SRSRecordStreaming.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "4bdc247c-0d14-4606-81f8-4823313b8f66",
   "metadata": {},
   "source": [
    "# SRS Record Streaming\n",
    "\n",
    "[API Documentation](https://s3.amazonaws.com/foliodocs/api/mod-source-record-storage/p/source-record-storage-stream.html)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "d424ab3c-3ccb-49e8-8841-50a4dad4e6d8",
   "metadata": {},
   "outputs": [],
   "source": [
    "import datetime\n",
    "import json\n",
    "\n",
    "import requests\n",
    "from folioclient import FolioClient"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "id": "5905d9e1-6c17-4919-b7e4-39066e682720",
   "metadata": {},
   "outputs": [],
   "source": [
    "dev_client = FolioClient(\n",
    " \"https://okapi-dev.stanford.edu\", \n",
    " \"sul\", \n",
    " \"sul_admin\", \n",
    " \"{ask-us-for-pwd}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "id": "742b0895-f93a-49c2-8972-08dd8748736b",
   "metadata": {},
   "outputs": [],
   "source": [
    "def record_stream(url, client):\n",
    "    session = requests.Session()\n",
    "    records = []\n",
    "    start = datetime.datetime.utcnow()\n",
    "    print(f\"Started stream of MARC JSON {start}\")\n",
    "    with session.get(url, headers=client.okapi_headers, stream=True) as response:\n",
    "        for i,line in enumerate(response.iter_lines()):\n",
    "            records.append(json.loads(line))\n",
    "    end = datetime.datetime.utcnow()\n",
    "    print(f\"Finished stream of MARC JSON {end}, total time {(end-start).seconds / 60.}\")\n",
    "    return records\n",
    "        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "id": "846abcb3-b37b-4147-bdad-c3930299dc3e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Started stream of MARC JSON 2022-07-12 20:54:57.507175\n",
      "Finished stream of MARC JSON 2022-07-12 20:56:56.378246, total time 1.9666666666666666\n"
     ]
    }
   ],
   "source": [
    "first_recs = record_stream(f\"{dev_client.okapi_url}/source-storage/stream/source-records?updatedAfter=2022-07-01&limit=100000&suppressFromDiscovery=False\", dev_client)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "id": "8d608278-b793-4d55-8c62-204bf5d13695",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "100000"
      ]
     },
     "execution_count": 80,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(first_recs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "id": "19dd6267-3434-42d4-a255-035537de3618",
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'recordId': '0000da5b-370d-5f2d-aa78-f6fb544038b6',\n",
       " 'snapshotId': 'a044ccbc-76dd-42fb-9ad8-9d9404cb88be',\n",
       " 'recordType': 'MARC_BIB',\n",
       " 'parsedRecord': {'id': '0000da5b-370d-5f2d-aa78-f6fb544038b6',\n",
       "  'content': {'fields': [{'001': 'a10451893'},\n",
       "    {'003': 'SIRSI'},\n",
       "    {'005': '20160716013742.0'},\n",
       "    {'007': 'a| |||||'},\n",
       "    {'007': 'aj|canzn'},\n",
       "    {'008': '140415s1910    cau|||||| a  || | ||eng|d'},\n",
       "    {'040': {'ind1': ' ',\n",
       "      'ind2': ' ',\n",
       "      'subfields': [{'a': 'CSt'},\n",
       "       {'b': 'eng'},\n",
       "       {'c': 'CSt'},\n",
       "       {'d': 'UtOrBLW'}]}},\n",
       "    {'050': {'ind1': ' ',\n",
       "      'ind2': '4',\n",
       "      'subfields': [{'a': 'G4364.R6 G46 [1910] .N4'}]}},\n",
       "    {'100': {'ind1': '1',\n",
       "      'ind2': ' ',\n",
       "      'subfields': [{'a': 'New Richmond Land Company.'}]}},\n",
