boris-behav-obs 8.9.16__py3-none-any.whl → 9.7.6__py3-none-any.whl

boris/db_functions.py

@@ -1,7 +1,7 @@
 """
 BORIS
 Behavioral Observation Research Interactive Software
-Copyright 2012-2023 Olivier Friard
+Copyright 2012-2025 Olivier Friard
 
 
   This program is free software; you can redistribute it and/or modify
@@ -26,6 +26,7 @@ import logging
 from typing import Optional, Tuple
 from . import config as cfg
 from . import project_functions
+from . import event_operations
 
 
 def load_events_in_db(
@@ -55,8 +56,7 @@ def load_events_in_db(
     state_behaviors_codes = [
         pj[cfg.ETHOGRAM][x][cfg.BEHAVIOR_CODE]
         for x in pj[cfg.ETHOGRAM]
-        if cfg.STATE in pj[cfg.ETHOGRAM][x][cfg.TYPE].upper()
-        and pj[cfg.ETHOGRAM][x][cfg.BEHAVIOR_CODE] in selected_behaviors
+        if cfg.STATE in pj[cfg.ETHOGRAM][x][cfg.TYPE].upper() and pj[cfg.ETHOGRAM][x][cfg.BEHAVIOR_CODE] in selected_behaviors
     ]
 
     # selected behaviors defined as point event
@@ -70,6 +70,7 @@ def load_events_in_db(
     """
 
     db = sqlite3.connect(":memory:", isolation_level=None)
+
     """
     import os
     os.system("rm /tmp/ramdisk/events.sqlite")
@@ -98,25 +99,19 @@ def load_events_in_db(
     cursor.execute("CREATE INDEX modifiers_idx ON events(modifiers)")
 
     for subject_to_analyze in selected_subjects:
-
         for obs_id in selected_observations:
-
             for event in pj[cfg.OBSERVATIONS][obs_id][cfg.EVENTS]:
-
                 if event[cfg.EVENT_BEHAVIOR_FIELD_IDX] in selected_behaviors:
-
                     # extract time, code, modifier and comment (time:0, subject:1, code:2, modifier:3, comment:4)
                     if (subject_to_analyze == cfg.NO_FOCAL_SUBJECT and event[cfg.EVENT_SUBJECT_FIELD_IDX] == "") or (
                         event[cfg.EVENT_SUBJECT_FIELD_IDX] == subject_to_analyze
                     ):
-
                         if pj[cfg.OBSERVATIONS][obs_id][cfg.TYPE] in (cfg.MEDIA, cfg.LIVE):
-
                             cursor.execute(
                                 (
                                     "INSERT INTO events "
-                                    "(observation, subject, code, type, modifiers, occurence, comment) "
-                                    "VALUES (?,?,?,?,?,?,?)"
+                                    "(observation, subject, code, type, modifiers, occurence, comment, image_index) "
+                                    "VALUES (?,?,?,?,?,?,?,?)"
                                 ),
                                 (
                                     obs_id,
@@ -124,14 +119,12 @@ def load_events_in_db(
                                     if event[cfg.EVENT_SUBJECT_FIELD_IDX] == ""
                                     else event[cfg.EVENT_SUBJECT_FIELD_IDX],
                                     event[cfg.EVENT_BEHAVIOR_FIELD_IDX],
-                                    cfg.STATE
-                                    if event[cfg.EVENT_BEHAVIOR_FIELD_IDX] in state_behaviors_codes
-                                    else cfg.POINT,
+                                    cfg.STATE if event[cfg.EVENT_BEHAVIOR_FIELD_IDX] in state_behaviors_codes else cfg.POINT,
                                     event[cfg.EVENT_MODIFIER_FIELD_IDX],
-                                    float(event[cfg.EVENT_TIME_FIELD_IDX])
-                                    if not event[cfg.EVENT_TIME_FIELD_IDX].is_nan()
-                                    else None,
+                                    float(event[cfg.EVENT_TIME_FIELD_IDX]) if not event[cfg.EVENT_TIME_FIELD_IDX].is_nan() else None,
                                     event[cfg.EVENT_COMMENT_FIELD_IDX],
+                                    # frame index or NA
+                                    event_operations.read_event_field(event, pj[cfg.OBSERVATIONS][obs_id][cfg.TYPE], cfg.FRAME_INDEX),
                                 ),
                             )
 
