@godscene/shared 1.7.11

package/dist/es/extractor/web-extractor.mjs

@@ -0,0 +1,393 @@
+import { CONTAINER_MINI_HEIGHT, CONTAINER_MINI_WIDTH, NodeType } from "../constants/index.mjs";
+import { isAElement, isButtonElement, isContainerElement, isFormElement, isImgElement, isTextElement } from "./dom-util.mjs";
+import { descriptionOfTree } from "./tree.mjs";
+import { elementRect, getNodeAttributes, getPseudoElementContent, getRect, getTopDocument, logger, midsceneGenerateHash, setDebugMode } from "./util.mjs";
+import { onWindowMessage, postWindowMessage } from "./cs_postmessage.mjs";
+import { getContainerPath, getLocators } from "./customLocator.mjs";
+let indexId = 0;
+function tagNameOfNode(node) {
+    let tagName = '';
+    if (node instanceof HTMLElement) tagName = node.tagName?.toLowerCase();
+    else {
+        const parentElement = node.parentElement;
+        if (parentElement && parentElement instanceof HTMLElement) tagName = parentElement.tagName?.toLowerCase();
+    }
+    return tagName ? `<${tagName}>` : '';
+}
+function collectElementInfo(node, currentWindow, currentDocument, baseZoom = 1, basePoint = {
+    left: 0,
+    top: 0
+}, isContainer = false) {
+    const rect = elementRect(node, currentWindow, currentDocument, baseZoom);
+    if (!rect) return null;
+    if (rect.width < CONTAINER_MINI_WIDTH || rect.height < CONTAINER_MINI_HEIGHT) return null;
+    if (0 !== basePoint.left || 0 !== basePoint.top) {
+        rect.left += basePoint.left;
+        rect.top += basePoint.top;
+    }
+    if (rect.height >= window.innerHeight && rect.width >= window.innerWidth) return null;
+    if (isFormElement(node)) {
+        const attributes = getNodeAttributes(node, currentWindow);
+        let valueContent = attributes.value || attributes.placeholder || node.textContent || '';
+        const nodeHashId = midsceneGenerateHash(node, valueContent, rect);
+        const tagName = node.tagName.toLowerCase();
+        if ('select' === node.tagName.toLowerCase()) {
+            const selectedOption = node.options[node.selectedIndex];
+            valueContent = selectedOption?.textContent || '';
+        }
+        if (('input' === node.tagName.toLowerCase() || 'textarea' === node.tagName.toLowerCase()) && node.value) valueContent = node.value;
+        const elementInfo = {
+            id: nodeHashId,
+            nodeHashId,
+            nodeType: NodeType.FORM_ITEM,
+            indexId: indexId++,
+            attributes: {
+                ...attributes,
+                htmlTagName: `<${tagName}>`,
+                nodeType: NodeType.FORM_ITEM
+            },
+            content: valueContent.trim(),
+            rect,
+            center: [
+                Math.round(rect.left + rect.width / 2),
+                Math.round(rect.top + rect.height / 2)
+            ],
+            zoom: rect.zoom,
+            isVisible: rect.isVisible
+        };
+        return elementInfo;
+    }
+    if (isButtonElement(node)) {
+        const rect = mergeElementAndChildrenRects(node, currentWindow, currentDocument, baseZoom);
+        if (!rect) return null;
+        const attributes = getNodeAttributes(node, currentWindow);
+        const pseudo = getPseudoElementContent(node, currentWindow);
+        const content = node.innerText || pseudo.before || pseudo.after || '';
+        const nodeHashId = midsceneGenerateHash(node, content, rect);
+        const elementInfo = {
+            id: nodeHashId,
+            indexId: indexId++,
+            nodeHashId,
+            nodeType: NodeType.BUTTON,
+            attributes: {
+                ...attributes,
+                htmlTagName: tagNameOfNode(node),
+                nodeType: NodeType.BUTTON
+            },
+            content,
+            rect,
+            center: [
+                Math.round(rect.left + rect.width / 2),
+                Math.round(rect.top + rect.height / 2)
+            ],
+            zoom: rect.zoom,
+            isVisible: rect.isVisible
+        };