       "    {'245': {'ind1': '1',\n",
       "      'ind2': '0',\n",
       "      'subfields': [{'a': \"Map of Richmond, California. New Richmond Land Company, 801-803 Monadnock Building, San Francisco. Wall's Harbor Center Tract, The Value Center. Bolte & Braden Company, San Francisco\"},\n",
       "       {'h': '[cartographic material].'}]}},\n",
       "    {'255': {'ind1': ' ',\n",
       "      'ind2': ' ',\n",
       "      'subfields': [{'a': 'Scale 1:19,200.'}]}},\n",
       "    {'264': {'ind1': ' ',\n",
       "      'ind2': '1',\n",
       "      'subfields': [{'a': 'San Francisco :'},\n",
       "       {'b': 'Bolte & Braden Company,'},\n",
       "       {'c': '[1910]'}]}},\n",
       "    {'300': {'ind1': ' ',\n",
       "      'ind2': ' ',\n",
       "      'subfields': [{'a': '1 map ;'}, {'c': '43.5 x 62 cm'}]}},\n",
       "    {'336': {'ind1': ' ',\n",
       "      'ind2': ' ',\n",
       "      'subfields': [{'a': 'cartographic image'}, {'2': 'rdacontent'}]}},\n",
       "    {'337': {'ind1': ' ',\n",
       "      'ind2': ' ',\n",
       "      'subfields': [{'a': 'unmediated'}, {'2': 'rdamedia'}]}},\n",
       "    {'338': {'ind1': ' ',\n",
       "      'ind2': ' ',\n",
       "      'subfields': [{'a': 'sheet'}, {'2': 'rdacarrier'}]}},\n",
       "    {'500': {'ind1': ' ',\n",
       "      'ind2': ' ',\n",
       "      'subfields': [{'a': \"A real estate development map. Date is estimated. Map shows the Wall's Harbor Center Tract in red, and the City Hall Site and Civic Center in green.\"}]}},\n",
       "    {'500': {'ind1': ' ', 'ind2': ' ', 'subfields': [{'a': 'Separate Map.'}]}},\n",
       "    {'650': {'ind1': ' ',\n",
       "      'ind2': '0',\n",
       "      'subfields': [{'a': 'Real estate development'},\n",
       "       {'z': 'California'},\n",
       "       {'z': 'Richmond'},\n",
       "       {'x': 'Maps'},\n",
       "       {'y': '20th century.'}]}},\n",
       "    {'650': {'ind1': ' ',\n",
       "      'ind2': '0',\n",
       "      'subfields': [{'a': 'City blocks'},\n",
       "       {'z': 'California'},\n",
       "       {'z': 'Richmond'},\n",
       "       {'x': 'Maps'},\n",
       "       {'y': '20th century.'}]}},\n",
       "    {'651': {'ind1': ' ',\n",
       "      'ind2': '0',\n",
       "      'subfields': [{'a': 'California'},\n",
       "       {'v': 'Maps'},\n",
       "       {'y': '20th century.'}]}},\n",
       "    {'651': {'ind1': ' ',\n",
       "      'ind2': '0',\n",
       "      'subfields': [{'a': 'Richmond (Calif.)'},\n",
       "       {'x': 'Maps'},\n",
       "       {'y': '20th century.'}]}},\n",
       "    {'655': {'ind1': ' ',\n",
       "      'ind2': '7',\n",
       "      'subfields': [{'a': 'Maps'}, {'2': 'lcgft'}]}},\n",
       "    {'655': {'ind1': ' ',\n",
       "      'ind2': '7',\n",
       "      'subfields': [{'a': 'Cadastral maps.'}, {'2': 'lcgft'}]}},\n",
       "    {'590': {'ind1': ' ',\n",
       "      'ind2': ' ',\n",
       "      'subfields': [{'a': 'Pub list no.: 5312.000.'}]}},\n",
       "    {'690': {'ind1': ' ',\n",
       "      'ind2': '4',\n",
       "      'subfields': [{'a': 'The David Rumsey Map Collection.'}]}},\n",
       "    {'856': {'ind1': '4',\n",
       "      'ind2': '0',\n",
       "      'subfields': [{'u': 'http://www.davidrumsey.com/luna/servlet/view/search?q=pub_list_no=5312.000%20LIMIT:RUMSEY~8~1&sort=Pub_List_No_InitialSort,Pub_Date,Pub_List_No,Series_No'}]}},\n",
       "    {'856': {'ind1': '4',\n",
       "      'ind2': '1',\n",
