Update playwright_model.py
Browse files- playwright_model.py +45 -39
playwright_model.py
CHANGED
|
@@ -13,17 +13,16 @@ async def capture_workflows(public_url: str, pdf_filename: str = "workflow_scree
|
|
| 13 |
async with async_playwright() as p:
|
| 14 |
browser = await p.chromium.launch(headless=True)
|
| 15 |
page = await browser.new_page()
|
| 16 |
-
print(f"
|
| 17 |
await page.goto(public_url, wait_until="load")
|
| 18 |
|
| 19 |
-
# ✅ Inject
|
| 20 |
js_logic = """
|
| 21 |
(function(){
|
| 22 |
const menus=[...document.querySelectorAll('.menu-item')];
|
| 23 |
const screens=[...document.querySelectorAll('.screen')];
|
| 24 |
window.__ordered=[];
|
| 25 |
const seen=new Set();
|
| 26 |
-
|
| 27 |
for(const m of menus){
|
| 28 |
let id=m.dataset.screen||m.dataset.target;
|
| 29 |
if(!id){
|
|
@@ -34,20 +33,20 @@ async def capture_workflows(public_url: str, pdf_filename: str = "workflow_scree
|
|
| 34 |
if(s && !seen.has(s)){seen.add(s);window.__ordered.push({menu:m,screen:s});}
|
| 35 |
}
|
| 36 |
for(const s of screens) if(!seen.has(s)) window.__ordered.push({menu:null,screen:s});
|
| 37 |
-
|
| 38 |
window.__visitedWorkflows=[];
|
| 39 |
window.__currentIndex=0;
|
| 40 |
window.__done=false;
|
| 41 |
|
| 42 |
-
|
| 43 |
-
|
| 44 |
-
const
|
|
|
|
| 45 |
for(const t of tabs){
|
| 46 |
-
const
|
| 47 |
-
if(
|
| 48 |
}
|
| 49 |
-
return
|
| 50 |
-
}
|
| 51 |
|
| 52 |
window.__captureNext=async function(){
|
| 53 |
if(window.__done) return false;
|
|
@@ -55,52 +54,38 @@ async def capture_workflows(public_url: str, pdf_filename: str = "workflow_scree
|
|
| 55 |
const pair=window.__ordered[window.__currentIndex];
|
| 56 |
const {menu,screen}=pair;
|
| 57 |
if(!screen){window.__done=true;return false;}
|
| 58 |
-
|
| 59 |
const wfName=screen.id || screen.getAttribute('data-name') || ('screen_'+window.__currentIndex);
|
| 60 |
if(window.__visitedWorkflows.includes(wfName)){
|
| 61 |
window.__currentIndex++;
|
| 62 |
return window.__captureNext();
|
| 63 |
}
|
| 64 |
-
|
| 65 |
window.__visitedWorkflows.push(wfName);
|
| 66 |
document.querySelectorAll('.screen').forEach(s=>s.classList.remove('active'));
|
| 67 |
document.querySelectorAll('.menu-item').forEach(m=>m.classList.remove('active'));
|
| 68 |
screen.classList.add('active');
|
| 69 |
if(menu) menu.classList.add('active');
|
| 70 |
screen.scrollIntoView({behavior:'smooth',block:'center'});
|
| 71 |
-
|
| 72 |
-
const mainTitle = screen.querySelector('.page-title, .header-title, h2')?.textContent?.trim() || wfName;
|
| 73 |
-
document.title = mainTitle;
|
| 74 |
-
|
| 75 |
-
const subs = await getSubScreens(screen);
|
| 76 |
window.__currentIndex++;
|
| 77 |
-
|
|
|
|
| 78 |
};
|
| 79 |
|
| 80 |
-
|
| 81 |
-
|
| 82 |
-
|
| 83 |
-
const
|
| 84 |
-
|
| 85 |
-
if(tab){
|
| 86 |
-
tab.click();
|
| 87 |
-
await new Promise(r=>setTimeout(r,700));
|
| 88 |
-
const title = screen.querySelector('.page-title, .header-title, h2')?.textContent?.trim() || mainName;
|
| 89 |
-
document.title = title + ' - ' + subName;
|
| 90 |
-
return true;
|
| 91 |
-
}
|
| 92 |
return false;
|
| 93 |
};
|
| 94 |
})();
|
| 95 |
"""
|
| 96 |
-
|
| 97 |
