From 2de0954bd3a9d9716c3c007c488c291f6dd07da8 Mon Sep 17 00:00:00 2001
From: veenstrajelmer <60435591+veenstrajelmer@users.noreply.github.com>
Date: Wed, 5 Feb 2025 00:18:57 +0100
Subject: [PATCH] added uhslc as backup for gshhs zipfile (#1112)

* added uhslc as backup for gshhs zipfile

* updated whatsnew
---
 dfm_tools/data.py | 29 ++++++++++++++++++++---------
 docs/whats-new.md |  2 ++
 2 files changed, 22 insertions(+), 9 deletions(-)

diff --git a/dfm_tools/data.py b/dfm_tools/data.py
index 1a3a773f2..11f2b0045 100644
--- a/dfm_tools/data.py
+++ b/dfm_tools/data.py
@@ -229,22 +229,33 @@ def gshhs_coastlines_shp() -> str:
     
     dir_testdata = get_dir_testdata()
     
-    fname = 'gshhg-shp-2.3.7.zip'
-    filepath_zip = os.path.join(dir_testdata,fname)
-    dir_gshhs = os.path.join(dir_testdata,'gshhg-shp-2.3.7')
+    fname = 'gshhg-shp-2.3.7'
+    filepath_zip = os.path.join(dir_testdata, f"{fname}.zip")
+    dir_gshhs = os.path.join(dir_testdata, fname)
+    
+    def download_gshhs(url):
+        url_base = url.split("/")[2]
+        fname_zip = url.split('/')[-1]
+        print(f'downloading "{fname_zip}" from {url_base} to cachedir')
+        # use allow_redirects=False to enforce this url
+        resp = requests.get(url, allow_redirects=False)
+        resp.raise_for_status() #raise HTTPError if url not exists
+        return resp
     
     #download zipfile if not present
     if not os.path.exists(filepath_zip) and not os.path.exists(dir_gshhs):
-        file_url = f'https://www.ngdc.noaa.gov/mgg/shorelines/data/gshhg/latest/{fname}'
-        print(f'downloading "{fname}" from www.ngdc.noaa.gov to cachedir')
-        r = requests.get(file_url, allow_redirects=True)
-        r.raise_for_status() #raise HTTPError if url not exists
+        file_url = f'https://www.ngdc.noaa.gov/mgg/shorelines/data/gshhg/latest/{fname}.zip'
+        resp = download_gshhs(file_url)
+        if resp.is_redirect:
+            print("failed (redirected), trying different source")
+            file_url = f'https://www.soest.hawaii.edu/pwessel/gshhg/{fname}.zip'
+            resp = download_gshhs(file_url)
         with open(filepath_zip, 'wb') as f:
-            f.write(r.content)
+            f.write(resp.content)
 
     #unzip zipfile if unzipped folder not present
     if not os.path.exists(dir_gshhs):
-        print(f'unzipping "{fname}"')
+        print(f'unzipping "{fname}.zip"')
         with zipfile.ZipFile(filepath_zip, 'r') as zip_ref:
             zip_ref.extractall(dir_gshhs)
     
diff --git a/docs/whats-new.md b/docs/whats-new.md
index 3289246ea..1d76ddaa4 100644
--- a/docs/whats-new.md
+++ b/docs/whats-new.md
@@ -5,6 +5,8 @@
 ### Feat
 - usage of outside time buffer in `dfmt.cmems_nc_to_ini()` so noon-centered or monthly timestamps are also supported in [#1087](https://github.com/Deltares/dfm_tools/pull/1087)
 - correct CMEMS daily mean data ("P1D-m") from midnight to noon by adding a 12-hour offset in `dfmt.download_CMEMS()` in [#1088](https://github.com/Deltares/dfm_tools/pull/1088)
+- updated cdsapi request to new format in [#1103](https://github.com/Deltares/dfm_tools/pull/1103)
+- added UHSLC backup for GSHHS data in [#1112](https://github.com/Deltares/dfm_tools/pull/1112)
 
 ### Fix
 - made p-drive paths for tide models and gesla3 work on linux also in [#1083](https://github.com/Deltares/dfm_tools/pull/1083) and [#1085](https://github.com/Deltares/dfm_tools/pull/1085)