vodorod/parcer
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat: remove container tags

Browse Source
This commit is contained in:
amhsirak
2024-12-15 04:58:13 +05:30
parent 4469325925
commit e70145219e
1 changed files with 0 additions and 37 deletions
Download Patch File Download Diff File Expand all files Collapse all files

37
server/src/workflow-management/selector.ts
View File

@@ -74,22 +74,10 @@ export const getElementInformation = async (
if (originalEl) {
let element = originalEl;
const containerTags = ['DIV', 'SECTION', 'ARTICLE', 'MAIN', 'HEADER', 'FOOTER', 'NAV', 'ASIDE',
'ADDRESS', 'BLOCKQUOTE', 'DETAILS', 'DIALOG', 'FIGURE', 'FIGCAPTION', 'MAIN', 'MARK', 'SUMMARY', 'TIME',
'TABLE', 'THEAD', 'TBODY', 'TFOOT', 'TR', 'TH', 'TD', 'CAPTION', 'COLGROUP', 'COL', 'FORM', 'FIELDSET',
'LEGEND', 'LABEL', 'INPUT', 'BUTTON', 'SELECT', 'DATALIST', 'OPTGROUP', 'OPTION', 'TEXTAREA', 'OUTPUT',
'PROGRESS', 'METER', 'DETAILS', 'SUMMARY', 'MENU', 'MENUITEM', 'MENUITEM', 'APPLET', 'EMBED', 'OBJECT',
'PARAM', 'VIDEO', 'AUDIO', 'SOURCE', 'TRACK', 'CANVAS', 'MAP', 'AREA', 'SVG', 'IFRAME', 'FRAME', 'FRAMESET',
'LI', 'UL', 'OL', 'DL', 'DT', 'DD', 'HR', 'P', 'PRE', 'LISTING', 'PLAINTEXT', 'A'
];
while (element.parentElement) {
const parentRect = element.parentElement.getBoundingClientRect();
const childRect = element.getBoundingClientRect();
if (!containerTags.includes(element.parentElement.tagName)) {
break;
}
const fullyContained =
parentRect.left <= childRect.left &&
parentRect.right >= childRect.right &&
@@ -202,22 +190,10 @@ export const getRect = async (page: Page, coordinates: Coordinates, listSelector
if (originalEl) {
let element = originalEl;
const containerTags = ['DIV', 'SECTION', 'ARTICLE', 'MAIN', 'HEADER', 'FOOTER', 'NAV', 'ASIDE',
'ADDRESS', 'BLOCKQUOTE', 'DETAILS', 'DIALOG', 'FIGURE', 'FIGCAPTION', 'MAIN', 'MARK', 'SUMMARY', 'TIME',
'TABLE', 'THEAD', 'TBODY', 'TFOOT', 'TR', 'TH', 'TD', 'CAPTION', 'COLGROUP', 'COL', 'FORM', 'FIELDSET',
'LEGEND', 'LABEL', 'INPUT', 'BUTTON', 'SELECT', 'DATALIST', 'OPTGROUP', 'OPTION', 'TEXTAREA', 'OUTPUT',
'PROGRESS', 'METER', 'DETAILS', 'SUMMARY', 'MENU', 'MENUITEM', 'MENUITEM', 'APPLET', 'EMBED', 'OBJECT',
'PARAM', 'VIDEO', 'AUDIO', 'SOURCE', 'TRACK', 'CANVAS', 'MAP', 'AREA', 'SVG', 'IFRAME', 'FRAME', 'FRAMESET',
'LI', 'UL', 'OL', 'DL', 'DT', 'DD', 'HR', 'P', 'PRE', 'LISTING', 'PLAINTEXT', 'A'
];
while (element.parentElement) {
const parentRect = element.parentElement.getBoundingClientRect();
const childRect = element.getBoundingClientRect();
if (!containerTags.includes(element.parentElement.tagName)) {
break;
}
const fullyContained =
parentRect.left <= childRect.left &&
parentRect.right >= childRect.right &&
@@ -914,23 +890,10 @@ export const getNonUniqueSelectors = async (page: Page, coordinates: Coordinates
let element = originalEl;
const containerTags = ['DIV', 'SECTION', 'ARTICLE', 'MAIN', 'HEADER', 'FOOTER', 'NAV', 'ASIDE',
'ADDRESS', 'BLOCKQUOTE', 'DETAILS', 'DIALOG', 'FIGURE', 'FIGCAPTION', 'MAIN', 'MARK', 'SUMMARY', 'TIME',
'TABLE', 'THEAD', 'TBODY', 'TFOOT', 'TR', 'TH', 'TD', 'CAPTION', 'COLGROUP', 'COL', 'FORM', 'FIELDSET',
'LEGEND', 'LABEL', 'INPUT', 'BUTTON', 'SELECT', 'DATALIST', 'OPTGROUP', 'OPTION', 'TEXTAREA', 'OUTPUT',
'PROGRESS', 'METER', 'DETAILS', 'SUMMARY', 'MENU', 'MENUITEM', 'MENUITEM', 'APPLET', 'EMBED', 'OBJECT',
'PARAM', 'VIDEO', 'AUDIO', 'SOURCE', 'TRACK', 'CANVAS', 'MAP', 'AREA', 'SVG', 'IFRAME', 'FRAME', 'FRAMESET',
'LI', 'UL', 'OL', 'DL', 'DT', 'DD', 'HR', 'P', 'PRE', 'LISTING', 'PLAINTEXT', 'A'
];
while (element.parentElement) {
const parentRect = element.parentElement.getBoundingClientRect();
const childRect = element.getBoundingClientRect();
if (!containerTags.includes(element.parentElement.tagName)) {
break;
}
const fullyContained =
parentRect.left <= childRect.left &&
parentRect.right >= childRect.right &&