support brand checking of contact #404

JasLin · 2017-04-12T07:57:08Z

@see https://github.com/Urinx/WeixinBot/blob/master/README.md

公众号/服务号	以@开头，但其VerifyFlag & 8 != 0 

VerifyFlag: 
         一般公众号/服务号：8 
         微信自家的服务号：24 
         微信官方账号微信团队：56

upgrade to latest code

…m, 2. some times we can got owneruin or ChatRoomOwner . as those reason, first member from memberlist should not be here any more

merge base

coveralls · 2017-04-12T08:20:24Z

Coverage decreased (-0.03%) to 54.759% when pulling 0e60d4e on JasLin:master into 7fcb128 on Chatie:master.

huan

Good work bro, It seems like a really tricky magic number inside.

LGTM and please get at least 3 approvements from our contributors before merge for enhancing our community communication.

Thanks!

huan · 2017-04-12T10:08:08Z

tslint.json

@@ -36,7 +36,7 @@
    "member-ordering": [false],
    "no-any": false,
    "no-arg": true,
-    "no-bitwise": true,


I'd like to keep this because the bitwise operation is very rare in our code.

yes, u r right:)

huan · 2017-04-12T10:09:06Z

src/contact.ts

+       * @see 1. https://github.com/nodeWechat/wechat4u/blob/master/src/interface/contact.js#L65
+       * @see 2. https://github.com/Urinx/WeixinBot/blob/master/README.md
+       */
+      brand:      !!rawObj.UserName && !rawObj.UserName.startsWith('@@') && (rawObj.VerifyFlag & 8) !== 0,


I'd like to use this line to disdable tslint temporary

// tslint:disable-next-line

// tslint:disable-next-line

will disables all rules for the following lines, it seems it's not a good choose for this case.

how about this?

!!rawObj.UserName && !rawObj.UserName.startsWith('@@') && [8, 24, 56].indexOf(rawObj.VerifyFlag) > -1,

I thought that will only disable the following ONE line?

Could you please have a check about it, because the new version of your code is difficult to read...

yes, ...next-line... u r right:)

huan · 2017-04-12T10:12:36Z

src/contact.ts

+   * ```
+   */
+  public brand(): boolean|null {
+    if (!this.obj) return null


Our framework should always to make sure a contact is ready, which means this.obj should not be empty.

If so, it's more like a bug and should throw an exception.

So my suggestion is to change this function to:

public brand(): boolean { return this.obj && this.obj.brand }

public brand(): boolean { return !!this.obj && this.obj.brand }

Yes your code is correct. :)

huan · 2017-04-12T10:14:16Z

example/contact-bot.ts

+   */
+  for (let i = 0; i < contactList.length; i++) {
+    const contact = contactList[i]
+    if (contact.brand() === null) {


The contact here should already ready()-ed, because Wechaty should make sure it is.

If it is not, then Wechaty must have a bug in somewhere...

JasLin · 2017-04-12T11:51:47Z

@lijiarui @Gcaufy @xinbenlv @dcsan @imWildCat @mukaiu @Samurais
hi guys , anybody have time to review this feature?

…se to false. 2. contact is ready always

coveralls · 2017-04-12T15:04:19Z

Coverage decreased (-0.02%) to 54.775% when pulling 33b7e4a on JasLin:master into 7fcb128 on Chatie:master.

coveralls · 2017-04-13T04:51:04Z

Coverage decreased (-0.02%) to 54.775% when pulling 0b570d8 on JasLin:master into b225722 on Chatie:master.

coveralls · 2017-04-13T04:51:04Z

Coverage decreased (-0.02%) to 54.775% when pulling 0b570d8 on JasLin:master into b225722 on Chatie:master.

lijiarui

Cool!

Btw, could you help to review the add mention and room leave event feature for me ? Thanks!

JasLin · 2017-04-13T14:16:09Z

@lijiarui thanks , i am trying to review them

lijiarui · 2017-04-14T12:24:21Z

@dcsan I guess maybe you need this feature most, could you help to review this please? Thanks

dcsan · 2017-04-16T10:59:32Z

@lijiarui I'm not sure what this feature does?

