{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "import requests\n",
    "import json\n",
    "QUAY_API_URL = 'https://quay.io/api/v1/repository'\n",
    "\n",
    "def get_all_repos(url):\n",
    "    repos = []\n",
    "    r = requests.get(url)\n",
    "    data = r.json()\n",
    "    repos.extend(data['repositories'])\n",
    "    while True:\n",
    "        next_link = r.headers.get('Link', None)\n",
    "        if not next_link:\n",
    "            return repos\n",
    "        r = requests.get(next_link.split(';')[0], params={'n': 50, 'next': repos[-1]})\n",
    "        data = r.json()\n",
    "        repos.extend(data['repositories'])\n",
    "r = get_all_repos('https://quay.io/v2/_catalog')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "all_quay_repos = len(r)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {
    "collapsed": false
   },
   "outputs": [],
   "source": [
    "mulled = [repo for repo in r if repo.startswith('biocontainers/')]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1642"
      ]
     },
     "execution_count": 63,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "mulled_repos = len(mulled)\n",
    "mulled_repos"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "biocontainers/align-it 489\n",
      "biocontainers/arvados-cwl-runner 502\n",
      "biocontainers/awk 511\n",
      "biocontainers/azure-cli 512\n",
      "biocontainers/bambamc 514\n",
      "biocontainers/bamutil 514\n",
      "biocontainers/bcalm 516\n",
      "biocontainers/bcbio-nextgen-vm 529\n",
      "biocontainers/bcbio-rnaseq 530\n",
      "biocontainers/bcl2fastq-nextseq 534\n",
      "biocontainers/beagle-lib 536\n",
      "biocontainers/beast 536\n",
      "biocontainers/bio_assembly_refinement 541\n",
      "biocontainers/bioawk 541\n",
      "biocontainers/bioconductor-bsgenome.mmusculus.ucsc.mm9 565\n",
      "biocontainers/bioconductor-bubbletree 565\n",
      "biocontainers/bioconductor-chipcomp 571\n",
      "biocontainers/bioconductor-ebimage 587\n",
      "biocontainers/bioconductor-ggbio 602\n",
      "biocontainers/bioconductor-minfi 629\n",
      "biocontainers/bioconductor-motifrg 631\n",
      "biocontainers/bioconductor-rcas 648\n",
      "biocontainers/biofluff 687\n",
      "biocontainers/biokit 687\n",
      "biocontainers/bio-vcf 688\n",
      "biocontainers/blast 689\n",
      "biocontainers/bmfilter 694\n",
      "biocontainers/bmtool 695\n",
      "biocontainers/burrito-fillings 700\n",
      "biocontainers/bwakit 701\n",
      "biocontainers/byobu 702\n",
      "biocontainers/cap-mirseq 704\n",
      "biocontainers/c-ares 704\n",
      "biocontainers/cgat-report 706\n",
      "biocontainers/chromhmm 710\n",
      "biocontainers/ciso8601 715\n",
      "biocontainers/cityhash 715\n",
      "biocontainers/cmfinder 719\n",
      "biocontainers/collectl 720\n",
      "biocontainers/cuneiform 726\n",
      "biocontainers/dammit 731\n",
      "biocontainers/dfam 741\n",
      "biocontainers/dinopy 743\n",
      "biocontainers/discovar-denovo 744\n",
      "biocontainers/dominate 745\n",
      "biocontainers/dreamtools 745\n",
      "biocontainers/ea-utils 749\n",
      "biocontainers/eigen 749\n",
      "biocontainers/eigensoft 749\n",
      "biocontainers/emperor 750\n",
      "biocontainers/ete2 754\n",
      "biocontainers/express 757\n",
      "biocontainers/fast5 759\n",
      "biocontainers/fastlmm 761\n",
      "biocontainers/fastqc 762\n",
      "biocontainers/fastq-join 762\n",
      "biocontainers/ficus 771\n",
      "biocontainers/fitter 773\n",
      "biocontainers/flash 773\n",
      "biocontainers/flask-basicauth 774\n",
      "biocontainers/flask-bootstrap 774\n",
      "biocontainers/flux-simulator 781\n",
