () => { + const inputOutput = 'Country improv: give me a city<br/>Audience: Memphis<br/>Improv troupe: come on, a better one<br/>Audience: el paso' + expect(convertHtmlToTree(inputOutput)).to.eql([ + 'Country improv: give me a city', + [ + '<br/>' + ], + 'Audience: Memphis', + [ + '<br/>' + ], + 'Improv troupe: come on, a better one', + [ + '<br/>' + ], + 'Audience: el paso' + ]) + }) + }) +}) diff --git a/test/unit/specs/services/html_converter/utility.spec.js b/test/unit/specs/services/html_converter/utility.spec.js new file mode 100644 index 00000000..cf6fd99b --- /dev/null +++ b/test/unit/specs/services/html_converter/utility.spec.js @@ -0,0 +1,37 @@ +import { processTextForEmoji, getAttrs } from 'src/services/html_converter/utility.service.js' + +describe('html_converter utility', () => { + describe('processTextForEmoji', () => { + it('processes all emoji in text', () => { + const input = 'Hello from finland! :lol: We have best water! :lmao:' + const emojis = [ + { shortcode: 'lol', src: 'LOL' }, + { shortcode: 'lmao', src: 'LMAO' } + ] + const processor = ({ shortcode, src }) => ({ shortcode, src }) + expect(processTextForEmoji(input, emojis, processor)).to.eql([ + 'Hello from finland! ', + { shortcode: 'lol', src: 'LOL' }, + ' We have best water! ', + { shortcode: 'lmao', src: 'LMAO' } + ]) + }) + it('leaves text as is', () => { + const input = 'Number one: that\'s terror' + const emojis = [] + const processor = ({ shortcode, src }) => ({ shortcode, src }) + expect(processTextForEmoji(input, emojis, processor)).to.eql([ + 'Number one: that\'s terror' + ]) + }) + }) + + describe('getAttrs', () => { + it('extracts arguments from tag', () => { + const input = '<img src="boop" cool ebin=\'true\'>' + const output = { src: 'boop', cool: true, ebin: 'true' } + + expect(getAttrs(input)).to.eql(output) + }) + }) +}) |