fix(collect): recognize short Shopee product titles

This commit is contained in:
chengma
2026-07-14 16:31:35 +08:00
parent 2f6b9f7af4
commit 250c7e27eb
5 changed files with 378 additions and 70 deletions
+216 -59
View File
@@ -18,8 +18,19 @@ from .cdp import (
DEFAULT_REGION_HOST = "seller.shopee.tw"
TITLE_XPATH = "//input[@class='eds-input__input' and string-length(@modelvalue)>24]"
TITLE_FIELD_SELECTOR = '[data-product-edit-field-unique-id="name"]'
IMAGE_FIELD_SELECTOR = '[data-product-edit-field-unique-id="images"]'
TITLE_XPATH = (
"//*[@data-product-edit-field-unique-id='name']"
"//input[contains(concat(' ', normalize-space(@class), ' '), ' eds-input__input ')]"
)
LEGACY_TITLE_XPATH = "//input[@class='eds-input__input' and string-length(@modelvalue)>24]"
ITEMBOX_XPATH = (
"//*[@data-product-edit-field-unique-id='images']"
"//div[contains(concat(' ', normalize-space(@class), ' '), "
"' shopee-image-manager__itembox ') and @data-draggable='true']"
)
LEGACY_ITEMBOX_XPATH = (
"//div[@class='container']/div[@class='can-drag shopee-image-manager__itembox' "
"and @data-draggable='true']"
)
@@ -35,22 +46,67 @@ LOGIN_PATH_MARKERS = (
"seller/accounts/signin",
)
JS_EDITOR_FIELD_HELPERS = "".join(
(
"function cmVisible(e){if(!e)return false;var r=e.getBoundingClientRect();"
"var s=getComputedStyle(e);return r.width>0&&r.height>0&&"
"s.visibility!=='hidden'&&s.display!=='none';}",
"function cmTitleCandidate(){",
f"var root=document.querySelector({json.dumps(TITLE_FIELD_SELECTOR)});",
"if(root){var primary=[].slice.call(root.querySelectorAll('input.eds-input__input'))"
".filter(cmVisible);return {node:primary.length===1?primary[0]:null,count:primary.length,"
"source:'business',root_exists:true};}",
f"var snap=document.evaluate({json.dumps(LEGACY_TITLE_XPATH)},document,null,"
"XPathResult.ORDERED_NODE_SNAPSHOT_TYPE,null);",
"var legacy=[];for(var i=0;i<snap.snapshotLength;i++){var item=snap.snapshotItem(i);"
"if(cmVisible(item))legacy.push(item);}",
"return {node:legacy.length===1?legacy[0]:null,count:legacy.length,"
"source:'legacy',root_exists:false};}",
"function cmMainImageContext(){",
f"var root=document.querySelector({json.dumps(IMAGE_FIELD_SELECTOR)});",
"if(root)return {node:root.querySelector('.shopee-image-manager'),"
"source:'business',root_exists:true};",
"return {node:document.querySelector('.shopee-image-manager'),"
"source:'legacy',root_exists:false};}",
"function cmMainImageItems(context){",
"if(context&&context.root_exists){var manager=context.node;return manager?"
"[].slice.call(manager.querySelectorAll("
"'.shopee-image-manager__itembox.can-drag[data-draggable=\"true\"]')):[];}",
f"var snap=document.evaluate({json.dumps(LEGACY_ITEMBOX_XPATH)},document,null,"
"XPathResult.ORDERED_NODE_SNAPSHOT_TYPE,null);",
"var items=[];for(var i=0;i<snap.snapshotLength;i++)items.push(snap.snapshotItem(i));"
"return items;}",
"function cmMainUploadInput(context){var manager=context&&context.node;return manager?"
"manager.querySelector('.shopee-image-manager__upload input[type=file]'):null;}",
"function cmMainUploadBox(context){var manager=context&&context.node;return manager?"