      "biocontainers/fuma 785\n",
      "biocontainers/gat 791\n",
      "biocontainers/gcs_oauth2_boto_plugin 795\n",
      "biocontainers/genblasta 797\n",
      "biocontainers/genomebaser 811\n",
      "biocontainers/gettext 814\n",
      "biocontainers/ggplot 817\n",
      "biocontainers/ghc 817\n",
      "biocontainers/gimmemotifs 818\n",
      "biocontainers/gnu-getopt 824\n",
      "biocontainers/gnuplot 824\n",
      "biocontainers/graph_embed 830\n",
      "biocontainers/graphicsmagick 830\n",
      "biocontainers/graphlan 830\n",
      "biocontainers/graphmap 831\n",
      "biocontainers/graphviz 831\n",
      "biocontainers/gromacs_no_mpi 834\n",
      "biocontainers/group_humann2_uniref_abundances_to_go 834\n",
      "biocontainers/hall-lab-svtools 842\n",
      "biocontainers/hapbin 842\n",
      "biocontainers/hivtrace 863\n",
      "biocontainers/hmmlearn 864\n",
      "biocontainers/htseq 865\n",
      "biocontainers/icqsol 873\n",
      "biocontainers/idr 874\n",
      "biocontainers/igdiscover 875\n",
      "biocontainers/intervals 879\n",
      "biocontainers/intervalstats 879\n",
      "biocontainers/ipython-cluster-helper 879\n",
      "biocontainers/joblib 885\n",
      "biocontainers/julia 885\n",
      "biocontainers/kat 885\n",
      "biocontainers/leptonica 891\n",
      "biocontainers/libbambamc 891\n",
      "biocontainers/libffi 892\n",
      "biocontainers/libmagic 894\n",
      "biocontainers/libsbml 894\n",
      "biocontainers/libssh2 894\n",
      "biocontainers/libsvm 894\n",
      "biocontainers/libxml2 894\n",
      "biocontainers/lightning 894\n",
      "biocontainers/limix 894\n",
      "biocontainers/lrzip 897\n",
      "biocontainers/matplotlib-venn 906\n",
      "biocontainers/medusa 909\n",
      "biocontainers/meme 911\n",
      "biocontainers/metaseq 914\n",
      "biocontainers/metaseq-all 914\n",
      "biocontainers/mgkit 918\n",
      "biocontainers/miniasm 922\n",
      "biocontainers/minimap 922\n",
      "biocontainers/moods 927\n",
      "biocontainers/mreps 930\n",
      "biocontainers/multiqc 935\n",
      "biocontainers/multiqc-bcbio 935\n",
      "biocontainers/mummer 935\n",
      "biocontainers/mysqlclient 938\n",
      "biocontainers/mysql-connector-c 938\n",
      "biocontainers/nanopolish 941\n",
      "biocontainers/ncbi-vdb 949\n",
      "biocontainers/neptune 949\n",
      "biocontainers/nextflow 952\n",
      "biocontainers/nglview 952\n",
      "biocontainers/ngs-sdk 952\n",
      "biocontainers/oligotyping 957\n",
      "biocontainers/oncofuse 957\n",
      "biocontainers/optitype 960\n",
      "biocontainers/osra 961\n",
      "biocontainers/p7zip 961\n",
      "biocontainers/pandoc 961\n",
      "biocontainers/peakranger 966\n",
      "biocontainers/pear 967\n",
      "biocontainers/pedagree 967\n",
      "biocontainers/peddy 967\n",
      "biocontainers/perl-appconfig 974\n",
      "biocontainers/perl-archive-extract 975\n",
      "biocontainers/perl-bio-db-sam 979\n",
      "biocontainers/perl-bio-phylo 982\n",
      "biocontainers/perl-carp 991\n",
      "biocontainers/perl-config-general 1001\n",
      "biocontainers/perl-data-stag 1006\n",
      "biocontainers/perl-dbd-mysql 1007\n",
      "biocontainers/perl-extutils-makemaker 1024\n",
      "biocontainers/perl-file-path 1027\n",
      "biocontainers/perl-font-afm 1035\n",
      "biocontainers/perl-gdgraph-histogram 1039\n",
      "biocontainers/perl-getopt-long 1040\n",
      "biocontainers/perl-html-parser 1048\n",
      "biocontainers/perl-io 1061\n",
      "biocontainers/perl-ipc-run 1068\n",
      "biocontainers/perl-lwp-protocol-https 1076\n",
      "biocontainers/perl-math-base-convert 1078\n",
      "biocontainers/perl-module-build 1090\n",
      "biocontainers/perl-params-validate 1103\n",
