From 27721f020fe91944f732a0a7db211b51969d3ccd Mon Sep 17 00:00:00 2001
From: Sudhisha K <sudhisha@google.com>
Date: Thu, 2 Jan 2025 06:13:36 +0000
Subject: [PATCH] CDC_Environmental_Health_Toxicology_Refresh

---
 .../manifest.json                             | 104 ++++++++++++++++++
 .../parse_air_quality.py                      |   1 -
 2 files changed, 104 insertions(+), 1 deletion(-)
 create mode 100644 scripts/us_cdc/environmental_health_toxicology/manifest.json

diff --git a/scripts/us_cdc/environmental_health_toxicology/manifest.json b/scripts/us_cdc/environmental_health_toxicology/manifest.json
new file mode 100644
index 000000000..d23e915ef
--- /dev/null
+++ b/scripts/us_cdc/environmental_health_toxicology/manifest.json
@@ -0,0 +1,104 @@
+{
+    "import_specifications": [
+      {
+        "import_name": "CDC_PM25CensusTract",
+        "curator_emails": [
+          "sudhisha@google.com"
+        ],
+        "provenance_url": "https://data.cdc.gov/browse?category=Environmental+Health+%26+Toxicology",
+        "provenance_description": "Environmental Health Toxicology",
+        "scripts": [
+          "parse_air_quality.py CDC_PM25CensusTract"
+        ],
+        "import_inputs": [
+          {
+            "template_mcf": "PM25CensusTractPollution.tmcf",
+            "cleaned_csv": "output/PM2.5CensusTract_0.csv"
+          },
+          {
+            "template_mcf": "PM25CensusTractPollution.tmcf",
+            "cleaned_csv": "output/PM2.5CensusTract_1.csv"
+          },
+          {
+            "template_mcf": "PM25CensusTractPollution.tmcf",
+            "cleaned_csv": "output/PM2.5CensusTract_2.csv"
+          },
+          {
+            "template_mcf": "PM25CensusTractPollution.tmcf",
+            "cleaned_csv": "output/PM2.5CensusTract_3.csv"
+          }
+          
+        ],
+        "cron_schedule": "0 1  2 * *"
+      },
+      {
+        "import_name": "CDC_OzoneCensusTract",
+        "curator_emails": [
+          "sudhisha@google.com"
+        ],
+        "provenance_url": "https://data.cdc.gov/browse?category=Environmental+Health+%26+Toxicology",
+        "provenance_description": "Environmental Health Toxicology",
+        "scripts": [
+          "parse_air_quality.py CDC_OzoneCensusTract"
+        ],
+        "import_inputs": [
+          {
+            "template_mcf": "OzoneCensusTractPollution.tmcf",
+            "cleaned_csv": "output/Census_Tract_Level_Ozone_Concentrations_0.csv"
+          },
+          {
+            "template_mcf": "OzoneCensusTractPollution.tmcf",
+            "cleaned_csv": "output/PM2.Census_Tract_Level_Ozone_Concentrations_1.csv"
+          },
+          {
+            "template_mcf": "OzoneCensusTractPollution.tmcf",
+            "cleaned_csv": "output/Census_Tract_Level_Ozone_Concentrations_2.csv"
+          },
+          {
+            "template_mcf": "OzoneCensusTractPollution.tmcf",
+            "cleaned_csv": "output/Census_Tract_Level_Ozone_Concentrations_3.csv"
+          }
+          
+        ],
+        "cron_schedule": "0 1  3 * *"
+      },
+      {
+        "import_name": "CDC_PM25County",
+        "curator_emails": [
+          "sudhisha@google.com"
+        ],
+        "provenance_url": "https://data.cdc.gov/browse?category=Environmental+Health+%26+Toxicology",
+        "provenance_description": "Environmental Health Toxicology",
+        "scripts": [
+          "parse_air_quality.py CDC_PM25County"
+        ],
+        "import_inputs": [
+          {
+            "template_mcf": "PM25CountyPollution.tmcf",
+            "cleaned_csv": "output/PM25county.csv"
+          }
+          
+        ],
+        "cron_schedule": "0 1  4 * *"
+      },
+      {
+        "import_name": "CDC_OzoneCounty",
+        "curator_emails": [
+          "sudhisha@google.com"
+        ],
+        "provenance_url": "https://data.cdc.gov/browse?category=Environmental+Health+%26+Toxicology",
+        "provenance_description": "Environmental Health Toxicology",
+        "scripts": [
+          "parse_air_quality.py CDC_OzoneCounty"
+        ],
+        "import_inputs": [
+          {
+            "template_mcf": "OzoneCountyPollution.tmcf",
+            "cleaned_csv": "output/OzoneCounty.csv"
+          }
+          
+        ],
+        "cron_schedule": "0 1  5 * *"
+      }
+    ]
+  }
diff --git a/scripts/us_cdc/environmental_health_toxicology/parse_air_quality.py b/scripts/us_cdc/environmental_health_toxicology/parse_air_quality.py
index 61571aeac..2a0dba061 100644
--- a/scripts/us_cdc/environmental_health_toxicology/parse_air_quality.py
+++ b/scripts/us_cdc/environmental_health_toxicology/parse_air_quality.py
@@ -207,7 +207,6 @@ def clean_air_quality_data(file_path, output_file, importname):
                         elif "County" in file and "Ozone" in file:
                             data["statefips"] = data["statefips"].astype(
                                 str).str.zfill(2)
-                            print("checkingggg")
                             data["countyfips"] = data["countyfips"].astype(
                                 str).str.zfill(3)
                             data["dcid"] = "geoId/" + data["statefips"] + data[