XSOARValueMetrics
Collects metrics for a time window and the top 20 incident types in that period. Creates a small CSV with four tables 1) All Incidents 2) Closed Incidents 3) Open Duration 4) SLA timer durations. The time window is expected to be a complete month specified by the "firstday" and "lastday" arguments. If partial months are used, the open durations and SLA metrics is the average of the last set of incidents found, while incident counts are incremented. The "slatimers" argument is a CSV list of custom SLA timer fields to include in the metrics. Example: slatimers="customsla1,customsla2,customsla3" The "filters" argument is a CSV list thats support the following field names to filter incidents on: status, notstatus, type, severity, owner: "severity" values are: unknown, information, low, medium, high, critical "status or notstatus" values are: pending, active, done, archive "type" is the name of a single incident type "owner" is the name of a single incident owner Example: filters ="type=typea,status=done,severity=high" If the "query" parameter is passed, the "filters" argument is ignored. The "query" parameter is a Lucene/Bleve search string in the incidents search box. The "query" string is used to select which incidents - do not specify any dates. These are controlled by the "firstday" and "lastday" parameters. If the "windowstart" and "windowend" parameters are passed with the name of timer fields, the duration is calculated from windowend.endDate - windowstart.startDate for the "UserWindow" SLA metric. The "mode" argument controls saving monthly statistics in this year's XSOAR list (a JSON object) as specified in the "thisyearlist" argument. The default mode is "increment" and expects the time windows for each query to be contiguous with no gaps or overlaps in the time window specified by the "firstday" and "lastday" arguments. If the time windows overlap, then incidents will be double counted. If there are gaps between time windows, then incidents may be missed. If the query needs to run and not update the saved statistics, use "mode=noupdate". In the event a month in the saved statistics becomes corrupted, this is corrected by using "mode=initialize" with the first day and the last day of the month to reset the values. The "computeduration" argument allows the overall incident open duration to be computed from the created and closed dates versus using the openDuration field. Set to "yes" to use computed duration. This is helpful when incidents do not have valid data in the openDuration field.
python · Community Common Dashboards
Source
import pandas as pd from datetime import datetime from collections import defaultdict from calendar import monthrange import csv import demistomock as demisto # noqa: F401 from CommonServerPython import * # noqa: F401 MAXINC = 2000 XDEBUG = True MONTHS_ABBR = [datetime.strptime(str(month), "%m").strftime("%b") for month in range(1, 13)] SEVERITY = {"unknown": 0, "information": 0.5, "low": 1, "medium": 2, "high": 3, "critical": 4} STATUS = {"pending": 0, "active": 1, "done": 2, "archive": 3} def LogMessage(message: str) -> str: if XDEBUG: timestr = datetime.now().strftime("%Y-%m-%d %H:%M:%S") return f"{timestr} | {message}\n" return "" def IncidentRecord(inc: dict, slatimers: list, windowstart: str, windowend: str, computeduration: str) -> dict: if computeduration == "yes": deltatime = ToDatetime(str(inc.get("closed"))) - ToDatetime(str(inc.get("created"))) duration = int(deltatime.total_seconds()) else: duration = int(inc.get("openDuration", 0)) record = { "type": inc.get("type"), "status": inc.get("status"), "created": inc.get("created"), "occurred": inc.get("occurred"), "duration": duration, "contime": "-1", "dettime": "-1", "remtime": "-1", "asstime": "-1", "tritime": "-1", "UserWindow": "-1", } for timer in slatimers: record[timer] = "-1" fields = ["containmentsla", "detectionsla", "remediationsla", "timetoassignment", "triagesla"] timers = ["contime", "dettime", "remtime", "asstime", "tritime", "UserWindow"] if inc.get("status") == STATUS["done"] and isinstance(inc.get("CustomFields"), dict): for field, timer in zip(fields, timers): if field in inc["CustomFields"] and inc["CustomFields"][field]["runStatus"] == "ended": record[timer] = inc["CustomFields"][field]["totalDuration"] for timer in slatimers: if timer in inc["CustomFields"] and inc["CustomFields"][timer]["runStatus"] == "ended": record[timer] = inc["CustomFields"][timer]["totalDuration"] if ( windowstart != "" and windowend != "" and windowstart in inc["CustomFields"] and windowend in inc["CustomFields"] and inc["CustomFields"][windowstart]["runStatus"] == "ended" and inc["CustomFields"][windowend]["runStatus"] == "ended" ): winduration = ToDatetime(inc["CustomFields"][windowend]["endDate"]) - ToDatetime( inc["CustomFields"][windowstart]["startDate"] ) record["UserWindow"] = winduration.total_seconds() return record def