"manager.querySelector('.shopee-image-manager__upload,[class*=image-manager__upload]'):null;}",
)
)
JS_READY = (
"(function(){"
f"var s=document.evaluate({json.dumps(ITEMBOX_XPATH)},document,null,"
"XPathResult.ORDERED_NODE_SNAPSHOT_TYPE,null);"
f"var r=document.evaluate({json.dumps(TITLE_XPATH)},document,null,"
"XPathResult.FIRST_ORDERED_NODE_TYPE,null);"
"var up=document.querySelector('.shopee-image-manager__upload input[type=file]');"
"return (s.snapshotLength>0 && !!r.singleNodeValue && !!up);})()"
+ JS_EDITOR_FIELD_HELPERS
+ "var title=cmTitleCandidate();var image=cmMainImageContext();"
"var items=cmMainImageItems(image);var up=cmMainUploadInput(image);"
"var uploads=image.node?image.node.querySelectorAll("
"'.shopee-image-manager__upload input[type=file]'):[];"
"var srcs=items.map(function(item){var img=item.querySelector('img');return img?img.src:null;})"
".filter(Boolean);"
"var ready=title.count===1&&items.length>0&&uploads.length===1;"
"return JSON.stringify({ready:ready,title_count:title.count,title_source:title.source,"
"title_field_exists:title.root_exists,title_ambiguous:title.count>1,"
"image_field_exists:image.root_exists,image_manager_exists:!!image.node,"
"image_count:items.length,cdn_count:srcs.filter(function(s){return /susercontent/.test(s);}).length,"
"blob_count:srcs.filter(function(s){return /^blob:/.test(s);}).length,"
"upload_input_exists:!!up,upload_input_count:uploads.length,current_url:location.href});})()"
)
JS_RECTS = (
"(function(){"
f"var s=document.evaluate({json.dumps(ITEMBOX_XPATH)},document,null,"
"XPathResult.ORDERED_NODE_SNAPSHOT_TYPE,null);"
"var a=[];for(var i=0;i<s.snapshotLength;i++){"
"var el=s.snapshotItem(i);var r=el.getBoundingClientRect();var im=el.querySelector('img');"
+ JS_EDITOR_FIELD_HELPERS
+ "var image=cmMainImageContext();var items=cmMainImageItems(image);"
"var a=[];for(var i=0;i<items.length;i++){"
"var el=items[i];var r=el.getBoundingClientRect();var im=el.querySelector('img');"
"a.push({i:i,x:r.left+r.width/2,y:r.top+r.height/2,left:r.left,top:r.top,"
"w:r.width,h:r.height,src:im?im.src:null});}"
"return JSON.stringify(a);})()"
@@ -59,14 +115,12 @@ JS_RECTS = (
JS_UPLOAD_STATE = (
"(function(){"
"function visible(e){if(!e)return false;var r=e.getBoundingClientRect();var s=getComputedStyle(e);"
"return r.width>0&&r.height>0&&s.visibility!=='hidden'&&s.display!=='none';}"
+ JS_EDITOR_FIELD_HELPERS
+ "function visible(e){return cmVisible(e);}"
"function text(e){return ((e&&((e.innerText||e.textContent)||''))||'').trim();}"
"var manager=document.querySelector('.shopee-image-manager');"
"var items=manager?[].slice.call(manager.querySelectorAll('.shopee-image-manager__itembox,[class*=image-manager__itembox]')):[];"
"var image=cmMainImageContext();var manager=image.node;var items=cmMainImageItems(image);"
"var imgs=items.map(function(item){var img=item.querySelector('img');return img?img.src:null;}).filter(Boolean);"
"var upload=manager?manager.querySelector('.shopee-image-manager__upload input[type=file]'):null;"
"var uploadBox=manager?manager.querySelector('.shopee-image-manager__upload,[class*=image-manager__upload]'):null;"
"var upload=cmMainUploadInput(image);var uploadBox=cmMainUploadBox(image);"
"var busy=manager?[].slice.call(manager.querySelectorAll('[class*=loading],[class*=Loading],[class*=spinner],[class*=Spinner],[class*=progress],[class*=Progress],[class*=uploading],[class*=Uploading]')).filter(visible):[];"
"var errRe=/(失敗|失败|錯誤|错误|不支援|不支持|格式|大小|尺寸|超過|超过|重複|重复|duplicate|error|fail|invalid|unsupported)/i;"
"var uploadRe=/(圖片|图片|封面|照片|相片|圖像|图像|image|photo|cover|upload|上傳|上传|檔案|文件|file|格式|大小|尺寸|像素|解析度|分辨率|超過|超过|重複|重复|duplicate)/i;"
@@ -80,11 +134,13 @@ JS_UPLOAD_STATE = (
"var errorToasts=pageErrorToasts.filter(function(t){return uploadRe.test(t);}).slice(0,8);"
