Merge pull request #2 from raynormast/dev

Dev

.gitignore

@@ -5,3 +5,10 @@ data/20221217.json
 config/domains-federated
 config/domains-local
 config/hashtags
+config/hosts
+.gitignore
+data/20221219.json
+.gitignore
+config/hashtag-urls.txt
+config/urls.txt
+data/20221219.uris.txt

TODO.md

@@ -0,0 +1,3 @@
+1. Create language filter
+2. Create bot filter
+3. 

config/domains-federated.sample

@@ -1,7 +1,12 @@
+## Comments require two #'s,, like ##
+
 ## Fake Firehose will only take all public posts from these domains
 ## This is the true firehose, use it carefully or blow up your server
 
-### International English (if you aren't from the US) ###
+## Hashtags to follow by instance
+mastodon.social #JohnMastodon #silentsunday #mastoadmin
+
+## International English (if you aren't from the US) ###
 mastodon.scot
 aus.social
 mastodon.nz

config/domains-local.sample

@@ -9,3 +9,6 @@ fosstodon.org
 
 ### News & Politics ###
 journa.host
+
+### Cool People ### 
+raynor.haus

config/hashtags.sample

@@ -0,0 +1,11 @@
+## If this file is not empty then EVERY host from an earlier file
+## including duplicates, will stream to look for these hashtags
+
+## You can list hashtags with a leading "#" ...
+
+cybersecurity
+mastoadmin
+
+## Or you can list hashtags with the "#"
+
+#mastodon

readme.md

@@ -18,8 +18,55 @@ If you want the full on public feed from an instance, put it in the domains-fede
 
 If you only want the local feed from an instance, put it on the domains-local file, one domain per line
 
-If you want to follow a hashtag you're out of luck because I didn't get that far. But it will go into the hashtags file.
+If you want to follow a hash tag you either either add a hashtag after an instance in `domains-federated` or `domains-local`
 
+For example: if in `domains-fedarated` you put `mastodon.social #JohnMastodon` a stream will open to watch for the hashtag #JohnMastodon on the public
+stream from mastodon.social
+
+Another example:  if in `domains-local` you put `infosec.exchange #hacker` a stream will open to watch for the hashtag #hacker on the _local_ stream from infosec.exchange
+
+## Docker
 Build docker
 
 Run docker
+
+### The hashtags file
+If you put ANY hashtags in here a stream will be opened for _every_ host in the `domains-federated` and `domains-local` file.
+
+Example:
+`domains-federated` content:
+
+```
+mastodon.social
+mas.to
+```
+
+`domains-local` content:
+
+```
+aus.social
+mastodon.nz
+```
+
+`hashtags` content:
+```
+JohnMastodon
+Mastodon
+```
+
+will result in the following streams all opening:
+```shell
+https://mastodon.social/api/v1/streaming/hashtag?tag=JohnMastodon
+https://mas.to/api/v1/streaming/hashtag?tag=JohnMastodon
+https://aus.social/api/v1/streaming/hashtag?tag=JohnMastodon
+https://mastodon.nz/api/v1/streaming/hashtag?tag=JohnMastodon
+https://mastodon.social/api/v1/streaming/hashtag?tag=Mastodon
+https://mas.to/api/v1/streaming/hashtag?tag=Mastodon
+https://aus.social/api/v1/streaming/hashtag?tag=Mastodon
+https://mastodon.nz/api/v1/streaming/hashtag?tag=Mastodon
+```
+
+If you had a total of 5 lines in `domains-federated` and `domains-local` plus 3 entries in `hashtags`
+there would 5x5x3 = 75 new streams.
+
+I mean, you can do it, but you won't need your central heating system any more.

rebuild.sh

@@ -0,0 +1,3 @@
+docker compose -f docker-compose.yml down
+docker build -t fakefirehose .
+docker compose -f docker-compose.yml up 

scripts/get-stream.sh

@@ -1,5 +1,6 @@
 host=$1
 type=$2
+hashtag=$1
 
 if [[ "$host" == "" ]]
 then
@@ -16,11 +17,7 @@ do
             fetch="https://$host/api/v1/streaming/public";;
         "local")
             fetch="https://$host/api/v1/streaming/public?local=true";;
-        "hashtags")
-            fetch="https://$host/api/v1/streaming/hashtag?tag=linux"
-            echo "Sorry, hash tags aren't implemented yet :("
-            exit 1
-            ;;
+
     esac
 
     echo "Starting to stream $fetch in 5 seconds"

