zimg TODO list about zimg HOT 47 OPEN

沙发自己坐了。

@欧飞程勇 提了两个需求：
1.空请求返回一个欢迎界面。（Done）
2.upload成功后返回xml或json风格的结果，而不是http格式。（Done）

from zimg.

Can zimg bind on a IP?

from zimg.

Can zimg bind on a IP?

具体是指？

from zimg.

1.zimg 支持lua脚本，配置和压图处理（Done）
2.支持ssdb存储，具备在线备份能力（Done）
3.上传逻辑优化（Done）
4.服务器系统信息通过lua文件获取，在HTTP包头中返回（Done）

from zimg.

支持更多的img处理模式，比如可以缩放时不是补边而是裁边的方式。

from zimg.

@momoplan 收到。引入lua脚本就是为了解决图片处理逻辑多样的问题，待存储稳定之后来做这个。

from zimg.

是不是应该有 remote image cache 等功能？

from zimg.

@viky 提出的需求：

1. 访问权限控制。（Done）
2. 不含参数的请求返回被压缩过的图片节省流量，原始图片使用p=0参数来获取。（Done）
3. 删除图片功能。（Done）

from zimg.

只返回md5吗
如果hash冲突了 图片会错乱吧 可否增加原始文件名或大小等数据.

from zimg.

@Smallthing 理论上说如果两张图片算出来的md5相同，后上传的会覆盖掉先上传的图片。
但是两张图片md5相同的概率太小了（1 / 2^64），普遍认为md5碰撞是不可能发生的，所以大可不必担心。
最后，增加文件名的困难在于需要增加一个存储文件名到图片数据的映射关系，这会大大降低查询速度，所以不会加入。图片原始大小可以通过拉原始图片获取。

from zimg.

1. @hunter邓 提出浏览器缓存功能。（Done）
2. 自定义返回header功能。（Done）

from zimg.

@buaazp
1.求编译打包后的文件比如rpm,deb文件，直接安装，因为我编译了好多次都没成功
2.提供更加现代的安装测试方式，比如 将zimg安装到docker中，这样就会有更多的人尝试使用

from zimg.

@buaazp
png文件在上传后压缩变为jpeg，原来背景透明部分变为黑色，能不能通过配置修改把png透明部分变为别的颜色，比如白色？谢谢

from zimg.

@adouggy
把你的图片上传到 http://demo.buaa.us/ 看一下还有没有这种问题，如果有的话麻烦把这张图片的MD5留言给我。

如果没问题的话，就等等新版本发布即可，新版本使用了新的图片处理引擎。

from zimg.

@buaazp
好的，谢谢。比如这张图：ef9a6b4fcb21839e8ed92a7348471ff0，如果不加p=0就变成一张同样的颜色了。
但是有些png背景就变成了白色。但有些背景会变成黑色。

from zimg.

@buaazp
先前有人提出过：Can zimg bind on a IP?
我觉得这个也有必要，比如能够绑定内网或者环回地址，这样只能够从nginx反向代理过去，对外不开端口会增加安全性。比如mysql的bind-address，或者tomcat、redis之类都有类似的功能。

from zimg.

@adouggy

你说的都可以在3.0版本里解决：

png图片问题：这个貌似是png转jpeg的时候无法避免的问题，所以可以设置默认输出图片格式为原始格式，这样就只会返回png
bind IP功能：有

from zimg.

@buaazp
太好了，3.0的milestone什么时候？
另外有没有专门讲解配置的文档？我可以系统的学习一下如何使用。
png转jpeg丢失掉透明度没有办法，如果zimg3.0能够提供指定透明部分颜色的选项就太好了。因为使用原图对于网络、客户端都开销太大了。

谢谢作者耐心细致的解答

from zimg.

@buaazp
还要打扰，传递图片会去除exif，这样图片的旋转信息就会丢失。
p=0可以获取原图和原始信息。
有没有可能让压缩过的图片也保留部分exif信息？因为旋转等某些参数在应用环境下还是很有用的。
多谢 :)

from zimg.

请问下：为什么不支持VC++编译呢？若想把它部署在win服务器下怎么解决？

from zimg.

