(Feat) OceanCurrent script for Argo profiles (#1195)

* (Feat) OceanCurrent script for Argo profiles

* (Feat) OceanCurrent script for Argo profiles: add README

ARGO/oceancurrent/README.md

@@ -0,0 +1,48 @@
+# ARGO Profiles JSON Generator for the Ocean Current website
+
+This script processes directories containing `.gif` files and generates a `profiles.json` file per argo platform directory with metadata extracted from the filenames. The filenames follow a specific pattern, and the extracted information includes the date, profile number, and cycle number.
+
+## Directory Structure
+
+The script expects a directory structure as follows:
+
+/mnt/oceancurrent/website/profiles/
+├── platform_code1/
+│ ├── YYYYMMDD_profileNumber_cycleNumber.gif
+│ ├── YYYYMMDD_profileNumber_cycleNumber.html
+│ └── ...
+├── platform_code2/
+│ ├── YYYYMMDD_profileNumber_cycleNumber.gif
+│ ├── YYYYMMDD_profileNumber_cycleNumber.html
+│ └── ...
+└── ...
+
+
+Where:
+- `YYYYMMDD` is the date.
+- `profileNumber` is the profile number.
+- `cycleNumber` is the cycle number.
+
+## Script Overview
+
+The script performs the following steps:
+1. Lists all platform codes in the specified `PROFILES_PATH` directory.
+2. For each platform code, lists all files in the corresponding directory.
+3. Filters the filenames to include only `.gif` files.
+4. Extracts the date, profile number, and cycle number from each filename using a regular expression.
+5. Creates a `profiles.json` file per directory containing the extracted information.
+
+## Usage
+
+### Running the Script
+
+1. Confirm the `PROFILES_PATH` variable in the script to point to the base directory containing the platform code directories.
+2. Run the script using Python as a cronjob:
+
+```bash
+python3 argo_oceancurrent.py
+```
+
+## Unittesting
+Run the unittests with ```pytest```
+

ARGO/oceancurrent/argo_oceancurrent.py

@@ -0,0 +1,47 @@
+#!/usr/bin/env python3
+import os
+import re
+import json
+
+# Specify the directory path
+PROFILES_PATH = "/mnt/oceancurrent/website/profiles/"
+
+
+def main():
+
+    platform_codes = os.listdir(PROFILES_PATH)
+
+    for platform_code in platform_codes:
+        # List all files and directories in the specified directory
+        profile_path = os.path.join(PROFILES_PATH, platform_code)
+        all_files = os.listdir(profile_path)
+
+        filenames = [f for f in all_files if os.path.isfile(os.path.join(profile_path, f))]
+
+        # Regular expression pattern to match the gtif filenames
+        pattern = re.compile(r'(\d{8})_(\d+)_(\d+)\.gif')
+
+        profiles = []
+        # Iterate over each gtif
+        for filename in filenames:
+            match = pattern.match(filename)
+            if match:
+                date, profile_number, cycle_number = match.groups()
+                profiles.append({
+                    "date": date,
+                    "cycle": cycle_number,
+                    "filename": filename
+                })
+
+        profiles_json = json.dumps(profiles, indent=4)
+
+        # Write the JSON string to a file in the path of the profile
+        profile_json_path = os.path.join(profile_path, "profiles.json")
+
+        with open(profile_json_path, "w") as f:
+            f.write(profiles_json)
+
+        print(f"{profile_json_path} created successfully.")
+
+if __name__ == '__main__':
+    main()

