tests + updates

This commit is contained in:
Henry Jameson 2019-11-14 22:40:20 +02:00
parent 51ea295704
commit bd2a682b83
2 changed files with 111 additions and 4 deletions

View File

@ -2,6 +2,8 @@
* This is a tiny purpose-built HTML parser/processor. This basically detects any type of visual newline and
* allows it to be processed, useful for greentexting, mostly
*
* known issue: doesn't handle CDATA so nested CDATA might not work well
*
* @param {Object} input - input data
* @param {(string) => string} processor - function that will be called on every line
* @return {string} processed html
@ -22,11 +24,15 @@ export const processHtml = (html, processor) => {
}
const flush = () => { // Processes current line buffer, adds it to output buffer and clears line buffer
if (textBuffer.trim().length > 0) {
buffer += processor(textBuffer)
} else {
buffer += textBuffer
}
textBuffer = ''
}
const handleBr = (tag) => { // handles single newlines/linebreaks
const handleBr = (tag) => { // handles single newlines/linebreaks/selfclosing
flush()
buffer += tag
}
@ -59,10 +65,12 @@ export const processHtml = (html, processor) => {
if (handledTags.has(tagName)) {
if (tagName === 'br') {
handleBr(tagFull)
}
if (openCloseTags.has(tagFull)) {
} else if (openCloseTags.has(tagName)) {
if (tagFull[1] === '/') {
handleClose(tagFull)
} else if (tagFull[tagFull.length - 2] === '/') {
// self-closing
handleBr(tagFull)
} else {
handleOpen(tagFull)
}
@ -76,6 +84,9 @@ export const processHtml = (html, processor) => {
textBuffer += char
}
}
if (tagBuffer) {
textBuffer += tagBuffer
}
flush()

View File

@ -0,0 +1,96 @@
import { processHtml } from 'src/services/tiny_post_html_processor/tiny_post_html_processor.service.js'
describe('TinyPostHTMLProcessor', () => {
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/>'
expect(processHtml(inputOutput, processorKeep)).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>'
expect(processHtml(inputOutput, processorKeep)).to.eql(inputOutput)
})
it('fed with very broken HTML with broken composition', () => {
const inputOutput = '</p> lmao what </div> whats going on <div> wha <p>'
expect(processHtml(inputOutput, processorKeep)).to.eql(inputOutput)
})
it('fed with sorta valid HTML but tags aren\'t closed', () => {
const inputOutput = 'just leaving a <div> hanging'
expect(processHtml(inputOutput, processorKeep)).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='
expect(processHtml(inputOutput, processorKeep)).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>'
expect(processHtml(inputOutput, processorKeep)).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"/> ?'
expect(processHtml(inputOutput, processorKeep)).to.eql(inputOutput)
})
it('fed with valid XHTML containing a CDATA', () => {
const inputOutput = 'Yes, it is me, <![CDATA[DIO]]>'
expect(processHtml(inputOutput, processorKeep)).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/>'
expect(processHtml(input, processorReplace)).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 = '_'
expect(processHtml(input, processorReplace)).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 = '</p>_</div>_<div>_<p>'
expect(processHtml(input, processorReplace)).to.eql(output)
})
it('fed with sorta valid HTML but tags aren\'t closed', () => {
const input = 'just leaving a <div> hanging'
const output = '_<div>_'
expect(processHtml(input, processorReplace)).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 = '_'
expect(processHtml(input, processorReplace)).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>'
expect(processHtml(input, processorReplace)).to.eql(output)
})
it('fed with maybe valid HTML? 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"/>_'
expect(processHtml(input, processorReplace)).to.eql(output)
})
it('fed with valid XHTML containing a CDATA', () => {
const input = 'Yes, it is me, <![CDATA[DIO]]>'
const output = '_'
expect(processHtml(input, processorReplace)).to.eql(output)
})
})
})