extract-content-mercury-utils

Documentation / extractor/html-to-content/extract-content/extract-content-mercury-utils

normalizeSpaces()

function normalizeSpaces(text: any): any;

Defined in: packages/ai-research-agent/src/extractor/html-to-content/extract-content/extract-content-mercury-utils.js:288

Parameters

Parameter	Type
text	any

Returns

any

---

paragraphize()

function paragraphize(
   node: any, 
   document: any, 
   br: boolean): any;

Defined in: packages/ai-research-agent/src/extractor/html-to-content/extract-content/extract-content-mercury-utils.js:293

Parameters

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

Returns

any

---

getAttrs()

function getAttrs(node: any): any;

Defined in: packages/ai-research-agent/src/extractor/html-to-content/extract-content/extract-content-mercury-utils.js:316

Parameters

Parameter	Type
node	any

Returns

any

---

convertNodeTo()

function convertNodeTo(
   node: any, 
   document: any, 
   tag: string): any;

Defined in: packages/ai-research-agent/src/extractor/html-to-content/extract-content/extract-content-mercury-utils.js:326

Parameters

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

Returns

any

---

brsToPs()

function brsToPs(document: any): any;

Defined in: packages/ai-research-agent/src/extractor/html-to-content/extract-content/extract-content-mercury-utils.js:349

Parameters

Parameter	Type
document	any

Returns

any

---

convertToParagraphs()

function convertToParagraphs(document: any): any;

Defined in: packages/ai-research-agent/src/extractor/html-to-content/extract-content/extract-content-mercury-utils.js:392

Parameters

Parameter	Type
document	any

Returns

any

---

cleanImages()

function cleanImages(article: any, document: any): any;

Defined in: packages/ai-research-agent/src/extractor/html-to-content/extract-content/extract-content-mercury-utils.js:424

Parameters

Parameter	Type
article	any
document	any

Returns

any

---

stripJunkTags()

function stripJunkTags(
   article: any, 
   document: any, 
   tags: any[]): any;

Defined in: packages/ai-research-agent/src/extractor/html-to-content/extract-content/extract-content-mercury-utils.js:434

Parameters

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

Returns

any

---

cleanHOnes()

function cleanHOnes(article: any, document: any): any;

Defined in: packages/ai-research-agent/src/extractor/html-to-content/extract-content/extract-content-mercury-utils.js:449

Parameters

Parameter	Type
article	any
document	any

Returns

any

---

cleanAttributes()

function cleanAttributes(article: any, document: any): any;

Defined in: packages/ai-research-agent/src/extractor/html-to-content/extract-content/extract-content-mercury-utils.js:464

Parameters

Parameter	Type
article	any
document	any

Returns

any

---

removeEmpty()

function removeEmpty(article: any): any;

Defined in: packages/ai-research-agent/src/extractor/html-to-content/extract-content/extract-content-mercury-utils.js:495

Parameters

Parameter	Type
article	any

Returns

any

---

rewriteTopLevel()

function rewriteTopLevel(article: any, document: any): any;

Defined in: packages/ai-research-agent/src/extractor/html-to-content/extract-content/extract-content-mercury-utils.js:565

Parameters

Parameter	Type
article	any
document	any

Returns

any

---

textLength()

function textLength(text: any): any;

Defined in: packages/ai-research-agent/src/extractor/html-to-content/extract-content/extract-content-mercury-utils.js:573

Parameters

Parameter	Type
text	any

Returns

any

---

linkDensity()

function linkDensity(node: any): number;

Defined in: packages/ai-research-agent/src/extractor/html-to-content/extract-content/extract-content-mercury-utils.js:578

Parameters

Parameter	Type
node	any

Returns

number

---

stripTags()

function stripTags(text: any, document: any): any;

Defined in: packages/ai-research-agent/src/extractor/html-to-content/extract-content/extract-content-mercury-utils.js:597

Parameters

Parameter	Type
text	any
document	any

Returns

any

---

stripUnlikelyCandidates()

function stripUnlikelyCandidates(document: any): any;

Defined in: packages/ai-research-agent/src/extractor/html-to-content/extract-content/extract-content-mercury-utils.js:605

Parameters

Parameter	Type
document	any

Returns

any

---

withinComment()

function withinComment(node: any): boolean;

Defined in: packages/ai-research-agent/src/extractor/html-to-content/extract-content/extract-content-mercury-utils.js:624

Parameters

Parameter	Type
node	any

Returns

boolean

---

nodeIsSufficient()

function nodeIsSufficient(node: any): boolean;

Defined in: packages/ai-research-agent/src/extractor/html-to-content/extract-content/extract-content-mercury-utils.js:640

Parameters

Parameter	Type
node	any

Returns

boolean

---

isWordpress()

function isWordpress(document: any): boolean;

Defined in: packages/ai-research-agent/src/extractor/html-to-content/extract-content/extract-content-mercury-utils.js:645

Parameters

Parameter	Type
document	any

Returns

boolean

---

setAttr()

function setAttr(
   node: any, 
   attr: any, 
   val: any): any;

Defined in: packages/ai-research-agent/src/extractor/html-to-content/extract-content/extract-content-mercury-utils.js:650

Parameters

Parameter	Type
node	any
attr	any
val	any

Returns

any

---

removeUnlessContent()

function removeUnlessContent(node: any, weight: any): void;

Defined in: packages/ai-research-agent/src/extractor/html-to-content/extract-content/extract-content-mercury-utils.js:508

Parameters

Parameter	Type
node	any
weight	any

Returns

void

---

setAttrs()

function setAttrs(node: any, attrs: any): any;

Defined in: packages/ai-research-agent/src/extractor/html-to-content/extract-content/extract-content-mercury-utils.js:656

Parameters

Parameter	Type
node	any
attrs	any

Returns

any