ARGO/oceancurrent/test_argo_oceancurrent.py

@@ -0,0 +1,85 @@
+import os
+import json
+import shutil
+import tempfile
+import unittest
+from unittest.mock import patch
+
+from argo_oceancurrent import main, PROFILES_PATH
+
+
+class TestProfilesJsonGeneration(unittest.TestCase):
+
+    def setUp(self):
+        # Create a temporary directory
+        self.test_dir = tempfile.mkdtemp()
+        self.profiles_test_dir = os.path.join(self.test_dir, 'mnt/oceancurrent/website/profiles')
+
+        # Path to the existing test files
+        self.existing_test_files_path = os.path.join(os.path.dirname(__file__), 'tests')
+
+        # Copy all test files to the temporary directory
+        for item in os.listdir(self.existing_test_files_path):
+            s = os.path.join(self.existing_test_files_path, item)
+            d = os.path.join(self.test_dir, item)
+            if os.path.isdir(s):
+                shutil.copytree(s, d, False, None)
+            else:
+                shutil.copy2(s, d)
+
+    def tearDown(self):
+        # Remove the temporary directory and all its contents
+        shutil.rmtree(self.test_dir)
+
+    def test_profiles_json_generation(self):
+        with patch('argo_oceancurrent.PROFILES_PATH', new=self.profiles_test_dir):
+            # Run the main from argo_oceancurrent to create the profiles.json
+            main()
+
+            # Verify the generated profiles.json for one platform code
+            platform_code = '7901108'
+            generated_json_path = os.path.join(self.profiles_test_dir, platform_code, "profiles.json")
+            self.assertTrue(os.path.exists(generated_json_path), f"{generated_json_path} was not created")
+
+            with open(generated_json_path, 'r') as f:
+                generated_json = json.load(f)
+
+            # Define the expected JSON content based on the test files
+            expected_json = [
+                {
+                    "date": "20240415",
+                    "cycle": "4",
+                    "filename": "20240415_7901108_4.gif"
+                },
+                {
+                    "date": "20240506",
+                    "cycle": "6",
+                    "filename": "20240506_7901108_6.gif"
+                },
+                {
+                    "date": "20240516",
+                    "cycle": "7",
+                    "filename": "20240516_7901108_7.gif"
+                },
+                {
+                    "date": "20240405",
+                    "cycle": "3",
+                    "filename": "20240405_7901108_3.gif"
+                },
+                {
+                    "date": "20240426",
+                    "cycle": "5",
+                    "filename": "20240426_7901108_5.gif"
+                },
+                {
+                    "date": "20240526",
+                    "cycle": "8",
+                    "filename": "20240526_7901108_8.gif"
+                }
+            ]
+
+            self.assertEqual(generated_json, expected_json, f"The generated profiles.json content in {platform_code} is incorrect")
+
+if __name__ == '__main__':
+    unittest.main()
+

ARGO/oceancurrent/tests/mnt/oceancurrent/website/profiles/7901107/20240407_7901107_3.html

@@ -0,0 +1,7 @@
+<html> <title> 7901107 </title>
+<!-- 20240407_7901107_3.gif 20240407_7901107_3 20240417_7901107_4 -->
+<BODY BGCOLOR = "#FFFFFF">
+        <a name="top"></a> 
+        <P align = "left">
+<a href=../../index.htm>[<em>OceanCurrent</em>]</a><a href=20240407_7901107_3.html>[PREV]</a> <a href=20240417_7901107_4.html>[NEXT]</a> <a href=./index.html>[DATE INDEX]</a> <a href=../whatsshown.htm>[WHAT'S SHOWN]</a> <a href=http://imos.org.au/imostermsofuse.html>[Conditions of use]</a> <BR>
+<IMG src=20240407_7901107_3.gif></P></BODY> </html>

ARGO/oceancurrent/tests/mnt/oceancurrent/website/profiles/7901107/20240417_7901107_4.html

@@ -0,0 +1,7 @@
+<html> <title> 7901107 </title>
+<!-- 20240417_7901107_4.gif 20240407_7901107_3 20240427_7901107_5 -->
+<BODY BGCOLOR = "#FFFFFF">
+        <a name="top"></a> 
+        <P align = "left">
+<a href=../../index.htm>[<em>OceanCurrent</em>]</a><a href=20240407_7901107_3.html>[PREV]</a> <a href=20240427_7901107_5.html>[NEXT]</a> <a href=./index.html>[DATE INDEX]</a> <a href=../whatsshown.htm>[WHAT'S SHOWN]</a> <a href=http://imos.org.au/imostermsofuse.html>[Conditions of use]</a> <BR>
+<IMG src=20240417_7901107_4.gif></P></BODY> </html>

ARGO/oceancurrent/tests/mnt/oceancurrent/website/profiles/7901107/20240427_7901107_5.html