@wqvbjhc 不支持windows

from zimg.

@buaazp ,很期待这版的功能。加油

from zimg.

恭喜发布3.1！

from zimg.

Hi all, zimg v3 has been released. All these requirements below have done.

• @zz : Can zimg bind on a IP? (Done)
• @momoplan : 支持更多的img处理模式，比如可以缩放时不是补边而是裁边的方式。(Done)
• @zz : 是不是应该有 remote image cache 等功能？(Done)
• @qiutiaoming : 求编译打包后的文件比如rpm,deb文件 (Done)
• @adouggy : 有没有可能让压缩过的图片也保留部分exif信息？(Done)

Furthermore, zimg v3 has many other wonderful features in image processing. Goto zimg.buaa.us for more information.

from zimg.

Would like to have user verification for authorized end users only by
value verification, cookie, sending to another existing central server, if match, return true and allow upload, once successfully uploaded, return send the value of

1,5f189d8ec57f5a5a0d3dcba47fa797e2,29615

boolean (1 or 0, success or failure), filename, size

saving bandwidth on trip to server.

Configuration of file

central_server = zimg.domainname.com

I can contribute by writing the connecting api (on the central server) for verification of cookie in php if you can produce the above connector to the api.

I have read the source code. You write very nicely. Rather elegantly.

I have no idea how to read cookies with your program. If you can code a module on this part I can help you code php or perl API connector. Appreciate it.

I have one question to ask, the ssdb connector... can it be easily written to use hashset instead of SET as inside the c source code? just not sure how to use hset instead of the below.

Can i change this line to
redisReply r = (redisReply)redisCommand(c, "SET %s %b", cache_key, buff, len);

redisReply r = (redisReply)redisCommand(c, "HSET %s %s %b", cache_key1, cache_key2, buff, len); ??? Can you provide the two modifications for hset and hget for this zimg?

That's all I need to know. Thanks a lot!

int save_img_ssdb(redisContext* c, const char *cache_key, const char *buff, const size_t len)
{
if(c == NULL)
return -1;

redisReply *r = (redisReply*)redisCommand(c, "SET %s %b", cache_key, buff, len);
if( NULL == r)
{
    LOG_PRINT(LOG_DEBUG, "Execut ssdb command failure");
    return -1;
}
if( !(r->type == REDIS_REPLY_STATUS && strcasecmp(r->str,"OK")==0))
{
    LOG_PRINT(LOG_DEBUG, "Failed to execute save [%s] to ssdb.", cache_key);
    freeReplyObject(r);
    return -1;
}
freeReplyObject(r);
LOG_PRINT(LOG_DEBUG, "Succeed to save [%s] to ssdb. length = [%d].", cache_key, len);

return 1;

}

from zimg.

oh two more things...

1. image reduction (resize smaller) on browser side before sending
2. maximum file size in (kb) or dimension of file send resize on server side.

That's all

from zimg.

An interesting feature would be to monitor average RTT between the client and server; create a lower resolution version of any new requests, and cache them for later!

from zimg.

@ellzey That's great! Hope the new version released soon.

from zimg.

@buaazp interestingly enough; I'm refactoring a bit on zimg right now. It's something to do, and hopefully be a good sport about testing new libevent & evhtp features.

from zimg.

@ellzey It's so helpful to both zimg and evhtp!

from zimg.

@ellzey And I did some modifications in evhtp:

• add a global variable
• create and delete the pthread_key

I use this pthread_key in zscale.c. I think I should tell you this.

from zimg.

Not quite sure this is the right place for these things to go. But that can be figured out.

I would check out the current develop branch here: https://github.com/ellzey/libevhtp/tree/develop the evthr stuff has been super simplified and lots of people have been contributing cool stuff.

Only problem is, I'm trying to work on https://github.com/ellzey/libevhtp/tree/libevhtp2 and all of these additions and fixes are a pain to keep in sync with a completely different architecture that evhtp2 is.

from zimg.