"return JSON.stringify({"
"manager_exists:!!manager,"
"image_field_exists:image.root_exists,"
"count:items.length,"
"srcs:imgs.slice(0,12),"
"blob_count:imgs.filter(function(s){return /^blob:/.test(s);}).length,"
"cdn_count:imgs.filter(function(s){return /susercontent/.test(s);}).length,"
"upload_input_exists:!!upload,"
"upload_input_count:manager?manager.querySelectorAll('.shopee-image-manager__upload input[type=file]').length:0,"
"upload_input_disabled:!!(upload&&(upload.disabled||upload.getAttribute('aria-disabled')==='true')),"
"upload_tile_visible:visible(uploadBox),"
"busy_count:busy.length,"
@@ -99,10 +155,9 @@ JS_UPLOAD_STATE = (
JS_CLICK_UPLOAD_TILE = (
"(function(){"
"function visible(e){if(!e)return false;var r=e.getBoundingClientRect();var s=getComputedStyle(e);"
"return r.width>0&&r.height>0&&s.visibility!=='hidden'&&s.display!=='none';}"
"var manager=document.querySelector('.shopee-image-manager');"
"var box=manager?manager.querySelector('.shopee-image-manager__upload,[class*=image-manager__upload]'):null;"
+ JS_EDITOR_FIELD_HELPERS
+ "function visible(e){return cmVisible(e);}"
"var image=cmMainImageContext();var box=cmMainUploadBox(image);"
"if(!box)return JSON.stringify({clicked:false,reason:'NO_UPLOAD_TILE'});"
"if(!visible(box))return JSON.stringify({clicked:false,reason:'UPLOAD_TILE_HIDDEN'});"
"box.scrollIntoView({block:'center',inline:'center'});"
@@ -112,18 +167,41 @@ JS_CLICK_UPLOAD_TILE = (
)
JS_TITLE_STATE = (
"(function(){"
f"var r=document.evaluate({json.dumps(TITLE_XPATH)},document,null,"
"XPathResult.FIRST_ORDERED_NODE_TYPE,null);"
"var el=r.singleNodeValue;if(!el)return null;"
"return JSON.stringify({value:el.value,modelvalue:el.getAttribute('modelvalue')});})()"
+ JS_EDITOR_FIELD_HELPERS
+ "var title=cmTitleCandidate();var el=title.node;"
"return JSON.stringify({found:!!el,candidate_count:title.count,source:title.source,"
"field_exists:title.root_exists,value:el?el.value:null,"
"modelvalue:el?el.getAttribute('modelvalue'):null});})()"
)
JS_FIRST_COVER = (
"(function(){"
f"var s=document.evaluate({json.dumps(ITEMBOX_XPATH)},document,null,"
"XPathResult.ORDERED_NODE_SNAPSHOT_TYPE,null);"
"if(!s.snapshotLength)return null;"
"var img=s.snapshotItem(0).querySelector('img');return img?img.src:null;})()"
+ JS_EDITOR_FIELD_HELPERS
+ "var image=cmMainImageContext();var items=cmMainImageItems(image);"
"if(!items.length)return null;"
"var img=items[0].querySelector('img');return img?img.src:null;})()"
)
JS_SCROLL_MAIN_IMAGE_MANAGER = (
"(function(){"
+ JS_EDITOR_FIELD_HELPERS
+ "var image=cmMainImageContext();var manager=image.node;"
"if(manager)manager.scrollIntoView({block:'center'});return !!manager;})()"
)
JS_MAIN_UPLOAD_INPUT = (
"(function(){"
+ JS_EDITOR_FIELD_HELPERS
+ "var image=cmMainImageContext();return cmMainUploadInput(image);})()"
)
JS_DISPATCH_MAIN_UPLOAD_INPUT_EVENTS = (
"(function(){"
+ JS_EDITOR_FIELD_HELPERS
+ "var image=cmMainImageContext();var up=cmMainUploadInput(image);"
"if(!up)return false;"
"up.dispatchEvent(new Event('input',{bubbles:true}));"
"up.dispatchEvent(new Event('change',{bubbles:true}));return true;})()"
)
JS_FIND_UPDATE = (
@@ -229,7 +307,11 @@ JS_PAGE_TOASTS = (
"function record(e){var root=rootOf(e);var value=text(root)||text(e);if(!value)return null;return {text:value,html:((root.outerHTML||e.outerHTML)||'').slice(0,2000),url:location.href,visible:visible(root)||visible(e),created_at:(new Date()).toISOString()};}"
"var cached=(window.__cmshopee_toasts||[]).slice(-10);"
"var current=Array.from(document.querySelectorAll('.eds-toasts .eds-toast,.eds-toast,.eds-toast__content,[role=alert],[class*=toast],[class*=Toast],[class*=message],[class*=Message],[class*=notice]')).map(record).filter(Boolean);"