+        return elementInfo;
+    }
+    if (isImgElement(node)) {
+        const attributes = getNodeAttributes(node, currentWindow);
+        const nodeHashId = midsceneGenerateHash(node, '', rect);
+        const elementInfo = {
+            id: nodeHashId,
+            indexId: indexId++,
+            nodeHashId,
+            attributes: {
+                ...attributes,
+                ...node.nodeName?.toLowerCase() === 'svg' ? {
+                    svgContent: 'true'
+                } : {},
+                nodeType: NodeType.IMG,
+                htmlTagName: tagNameOfNode(node)
+            },
+            nodeType: NodeType.IMG,
+            content: '',
+            rect,
+            center: [
+                Math.round(rect.left + rect.width / 2),
+                Math.round(rect.top + rect.height / 2)
+            ],
+            zoom: rect.zoom,
+            isVisible: rect.isVisible
+        };
+        return elementInfo;
+    }
+    if (isTextElement(node)) {
+        const text = node.textContent?.trim().replace(/\n+/g, ' ');
+        if (!text) return null;
+        const attributes = getNodeAttributes(node, currentWindow);
+        const attributeKeys = Object.keys(attributes);
+        if (!text.trim() && 0 === attributeKeys.length) return null;
+        const nodeHashId = midsceneGenerateHash(node, text, rect);
+        const elementInfo = {
+            id: nodeHashId,
+            indexId: indexId++,
+            nodeHashId,
+            nodeType: NodeType.TEXT,
+            attributes: {
+                ...attributes,
+                nodeType: NodeType.TEXT,
+                htmlTagName: tagNameOfNode(node)
+            },
+            center: [
+                Math.round(rect.left + rect.width / 2),
+                Math.round(rect.top + rect.height / 2)
+            ],
+            content: text,
+            rect,
+            zoom: rect.zoom,
+            isVisible: rect.isVisible
+        };
+        return elementInfo;
+    }
+    if (isAElement(node)) {
+        const attributes = getNodeAttributes(node, currentWindow);
+        const pseudo = getPseudoElementContent(node, currentWindow);
+        const content = node.innerText || pseudo.before || pseudo.after || '';
+        const nodeHashId = midsceneGenerateHash(node, content, rect);
+        const elementInfo = {
+            id: nodeHashId,
+            indexId: indexId++,
+            nodeHashId,
+            nodeType: NodeType.A,
+            attributes: {
+                ...attributes,
+                htmlTagName: tagNameOfNode(node),
+                nodeType: NodeType.A
+            },
+            content,
+            rect,
+            center: [
+                Math.round(rect.left + rect.width / 2),
+                Math.round(rect.top + rect.height / 2)
+            ],
+            zoom: rect.zoom,
+            isVisible: rect.isVisible
+        };
+        return elementInfo;
+    }
+    if (isContainerElement(node) || isContainer) {
+        const attributes = getNodeAttributes(node, currentWindow);
+        const nodeHashId = midsceneGenerateHash(node, '', rect);
+        const elementInfo = {
+            id: nodeHashId,
+            nodeHashId,
+            indexId: indexId++,
+            nodeType: NodeType.CONTAINER,
+            attributes: {
+                ...attributes,
+                nodeType: NodeType.CONTAINER,
+                htmlTagName: tagNameOfNode(node)
+            },
+            content: '',
+            rect,
+            center: [
+                Math.round(rect.left + rect.width / 2),
+                Math.round(rect.top + rect.height / 2)
+            ],
+            zoom: rect.zoom,
+            isVisible: rect.isVisible
+        };
+        return elementInfo;
+    }
+    return null;
+}
+async function extractTextWithPosition(initNode, debugMode = false) {
+    const elementNode = await extractTreeNode(initNode, debugMode);
+    const elementInfoArray = [];
+    function dfsTopChildren(node) {