Check if it's a brand contact

what is a brand in this context? It usually means a product brand. Maybe you mean brand new ?
Sorry but I'm unclear.

lijiarui · 2017-04-16T11:23:09Z

@dcsan brand here means wechat oa account

Generally ,wechaty cannot identify the contact is a real contact or a wechat oa account, but this pr can fix this problem.

@JasLin
I suggest to change brand to wechat oa here.

huan · 2017-04-16T11:28:02Z

Tencent has this title at https://admin.wechat.com/:
Wechat Official Account Admin Platform

I suggest we might consider a rename, like from brand() to official()?
Another thing is: should we include the 24 and 56 type in our brand() / official()(currently we only check for 8)? I think we might include them because then we can use !contact.official() to identify a personal user account.

VerifyFlag:
一般公众号/服务号：8
微信自家的服务号：24
微信官方账号微信团队：56

dcsan · 2017-04-17T02:42:58Z

ok mingbai! so this is in the case where chatie bot is following an official account.
I'm not clear exactly what the use case would be for that - maybe testing other OAs?

so we have:

msg.room() - if it is a chat GROUP or a 1:1 (DM) chat
msg.brand() - if it's an OfficialAccount or ( a normal user OR group )

then service accounts, subscription accounts, personal OAs - not sure if the wechaty client gets any information about the type of brand/ official account, or if it matters.

msg.contact() - isn't really relevant for groups. maybe msg.sender() ?

then we could have a msg.sender().type() that returns

brand | user | group

as a string?

If we use enums btw I think we need to make these more easily exportable/importable to work with JS. Now we have async/await in node7 I've switched back to JS myself for most new projects.

huan · 2017-04-17T04:16:11Z

@dcsan IMO stay with TypeScript is the bright decision, I love it to death now. :P

And for this VerifyFlag I think the most useful part is to identify if a contact is a personal account.

Now we can get all contact after login by Contact.findAll(), but the contactList is a mix of personal users and official accounts.

Add a contact.official() could sperate them easily by:

if (contact.official()) {
  // is Official Account
} else {
  // is Personal Account
}

Or

let contactList = await Contact.findAll()

let officialAccountList = contactList.filter(c => c.official())
let personalAccountList = contactList.filter(c => !c.officlal())

dcsan · 2017-04-17T05:00:38Z

I think it's also useful to know if its a group(room) or an individual user, as some of the behavior is different, eg I don't think there is a contact for a room? (and not sure if there is a msg.room() or a topic() for an individual user. So you have to be careful with the logic or code can break, the types are not polymorphic.

I think i have stuff like:

topic = msg.room() ? msg.room().topic() : 'dm'

So when you have three major possible types, a boolean check method isn't as useful?

Personally I like Typescript a lot, but it is hard just getting new developers up and running, especially we have some team in Russia and other places. typescript, tslint, wechat api, docker... hours of setup to even get to helloworld.

huan · 2017-04-17T08:37:48Z

Yes, I agree with you that TypeScript has a huge study curve. It seems things getting better after TypeScript v2.3 because they became treating unknown things to be any by default instead of throw errors, which will make JS developer easier to getting start.

And I do not fully understand you, as in the reply you said:

to know if its a group(room) or an individual user

To clarify, In Wechaty:

room(group) will be an instance of Room and,
contact(individual user) will be an instance of Contact. Room and Contact are totally two different Class(abstract);
There is no contact for a room. Contact is for one, Room is for many;
There is a message.room(): Room method to identify which room this message comes from;
There is no contact.topic() because that's meaningless;

For your code:

topic = msg.room() ? msg.room().topic() : 'dm'

I'd like to rewrite it to:

const room = msg.room()
let topic: string
if (room) { 
  // this message is received in a room
  assert(room instanceof Room)
  topic = room.topic()
} else {
  // this message is received from a indivial user