scripts/start-firehose.sh

@@ -1,25 +1,98 @@
 #!/bin/bash
 
-cat /config/domains-federated|grep -v "#"|while read -r host
+echo > /config/urls.txt
+echo > /config/hosts
+
+# Get federated hosts and begin to stream them
+cat /config/domains-federated | grep -v "##" | while read -r line
 do
-   if [[ "$host" != "" ]]
-   then
-      /scripts/get-stream.sh $host "federated" &
+   #filter out empty lines
+   if [[ "$line" != "" ]]; then
+      echo "Opening federated line $line"
+
+      #Check for hashtags
+      if [[ "$line" == *" #"* ]]; then
+
+         echo "$line has hashtags!"
+
+         # Get just the first field of the line, which is the host
+         host=`echo $line | cut -d " " -f 1`
+         tags=`echo $line | cut -d " " -f 1 --complement|tr "#" "\n "`
+         for tag in $tags
+         do
+            if [[ $tag != "" ]]; then
+            echo "Found tag $tag"
+            # Create a url to fetch for each tag
+            echo "https://$host/api/v1/streaming/hashtag?tag=$tag $host" >> /config/urls.txt
+            fi
+         done
+      elif [[ "$line" != *" #"* ]]; then
+         echo "$line didn't have hashtags"
+         host=$line
+         echo "https://$line/api/v1/streaming/public $line" >> /config/urls.txt
+      fi 
+      echo $host >> /config/hosts
    fi
 done
 
-cat /config/domains-local|grep -v "#"|while read -r host
+
+# Get local hosts and begin to stream them
+cat /config/domains-local | grep -v "##" | while read -r line
 do
-   if [[ "$host" != "" ]]
-   then
-      /scripts/get-stream.sh $host "local" &
+   #filter out empty lines
+   if [[ "$line" != "" ]]; then
+      echo "Opening federated line $line"
+
+      #Check for hashtags
+      if [[ "$line" == *" #"* ]]; then
+
+         echo "$line has hashtags!"
+
+         # Get just the first field of the line, which is the host
+         host=`echo $line | cut -d " " -f 1`
+         tags=`echo $line | cut -d " " -f 1 --complement|tr "#" "\n "`
+         for tag in $tags
+         do
+            if [[ $tag != "" ]]; then
+            echo "Found tag $tag"
+            # Create a url to fetch for each tag
+            echo "https://$host/api/v1/streaming/hashtag/local?tag=$tag $host" >> /config/urls.txt
             fi
          done
+      elif [[ "$line" != *" #"* ]]; then
+         echo "$line didn't have hashtags"
+         host=$line
+         echo "https://$line/api/v1/streaming/local $line" >> /config/urls.txt
+      fi
+      echo $host >> /config/hosts
+   fi
+done
+
+cat /config/hashtags | grep -v "##" | while read -r hashtag; do
+   hashtag=`echo $hashtag | cut -d "#" -f 2`
+   sort /config/hosts | uniq -u |while read -r host; do
+      if [[ $hashtag != "" && "$host" != "" ]]; then
+         echo "https://$host/api/v1/streaming/hashtag?tag=$hashtag $host" >> /config/hashtag-urls.txt
+      fi
+   done
+done
+
+cat /config/hashtag-urls.txt >> /config/urls.txt
+
+cat /config/urls.txt | while read -r url
+do
+   echo "Opening $url to stream"
+   sleep 1s
+   ./stream-url.sh $url &
+done
 
 if [[ $runFirehose == true ]]
 then
    /scripts/run-firehose.sh &
 fi
 
-## Don't let the container exit
-while true; do sleep 1; done
+## We don't have a health check, so just exit after an hour
+# If your docker file has restart: always on this should gracefully exit, and 
+# then restart
+sleep 1h
+exit 0

scripts/stream-url.sh

@@ -0,0 +1,37 @@
+url=$1 #A proper URL is all that should be sent to this script
+host=$2
+
+if [[ "$url" == "" ]]
+then
+    echo "Empty url, skipping" # Exit if an empty URL was sent
+    exit 2
+fi
+
+while true # Loop endlessly
+do
+    today=`date +"%Y%m%d"`
+
+    echo "Starting to stream $url in 5 seconds"
+
+    sleep 5s;
+
+    curl -X "GET" "$url" \
+         --no-progress-meter | \
+        tee -a "/data/$today.json" | \
+        grep url | \
+        sed 's/data://g' | \
+
+     while read -r line
+     do
+
+         if [[ $line == *"uri"* ]]
+         then
+            url=`echo $line | jq .url| sed 's/\"//g'` 
+            uri=`echo $line | jq .uri| sed 's/\"//g'`
+
+            echo "STREAMING from $host $url"
+            echo $uri >> "/data/$today.uris.txt"
+
+        fi
+    done
+done