       "      'subfields': [{'u': 'http://purl.stanford.edu/yv535wk5017'},\n",
       "       {'x': 'SDR-PURL'},\n",
       "       {'x': 'item'},\n",
       "       {'x': 'file:yv535wk5017%2F5312000.jp2'},\n",
       "       {'x': 'collection:xh235dd9059::David Rumsey Map Collection at Stanford University Libraries'}]}},\n",
       "    {'035': {'ind1': ' ',\n",
       "      'ind2': ' ',\n",
       "      'subfields': [{'a': '(OCoLC-M)953571215'}]}},\n",
       "    {'596': {'ind1': ' ', 'ind2': ' ', 'subfields': [{'a': '13'}]}},\n",
       "    {'918': {'ind1': ' ', 'ind2': ' ', 'subfields': [{'a': '10451893'}]}},\n",
       "    {'999': {'ind1': 'f',\n",
       "      'ind2': 'f',\n",
       "      'subfields': [{'i': 'b6919f84-f7f3-5957-84d6-fcf603aaacbb'},\n",
       "       {'s': '0000da5b-370d-5f2d-aa78-f6fb544038b6'}]}}],\n",
       "   'leader': '01834nem a2200397uu 4500'}},\n",
       " 'deleted': False,\n",
       " 'externalIdsHolder': {'instanceId': 'b6919f84-f7f3-5957-84d6-fcf603aaacbb',\n",
       "  'instanceHrid': 'a10451893'},\n",
       " 'additionalInfo': {'suppressDiscovery': False},\n",
       " 'metadata': {'createdDate': 1657579301983,\n",
       "  'createdByUserId': 'd6507bbe-e985-4baa-beb4-d7ca4d6c279f',\n",
       "  'updatedDate': 1657579303250,\n",
       "  'updatedByUserId': 'd6507bbe-e985-4baa-beb4-d7ca4d6c279f'}}"
      ]
     },
     "execution_count": 83,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "first_recs[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "id": "8a992fad-e7b1-488b-a636-b15528d1a838",
   "metadata": {},
   "outputs": [],
   "source": [
    "with open(\"folio-srs-10k.json\", \"w+\") as fo:\n",
    "    json.dump(first_recs, fo)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "id": "4ed2d027-b6f7-4474-aa9b-f41180753804",
   "metadata": {},
   "outputs": [],
   "source": [
    "from pymarc import parse_json_to_array"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "id": "681fd0ce-29c3-4605-8477-57737c2cf7dc",
   "metadata": {},
   "outputs": [],
   "source": [
    "rtac_url = f\"{dev_client.okapi_url}/rtac-batch\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "id": "e4c01205-fef1-488a-bd61-6fcf1f020866",
   "metadata": {},
   "outputs": [],
   "source": [
    "result =requests.post(rtac_url,\n",
    "                      headers=dev_client.okapi_headers,\n",
    "                      json={ \"instanceIds\": ['b6919f84-f7f3-5957-84d6-fcf603aaacbb'],\n",
    "                             \"fullPeriodicals\" : 'false'})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "id": "fe6ff772-e6f5-4713-9159-416af159e2bc",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "200"
      ]
     },
     "execution_count": 86,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "result.status_code"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "id": "4eb5d26f-8010-4478-80a5-47d8deb9d99b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'holdings': [{'instanceId': 'b6919f84-f7f3-5957-84d6-fcf603aaacbb',\n",
       "   'holdings': [{'id': 'e90af420-7ed2-54ba-b297-0c871d676b7f',\n",
       "     'location': 'Map Center (W7 large box)',\n",
       "     'callNumber': 'G4364.R6 G46 [1910] .N4',\n",
       "     'status': 'Available',\n",
       "     'permanentLoanType': 'Can circulate'}]}]}"
      ]
     },
     "execution_count": 87,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "result.json()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "c8296473-6937-4102-96b5-41cf1e05a1b3",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}