Merge pull request #1508 from cmu-delphi/release/delphi-epidata-4.1.25

Release Delphi Epidata 4.1.25

Author: minhkhul

.bumpversion.cfg

@@ -1,5 +1,5 @@
 [bumpversion]
-current_version = 4.1.24
+current_version = 4.1.25
 commit = False
 tag = False
 

.git-blame-ignore-revs

@@ -20,3 +20,5 @@ b9ceb400d9248c8271e8342275664ac5524e335d
 07ed83e5768f717ab0f9a62a9209e4e2cffa058d
 # style(black): format wiki acquisition
 923852eafa86b8f8b182d499489249ba8f815843
+# lint: trailing whitespace changes
+81179c5f144b8f25421e799e823e18cde43c84f9

.github/PULL_REQUEST_TEMPLATE.md

@@ -1,4 +1,4 @@
-closes|addresses <!--list issues closed or partially-addressed by this PR -->
+addresses issue(s) #ISSUE <!--list issue(s) associated with this PR -->
 
 ### Summary:
 

.github/workflows/update_gdocs_data.yml

@@ -21,7 +21,10 @@ jobs:
           restore-keys: |
             ${{ runner.os }}-pipd-
       - name: Install Dependencies
-        run: pip install -r requirements.dev.txt
+        run: |
+          pip -V
+          python -m pip install pip==22.0.2
+          pip install -r requirements.dev.txt
       - name: Update Docs
         run: inv update-gdoc
       - name: Create pull request into dev

dev/local/setup.cfg

@@ -1,6 +1,6 @@
 [metadata]
 name = Delphi Development
-version = 4.1.24
+version = 4.1.25
 
 [options]
 packages =

docs/api/covidcast-signals/covid-act-now.md

@@ -15,11 +15,13 @@ grand_parent: COVIDcast Main Endpoint
 * **Time type:** day (see [date format docs](../covidcast_times.md))
 * **License:** [CC BY-NC](../covidcast_licensing.md#creative-commons-attribution-noncommercial)
 
-The COVID Act Now (CAN) data source provides COVID-19 testing statistics, such as positivity rates and total tests performed.
-The county-level positivity rates and test totals are pulled directly from CAN.
-While CAN provides this data potentially from multiple sources, we only use data sourced from the
+The [COVID Act Now (CAN)](https://covidactnow.org/) data source provides COVID-19 testing statistics, such as positivity rates and total tests performed.
+The county-level positivity rates and test totals are pulled directly from CAN using [their API](https://covidactnow.org/data-api).
+While CAN provides this data potentially from multiple sources, we only use data that CAN sources from the
 [CDC's COVID-19 Integrated County View](https://covid.cdc.gov/covid-data-tracker/#county-view).
 
+Delphi's mirror of the CAN data was deactivated in December 2021 (last issue 2021-12-10) in favor of the [DSEW CPR data](./dsew-cpr.md), which reports the same information under the `covid_naat_pct_positive_7dav` signal.
+
 
 | Signal                         | Description                                                    |
 |--------------------------------|----------------------------------------------------------------|
@@ -34,9 +36,9 @@ While CAN provides this data potentially from multiple sources, we only use data
 
 ## Estimation
 
-The quantities received from CAN / CDC are the county-level positivity rate and total tests, 
-which are based on the counts of PCR specimens tested.
-In particular, they are also already smoothed with a 7-day-average.
+We receive county-level positivity rate and total tests from CAN, originating from the CDC.
+These quantiles are based on the counts of PCR specimens tested.
+They are also already smoothed with a 7-day-average.
 
 For a fixed location $$i$$ and time $$t$$, let $$Y_{it}$$ denote the number of PCR specimens 
 tested that have a positive result. Let $$N_{it}$$ denote the total number of PCR specimens tested.
@@ -79,38 +81,41 @@ $$
 
 ### Smoothing
 
-No additional smoothing is done to avoid double-smoothing, since the data pulled from CAN / CDC 
+No additional smoothing is done to avoid double-smoothing, since the CAN data
 is already smoothed with a 7-day-average.
 
 ## Limitations
 
-Estimates for geographical levels beyond counties may be inaccurate due to how aggregations 
-are done on smoothed values instead of the raw values. Ideally we would aggregate raw values 
+Estimates for geographical levels beyond counties may be inaccurate because our aggregations
+are performed on smoothed values instead of the raw values.
+Ideally we would aggregate raw values
 then smooth, but the raw values are not accessible in this case.
 
-The positivity rate here should not be interpreted as the population positivity rate as 
+The reported test positivity rate should not be interpreted as the population positivity rate as
 the testing performed are typically not randomly sampled, especially for early data 
 with lower testing volumes.
 
 A few counties, most notably in California, are also not covered by this data source.
 