@@ -0,0 +1,7 @@
+<html> <title> 7901107 </title>
+<!-- 20240427_7901107_5.gif 20240417_7901107_4 20240508_7901107_6 -->
+<BODY BGCOLOR = "#FFFFFF">
+        <a name="top"></a> 
+        <P align = "left">
+<a href=../../index.htm>[<em>OceanCurrent</em>]</a><a href=20240417_7901107_4.html>[PREV]</a> <a href=20240508_7901107_6.html>[NEXT]</a> <a href=./index.html>[DATE INDEX]</a> <a href=../whatsshown.htm>[WHAT'S SHOWN]</a> <a href=http://imos.org.au/imostermsofuse.html>[Conditions of use]</a> <BR>
+<IMG src=20240427_7901107_5.gif></P></BODY> </html>

ARGO/oceancurrent/tests/mnt/oceancurrent/website/profiles/7901107/20240508_7901107_6.html

@@ -0,0 +1,7 @@
+<html> <title> 7901107 </title>
+<!-- 20240508_7901107_6.gif 20240427_7901107_5 20240518_7901107_7 -->
+<BODY BGCOLOR = "#FFFFFF">
+        <a name="top"></a> 
+        <P align = "left">
+<a href=../../index.htm>[<em>OceanCurrent</em>]</a><a href=20240427_7901107_5.html>[PREV]</a> <a href=20240518_7901107_7.html>[NEXT]</a> <a href=./index.html>[DATE INDEX]</a> <a href=../whatsshown.htm>[WHAT'S SHOWN]</a> <a href=http://imos.org.au/imostermsofuse.html>[Conditions of use]</a> <BR>
+<IMG src=20240508_7901107_6.gif></P></BODY> </html>

ARGO/oceancurrent/tests/mnt/oceancurrent/website/profiles/7901107/20240518_7901107_7.html

@@ -0,0 +1,7 @@
+<html> <title> 7901107 </title>
+<!-- 20240518_7901107_7.gif 20240508_7901107_6 20240528_7901107_8 -->
+<BODY BGCOLOR = "#FFFFFF">
+        <a name="top"></a> 
+        <P align = "left">
+<a href=../../index.htm>[<em>OceanCurrent</em>]</a><a href=20240508_7901107_6.html>[PREV]</a> <a href=20240528_7901107_8.html>[NEXT]</a> <a href=./index.html>[DATE INDEX]</a> <a href=../whatsshown.htm>[WHAT'S SHOWN]</a> <a href=http://imos.org.au/imostermsofuse.html>[Conditions of use]</a> <BR>
+<IMG src=20240518_7901107_7.gif></P></BODY> </html>

ARGO/oceancurrent/tests/mnt/oceancurrent/website/profiles/7901107/20240528_7901107_8.html

@@ -0,0 +1,7 @@
+<html> <title> 7901107 </title>
+<!-- 20240528_7901107_8.gif 20240518_7901107_7 20240528_7901107_8 -->
+<BODY BGCOLOR = "#FFFFFF">
+        <a name="top"></a> 
+        <P align = "left">
+<a href=../../index.htm>[<em>OceanCurrent</em>]</a><a href=20240518_7901107_7.html>[PREV]</a> <a href=20240528_7901107_8.html>[NEXT]</a> <a href=./index.html>[DATE INDEX]</a> <a href=../whatsshown.htm>[WHAT'S SHOWN]</a> <a href=http://imos.org.au/imostermsofuse.html>[Conditions of use]</a> <BR>
+<IMG src=20240528_7901107_8.gif></P></BODY> </html>

ARGO/oceancurrent/tests/mnt/oceancurrent/website/profiles/7901107/index.html

@@ -0,0 +1,13 @@
+<html> <title> 7901107 </title>
+<BODY BGCOLOR = "#FFFFFF">
+<a href="../../index.htm">[<em>OceanCurrent</em>]</a> <a href="../map/latest.html">[Argo map]</a><br>
+<br><a name=2024><b>----- 2024 -----</b></a>
+<br><b> Apr</b>
+<a href=20240407_7901107_3.html>07</a>
+<a href=20240417_7901107_4.html>17</a>
+<a href=20240427_7901107_5.html>27</a>
+<br><b> May</b>
+<a href=20240508_7901107_6.html>08</a>
+<a href=20240518_7901107_7.html>18</a>
+<a href=20240528_7901107_8.html>28</a>
+</BODY> </html>