  // WHY do you set room topic 
  // when you are receiving a direct message 
  // from an individual user?
  topic = 'dm' // ???
}

dcsan · 2017-04-17T09:21:07Z

hi - thanks for the above.

    topic = msg.room() ? msg.room().topic() : 'dm'
    // WHY do you set room topic 
    // when you are receiving a direct message 
    // from an individual user?

yeah, it's a bit hacky agreed. it is so we can render all messages and handle group|dm display based on a single (cached) property rather than another chain of method calls.
But they do still have a different class hierarchy so it's not that correct.
Now we have to check two methods:

if ( msg.room() === null && msg.brand() === null ) then it's a DM

If we had msg.type() we could do:

switch (msg.type() ):
   case 'room':
      // responding in a group
   case 'dm':
     //..
   case 'oa':
    // brand / official account

Just a thought. This maybe still a hacky way to do this, as they are still different types of objects. Perhaps we'd be better with something like the Message < MediaMessage hierarchy.

re TypeScript, I think the language is easy.
it's all the setup of the tooling, linter, compiler (global tsc? compile output to where? .tsconfig settings?) especially for new apps. We are trying to break out more into microservices now, so we have a few smaller apps. this creates more setup overhead with TS.

huan · 2017-04-20T05:31:03Z

@JasLin Hi bro,

I suggest:

we rename brand() to official() and
include all the VerifyFlag(8/24/56) and
launch this PR for now.

Because we could use this method to identify whether the contact is an individual personal user, future design/changes could be discussed later.

How do you think?

JasLin · 2017-04-20T15:43:17Z

@zixia sorry check this so later，brand() is come from wxwebapp source code ，but i love official() too.

I will try to finish this tomorrow. And, happy birthday bro!:)

JasLin · 2017-04-21T05:12:10Z

@zixia Hi bro, I just check the wxwebapp's source and the rawObj again . the following info is confirmed

we don't need to include all verifyFlag , use !!(VerifyFlag & 8) can identify all brand/offical account , check isBrandContact method from source
VerifyFlag should be define like this exactly:

VerifyFlag:
~~一般公众号/服务号：8~~ 个人公众号: 8
~~微信自家的服务号：24~~ 企业公众号： 24
微信官方账号微信团队：56

and two more thing :

should we add a method personal() {return !official()}to identify personal() account for convenient?
should we add special() to identify special account ? check the source isSpContact

coveralls · 2017-04-21T08:37:16Z

Coverage increased (+0.7%) to 54.869% when pulling a48e386 on JasLin:master into 93b1704 on Chatie:master.

coveralls · 2017-04-21T08:54:15Z

Coverage increased (+0.7%) to 54.897% when pulling a48e386 on JasLin:master into 93b1704 on Chatie:master.

huan · 2017-04-21T10:04:48Z

VerifyFlag:
个人公众号: 8
企业公众号： 24
微信官方账号微信团队：56

I think we could translate them into bit position:

Type	10	2	Bit
Official Account	8	0000 1000	1 << 4
Business Verified	24	0001 1000	1 << 5
Tencent Verified	56	0011 1000	1 << 6

How about let's redesign official() like this:

enum OfficialType {
  OFFICIAL,
  BUSINESS,
  TENCENT,
}

official(type = OfficialType.OFFICIAL): boolean

By default, official() will return whether the contact is an official account(1<<4 bit set). If we want to know whether it's a business verified, we could call official(OfficialType.BUSINESS) or official(OfficialType.TENCENT), and we could extend to more types for the official account if needed in the future.

~~Abut the personal() method, I have no feeling about it. :P~~
~~About the isSpContact(), I have no idea about what it is. I also see other names isShieldContact() etc, I think we could talk about them in another issue?~~

Update: I saw you have already implemented the personal() & special(), it's ok for me, we can landing it first. :)

huan

Please use https://github.com/Chatie/webwx-app-tracker to reference web wechat source.

Thanks!

huan · 2017-04-21T17:08:25Z

src/contact.ts

@@ -61,6 +64,16 @@ export type ContactQueryFilter = {
 }

 /**
+ * @see https://github.com/Chatie/wechaty/blob/master/doc/webwxapp.js#L4057


Please do not use wechaty/doc, because this directory is deprecated.

