extract-content-mercury-utils

ai-research-agent / extractor/html-to-content/extract-content/extract-content-mercury-utils

Functions

brsToPs()

function brsToPs(document): any

Parameters

Parameter	Type
document	any

Returns

any

---

cleanAttributes()

function cleanAttributes(article, document): any

Parameters

Parameter	Type
article	any
document	any

Returns

any

---

cleanHOnes()

function cleanHOnes(article, document): any

Parameters

Parameter	Type
article	any
document	any

Returns

any

---

cleanImages()

function cleanImages(article, document): any

Parameters

Parameter	Type
article	any
document	any

Returns

any

---

convertNodeTo()

function convertNodeTo(
   node, 
   document, 
   tag): any

Parameters

Parameter	Type	Default value
node	any	undefined
document	any	undefined
tag	string	"p"

Returns

any

---

convertToParagraphs()

function convertToParagraphs(document): any

Parameters

Parameter	Type
document	any

Returns

any

---

getAttrs()

function getAttrs(node): any

Parameters

Parameter	Type
node	any

Returns

any

---

isWordpress()

function isWordpress(document): boolean

Parameters

Parameter	Type
document	any

Returns

boolean

---

linkDensity()

function linkDensity(node): number

Parameters

Parameter	Type
node	any

Returns

number

---

nodeIsSufficient()

function nodeIsSufficient(node): boolean

Parameters

Parameter	Type
node	any

Returns

boolean

---

normalizeSpaces()

function normalizeSpaces(text): any

Parameters

Parameter	Type
text	any

Returns

any

---

paragraphize()

function paragraphize(
   node, 
   document, 
   br): any

Parameters

Parameter	Type	Default value
node	any	undefined
document	any	undefined
br	boolean	false

Returns

any

---

removeEmpty()

function removeEmpty(article): any

Parameters

Parameter	Type
article	any

Returns

any

---

removeUnlessContent()

function removeUnlessContent(node, weight): void

Parameters

Parameter	Type
node	any
weight	any

Returns

void

---

rewriteTopLevel()

function rewriteTopLevel(article, document): any

Parameters

Parameter	Type
article	any
document	any

Returns

any

---

setAttr()

function setAttr(
   node, 
   attr, 
   val): any

Parameters

Parameter	Type
node	any
attr	any
val	any

Returns

any

---

setAttrs()

function setAttrs(node, attrs): any

Parameters

Parameter	Type
node	any
attrs	any

Returns

any

---

stripJunkTags()

function stripJunkTags(
   article, 
   document, 
   tags): any

Parameters

Parameter	Type	Default value
article	any	undefined
document	any	undefined
tags	any[]	[]

Returns

any

---

stripTags()

function stripTags(text, document): any

Parameters

Parameter	Type
text	any
document	any

Returns

any

---

stripUnlikelyCandidates()

function stripUnlikelyCandidates(document): any

Parameters

Parameter	Type
document	any

Returns

any

---

textLength()

function textLength(text): any

Parameters

Parameter	Type
text	any

Returns

any

---

withinComment()

function withinComment(node): boolean

Parameters

Parameter	Type
node	any

Returns

boolean