@@ -148,13 +141,9 @@ def load_events_in_db(
                                     if event[cfg.EVENT_SUBJECT_FIELD_IDX] == ""
                                     else event[cfg.EVENT_SUBJECT_FIELD_IDX],
                                     event[cfg.EVENT_BEHAVIOR_FIELD_IDX],
-                                    cfg.STATE
-                                    if event[cfg.EVENT_BEHAVIOR_FIELD_IDX] in state_behaviors_codes
-                                    else cfg.POINT,
+                                    cfg.STATE if event[cfg.EVENT_BEHAVIOR_FIELD_IDX] in state_behaviors_codes else cfg.POINT,
                                     event[cfg.EVENT_MODIFIER_FIELD_IDX],
-                                    float(event[cfg.EVENT_TIME_FIELD_IDX])
-                                    if not event[cfg.EVENT_TIME_FIELD_IDX].is_nan()
-                                    else None,
+                                    float(event[cfg.EVENT_TIME_FIELD_IDX]) if not event[cfg.EVENT_TIME_FIELD_IDX].is_nan() else None,
                                     event[cfg.EVENT_COMMENT_FIELD_IDX],
                                     event[cfg.PJ_OBS_FIELDS[cfg.IMAGES][cfg.IMAGE_INDEX]],
                                     event[cfg.PJ_OBS_FIELDS[cfg.IMAGES][cfg.IMAGE_PATH]],
@@ -184,7 +173,7 @@ def load_aggregated_events_in_db(
 
     """
 
-    logging.debug(f"function: load_aggregated_events_in_db")
+    logging.debug("function: load_aggregated_events_in_db")
 
     # if no observation selected select all
     if not selected_observations:
@@ -192,20 +181,16 @@ def load_aggregated_events_in_db(
 
     # if no subject selected select all
     if not selected_subjects:
-        selected_subjects = sorted(
-            [pj[cfg.SUBJECTS][x][cfg.SUBJECT_NAME] for x in pj[cfg.SUBJECTS]] + [cfg.NO_FOCAL_SUBJECT]
-        )
+        selected_subjects = sorted([pj[cfg.SUBJECTS][x][cfg.SUBJECT_NAME] for x in pj[cfg.SUBJECTS]] + [cfg.NO_FOCAL_SUBJECT])
 
     # if no behavior selected select all
     if not selected_behaviors:
         selected_behaviors = sorted([pj[cfg.ETHOGRAM][x][cfg.BEHAVIOR_CODE] for x in pj[cfg.ETHOGRAM]])
 
     # check if state events are paired
-    out = ""
+    out: str = ""
     for obs_id in selected_observations:
-        r, msg = project_functions.check_state_events_obs(
-            obs_id, pj[cfg.ETHOGRAM], pj[cfg.OBSERVATIONS][obs_id], cfg.HHMMSS
-        )
+        r, msg = project_functions.check_state_events_obs(obs_id, pj[cfg.ETHOGRAM], pj[cfg.OBSERVATIONS][obs_id], cfg.HHMMSS)
         if not r:
             out += f"Observation: <strong>{obs_id}</strong><br>{msg}<br>"
     if out:
@@ -215,24 +200,23 @@ def load_aggregated_events_in_db(
     state_behaviors_codes = [
         pj[cfg.ETHOGRAM][x][cfg.BEHAVIOR_CODE]
         for x in pj[cfg.ETHOGRAM]
-        if cfg.STATE in pj[cfg.ETHOGRAM][x][cfg.TYPE].upper()
-        and pj[cfg.ETHOGRAM][x][cfg.BEHAVIOR_CODE] in selected_behaviors
+        if cfg.STATE in pj[cfg.ETHOGRAM][x][cfg.TYPE].upper() and pj[cfg.ETHOGRAM][x][cfg.BEHAVIOR_CODE] in selected_behaviors
     ]
 
     # selected behaviors defined as point event
     point_behaviors_codes = [
         pj[cfg.ETHOGRAM][x][cfg.BEHAVIOR_CODE]
         for x in pj[cfg.ETHOGRAM]
-        if cfg.POINT in pj[cfg.ETHOGRAM][x][cfg.TYPE].upper()
-        and pj[cfg.ETHOGRAM][x][cfg.BEHAVIOR_CODE] in selected_behaviors
+        if cfg.POINT in pj[cfg.ETHOGRAM][x][cfg.TYPE].upper() and pj[cfg.ETHOGRAM][x][cfg.BEHAVIOR_CODE] in selected_behaviors
     ]
 
     db = sqlite3.connect(":memory:")
 
+    # only for debugging
     """
     import os
-    os.system("rm /tmp/ramdisk/aggreg.sqlite")
-    db = sqlite3.connect("/tmp/ramdisk/aggreg.sqlite", isolation_level=None)
+    os.system("rm /tmp/ramdisk/aggregated_events.sqlite")
+    db = sqlite3.connect("/tmp/ramdisk/aggregated_events.sqlite", isolation_level=None)
     """
 
     db.row_factory = sqlite3.Row
@@ -272,12 +256,10 @@ def load_aggregated_events_in_db(
     )
 
     for obs_id in selected_observations:
-
         cursor1 = load_events_in_db(pj, selected_subjects, [obs_id], selected_behaviors)
 
         for subject in selected_subjects:
             for behavior in selected_behaviors:
-
                 cursor1.execute(
                     "SELECT DISTINCT modifiers FROM events WHERE subject=? AND code=? ORDER BY modifiers",
                     (
@@ -289,7 +271,6 @@ def load_aggregated_events_in_db(
                 rows_distinct_modifiers = list(x[0] for x in cursor1.fetchall())
 