Should change refer to https://github.com/Chatie/webwx-app-tracker with the exactly commit hash.

huan · 2017-04-21T17:09:11Z

src/contact.ts

@@ -110,6 +123,16 @@ export class Contact implements Sayable {
      star:       !!rawObj.StarFriend,
      stranger:   !!rawObj.stranger, // assign by injectio.js
      avatar:     rawObj.HeadImgUrl,
+      /**
+       * @see 1. https://github.com/Chatie/wechaty/blob/master/doc/webwxapp.js#L3368


Please change refer to https://github.com/Chatie/webwx-app-tracker with the exactly commit hash.

huan · 2017-04-21T17:09:32Z

src/contact.ts

+      // tslint:disable-next-line
+      official:      !!rawObj.UserName && !rawObj.UserName.startsWith('@@') && !!(rawObj.VerifyFlag & 8),
+      /**
+       * @see 1. https://github.com/Chatie/wechaty/blob/master/doc/webwxapp.js#L4187


Please change refer to https://github.com/Chatie/webwx-app-tracker with the exactly commit hash.

huan · 2017-04-21T17:10:00Z

src/contact.ts

+   * 'masssendapp', 'meishiapp', 'feedsapp', 'voip', 'blogappweixin', 'weixin', 'brandsessionholder',
+   * 'weixinreminder', 'wxid_novlwrv3lqwv11', 'gh_22b87fa7cb3c', 'officialaccounts', 'notification_messages',
+   * ```
+   * @see https://github.com/Chatie/wechaty/blob/master/doc/webwxapp.js#L4057


Please change refer to https://github.com/Chatie/webwx-app-tracker with the exactly commit hash.

had changed it with the latest commit

JasLin · 2017-04-22T15:40:29Z

actually , special contacts contain most of Tencent contacts , but i haven't checked that whether all the special|Tencent contact's verifyFlag is 56, because i can not find out a way to load all those contacts to my WeChat.
and , check the source code of wxwebapp, there is no verifyFlag defined as 24 and 56
, so i guess that maybe Urinx's analysis about 24 and 56 is just a guess too:) , as u see in the document ,he think 微信自家的服务号 's flag is 24 , but in my bot, I found that most of 企业公众号's flag is 24 instead. since that , i am not sure about this exactly. but we can sure about that using !!(VerifyFlag & 8) can identify all brand/offical contacts

here are all flags defined in source code:

            CONTACTFLAG_CONTACT: 1,
            CONTACTFLAG_CHATCONTACT: 2,
            CONTACTFLAG_CHATROOMCONTACT: 4,
            CONTACTFLAG_BLACKLISTCONTACT: 8,
            CONTACTFLAG_DOMAINCONTACT: 16,
            CONTACTFLAG_HIDECONTACT: 32,
            CONTACTFLAG_FAVOURCONTACT: 64,
            CONTACTFLAG_3RDAPPCONTACT: 128,
            CONTACTFLAG_SNSBLACKLISTCONTACT: 256,
            CONTACTFLAG_NOTIFYCLOSECONTACT: 512,
            CONTACTFLAG_TOPCONTACT: 2048,
            ... ...
            MM_USERATTRVERIFYFALG_BIZ: 1
            , MM_USERATTRVERIFYFALG_FAMOUS: 2
            , MM_USERATTRVERIFYFALG_BIZ_BIG: 4
            , MM_USERATTRVERIFYFALG_BIZ_BRAND: 8
            , MM_USERATTRVERIFYFALG_BIZ_VERIFIED: 16

so I suggest we design official() at this time , after we find out what are 24, 56 exactly means , then we can change office() to official(type?): boolean.

personal() is just a convenient method for identifying individual personal contacts. this method maybe use more often than official()

coveralls · 2017-04-22T16:09:52Z

Coverage increased (+0.7%) to 54.869% when pulling 68ebd91 on JasLin:master into 93b1704 on Chatie:master.

huan

Hi @JasLin ,

I hope you have a good weekend!