大部分需求场景，例如：
原图尺寸：900x600
容器大小：300x300
需要展示全图，并进行按容器缩放。
Example 1: Get the image with width=300, and equal proportion.
1.1、处理后：300x200 [ok]
但是如果原图尺寸：600x900
1.2、处理后：300x450 [error]
Example 2: Get the image in 300_300 and the redundant part will discard.
处理后：300x300，但是图片被裁剪 [error]
Example 3: Get the center part of image in certain resolution
处理后： 图片被裁剪 [error]
Example 4: Get the resized image with percent
处理后： 图片被裁剪 [error]
Example 5: Get 300_300 image but the image will be stretched
处理后： 图片显示全图，但是被拉伸 [error]
Example 6: Get the certain part of the image
处理后： 图片被裁剪 [error]

而想要的是
需要判断图片的宽和高，以最大宽度或高度为最大值，进行等比缩放图片（不裁剪）
请求参数：max_width=300，max_height=300
请求url：http://demo.buaa.us/5f189d8ec57f5a5a0d3dcba47fa797e2?w=300&h=300&p=xxx
若图片尺寸：900 x 600，缩放后：300 x 200
若图片尺寸： 600 x 900，缩放后：200 x 300

望优先采纳，BTW，zimg很强大，期待越来越好(^__^)

from zimg.

@fanrong33

Now zimg supported max_size(via using p=4) method to meet your requirements.

Just pull the newest codes and rebuild them to test.

from zimg.

Hi there, I am writing a django storage for zimg, and to achieve a complete implementation, we need to be able to extract below information from a zimg hosted file:

1. created_time, the datetime when the image is uploaded to a zimg host
2. accessed_time, the datetime when the zimg hosted image is accessed (this could be trivial and not very necessary, entirely up to you and of course the zimg philosophy)

Please do consider this request

from zimg.

Actually, I do suggest to have standard resource philosphy based restful API in place for zimg.

All the image operations should be only available at below endpoints:

http://zimg.host/images/, HTTP method POST <-> Upload
http://zimg.host/images/, HTTP method GET <-> List hosted images (perhaps pagination is needed here)
http://zimg.host/images/, HTTP method HEAD <-> Return statistical information like total file counts, total file size, etc.
http://zimg.host/images/, HTTP method OPTIONS <-> Return a json to tell the client what the allowed verbs are (GET, POST, HEAD)

http://zimg.host/images/[md5], HTTP method GET <-> Download
http://zimg.host/images/[md5], HTTP method DELETE <-> Delete
http://zimg.host/images/[md5], HTTP method HEAD <-> Retrieve information
http://zimg.host/images/[md5], HTTP method OPTIONS <-> Return a json to tell the client what the allowed verbs are (GET, HEAD, DELETE)

And, for all above operations, make sure the appropriate HTTP status code is returned with the response. (404 for resource not found, 401 for admin operation denied due to insufficient permission, ...)

from zimg.

1 能否支持lua通过请求参数自定义保存路径及保存文件名？
2 能否支持lua通过请求参数自定义图片文件路径？

from zimg.

@sillydong 目前不能，这需求有点难

from zimg.

@buaazp 就像process中is_img函数一样，也定义一个函数将请求参数传递过去获取生成的保存路径呢？

from zimg.

@walnutist Good idea! The RESTful api will be implemented in the future.

from zimg.

@walnutist +1

200 is not the best status code for "File type is not supported! Upload Failed!"

Should I open a separate issue?

from zimg.

@niko

A full RESTful api will come with v4.0, so you needn't to open a new issue.

from zimg.

I heard a rumor evhtp is going away! What can I do to make that not happen?

from zimg.

@ellzey
Bad news! 😱what's happening?

from zimg.

nothing, I just saw a paper saying that
On Mon, Oct 19, 2015 at 8:52 PM 招牌疯子 [email protected] wrote:

@ellzey https://github.com/ellzey
Bad news! 😱what's happening?

—
Reply to this email directly or view it on GitHub
#5 (comment).

from zimg.

权限控制可以用的，但是如何配置多个ip呢
upload_rule = 'allow 192.168.59.*/24'
这样配置不生效
upload_rule = 'allow 192.168.59.11/24 192.168.59.12/24 '
这样也不行

还请大神给提示下

from zimg.

@hongjunsu 去开一个新的 issue，不要在这个主题下提。

from zimg.