"var seen={};var out=[];cached.concat(current).forEach(function(item){var key=[item.text,item.html,item.url].join('|');if(seen[key])return;seen[key]=true;out.push(item);});"
"var positions={};var out=[];"
"cached.forEach(function(item){var copy=Object.assign({},item,{visible:false});var key=[copy.text,copy.url].join('|');"
"if(positions[key]!==undefined){out[positions[key]]=copy;return;}positions[key]=out.length;out.push(copy);});"
"current.forEach(function(item){var key=[item.text,item.url].join('|');"
"if(positions[key]!==undefined){out[positions[key]]=item;return;}positions[key]=out.length;out.push(item);});"
"return JSON.stringify(out.slice(-10));"
"})()"
)
@@ -255,10 +337,10 @@ JS_POST_UPDATE_STATE = (
JS_CLICK_FIRST_DELETE = (
"(function(){"
f"var s=document.evaluate({json.dumps(ITEMBOX_XPATH)},document,null,"
"XPathResult.ORDERED_NODE_SNAPSHOT_TYPE,null);"
"if(!s.snapshotLength)return JSON.stringify({clicked:false,reason:'NO_IMAGE'});"
"var item=s.snapshotItem(0);"
+ JS_EDITOR_FIELD_HELPERS
+ "var image=cmMainImageContext();var items=cmMainImageItems(image);"
"if(!items.length)return JSON.stringify({clicked:false,reason:'NO_IMAGE'});"
"var item=items[0];"
"['mouseover','mouseenter','mousemove'].forEach(function(n){"
"item.dispatchEvent(new MouseEvent(n,{bubbles:true,view:window}));});"
"var selectors=['.shopee-image-manager__icon--delete','[class*=\"icon--delete\"]','[class*=\"delete\"]'];"
@@ -456,15 +538,29 @@ def _product_unavailable_toast_text(toasts):
return None
def _open_product_ready_error(cdp, fallback):
def _same_page_url(left, right):
try:
left_url = urlparse(str(left or ""))
right_url = urlparse(str(right or ""))
except Exception:
return False
return bool(
left_url.netloc
and left_url.netloc == right_url.netloc
and left_url.path.rstrip("/") == right_url.path.rstrip("/")
)
def _open_product_ready_error(cdp, fallback, current_url=""):
toasts = read_page_toasts(cdp)
invalid_text = _product_unavailable_toast_text(toasts)
if invalid_text:
return product_unavailable_error_message(invalid_text)
page_url = str(current_url or _current_url(cdp) or "")
for toast in reversed(toasts):
text = str(toast.get("text") or "").strip()
if text:
return f"{fallback}:{text}"
if text and toast.get("visible") and _same_page_url(toast.get("url"), page_url):
return f"{fallback};页面提示(可能无关):{text}"
return fallback
@@ -472,6 +568,58 @@ def _title_state(cdp):
return _json_value(cdp, JS_TITLE_STATE, default={}) or {}
def _ready_state(cdp):
raw = cdp.val(JS_READY)
if isinstance(raw, bool):
return {"ready": raw}
if isinstance(raw, dict):
return raw
if raw in (None, ""):
return {}
parsed = json.loads(raw)
return parsed if isinstance(parsed, dict) else {}
def _ready_error_message(fallback, state):
if not state:
return fallback
details = []
title_count = state.get("title_count")
if title_count == 1:
details.append("商品名称输入框正常")
elif isinstance(title_count, int) and title_count > 1:
details.append(f"商品名称输入框匹配到{title_count}个,无法确定唯一字段")
elif title_count is not None:
details.append("未识别商品名称输入框")
image_count = state.get("image_count")
if not state.get("image_manager_exists") and "image_manager_exists" in state:
details.append("未识别商品主图区域")
elif isinstance(image_count, int) and image_count > 0:
details.append(f"商品图片{image_count}张")
elif image_count is not None:
details.append("商品图片尚未加载")
upload_count = state.get("upload_input_count")
if upload_count == 1:
details.append("主图上传入口正常")
elif isinstance(upload_count, int) and upload_count > 1:
details.append(f"主图上传入口匹配到{upload_count}个,无法确定唯一入口")
elif "upload_input_exists" in state or upload_count is not None:
details.append("主图上传入口尚未加载")
current_url = str(state.get("current_url") or "")
if current_url:
try:
parsed = urlparse(current_url)