await page.evaluate(js_logic)
|
| 98 |
await asyncio.sleep(1)
|
| 99 |
|
| 100 |
screenshots = []
|
| 101 |
index = 0
|
| 102 |
|
| 103 |
-
# ✅ Loop through
|
| 104 |
while True:
|
| 105 |
result = await page.evaluate("window.__captureNext()")
|
| 106 |
if not result:
|
|
@@ -108,21 +93,42 @@ async def capture_workflows(public_url: str, pdf_filename: str = "workflow_scree
|
|
| 108 |
|
| 109 |
screen_name = result.get("screenName", f"screen_{index}")
|
| 110 |
sub_screens = result.get("subScreens", [])
|
| 111 |
-
|
| 112 |
-
# Capture main screen
|
| 113 |
screenshot_path = os.path.join(OUTPUT_DIR, f"{screen_name}.png")
|
| 114 |
print(f"📸 Capturing main screen: {screen_name}")
|
| 115 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 116 |
await asyncio.sleep(1.2)
|
| 117 |
await page.screenshot(path=screenshot_path, full_page=True)
|
| 118 |
screenshots.append(screenshot_path)
|
| 119 |
|
| 120 |
-
# Capture sub-screens
|
| 121 |
for sub in sub_screens:
|
| 122 |
-
print(f"
|
| 123 |
-
await page.evaluate(f"window.__clickSubScreen('{
|
| 124 |
await asyncio.sleep(1.0)
|
| 125 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 126 |
await page.screenshot(path=sub_path, full_page=True)
|
| 127 |
screenshots.append(sub_path)
|
| 128 |
|
|
|
|
| 13 |
async with async_playwright() as p:
|
| 14 |
browser = await p.chromium.launch(headless=True)
|
| 15 |
page = await browser.new_page()
|
| 16 |
+
print(f"Opening page: {public_url}")
|
| 17 |
await page.goto(public_url, wait_until="load")
|
| 18 |
|
| 19 |
+
# ✅ Inject workflow traversal + sub-screen logic
|
| 20 |
js_logic = """
|
| 21 |
(function(){
|
| 22 |
const menus=[...document.querySelectorAll('.menu-item')];
|
| 23 |
const screens=[...document.querySelectorAll('.screen')];
|
| 24 |
window.__ordered=[];
|
| 25 |
const seen=new Set();
|
|
|
|
| 26 |
for(const m of menus){
|
| 27 |
let id=m.dataset.screen||m.dataset.target;
|
| 28 |
if(!id){
|
|
|
|
| 33 |
if(s && !seen.has(s)){seen.add(s);window.__ordered.push({menu:m,screen:s});}
|
| 34 |
}
|
| 35 |
for(const s of screens) if(!seen.has(s)) window.__ordered.push({menu:null,screen:s});
|
|
|
|
| 36 |
window.__visitedWorkflows=[];
|
| 37 |
window.__currentIndex=0;
|
| 38 |
window.__done=false;
|
| 39 |
|
| 40 |
+
// helper: get all visible sub-screen tabs dynamically
|
| 41 |
+
window.__getSubScreens = function(screen){
|
| 42 |
+
const tabs=[...screen.querySelectorAll('.tab, .nav-link, [role="tab"], [data-tab], .sub-tab, .tab-item')];
|
| 43 |
+
const list=[];
|
| 44 |
for(const t of tabs){
|
| 45 |
+
const sub=t.textContent?.trim();
|
| 46 |
+
if(sub && !list.includes(sub)) list.push(sub);
|
| 47 |
}
|
| 48 |
+
return list;
|
| 49 |
+
};
|
| 50 |
|
| 51 |
window.__captureNext=async function(){
|
| 52 |
if(window.__done) return false;
|
|
|
|
| 54 |
const pair=window.__ordered[window.__currentIndex];
|
| 55 |
const {menu,screen}=pair;
|
| 56 |
if(!screen){window.__done=true;return false;}
|
|
|
|
| 57 |
const wfName=screen.id || screen.getAttribute('data-name') || ('screen_'+window.__currentIndex);
|
| 58 |
if(window.__visitedWorkflows.includes(wfName)){
|
| 59 |
window.__currentIndex++;
|