-Entries with zero total tests performed are also suppressed, even if it was actually the case that 
+Entries with zero total tests performed are suppressed, even if it was actually the case that
 no tests were performed for the day.
 
 ## Lag and Backfill
 
 The lag for these signals varies depending on the reporting patterns of individual counties.
 Most counties have their latest data report with a lag of 2 days, while others can take 9 days 
-or more in the case of California counties.
+or more, as is the case with California counties.
 
-These signals are also backfilled as backlogged test results could get assigned to older 7-day timeframes.
-Most recent test positivity rates do not change substantially with backfill (having a median delta of close to 0).
-However, most recent total tests performed is expected to increase in later data revisions (having a median increase of 7%).
+Revisions are sometimes made to the data. For example, backlogged test results can get assigned to past dates.
+The majority of recent test positivity rates do not change substantially with backfill (having a median delta of close to 0).
+However, the majority of recent total tests performed is expected to increase in later data revisions (having a median increase of 7%).
 Values more than 5 days in the past are expected to remain fairly static (with total tests performed 
 having a median increase of 1% of less), as most major revisions have already occurred.
 
 ## Source and Licensing
 
-County-level testing data is scraped by CAN from the 
+County-level testing data is scraped by [CAN](https://covidactnow.org/) from the
 [CDC's COVID-19 Integrated County View](https://covid.cdc.gov/covid-data-tracker/#county-view),
 and made available through [CAN's API](https://covidactnow.org/tools).
+
+The data is made available under a [CC BY-NC](../covidcast_licensing.md#creative-commons-attribution-noncommercial) license.

docs/api/covidcast-signals/hhs.md

@@ -1,6 +1,6 @@
 ---
 title: Department of Health & Human Services
-parent: Data Sources and Signals
+parent: Inactive Signals
 grand_parent: COVIDcast Main Endpoint
 ---
 

docs/api/covidcast-signals/indicator-combination renamed to docs/api/covidcast-signals/indicator-combination.md