      "biocontainers/perl-set-intervaltree 1127\n",
      "biocontainers/perl-term-ansicolor 1147\n",
      "biocontainers/perl-test 1148\n",
      "biocontainers/perl-test-base 1148\n",
      "biocontainers/perl-test-yaml 1164\n",
      "biocontainers/perl-xml-libxml 1187\n",
      "biocontainers/perl-xml-libxslt 1187\n",
      "biocontainers/perl-xml-parser-lite 1189\n",
      "biocontainers/phizz 1201\n",
      "biocontainers/pindel 1207\n",
      "biocontainers/plink2 1212\n",
      "biocontainers/pomegranate 1213\n",
      "biocontainers/potrace 1215\n",
      "biocontainers/prinseq 1217\n",
      "biocontainers/protrac 1221\n",
      "biocontainers/pyfasta 1233\n",
      "biocontainers/pyopt 1241\n",
      "biocontainers/pysnptools 1250\n",
      "biocontainers/python-hivclustering 1260\n",
      "biocontainers/python-igraph 1262\n",
      "biocontainers/python-intermine 1262\n",
      "biocontainers/python-wget 1266\n",
      "biocontainers/pyutilib 1267\n",
      "biocontainers/query_phenomizer 1274\n",
      "biocontainers/rainbow 1277\n",
      "biocontainers/rapclust 1279\n",
      "biocontainers/rapmap 1279\n",
      "biocontainers/r-bma 1297\n",
      "biocontainers/r-checkmate 1301\n",
      "biocontainers/r-coala 1301\n",
      "biocontainers/r-deldir 1307\n",
      "biocontainers/r-dt 1315\n",
      "biocontainers/requests-mock 1320\n",
      "biocontainers/retrying 1321\n",
      "biocontainers/r-flowr 1328\n",
      "biocontainers/r-gplots 1337\n",
      "biocontainers/r-gsubfn 1338\n",
      "biocontainers/r-htmlwidgets 1340\n",
      "biocontainers/r-jpeg 1343\n",
      "biocontainers/r-ks 1346\n",
      "biocontainers/rmblast 1354\n",
      "biocontainers/r-mixomics 1355\n",
      "biocontainers/r-pheatmap 1363\n",
      "biocontainers/r-polynom 1368\n",
      "biocontainers/r-rainbow 1376\n",
      "biocontainers/r-rBLAST 1376\n",
      "biocontainers/r-rehh 1382\n",
      "biocontainers/r-rgl 1383\n",
      "biocontainers/r-rjson 1383\n",
      "biocontainers/r-robustbase 1385\n",
      "biocontainers/rsem 1398\n",
      "biocontainers/r-spp 1408\n",
      "biocontainers/rtg-tools 1412\n",
      "biocontainers/r-tiff 1412\n",
      "biocontainers/r-tximport 1412\n",
      "biocontainers/rust 1418\n",
      "biocontainers/sailfish 1426\n",
      "biocontainers/salmon 1426\n",
      "biocontainers/samblaster 1426\n",
      "biocontainers/samsifter 1426\n",
      "biocontainers/scalpel 1426\n",
      "biocontainers/scikit-allel 1430\n",
      "biocontainers/scikit-bio 1430\n",
      "biocontainers/scoop 1430\n",
      "biocontainers/scrm 1432\n",
      "biocontainers/sed 1432\n",
      "biocontainers/sga 1444\n",
      "biocontainers/shape-it 1444\n",
      "biocontainers/sickle 1445\n",
      "biocontainers/sickle-trim 1445\n",
      "biocontainers/skewer 1448\n",
      "biocontainers/snap-aligner 1453\n",
      "biocontainers/soapdenovo2-gapcloser 1458\n",
      "biocontainers/socksipy-branch 1460\n",
      "biocontainers/spades 1461\n",
      "biocontainers/spanki 1461\n",
      "biocontainers/sphinx-argparse 1464\n",
      "biocontainers/sqlalchemy-utils 1468\n",
      "biocontainers/srprism 1471\n",
      "biocontainers/star-fusion 1472\n",
      "biocontainers/stride 1472\n",
      "biocontainers/strip-it 1472\n",
      "biocontainers/suma_package 1475\n",
      "biocontainers/svgutils 1476\n",
      "biocontainers/swarm 1481\n",
      "biocontainers/tables 1484\n",
      "biocontainers/tabview 1485\n",
      "biocontainers/tdrmapper 1489\n",
      "biocontainers/tesseract 1489\n",
      "biocontainers/tn93 1492\n",
      "biocontainers/tophat2 1493\n",
      "biocontainers/treemix 1500\n",
      "biocontainers/trim_galore 1502\n",