                 for distinct_modifiers in rows_distinct_modifiers:
-
                     cursor1.execute(
                         (
                             "SELECT occurence, comment, image_index, image_path FROM events "
@@ -300,7 +281,6 @@ def load_aggregated_events_in_db(
                     rows = list(cursor1.fetchall())
 
                     for idx, row in enumerate(rows):
-
                         if behavior in point_behaviors_codes:
                             data = (
                                 obs_id,

boris/dev.py

@@ -0,0 +1,134 @@
+import sys
+import time
+import pprint
+import pandas as pd
+
+from . import project_functions
+from . import config as cfg
+from . import utilities as util
+
+
+def event_val(event, field, obs_type: str = cfg.MEDIA):
+    try:
+        return event[cfg.PJ_OBS_FIELDS[obs_type][field]]
+    except:
+        return None
+
+
+_, _, pj, _ = project_functions.open_project_json(sys.argv[1])
+
+# pprint.pprint(list(pj[cfg.OBSERVATIONS].keys()))
+
+# print()
+
+"""
+obs_id = "images NO TIME"
+pprint.pprint(pj[cfg.OBSERVATIONS][obs_id])
+"""
+
+state_events_list = util.state_behavior_codes(pj[cfg.ETHOGRAM])
+
+df_def = {
+    "observation id": pd.Series(dtype="str"),
+    "observation type": pd.Series(dtype="str"),
+    "observation description": pd.Series(dtype="str"),
+    "subject": pd.Series(dtype="str"),
+    "behavior": pd.Series(dtype="str"),
+    "modifier": pd.Series(dtype="str"),
+    "start": pd.Series(dtype="float"),
+    "stop": pd.Series(dtype="float"),
+    "duration": pd.Series(dtype="float"),
+}
+df = pd.DataFrame(df_def)
+l = []
+# print(df.info)
+
+for obs_id in pj[cfg.OBSERVATIONS]:
+    # print(obs_id)
+    obs_type = pj[cfg.OBSERVATIONS][obs_id][cfg.TYPE]
+    obs_descr = pj[cfg.OBSERVATIONS][obs_id][cfg.DESCRIPTION]
+    mem_idx = []
+    for idx, event in enumerate(pj[cfg.OBSERVATIONS][obs_id][cfg.EVENTS]):
+        if idx in mem_idx:
+            continue
+        start = event_val(event, cfg.TIME)
+        subject = event_val(event, cfg.SUBJECT)
+        behavior = event_val(event, cfg.BEHAVIOR_CODE)
+        modifier = event_val(event, cfg.MODIFIER)
+        if behavior in state_events_list:
+            stop = None
+            for idx2, event2 in enumerate(pj[cfg.OBSERVATIONS][obs_id][cfg.EVENTS][idx + 1 :], start=idx + 1):
+                subject2 = event_val(event2, cfg.SUBJECT)
+                behavior2 = event_val(event2, cfg.BEHAVIOR_CODE)
+                modifier2 = event_val(event2, cfg.MODIFIER)
+                if subject == subject2 and behavior == behavior2 and modifier == modifier2:
+                    stop = event_val(event2, cfg.TIME)
+                    mem_idx.append(idx2)
+                    l.append([obs_id, obs_type, obs_descr, subject, behavior, modifier, start, stop, stop - start])
+                    break
+            if stop is None:
+                print(obs_id, " not paired")
+                # print(pj[cfg.OBSERVATIONS][obs_id][cfg.EVENTS])
+                # print(f"{l=}")
+                # print(f"{mem_idx=}")
+
+                sys.exit()
+        else:
+            l.append([obs_id, obs_type, obs_descr, subject, behavior, modifier, start, start, 0])
+
+# print(pd.DataFrame(l))
+df = pd.concat(
+    [
+        df,
+        pd.DataFrame(
+            l,
+            columns=[
+                "observation id",
+                "observation type",
+                "observation description",
+                "subject",
+                "behavior",
+                "modifier",
+                "start",
+                "stop",
+                "duration",
+            ],
+        ),
+    ]
+)
+del l
+print("=" * 30)
+print("describe")
+print(df.describe())
+print("=" * 30)
+
+# print(f'{df["subject"].value_counts()=}')
+# print(f'{df["subject"].nunique()=}')
+
+pd.set_option("display.max_rows", None, "display.max_columns", None)
+
+print("=" * 30)
+print("mean")
+r = df.groupby(["subject", "behavior"])["duration"].mean()
+print(r)
+print("=" * 30)
+
+
+r = df.groupby(["observation id", "subject", "behavior"])
+print(r["start"])
+
+
+"""
+# replace value (for selecting a time interval)
+t1 = time.time()
+df.loc[df["stop"] > 10, "stop"] = 10
+print(time.time() - t1)
+print(df)
+"""
+
+
+"""
+t1 = 1
+t2 = 2
+print(df.query(f"`start` <= {t1} & `stop` >= {t2} & `duration` != 0"))
+"""