@@ -49,7 +49,7 @@ $ [sudo] make test pdb=1
 $ [sudo] make test test=repos/delphi/delphi-epidata/integrations/acquisition
 ```
 
-You can read the commands executed by the Makefile [here](../dev/local/Makefile).
+You can read the commands executed by the Makefile [here](https://github.com/cmu-delphi/delphi-epidata/blob/dev/dev/local/Makefile).
 
 ## Rapid Iteration and Bind Mounts
 
@@ -87,8 +87,8 @@ You can test your changes manually by:
 
 What follows is a worked demonstration based on the `fluview` endpoint. Before
 starting, make sure that you have the `delphi_database_epidata`,
-`delphi_web_epidata`, and `delphi_redis` containers running; if you don't, see
-the Makefile instructions above.
+`delphi_web_epidata`, and `delphi_redis` containers running (with `docker ps`);
+if you don't, see the Makefile instructions above.
 
 First, let's insert some fake data into the `fluview` table:
 

docs/epidata_development.md

@@ -26,6 +26,10 @@ Pandemic"](https://www.pnas.org/topic/548) in *PNAS*:
 
 Research publications using the survey data include:
 
+- C.K. Ettman, E. Badillo-Goicoechea, E.A. Stuart (2024). [Financial
+  strain, schooling modality and mental health of US adults living
+  with children during the COVID-19 pandemic](https://doi.org/10.1136/jech-2023-221672).
+  *Journal of Epidemiology & Community Health*.
 - K. Sasse, R. Mahabir, O. Gkountouna, A. Crooks, A. Croitoru (2024).
   [Understanding the determinants of vaccine hesitancy in the United
   States: A comparison of social surveys and social media](https://doi.org/10.1371/journal.pone.0301488).
@@ -41,7 +45,7 @@ Research publications using the survey data include:
 - Z. Yang, R. Krishnan, and B. Li (2024). [The interplay between individual
   mobility, health risk, and economic choice: A holistic model for COVID-19
   policy intervention](https://doi.org/10.1287/ijds.2023.0013). *INFORMS
-  Journal on Data Science*.
+  Journal on Data Science* 3 (1), 6-27.
 - A. Srivastava, J. M. Ramirez, S. Díaz-Aranda, J. Aguilar, A. F. Anta, A. Ortega,
   and R. E. Lillo (2024). [Nowcasting temporal trends using indirect surveys](https://doi.org/10.1609/aaai.v38i20.30242).
   In *Proceedings of the 38th AAAI Conference on Artificial Intelligence* 38,

docs/symptom-survey/publications.md

@@ -2,8 +2,8 @@
 
 # standard library
 import time
+import json
 from json import JSONDecodeError
-from requests.models import Response
 from unittest.mock import MagicMock, patch
 
 # first party
@@ -306,6 +306,26 @@ def test_sandbox(self, get, post):
       Epidata.debug = False
       Epidata.sandbox = False
 
+  @patch('requests.get')
+  def test_version_check(self, get):
+    """Test that the _version_check() function correctly logs a version discrepancy."""
+    class MockJson:
+      def __init__(self, content, status_code):
+          self.content = content
+          self.status_code = status_code
+      def raise_for_status(self): pass
+      def json(self): return json.loads(self.content)
+    get.reset_mock()
+    get.return_value = MockJson(b'{"info": {"version": "0.0.1"}}', 200)
+
+    Epidata._version_check()
+
+    captured = self.capsys.readouterr()
+    output = captured.err.splitlines()
+    self.assertEqual(len(output), 1)
+    self.assertIn("Client version not up to date", output[0])
+    self.assertIn("\'latest_version\': \'0.0.1\'", output[0])
+
   def test_geo_value(self):
     """test different variants of geo types: single, *, multi."""
 

integrations/client/test_delphi_epidata.py

@@ -11,7 +11,7 @@
 
 # first party
 import delphi.operations.secrets as secrets
-from delphi.epidata.common.logger import get_structured_logger
+from delphi_utils import get_structured_logger
 
 Columndef = namedtuple("Columndef", "csv_name sql_name dtype")
 

src/acquisition/covid_hosp/common/database.py

@@ -13,10 +13,9 @@
 import pandas as pd
 
 # first party
-from delphi_utils import Nans
+from delphi_utils import get_structured_logger, Nans
 from delphi.utils.epiweek import delta_epiweeks
 from delphi.epidata.common.covidcast_row import CovidcastRow
-from delphi.epidata.common.logger import get_structured_logger
 
 DataFrameRow = NamedTuple('DFRow', [
   ('geo_id', str),

src/acquisition/covidcast/csv_importer.py

@@ -11,7 +11,7 @@
 from delphi.epidata.acquisition.covidcast.csv_importer import CsvImporter, PathDetails
 from delphi.epidata.acquisition.covidcast.database import Database, DBLoadStateException
 from delphi.epidata.acquisition.covidcast.file_archiver import FileArchiver
-from delphi.epidata.common.logger import get_structured_logger
+from delphi_utils import get_structured_logger
 
 
 def get_argument_parser():

src/acquisition/covidcast/csv_to_database.py

@@ -14,7 +14,7 @@
 
 # first party
 import delphi.operations.secrets as secrets
-from delphi.epidata.common.logger import get_structured_logger
+from delphi_utils import get_structured_logger
 from delphi.epidata.common.covidcast_row import CovidcastRow
 
 
@@ -117,28 +117,28 @@ def insert_or_update_batch(self, cc_rows: List[CovidcastRow], batch_size=2**20,
       get_structured_logger("insert_or_update_batch").fatal(err_msg)
       raise DBLoadStateException(err_msg)
 
-    # NOTE: `value_update_timestamp` is hardcoded to "NOW" (which is appropriate) and 
+    # NOTE: `value_update_timestamp` is hardcoded to "NOW" (which is appropriate) and
     #       `is_latest_issue` is hardcoded to 1 (which is temporary and addressed later in this method)
     insert_into_loader_sql = f'''
       INSERT INTO `{self.load_table}`
         (`source`, `signal`, `time_type`, `geo_type`, `time_value`, `geo_value`,
