Files
services/feeds/handler/crawl.go
Janos Dobronszki 40b71a9cf9 Autogenerate services.m3o.com (#37)
* Autogenerate services.m3o.com

* Openapi for all

* Gen

* Fix

* Whaat

* Fix dep

* Fix

* Hmm

* Install make

* Debug

* Debug 1

* Location -> locations

* Fix

* Intall protoc gen micro

* F

* F

* F

* Push

* Rename secret

* Fix npm install

* Fix script

* Fix v2

* Ignore errors

* Ignore v2

* F

* F

* F

* Docs index

* Add hugo theme

* Hugo tania fixes

* Change gen

* Change gen 2

* Install hugo

* Change gen

* Gen fix

* Change hugo install

* Change hugo install

* CNAME

* Change articles wording

* Tiny fix

* Fix gen

* Redoc it all

* Fix gen

* Fixing up protos

* Fix proto

* Fix gen

* Fix

* Trigger build

* Fix copy

* Openapi docs

* Flatten

* Changes

* No date vol2

* Changes

* Add make to chat

* Fixes

* Change

* api spec

* replace RSS

* fix link

* Dont continue on error

* increase the width

* use micro at master

* change box colours

* move some things

* Pushing new readmes to see how they look like

* Add skip file

* Readmes

* Nicer api link

* Remove stutter

* FIx mistake

* set service font weight

* Messages readme fix

* add other font bold

* Notes

* Remove post from url

* Revert "Remove post from url"

This reverts commit 5fea2c23d0bafa910f5dc4d4cc63f71f578530e3.

* move exampleSite to site

* replace exampleSite with site

* update readme

* use filename for post

* update index

* Add source urls

* set source as params

* set source as params

* Fix entries

* Generator in go

* Fixes to generator

* F

* Change doc gen

* FIx cname

* Fixing protos

* Change to makefiles

* Fix gen script

Co-authored-by: Asim Aslam <asim@aslam.me>
2021-01-19 16:59:25 +00:00

78 lines
1.6 KiB
Go

package handler
import (
"context"
"crypto/md5"
"fmt"
"net/url"
"github.com/SlyMarbo/rss"
log "github.com/micro/micro/v3/service/logger"
feeds "github.com/micro/services/feeds/proto"
posts "github.com/micro/services/posts/proto"
)
func (e *Feeds) fetchAll() {
fs := []*feeds.Feed{}
err := e.feeds.List(e.feedsNameIndex.ToQuery(nil), &fs)
if err != nil {
log.Errorf("Error listing feeds: %v", err)
return
}
if len(fs) == 0 {
log.Infof("No feeds to fetch")
return
}
for _, feed := range fs {
err = e.fetch(feed.Url)
if err != nil {
log.Errorf("Error saving post: %v", err)
}
}
}
func (e *Feeds) fetch(url string) error {
log.Infof("Fetching address %v", url)
fd, err := rss.Fetch(url)
if err != nil {
return fmt.Errorf("Error fetching address %v: %v", url, err)
}
domain := getDomain(url)
for _, item := range fd.Items {
id := fmt.Sprintf("%x", md5.Sum([]byte(item.ID)))
err = e.entries.Save(feeds.Entry{
Id: id,
Url: item.Link,
Title: item.Title,
Domain: domain,
Content: item.Summary,
Date: item.Date.Unix(),
})
if err != nil {
return fmt.Errorf("Error saving item: %v", err)
}
// @todo make this optional
_, err := e.postsService.Save(context.TODO(), &posts.SaveRequest{
Id: id,
Title: item.Title,
Content: item.Content,
Timestamp: item.Date.Unix(),
Metadata: map[string]string{
"domain": domain,
"link": item.Link,
},
})
if err != nil {
return err
}
}
return nil
}
func getDomain(address string) string {
uri, _ := url.Parse(address)
return uri.Host
}