current_url = f"{parsed.scheme}://{parsed.netloc}{parsed.path}"
except Exception:
pass
details.append(f"当前页面:{current_url}")
return f"{fallback}:{';'.join(details)}" if details else fallback
def _image_rects(cdp):
return _json_value(cdp, JS_RECTS, default=[]) or []
@@ -644,13 +792,14 @@ def _cover_upload_error_message(result):
def _wait_ready(cdp, timeout=60):
end = time.time() + timeout
while time.time() < end:
last_state = {}
first_probe = True
while first_probe or time.time() < end:
first_probe = False
try:
if cdp.val(JS_READY):
cdp.val(
"(function(){var m=document.querySelector('.shopee-image-manager');"
"if(m)m.scrollIntoView({block:'center'});return 1;})()"
)
last_state = _ready_state(cdp)
if last_state.get("ready"):
cdp.val(JS_SCROLL_MAIN_IMAGE_MANAGER)
time.sleep(0.5)
return True
except Exception:
@@ -658,8 +807,17 @@ def _wait_ready(cdp, timeout=60):
invalid_text = _product_unavailable_toast_text(read_page_toasts(cdp))
if invalid_text:
raise EditorError(product_unavailable_error_message(invalid_text))
if time.time() >= end:
break
time.sleep(1)
raise EditorError(_open_product_ready_error(cdp, "等待蝦皮商品编辑器就绪超时"))
fallback = _ready_error_message("等待蝦皮商品编辑器就绪超时", last_state)
raise EditorError(
_open_product_ready_error(
cdp,
fallback,
current_url=last_state.get("current_url") if last_state else "",
)
)
def _ensure_page_domains(cdp):
@@ -1073,9 +1231,9 @@ def change_title(cdp, new_title) -> dict:
expr = (
"(function(){"
f"var r=document.evaluate({json.dumps(TITLE_XPATH)},document,null,"
"XPathResult.FIRST_ORDERED_NODE_TYPE,null);var el=r.singleNodeValue;"
"if(!el)return 'NO_INPUT';"
+ JS_EDITOR_FIELD_HELPERS
+ "var title=cmTitleCandidate();var el=title.node;"
"if(!el)return title.count>1?'AMBIGUOUS_INPUT':'NO_INPUT';"
"var s=Object.getOwnPropertyDescriptor(window.HTMLInputElement.prototype,'value').set;"
f"el.focus();s.call(el,{json.dumps(str(new_title))});"
"el.dispatchEvent(new Event('input',{bubbles:true}));"
@@ -1087,7 +1245,14 @@ def change_title(cdp, new_title) -> dict:
value = state.get("value")
modelvalue = state.get("modelvalue")
ok = value == str(new_title) and modelvalue == str(new_title)
return {"ok": ok, "written": written, "value": value, "modelvalue": modelvalue}
return {
"ok": ok,
"written": written,
"value": value,
"modelvalue": modelvalue,
"candidate_count": state.get("candidate_count"),
"source": state.get("source"),
}
def replace_cover(cdp, image_win_path, old_cover_path=None, timeout=180) -> dict:
@@ -1101,10 +1266,7 @@ def replace_cover(cdp, image_win_path, old_cover_path=None, timeout=180) -> dict
if not os.path.exists(image_win_path):
raise FileNotFoundError(f"封面图片不存在: {image_win_path}")
cdp.val(
"(function(){var m=document.querySelector('.shopee-image-manager');"
"if(m)m.scrollIntoView({block:'center'});return !!m;})()"
)
cdp.val(JS_SCROLL_MAIN_IMAGE_MANAGER)
time.sleep(0.5)
stable = _wait_image_manager_stable(
cdp,
@@ -1172,7 +1334,7 @@ def replace_cover(cdp, image_win_path, old_cover_path=None, timeout=180) -> dict
"upload_state": _upload_state(cdp),
}
time.sleep(1)
oid = cdp.object_id("document.querySelector('.shopee-image-manager__upload input[type=file]')")
oid = cdp.object_id(JS_MAIN_UPLOAD_INPUT)
if not oid:
return {
"ok": False,
@@ -1184,12 +1346,7 @@ def replace_cover(cdp, image_win_path, old_cover_path=None, timeout=180) -> dict
}
cdp.send("DOM.setFileInputFiles", {"objectId": oid, "files": [image_win_path]})
changed = cdp.val(
"(function(){var up=document.querySelector('.shopee-image-manager__upload input[type=file]');"
"if(!up)return false;"
"up.dispatchEvent(new Event('input',{bubbles:true}));"
"up.dispatchEvent(new Event('change',{bubbles:true}));return true;})()"
)
changed = cdp.val(JS_DISPATCH_MAIN_UPLOAD_INPUT_EVENTS)
time.sleep(2)
new_src = None
last_state = _upload_state(cdp)