ARGO/oceancurrent/tests/mnt/oceancurrent/website/profiles/7901107/latest.html

@@ -0,0 +1,7 @@
+<html> <title> 7901107 </title>
+<!-- 20240528_7901107_8.gif 20240518_7901107_7 20240528_7901107_8 -->
+<BODY BGCOLOR = "#FFFFFF">
+        <a name="top"></a> 
+        <P align = "left">
+<a href=../../index.htm>[<em>OceanCurrent</em>]</a><a href=20240518_7901107_7.html>[PREV]</a> <a href=20240528_7901107_8.html>[NEXT]</a> <a href=./index.html>[DATE INDEX]</a> <a href=../whatsshown.htm>[WHAT'S SHOWN]</a> <a href=http://imos.org.au/imostermsofuse.html>[Conditions of use]</a> <BR>
+<IMG src=20240528_7901107_8.gif></P></BODY> </html>

ARGO/oceancurrent/tests/mnt/oceancurrent/website/profiles/7901108/20240405_7901108_3.html

@@ -0,0 +1,7 @@
+<html> <title> 7901108 </title>
+<!-- 20240405_7901108_3.gif 20240405_7901108_3 20240415_7901108_4 -->
+<BODY BGCOLOR = "#FFFFFF">
+        <a name="top"></a> 
+        <P align = "left">
+<a href=../../index.htm>[<em>OceanCurrent</em>]</a><a href=20240405_7901108_3.html>[PREV]</a> <a href=20240415_7901108_4.html>[NEXT]</a> <a href=./index.html>[DATE INDEX]</a> <a href=../whatsshown.htm>[WHAT'S SHOWN]</a> <a href=http://imos.org.au/imostermsofuse.html>[Conditions of use]</a> <BR>
+<IMG src=20240405_7901108_3.gif></P></BODY> </html>

ARGO/oceancurrent/tests/mnt/oceancurrent/website/profiles/7901108/20240415_7901108_4.html

@@ -0,0 +1,7 @@
+<html> <title> 7901108 </title>
+<!-- 20240415_7901108_4.gif 20240405_7901108_3 20240426_7901108_5 -->
+<BODY BGCOLOR = "#FFFFFF">
+        <a name="top"></a> 
+        <P align = "left">
+<a href=../../index.htm>[<em>OceanCurrent</em>]</a><a href=20240405_7901108_3.html>[PREV]</a> <a href=20240426_7901108_5.html>[NEXT]</a> <a href=./index.html>[DATE INDEX]</a> <a href=../whatsshown.htm>[WHAT'S SHOWN]</a> <a href=http://imos.org.au/imostermsofuse.html>[Conditions of use]</a> <BR>
+<IMG src=20240415_7901108_4.gif></P></BODY> </html>

ARGO/oceancurrent/tests/mnt/oceancurrent/website/profiles/7901108/20240426_7901108_5.html

@@ -0,0 +1,7 @@
+<html> <title> 7901108 </title>
+<!-- 20240426_7901108_5.gif 20240415_7901108_4 20240506_7901108_6 -->
+<BODY BGCOLOR = "#FFFFFF">
+        <a name="top"></a> 
+        <P align = "left">
+<a href=../../index.htm>[<em>OceanCurrent</em>]</a><a href=20240415_7901108_4.html>[PREV]</a> <a href=20240506_7901108_6.html>[NEXT]</a> <a href=./index.html>[DATE INDEX]</a> <a href=../whatsshown.htm>[WHAT'S SHOWN]</a> <a href=http://imos.org.au/imostermsofuse.html>[Conditions of use]</a> <BR>
+<IMG src=20240426_7901108_5.gif></P></BODY> </html>

ARGO/oceancurrent/tests/mnt/oceancurrent/website/profiles/7901108/20240506_7901108_6.html