+        if (node.node) elementInfoArray.push(node.node);
+        for(let i = 0; i < node.children.length; i++)dfsTopChildren(node.children[i]);
+    }
+    dfsTopChildren({
+        children: elementNode.children,
+        node: elementNode.node
+    });
+    return elementInfoArray;
+}
+async function extractTreeNodeAsString(initNode, visibleOnly = false, debugMode = false) {
+    const elementNode = await extractTreeNode(initNode, debugMode);
+    return descriptionOfTree(elementNode, void 0, false, visibleOnly);
+}
+async function extractTreeNode(initNode, debugMode = false, basePoint = {
+    left: 0,
+    top: 0
+}, baseZoom = 1) {
+    setDebugMode(debugMode);
+    indexId = 0;
+    const topDocument = getTopDocument();
+    const startNode = initNode || topDocument;
+    const topChildren = [];
+    function dfs(node, currentWindow, currentDocument, baseZoom = 1, basePoint = {
+        left: 0,
+        top: 0
+    }) {
+        if (!node) return null;
+        if (node.nodeType && 10 === node.nodeType) return null;
+        const elementInfo = collectElementInfo(node, currentWindow, currentDocument, baseZoom, basePoint, false);
+        if (node instanceof currentWindow.HTMLIFrameElement) {
+            if (node.contentWindow && node.contentWindow) return null;
+        }
+        const nodeInfo = {
+            node: elementInfo,
+            children: []
+        };
+        if (elementInfo?.nodeType === NodeType.BUTTON || elementInfo?.nodeType === NodeType.IMG || elementInfo?.nodeType === NodeType.TEXT || elementInfo?.nodeType === NodeType.FORM_ITEM || elementInfo?.nodeType === NodeType.CONTAINER) return nodeInfo;
+        const rect = getRect(node, baseZoom, currentWindow);
+        for(let i = 0; i < node.childNodes.length; i++){
+            logger('will dfs', node.childNodes[i]);
+            const childNodeInfo = dfs(node.childNodes[i], currentWindow, currentDocument, rect.zoom, basePoint);
+            if (Array.isArray(childNodeInfo)) nodeInfo.children.push(...childNodeInfo);
+            else if (childNodeInfo) nodeInfo.children.push(childNodeInfo);
+        }
+        if (null === nodeInfo.node) {
+            if (0 === nodeInfo.children.length) return null;
+            return nodeInfo.children;
+        }
+        return nodeInfo;
+    }
+    function getCssScale(ele) {
+        try {
+            const { width, height } = ele.getBoundingClientRect();
+            return {
+                cssScaleX: Number((width / ele.offsetWidth).toFixed(1)),
+                cssScaleY: Number((height / ele.offsetHeight).toFixed(1))
+            };
+        } catch (error) {
+            return {
+                cssScaleX: 1,
+                cssScaleY: 1
+            };
+        }
+    }
+    const rootNodeInfo = dfs(startNode, window, document, baseZoom, basePoint);
+    if (Array.isArray(rootNodeInfo)) topChildren.push(...rootNodeInfo);
+    else if (rootNodeInfo) topChildren.push(rootNodeInfo);
+    if (startNode === topDocument) {
+        const iframes = document.querySelectorAll('iframe');
+        for(let i = 0; i < iframes.length; i++){
+            const iframe = iframes[i];
+            const iframeInfo = collectElementInfo(iframe, window, document, 1);
+            const baseZoom = getCssScale(iframe).cssScaleX;
+            if (iframeInfo) {
+                const iframeChildren = await postWindowMessage(iframe.contentWindow, window, {
+                    action: 'extractTreeNode',
+                    data: {
+                        basePoint: {
+                            left: iframeInfo.rect.left,
+                            top: iframeInfo.rect.top
+                        },
+                        baseZoom
+                    }
+                });
+                if (Array.isArray(iframeChildren)) topChildren.push(...iframeChildren);
+                else if (iframeChildren) topChildren.push(iframeChildren);
+            }
+        }
+    }
+    return {
+        node: null,
+        children: topChildren
