refacto: change variable name from fifteen to twenty (#259)

dataforgoodfr · Oct 9, 2024 · 763e1b0 · 763e1b0 · github-actions · Oct 9, 2024
1 parent a5bbe03
commit 763e1b0
Show file tree

Hide file tree

Showing 3 changed files with 93 additions and 92 deletions.
diff --git a/quotaclimat/data_processing/mediatree/detect_keywords.py b/quotaclimat/data_processing/mediatree/detect_keywords.py
@@ -21,6 +21,7 @@
 dask.config.set({'dataframe.query-planning': True})
 
 indirectes = 'indirectes'
+DEFAULT_WINDOW_DURATION = 20
 
 def get_cts_in_ms_for_keywords(subtitle_duration: List[dict], keywords: List[dict], theme: str) -> List[dict]:
     result = []
@@ -117,7 +118,7 @@ def remove_stopwords(plaintext: str) -> str:
 def get_themes_keywords_duration(plaintext: str, subtitle_duration: List[str], start: datetime):
     keywords_with_timestamp = []
     number_of_elements_in_array = 17
-    default_window_in_seconds = 20
+    default_window_in_seconds = DEFAULT_WINDOW_DURATION
     plaitext_without_stopwords = remove_stopwords(plaintext)
     logging.debug(f"display datetime start {start}")
 
@@ -200,7 +201,7 @@ def get_themes_keywords_duration(plaintext: str, subtitle_duration: List[str], s
     else:
         return [None] * number_of_elements_in_array
 
-def get_keywords_with_timestamp_with_false_positive(keywords_with_timestamp, start, duration_seconds: int = 15):
+def get_keywords_with_timestamp_with_false_positive(keywords_with_timestamp, start, duration_seconds: int = 20):
     logging.debug(f"Using duration_seconds {duration_seconds}")
 
     # Shallow copy to avoid unnecessary deep copying (wip: for memory leak)
@@ -330,7 +331,7 @@ def transform_false_positive_keywords_to_positive(keywords_with_timestamp: List[
 
     return keywords_with_timestamp
 
-def tag_wanted_duration_second_window_number(keywords_with_timestamp: List[dict], start, duration_seconds: int = 15) -> List[dict]:
+def tag_wanted_duration_second_window_number(keywords_with_timestamp: List[dict], start, duration_seconds: int = 20) -> List[dict]:
     window_size_seconds = get_keyword_time_separation_ms(duration_seconds=duration_seconds)
     total_seconds_in_window = get_chunk_duration_api()
     number_of_windows = int(total_seconds_in_window // window_size_seconds)

diff --git a/quotaclimat/data_processing/mediatree/utils.py b/quotaclimat/data_processing/mediatree/utils.py
@@ -11,7 +11,7 @@
 EPOCH__5MIN_MARGIN = 300
 EPOCH__1MIN_MARGIN = 60 # to add margin for program
 
-def get_keyword_time_separation_ms(duration_seconds: int = 15):
+def get_keyword_time_separation_ms(duration_seconds: int = 20):
     return duration_seconds * 1000
 
 def get_chunk_duration_api():
File	Stmts	Miss	Cover	Missing
postgres
insert_data.py	43	7	84%	36–38, 56–58, 63
insert_existing_data_example.py	19	3	84%	25–27
postgres/schemas
models.py	155	10	94%	126–133, 145–146, 211–212, 226–227
quotaclimat/data_ingestion
scrap_sitemap.py	134	17	87%	27–28, 33–34, 66–71, 95–97, 138–140, 202, 223–228
quotaclimat/data_ingestion/ingest_db
ingest_sitemap_in_db.py	55	37	33%	21–42, 45–58, 62–73
quotaclimat/data_ingestion/scrap_html
scrap_description_article.py	36	3	92%	19–20, 32
quotaclimat/data_processing/mediatree
api_import.py	211	133	37%	44–48, 53–74, 78–81, 87, 90–132, 138–153, 158, 171–183, 187–193, 206–218, 221–225, 231, 266–267, 270–301, 304–306
channel_program.py	165	57	65%	25–27, 38–40, 57–58, 61–63, 102–103, 112, 128, 179–220
config.py	15	2	87%	7, 16
detect_keywords.py	210	8	96%	221, 279–286
update_pg_keywords.py	67	49	27%	15–108, 132, 135, 142–157, 180–206, 213
utils.py	79	25	68%	29–53, 56, 65, 86–87, 117–120
quotaclimat/utils
healthcheck_config.py	29	14	52%	22–24, 27–38
logger.py	24	11	54%	22–24, 28–37
sentry.py	11	2	82%	22–23
TOTAL	1279	378	70%