@@ -0,0 +1,7 @@
+<html> <title> 7901108 </title>
+<!-- 20240506_7901108_6.gif 20240426_7901108_5 20240516_7901108_7 -->
+<BODY BGCOLOR = "#FFFFFF">
+        <a name="top"></a> 
+        <P align = "left">
+<a href=../../index.htm>[<em>OceanCurrent</em>]</a><a href=20240426_7901108_5.html>[PREV]</a> <a href=20240516_7901108_7.html>[NEXT]</a> <a href=./index.html>[DATE INDEX]</a> <a href=../whatsshown.htm>[WHAT'S SHOWN]</a> <a href=http://imos.org.au/imostermsofuse.html>[Conditions of use]</a> <BR>
+<IMG src=20240506_7901108_6.gif></P></BODY> </html>

ARGO/oceancurrent/tests/mnt/oceancurrent/website/profiles/7901108/20240516_7901108_7.html

@@ -0,0 +1,7 @@
+<html> <title> 7901108 </title>
+<!-- 20240516_7901108_7.gif 20240506_7901108_6 20240526_7901108_8 -->
+<BODY BGCOLOR = "#FFFFFF">
+        <a name="top"></a> 
+        <P align = "left">
+<a href=../../index.htm>[<em>OceanCurrent</em>]</a><a href=20240506_7901108_6.html>[PREV]</a> <a href=20240526_7901108_8.html>[NEXT]</a> <a href=./index.html>[DATE INDEX]</a> <a href=../whatsshown.htm>[WHAT'S SHOWN]</a> <a href=http://imos.org.au/imostermsofuse.html>[Conditions of use]</a> <BR>
+<IMG src=20240516_7901108_7.gif></P></BODY> </html>

ARGO/oceancurrent/tests/mnt/oceancurrent/website/profiles/7901108/20240526_7901108_8.html

@@ -0,0 +1,7 @@
+<html> <title> 7901108 </title>
+<!-- 20240526_7901108_8.gif 20240516_7901108_7 20240526_7901108_8 -->
+<BODY BGCOLOR = "#FFFFFF">
+        <a name="top"></a> 
+        <P align = "left">
+<a href=../../index.htm>[<em>OceanCurrent</em>]</a><a href=20240516_7901108_7.html>[PREV]</a> <a href=20240526_7901108_8.html>[NEXT]</a> <a href=./index.html>[DATE INDEX]</a> <a href=../whatsshown.htm>[WHAT'S SHOWN]</a> <a href=http://imos.org.au/imostermsofuse.html>[Conditions of use]</a> <BR>
+<IMG src=20240526_7901108_8.gif></P></BODY> </html>

ARGO/oceancurrent/tests/mnt/oceancurrent/website/profiles/7901108/index.html

@@ -0,0 +1,13 @@
+<html> <title> 7901108 </title>
+<BODY BGCOLOR = "#FFFFFF">
+<a href="../../index.htm">[<em>OceanCurrent</em>]</a> <a href="../map/latest.html">[Argo map]</a><br>
+<br><a name=2024><b>----- 2024 -----</b></a>
+<br><b> Apr</b>
+<a href=20240405_7901108_3.html>05</a>
+<a href=20240415_7901108_4.html>15</a>
+<a href=20240426_7901108_5.html>26</a>
+<br><b> May</b>
+<a href=20240506_7901108_6.html>06</a>
+<a href=20240516_7901108_7.html>16</a>
+<a href=20240526_7901108_8.html>26</a>
+</BODY> </html>

ARGO/oceancurrent/tests/mnt/oceancurrent/website/profiles/7901108/latest.html

@@ -0,0 +1,7 @@
+<html> <title> 7901108 </title>
+<!-- 20240526_7901108_8.gif 20240516_7901108_7 20240526_7901108_8 -->
+<BODY BGCOLOR = "#FFFFFF">
+        <a name="top"></a> 
+        <P align = "left">
+<a href=../../index.htm>[<em>OceanCurrent</em>]</a><a href=20240516_7901108_7.html>[PREV]</a> <a href=20240526_7901108_8.html>[NEXT]</a> <a href=./index.html>[DATE INDEX]</a> <a href=../whatsshown.htm>[WHAT'S SHOWN]</a> <a href=http://imos.org.au/imostermsofuse.html>[Conditions of use]</a> <BR>
+<IMG src=20240526_7901108_8.gif></P></BODY> </html>