+    };
+}
+function mergeElementAndChildrenRects(node, currentWindow, currentDocument, baseZoom = 1) {
+    const selfRect = elementRect(node, currentWindow, currentDocument, baseZoom);
+    if (!selfRect) return null;
+    let minLeft = selfRect.left;
+    let minTop = selfRect.top;
+    let maxRight = selfRect.left + selfRect.width;
+    let maxBottom = selfRect.top + selfRect.height;
+    function traverse(child) {
+        for(let i = 0; i < child.childNodes.length; i++){
+            const sub = child.childNodes[i];
+            if (1 === sub.nodeType) {
+                const rect = elementRect(sub, currentWindow, currentDocument, baseZoom);
+                if (rect) {
+                    minLeft = Math.min(minLeft, rect.left);
+                    minTop = Math.min(minTop, rect.top);
+                    maxRight = Math.max(maxRight, rect.left + rect.width);
+                    maxBottom = Math.max(maxBottom, rect.top + rect.height);
+                }
+                traverse(sub);
+            }
+        }
+    }
+    traverse(node);
+    return {
+        ...selfRect,
+        left: minLeft,
+        top: minTop,
+        width: maxRight - minLeft,
+        height: maxBottom - minTop
+    };
+}
+if ("u" > typeof window) onWindowMessage(window, async ({ action, data }, { source })=>{
+    if ('extractTreeNode' === action) try {
+        return await extractTreeNode(document.body, false, data.basePoint, data.baseZoom);
+    } catch (error) {
+        return {
+            error: 'error message'
+        };
+    }
+    if ('getElementFromPoint' === action) try {
+        return await getElementFromPoint({
+            x: data.x,
+            y: data.y,
+            containerPaths: data.containerPaths
+        });
+    } catch (error) {
+        return {
+            error: 'error message'
+        };
+    }
+});
+const getElementOffset = (element)=>{
+    const pLeft = Number.parseInt(window.getComputedStyle(element, null).getPropertyValue('padding-left'), 10) || 0;
+    const pTop = Number.parseInt(window.getComputedStyle(element, null).getPropertyValue('padding-top'), 10) || 0;
+    const bLeft = Number.parseInt(window.getComputedStyle(element, null).getPropertyValue('border-left-width'), 10) || 0;
+    const bTop = Number.parseInt(window.getComputedStyle(element, null).getPropertyValue('border-top-width'), 10) || 0;
+    return {
+        x: pLeft + bLeft,
+        y: pTop + bTop
+    };
+};
+const web_extractor_getCssScale = (element)=>{
+    try {
+        const { width, height } = element.getBoundingClientRect();
+        return {
+            cssScaleX: Number((width / element.offsetWidth).toFixed(1)),
+            cssScaleY: Number((height / element.offsetHeight).toFixed(1))
+        };
+    } catch (error) {
+        return {
+            cssScaleX: 1,
+            cssScaleY: 1
+        };
+    }
+};
+const getElementFromPoint = async (args)=>{
+    if (!args.containerPaths) args.containerPaths = [];
+    const dom = document.elementFromPoint(args.x, args.y);
+    if (dom?.nodeName === 'IFRAME') {
+        const containerPath = getContainerPath(dom);
+        const bounding = dom.getBoundingClientRect();
+        const offset = getElementOffset(dom);
+        const { cssScaleX, cssScaleY } = web_extractor_getCssScale(dom);
+        return await postWindowMessage(dom.contentWindow, window, {
+            action: 'getElementFromPoint',
+            data: {
+                x: (args.x - bounding.x - offset.x) / cssScaleX,
+                y: (args.y - bounding.y - offset.y) / cssScaleY,
+                containerPaths: args.containerPaths.concat(containerPath)
+            }
+        });
+    }
+    {
+        const elementInfo = {
+            allPaths: getLocators(dom),
+            containerPaths: args.containerPaths
+        };
+        return elementInfo;
+    }
+};
+export { collectElementInfo, extractTextWithPosition, extractTreeNode, extractTreeNodeAsString, getElementFromPoint, mergeElementAndChildrenRects };