From f7cc17a44663c3cae7dded268fb9ffb0f0a053e8 Mon Sep 17 00:00:00 2001
From: Shanrahan16 <sarah.hanrahan@stfc.ac.uk>
Date: Fri, 1 Sep 2023 15:01:06 +0100
Subject: [PATCH] #35 instrument controlled vocab check

---
 checksit/rules/rule_funcs.py                  | 37 +++++++++++++++++++
 .../amof-image-global-attrs.yml               |  2 +-
 2 files changed, 38 insertions(+), 1 deletion(-)

diff --git a/checksit/rules/rule_funcs.py b/checksit/rules/rule_funcs.py
index bfd090e5..348b0203 100644
--- a/checksit/rules/rule_funcs.py
+++ b/checksit/rules/rule_funcs.py
@@ -2,9 +2,12 @@
 import re
 from datetime import datetime
 import requests
+import json
+import pandas as pd
 
 from . import processors
 from ..config import get_config
+from pandas import json_normalize
 
 conf = get_config()
 rule_splitter = conf["settings"].get("rule_splitter", "|")
@@ -187,6 +190,40 @@ def title_check(value, context, extras=None, label=""):
     return errors
 
 
+def title_instrument(value, context, extras=None, label=""):
+    """
+    A function to check if the instrument in the title is contained in the controlled vocabulary list
+    """
+    warnings = []
+
+    instrument = value.partition("_")[0]
+
+    # open JSON controlled vocab file:
+    f = open ('./checksit/vocabs/AMF_CVs/2.0.0/AMF_ncas_instrument.json', "r")
+ 
+    ## Reading from file:
+    data = json.loads(f.read())
+    
+    #print(type(data))   #delete
+    #descriptn = data.description.map(item,item.description)
+
+    #d = json.loads('./checksit/vocabs/AMF_CVs/2.0.0/AMF_ncas_instrument.json')
+    
+    #df = pd.json_normalize(data)
+    #print(df)  #delete
+    #print(type(df))   #delete
+
+    #if instrument not in data['ncas_instrument'][*]['description']:
+    #if instrument not in descriptn:
+    if instrument not in data['ncas_instrument']:
+        warnings.append(f"{label} '{value}' should be contained in the instrument controlled vocabulary list")
+
+    # Closing file
+    f.close()
+
+    return warnings
+
+
 def url_checker(value, context, extras=None, label=""):
     """
     A function to check if the url exists
diff --git a/specs/groups/ncas-image-v1.0/amof-image-global-attrs.yml b/specs/groups/ncas-image-v1.0/amof-image-global-attrs.yml
index 11a7439e..19c1a617 100644
--- a/specs/groups/ncas-image-v1.0/amof-image-global-attrs.yml
+++ b/specs/groups/ncas-image-v1.0/amof-image-global-attrs.yml
@@ -15,7 +15,7 @@ required-global-attrs:
       XMP-xmpRights:WebStatement: regex-warning:http://www.nationalarchives.gov.uk/doc/open-government-licence/
       XMP-photoshop:Credit: regex-warning:National\sCentre\sfor\sAtmospheric\sScience\s\(NCAS\)
       # Source of the image
-      XMP-dc:Title: regex-rule:title-data-product, rule-func:title_check
+      XMP-dc:Title: regex-rule:title-data-product, rule-func:title_check, rule-func-warning:title_instrument
       # Instrument Scientist’s Details
       XMP-dc:Creator: rule-func-warning:list_of_names
       XMP-iptcCore:CreatorWorkEmail: regex-rule:work-email