Spaces:
Sleeping
Sleeping
Update patentwiz/preprocess_data.py
Browse files- patentwiz/preprocess_data.py +18 -17
patentwiz/preprocess_data.py
CHANGED
@@ -81,29 +81,30 @@ def download_weekly_patents(year, month, day, logging):
|
|
81 |
)
|
82 |
return False
|
83 |
|
84 |
-
def filter_rf_patents(patents, keywords=None):
|
85 |
"""
|
86 |
-
Filters patents based on
|
87 |
Parameters:
|
88 |
-
patents (list):
|
89 |
keywords (list): Keywords to filter patents.
|
|
|
90 |
Returns:
|
91 |
-
list: Filtered patents
|
92 |
"""
|
93 |
if keywords is None:
|
94 |
-
keywords = [
|
95 |
-
|
96 |
-
|
97 |
-
|
98 |
-
|
99 |
-
|
100 |
-
|
101 |
-
|
102 |
-
|
103 |
-
|
104 |
-
|
105 |
-
|
106 |
-
|
107 |
|
108 |
def extract_patents(year, month, day, logging):
|
109 |
"""
|
|
|
81 |
)
|
82 |
return False
|
83 |
|
84 |
+
def filter_rf_patents(patents, keywords=None, fields=None):
|
85 |
"""
|
86 |
+
Filters patents based on keywords and specified fields.
|
87 |
Parameters:
|
88 |
+
patents (list): List of patents.
|
89 |
keywords (list): Keywords to filter patents.
|
90 |
+
fields (list): Fields to search for keywords (e.g., Title, Abstract, Claims).
|
91 |
Returns:
|
92 |
+
list: Filtered patents.
|
93 |
"""
|
94 |
if keywords is None:
|
95 |
+
keywords = ["Radio Frequency", "Antenna", "UAV", "Wireless Charging"] # Default keywords
|
96 |
+
if fields is None:
|
97 |
+
fields = ["Title", "Abstract"] # Default fields
|
98 |
+
|
99 |
+
filtered_patents = []
|
100 |
+
for patent in patents:
|
101 |
+
for field in fields:
|
102 |
+
field_content = patent.get(field.lower(), "")
|
103 |
+
if any(keyword.lower() in field_content.lower() for keyword in keywords):
|
104 |
+
filtered_patents.append(patent)
|
105 |
+
break
|
106 |
+
return filtered_patents
|
107 |
+
|
108 |
|
109 |
def extract_patents(year, month, day, logging):
|
110 |
"""
|