| 60 |
return window.__captureNext();
|
| 61 |
}
|
|
|
|
| 62 |
window.__visitedWorkflows.push(wfName);
|
| 63 |
document.querySelectorAll('.screen').forEach(s=>s.classList.remove('active'));
|
| 64 |
document.querySelectorAll('.menu-item').forEach(m=>m.classList.remove('active'));
|
| 65 |
screen.classList.add('active');
|
| 66 |
if(menu) menu.classList.add('active');
|
| 67 |
screen.scrollIntoView({behavior:'smooth',block:'center'});
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 68 |
window.__currentIndex++;
|
| 69 |
+
const subs = window.__getSubScreens(screen);
|
| 70 |
+
return {screenName:wfName, subScreens:subs};
|
| 71 |
};
|
| 72 |
|
| 73 |
+
// click sub-screen tab by text
|
| 74 |
+
window.__clickSubScreen = async function(name){
|
| 75 |
+
const tabs=[...document.querySelectorAll('.tab, .nav-link, [role="tab"], [data-tab], .sub-tab, .tab-item')];
|
| 76 |
+
const t=tabs.find(x=>x.textContent.trim()===name);
|
| 77 |
+
if(t){t.click(); return true;}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 78 |
return false;
|
| 79 |
};
|
| 80 |
})();
|
| 81 |
"""
|
|
|
|
| 82 |
await page.evaluate(js_logic)
|
| 83 |
await asyncio.sleep(1)
|
| 84 |
|
| 85 |
screenshots = []
|
| 86 |
index = 0
|
| 87 |
|
| 88 |
+
# ✅ Loop through each workflow
|
| 89 |
while True:
|
| 90 |
result = await page.evaluate("window.__captureNext()")
|
| 91 |
if not result:
|
|
|
|
| 93 |
|
| 94 |
screen_name = result.get("screenName", f"screen_{index}")
|
| 95 |
sub_screens = result.get("subScreens", [])
|
|
|
|
|
|
|
| 96 |
screenshot_path = os.path.join(OUTPUT_DIR, f"{screen_name}.png")
|
| 97 |
print(f"📸 Capturing main screen: {screen_name}")
|
| 98 |
|
| 99 |
+
# Update visible + browser tab title dynamically
|
| 100 |
+
await page.evaluate(
|
| 101 |
+
"""(name) => {
|
| 102 |
+
const titleEl = document.querySelector('.page-title, .header-title, main h2');
|
| 103 |
+
if (titleEl) titleEl.textContent = name;
|
| 104 |
+
document.title = name;
|
| 105 |
+
}""",
|
| 106 |
+
screen_name
|
| 107 |
+
)
|
| 108 |
+
|
| 109 |
await asyncio.sleep(1.2)
|
| 110 |
await page.screenshot(path=screenshot_path, full_page=True)
|
| 111 |
screenshots.append(screenshot_path)
|
| 112 |
|
| 113 |
+
# ✅ Capture sub-screens dynamically
|
| 114 |
for sub in sub_screens:
|
| 115 |
+
print(f" ↳ Capturing sub-screen: {sub}")
|
| 116 |
+
await page.evaluate(f"window.__clickSubScreen('{sub}')")
|
| 117 |
await asyncio.sleep(1.0)
|
| 118 |
+
sub_name_clean = sub.replace(" ", "_").lower()
|
| 119 |
+
sub_path = os.path.join(OUTPUT_DIR, f"{screen_name}_{sub_name_clean}.png")
|
| 120 |
+
|
| 121 |
+
# update tab title for sub-screen
|
| 122 |
+
await page.evaluate(
|
| 123 |
+
"""(mainName, subName) => {
|
| 124 |
+
const titleEl = document.querySelector('.page-title, .header-title, main h2');
|
| 125 |
+
const newTitle = mainName + ' - ' + subName;
|
| 126 |
+
if (titleEl) titleEl.textContent = newTitle;
|
| 127 |
+
document.title = newTitle;
|
| 128 |
+
}""",
|
| 129 |
+
screen_name, sub
|
| 130 |
+
)
|
| 131 |
+
|
| 132 |
await page.screenshot(path=sub_path, full_page=True)
|
| 133 |
screenshots.append(sub_path)
|
| 134 |
|