-        `value_updated_timestamp`, `value`, `stderr`, `sample_size`, `issue`, `lag`, 
+        `value_updated_timestamp`, `value`, `stderr`, `sample_size`, `issue`, `lag`,
         `is_latest_issue`, `missing_value`, `missing_stderr`, `missing_sample_size`)
       VALUES
-        (%s, %s, %s, %s, %s, %s, 
-        UNIX_TIMESTAMP(NOW()), %s, %s, %s, %s, %s, 
+        (%s, %s, %s, %s, %s, %s,
+        UNIX_TIMESTAMP(NOW()), %s, %s, %s, %s, %s,
         1, %s, %s, %s)
     '''
 
     # all load table entries are already marked "is_latest_issue".
     # if an entry in the load table is NOT in the latest table, it is clearly now the latest value for that key (so we do nothing (thanks to INNER join)).
     # if an entry *IS* in both load and latest tables, but latest table issue is newer, unmark is_latest_issue in load.
     fix_is_latest_issue_sql = f'''
-        UPDATE 
-            `{self.load_table}` JOIN `{self.latest_view}` 
-                USING (`source`, `signal`, `geo_type`, `geo_value`, `time_type`, `time_value`) 
-            SET `{self.load_table}`.`is_latest_issue`=0 
-            WHERE `{self.load_table}`.`issue` < `{self.latest_view}`.`issue` 
+        UPDATE
+            `{self.load_table}` JOIN `{self.latest_view}`
+                USING (`source`, `signal`, `geo_type`, `geo_value`, `time_type`, `time_value`)
+            SET `{self.load_table}`.`is_latest_issue`=0
+            WHERE `{self.load_table}`.`issue` < `{self.latest_view}`.`issue`
     '''
 
     # TODO: consider handling cc_rows as a generator instead of a list

src/acquisition/covidcast/database.py

@@ -6,7 +6,7 @@
 import shutil
 
 # first party
-from delphi.epidata.common.logger import get_structured_logger
+from delphi_utils import get_structured_logger
 
 class FileArchiver:
   """Archives files by moving and compressing."""

src/acquisition/covidcast/file_archiver.py

@@ -15,7 +15,7 @@ Epidata <- (function() {
   # API base url
   BASE_URL <- getOption('epidata.url', default = 'https://api.delphi.cmu.edu/epidata/')
 
-  client_version <- '4.1.24'
+  client_version <- '4.1.25'
 
   auth <- getOption("epidata.auth", default = NA)
 

src/client/delphi_epidata.R

@@ -22,7 +22,7 @@
   }
 })(this, function (exports, fetchImpl, jQuery) {
   const BASE_URL = "https://api.delphi.cmu.edu/epidata/";
-  const client_version = "4.1.24";
+  const client_version = "4.1.25";
 
   // Helper function to cast values and/or ranges to strings
   function _listitem(value) {

src/client/delphi_epidata.js

@@ -18,7 +18,7 @@
 
 from aiohttp import ClientSession, TCPConnector, BasicAuth
 
-__version__ = "4.1.24"
+__version__ = "4.1.25"
 
 _HEADERS = {"user-agent": "delphi_epidata/" + __version__ + " (Python)"}
 
@@ -43,8 +43,6 @@ class Epidata:
     BASE_URL = "https://api.delphi.cmu.edu/epidata"
     auth = None
 
-    client_version = __version__
-
     debug = False # if True, prints extra logging statements
     sandbox = False # if True, will not execute any queries
 
@@ -54,6 +52,25 @@ def log(evt, **kwargs):
         kwargs['timestamp'] = time.strftime("%Y-%m-%d %H:%M:%S %z")
         return sys.stderr.write(str(kwargs) + "\n")
 
+    # Check that this client's version matches the most recent available.  This
+    # is intended to run just once per program execution, on initial module load.
+    # See the bottom of this file for the ultimate call to this method.
+    @staticmethod
+    def _version_check():
+        try:
+            request = requests.get('https://pypi.org/pypi/delphi-epidata/json', timeout=5)
+            latest_version = request.json()['info']['version']
+        except Exception as e:
+            Epidata.log("Error getting latest client version", exception=str(e))
+            return
+
+        if latest_version != __version__:
+            Epidata.log(
+                "Client version not up to date",
+                client_version=__version__,
+                latest_version=latest_version
+            )
+
     # Helper function to cast values and/or ranges to strings
     @staticmethod
     def _listitem(value):
@@ -692,3 +709,10 @@ async def async_make_calls(param_combos):
         future = asyncio.ensure_future(async_make_calls(param_list))
         responses = loop.run_until_complete(future)
         return responses
+
+
+
+# This should only run once per program execution, on initial module load,
+# as a result of how Python's module system works:
+# https://docs.python.org/3/reference/import.html#the-module-cache
+Epidata._version_check()

src/client/delphi_epidata.py

@@ -2,7 +2,7 @@
   "name": "delphi_epidata",
   "description": "Delphi Epidata API Client",
   "authors": "Delphi Group",
-  "version": "4.1.24",
+  "version": "4.1.25",
   "license": "MIT",
   "homepage": "https://github.com/cmu-delphi/delphi-epidata",
   "bugs": {

src/client/packaging/npm/package.json

@@ -1,5 +1,5 @@
 [bumpversion]
-current_version = 4.1.24
+current_version = 4.1.25
 commit = False
 tag = False
 

src/client/packaging/pypi/.bumpversion.cfg

@@ -3,6 +3,15 @@
 All notable future changes to the `delphi_epidata` python client will be documented in this file.
 The format is based on [Keep a Changelog](http://keepachangelog.com/).
 
+## [4.1.25] - 2024-07-29
+
+### Includes
+- https://github.com/cmu-delphi/delphi-epidata/pull/1456
+- https://github.com/cmu-delphi/delphi-epidata/pull/1497
+
+### Changed
+- Added a one-time check which logs a warning when the newest client version does not match the client version in use.
+
 ## [4.1.24] - 2024-07-09
 
 ### Includes