Your new code looks good for me, and I agree with you about the magic number and method name opinion.

Please get 3 approving from our @Chatie/contributor before I could merge it.

Thanks!

JasLin · 2017-04-23T04:08:16Z

it seems we need more contributors ...

huan · 2017-04-23T04:22:08Z

Yes, I definitely agree with you bro... @_@

JasLin and others added 6 commits March 24, 2017 17:13

Merge pull request #5 from Chatie/master

62dc38b

upgrade to latest code

1. in wxwebapp,first item in memberList is a random user from the roo…

22d6c39

…m, 2. some times we can got owneruin or ChatRoomOwner . as those reason, first member from memberlist should not be here any more

Merge pull request #6 from Chatie/master

d6914b0

merge base

brand checking support

3ff7f85

example(test) for brand checking

9dac23a

allow bitwise operate for brand checking

0e60d4e

huan requested review from dcsan, xinbenlv, Gcaufy, imWildCat, hailiang-wang, lijiarui and mukaiu April 12, 2017 10:04

huan approved these changes Apr 12, 2017

View reviewed changes

JasLin added 2 commits April 12, 2017 22:20

1. use tslint:disable-next-line in source insteadof changing no-bitwi…

5e1ea2a

…se to false. 2. contact is ready always

contact is always ready thanks to framework

33b7e4a

merge upstream

0b570d8

lijiarui approved these changes Apr 13, 2017

View reviewed changes

JasLin mentioned this pull request Apr 14, 2017

Add mention #362

Merged

JasLin added 4 commits April 21, 2017 16:13

identify contact as official|special|personal

d8069e7

example of identify official|special|personal account

f2a3738

merge upstream

c44f67c

code clean

a48e386

huan requested changes Apr 21, 2017

View reviewed changes

change source refers to a exactly commit hash

68ebd91

huan approved these changes Apr 22, 2017

View reviewed changes

huan merged commit 54d199f into wechaty:master Apr 25, 2017

support brand checking of contact #404

support brand checking of contact #404

Conversation

JasLin commented Apr 12, 2017

coveralls commented Apr 12, 2017 • edited

huan left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

JasLin commented Apr 12, 2017 • edited

coveralls commented Apr 12, 2017 • edited

coveralls commented Apr 13, 2017

coveralls commented Apr 13, 2017

lijiarui left a comment

Choose a reason for hiding this comment

JasLin commented Apr 13, 2017

lijiarui commented Apr 14, 2017

dcsan commented Apr 16, 2017

lijiarui commented Apr 16, 2017 • edited

huan commented Apr 16, 2017 • edited

dcsan commented Apr 17, 2017 • edited

huan commented Apr 17, 2017 • edited

dcsan commented Apr 17, 2017

huan commented Apr 17, 2017

dcsan commented Apr 17, 2017

huan commented Apr 20, 2017 • edited

JasLin commented Apr 20, 2017

JasLin commented Apr 21, 2017

coveralls commented Apr 21, 2017

coveralls commented Apr 21, 2017

huan commented Apr 21, 2017 • edited

huan left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

JasLin Apr 22, 2017 • edited

Choose a reason for hiding this comment

JasLin commented Apr 22, 2017 • edited

coveralls commented Apr 22, 2017

huan left a comment

Choose a reason for hiding this comment

JasLin commented Apr 23, 2017

huan commented Apr 23, 2017 • edited

coveralls commented Apr 12, 2017 •

edited

JasLin commented Apr 12, 2017 •

edited

coveralls commented Apr 12, 2017 •

edited

lijiarui commented Apr 16, 2017 •

edited

huan commented Apr 16, 2017 •

edited

dcsan commented Apr 17, 2017 •

edited

huan commented Apr 17, 2017 •

edited

huan commented Apr 20, 2017 •

edited

huan commented Apr 21, 2017 •

edited

JasLin Apr 22, 2017 •

edited

JasLin commented Apr 22, 2017 •

edited

huan commented Apr 23, 2017 •

edited