LittleBock/extract.py

@@ -2,27 +2,24 @@
 """
 extracteur de données iSpindel JSON entreposées par Little Bock en HTML
 """
-import json, pathlib
+import json
 try:
     import lxml.html as LX
 except ModuleNotFoundError:
     import sys
     sys.exit("Le module 'lxml' est nécessaire.\n\thttp://pypi.org/lxml")
+import pathlib
 
 def jsproc(data):
-    """supprime les données inutiles d'une série"""
-    unwanted = ("color","opacity","yAxis","dashStyle", "tooltip",)
-    if data.get("data"):
-        for u in unwanted:
-            if u in data.keys():
-                if u == "tooltip":
-                    tt = data.get(u)
-                    suffix = tt.get("valueSuffix")
-                    if suffix: data.update({
-                        "name": " ".join([data["name"], suffix])
-                    })
-                del(data[u])
-        return data
+    """nettoie des données récursivement"""
+    unwanted = ["color","opacity","yAxis","dashStyle"]
+    for u in unwanted:
+        if u in data.keys():
+            del(data[u])
+    for k,v in data.items():
+        if isinstance(v,dict):
+            jsproc(v)
+    return data
 
 def hproc(path_in):
     """traite un document HTML"""
@@ -32,13 +29,11 @@ def hproc(path_in):
     if x: data1 = x[0].get('data-chart-options')
     if data1:
         for i in json.loads(data1).pop('series'):
-            data = jsproc(i)
-            if data: data2.append(data)
+            data2.append(jsproc(i))
     if len(data2) > 0:
         path_out = path_in.with_suffix('.json')
         with open(path_out,'w') as f:
-            f.write(json.dumps(data2,
-                sort_keys=False, ensure_ascii=False, indent=2))
+            f.write(json.dumps(data2, indent=2, sort_keys=False))
             print(f"INFO: {path_in.name} >> {path_out.name}")
 
 if __name__ == "__main__":