      "biocontainers/ucsc-ave 1512\n",
      "biocontainers/ucsc-blat 1517\n",
      "biocontainers/ucsc-calc 1517\n",
      "biocontainers/ucsc-chromgraphtobin 1525\n",
      "biocontainers/ucsc-genepredtobiggenepred 1555\n",
      "biocontainers/ucsc-getrna 1556\n",
      "biocontainers/ucsc-getrnapred 1556\n",
      "biocontainers/ucsc-gfclient 1556\n",
      "biocontainers/ucsc-gff3topsl 1557\n",
      "biocontainers/ucsc-gfserver 1557\n",
      "biocontainers/ucsc-gmtime 1557\n",
      "biocontainers/ucsc-headrest 1557\n",
      "biocontainers/ucsc-hgfindspec 1557\n",
      "biocontainers/ucsc-hgloadnet 1558\n",
      "biocontainers/ucsc-hgloadout 1558\n",
      "biocontainers/ucsc-hgloadoutjoined 1558\n",
      "biocontainers/ucsc-hgspeciesrna 1558\n",
      "biocontainers/ucsc-hgsqldump 1558\n",
      "biocontainers/ucsc-hgtrackdb 1558\n",
      "biocontainers/ucsc-htmlcheck 1558\n",
      "biocontainers/ucsc-hubcheck 1558\n",
      "biocontainers/ucsc-hubpubliccheck 1558\n",
      "biocontainers/ucsc-ixixx 1558\n",
      "biocontainers/ucsc-lavtoaxt 1558\n",
      "biocontainers/ucsc-ldhggene 1558\n",
      "biocontainers/ucsc-linestora 1558\n",
      "biocontainers/ucsc-localtime 1558\n",
      "biocontainers/ucsc-maketablelist 1558\n",
      "biocontainers/ucsc-mktime 1559\n",
      "biocontainers/ucsc-mrnatogene 1559\n",
      "biocontainers/ucsc-netsplit 1561\n",
      "biocontainers/ucsc-newprog 1563\n",
      "biocontainers/ucsc-overlapselect 1567\n",
      "biocontainers/umis 1572\n",
      "biocontainers/vardict 1576\n",
      "biocontainers/variantbam 1577\n",
      "biocontainers/vcf2db 1579\n",
      "biocontainers/vispr 1587\n",
      "biocontainers/vsearch 1589\n",
      "biocontainers/vt 1589\n",
      "biocontainers/whatshap 1596\n",
      "biocontainers/xmltodict 1601\n",
      "biocontainers/yahmm 1603\n",
      "biocontainers/perl-moose 1609\n",
      "biocontainers/perl-module-pluggable 1617\n",
      "biocontainers/pydotplus 1651\n",
      "biocontainers/pygraphviz 1651\n",
      "biocontainers/cwltest 1655\n",
      "biocontainers/xopen 1655\n",
      "biocontainers/No 1656\n",
      "biocontainers/tests 1656\n",
      "biocontainers/defined 1656\n",
      "biocontainers/cromwell 1662\n",
      "biocontainers/sbt 1662\n",
      "biocontainers/scala 1662\n",
      "biocontainers/extract_genome_region 1676\n",
      "biocontainers/snpsift 1676\n",
      "biocontainers/kronik 1677\n",
      "biocontainers/mgf-formatter 1677\n",
      "biocontainers/pyquant-ms 1677\n",
      "biocontainers/pyteomics 1677\n",
      "biocontainers/pyprophet 1684\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "1693"
      ]
     },
     "execution_count": 73,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\n",
    "tags_count = 0\n",
    "for repository_string in mulled:\n",
    "    url = \"%s/%s\" % (QUAY_API_URL, repository_string)\n",
    "    infos = requests.get(url, headers={'Accept-encoding': 'gzip'}, timeout=30)\n",
    "    json_decoder = json.JSONDecoder()\n",
    "    decoded_request = json_decoder.decode(infos.text)\n",
    "    tags = decoded_request.get('tags', False).keys()\n",
    "    if 'latest' in tags: tags.remove('latest')\n",
    "    tags_count += len(tags)\n",
    "    if len(tags) == 0: print repository_string, tags_count\n",
    "\n",
    "tags_count\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "# Quay.io repositories: 13316\n",
      "# Mulled repositories: 1642\n",
      "# Mulled images: 1693\n"
     ]
    }
   ],
   "source": [
    "print '# Quay.io repositories: %s' % all_quay_repos\n",
    "print '# Mulled repositories: %s' % mulled_repos\n",
    "print '# Mulled images: %s' % tags_count"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}