TopTwenty(incsumm: dict) -> list: sorted_inc = dict(sorted(incsumm.items(), key=lambda item: len(item[1]), reverse=True)) removeKeys = list(sorted_inc.keys())[20:] return removeKeys def MonthlyIncidents(removeKeys, monthly: dict) -> dict: for key in removeKeys: if key in monthly: del monthly[key] return monthly def BuildWindows(start_date_str, end_date_str): # Convert the input strings to datetime objects start_date = datetime.strptime(start_date_str, "%Y-%m-%d") end_date = datetime.strptime(end_date_str, "%Y-%m-%d") result_dates = [] current_date = start_date day = 1 # Increment the window and store the first and last dates until reaching the end date while current_date <= end_date: # Get the first day of the current month if current_date != start_date: start = current_date else: start = start_date day = start_date.day # Get the last day of the current window if current_date.year == end_date.year and current_date.month == end_date.month and current_date.day == end_date.day: last = end_date # Set the window to the next day, or last day of the month else: _, lastday = monthrange(current_date.year, current_date.month) day += 1 # check to ensure did not step past the end of the month if current_date.year == end_date.year and current_date.month == end_date.month and day >= end_date.day: day = end_date.day elif day >= lastday: day = lastday last = current_date.replace(day=day) # Create the start, end date tuples for each window traversed result_dates.append((start.strftime("%Y-%m-%d"), last.strftime("%Y-%m-%d"))) # Step to the next day of the window year = current_date.year month = current_date.month _, lastday = monthrange(year, month) day += 1 if day > lastday: if month == 12: year += 1 month = month % 12 + 1 day = 1 current_date = current_date.replace(year=year, month=month, day=day) return result_dates def ToDatetime(date: str): # 2023-12-02T23:20:47Z' or 2023-12-02T23:20:47.000Z' for pre python 3.11 and XSOAR 8 timestamps if date.endswith("Z"): isodate = date.split(".", 1)[0].replace("Z", "", 1) + "+00:00" return datetime.fromisoformat(isodate) else: count = date.count("-") if count == 2: return datetime.strptime(date.rsplit(".", 1)[0], "%Y-%m-%dT%H:%M:%S") else: date, tz_offset = date.rsplit("-", 1) return datetime.fromisoformat(f"{date.rsplit('.', 1)[0]}-{tz_offset}") def CloseMetrics(monthly_data: dict, records: list) -> str: # Process each record and count the closed incidents by month and 'type' for record in records: month = ToDatetime(record["created"]).strftime("%b") if record["type"] not in monthly_data[month]: monthly_data[month][record["type"]] = 0 if record["status"] == STATUS["done"]: monthly_data[month][record["type"]] += 1 return BuildCsv("Type", monthly_data) def CountMetrics(counts: dict, records: list) -> str: # Process each record and store the incident count by month and 'type' for record in records: month = ToDatetime(record["created"]).strftime("%b") if record["type"] not in counts[month]: counts[month][record["type"]] = 0 counts[month][record["type"]] += 1 return BuildCsv("Type", counts) def DurationMetrics(records: list) -> str: # Create a dictionary to hold the monthly data for each 'type' monthly_data: dict = defaultdict(lambda: defaultdict(list)) averages: dict = {month: {} for month in MONTHS_ABBR} # Process each record and aggregate the 'duration' data by month and 'type' for record in records: month = ToDatetime(record["created"]).strftime("%b") monthly_data[record["type"]][month].append(record["duration"]) # Calculate the average for each 'type' field for each month for inctype, months in monthly_data.items(): for month, durations in months.items(): averages[month][inctype] = int(sum(durations) / len(durations)) return BuildCsv("Type", averages) def SlaMetrics(records: list, slatimers: list) -> str: # Initialize a dictionary to hold the aggregated metrics by month fieldnames = ["contime", "dettime", "remtime", "asstime", "tritime", "UserWindow"] fieldnames.extend(slatimers) monthly_data: dict = {key: {month: [] for month in MONTHS_ABBR} for key in fieldnames} averages: dict = {month: {} for month in MONTHS_ABBR} for record in records: month = ToDatetime(record["created"]).strftime("%b") for field in fieldnames: if field in record and record[field] != -1: monthly_data[field][month].append(int(record[field])) for field, months in monthly_data.items(): for month, metrics in months.items(): length = len(metrics) total = 0 for m in metrics: if m == -1: length -= 1 else: total += m if length > 0: averages[month][field] = total / length else: averages[month][field] = 0 return BuildCsv("Metric", averages) def BuildCsv(key: str, data: dict) -> str: df = pd.DataFrame(data).fillna(0).astype(int) df[key] = df.index df = df.set_index(key) csv_data_string = df.to_csv() return csv_data_string def SplitRecords(records: list) -> tuple[list, list]: curyear = "" thisyear: list = [] lastyear: list = [] for r in records: year = r["created"].split("-")[0] if curyear == "": curyear = year if curyear == year: lastyear.append(r) else: thisyear.append(r) return lastyear, thisyear def GenerateTables(startday: str, endday: str, records: list, slatimers: list) -> tuple[str, dict, str, dict]: json_met: dict = {} json_met2: dict = {} json_met["YEAR"] = startday.split("-")[0] json_met2["YEAR"] = endday.split("-")[0] m = "" m2 = "" twoyears = json_met["YEAR"] != json_met2["YEAR"] if twoyears: records, records2 = SplitRecords(records) monthly: dict = {month: {} for month in MONTHS_ABBR} metrics = CountMetrics(monthly, records) json_met["Incidents"] = CsvToJson(metrics) m = metrics + "\n" if twoyears: monthly = {month: {} for month in MONTHS_ABBR} metrics2 = CountMetrics(monthly, records2) json_met2["Incidents"] = CsvToJson(metrics2) m2 = metrics2 + "\n" monthly: dict = {month: {} for month in MONTHS_ABBR} metrics = CloseMetrics(monthly, records) json_met["Closed Incidents"] = CsvToJson(metrics) m += metrics + "\n" if twoyears: monthly = {month: {} for month in MONTHS_ABBR} metrics2 = CloseMetrics(monthly, records2) json_met2["Closed Incidents"] = CsvToJson(metrics2) m2 += metrics2 + "\n" metrics = DurationMetrics(records) json_met["Incident Open Duration"] = CsvToJson(metrics) m += metrics + "\n" if twoyears: metrics2 = DurationMetrics(records2) json_met2["Incident Open Duration"] = CsvToJson(metrics2) m2 += metrics2 + "\n" metrics = SlaMetrics(records, slatimers) json_met["SLA Metrics"] = CsvToJson(metrics) m += metrics if twoyears: metrics2 = SlaMetrics(records2, slatimers) json_met2["SLA Metrics"] = CsvToJson(metrics2) m2 += metrics2 return m, json_met, m2, json_met2 def GetIncSmallWindow(w, page: int, curday: int, curhour: int, filters: dict, userquery: str): if userquery == "": query = { "page": page, "size": MAXINC, "fromdate": f"{w[curday]}T{curhour-4:02d}:00:00", "todate": f"{w[curday]}T{curhour-1:02d}:59:59", } query.update(filters) else: userquery += f" occurred:>={w[curday]}T{curhour-4:02d}:00:00 and occurred:<={w[curday]}T{curhour-1:02d}:59:59" query = {"page": page, "size": MAXINC, "query": userquery} return execute_command("getIncidents", query, extract_contents=False) def GetIncLargeWindow(w, page: int, filters: dict, userquery: str): if userquery == "": query = {"page": page, "size": MAXINC, "fromdate": f"{w[0]}T00:00:00", "todate": f"{w[1]}T23:59:59"} query.update(filters) else: userquery += f" occurred:>={w[0]}T00:00:00 and occurred:<={w[1]}T23:59:59" query = {"page": page, "size": MAXINC, "query": userquery} return execute_command("getIncidents", query, extract_contents=False) def ProcessResponse(w, response, monthly, period, inccount, slatimers, windowstart, windowend, computeduration): curmonth = w[0] if curmonth not in monthly: monthly[curmonth] = {} for inc in response[0]["Contents"]["data"]: rec = IncidentRecord(inc, slatimers, windowstart, windowend, computeduration) inccount += 1 inctype = rec["type"] if inctype not in monthly[curmonth]: monthly[curmonth][inctype] = [] monthly[curmonth][inctype].append(rec) if inctype not in period: period[inctype] = [] period[inctype].append(rec) return inccount, monthly, period def ValidArgs(args: dict) -> bool: array_args = ["status", "notstatus", "severity", "owner", "type"] return all(key in array_args for key, value in args.items()) def ValidFilter(fil: list) -> bool: if len(fil) != 2: return False k, v = fil filter_mappings: dict = {"status": STATUS, "notstatus": STATUS, "severity": SEVERITY} if k in filter_mappings: return v in filter_mappings[k] elif k in ["owner", "type"]: return True return False def BuildFilters(filters: list) -> dict: filtargs: dict = {} if len(filters) == 0: return filtargs filter_mappings: dict = {"status": STATUS, "notstatus": STATUS, "severity": SEVERITY} for f in filters: newfil = [item.strip() for item in f.split("=")] if not ValidFilter(newfil): continue key, val = newfil newval = filter_mappings.get(key, {}).get(val, val) if key == "severity": filtargs["level"] = newval else: filtargs[key] = newval return filtargs def CsvToJson(csv_text: str) -> dict: lines = csv_text.strip("\n").split("\n") reader = csv.reader(lines) data = list(reader) header = data[0] # Month labels series = [row[0] for row in data[1:]] # Series labels json_data = {} for i, row in enumerate(data[1:]): series_data = {} for j, cell in enumerate(row[1:]): month = header[j + 1] series_data[month] = cell json_data[series[i]] = series_data return json_data def RollYearList(thisyearlist: str, lastyearlist: str, curmetrics: dict): existing_metrics = LoadJsonList(thisyearlist) if "YEAR" in existing_metrics and existing_metrics["YEAR"] != curmetrics["YEAR"]: SaveJsonList(lastyearlist, existing_metrics) existing_metrics = {} SaveJsonList(thisyearlist, existing_metrics) def UpdateMetricsList(listname: str, curmetrics: dict, mode: str): existing_metrics = LoadJsonList(listname) for key, val in curmetrics.items(): if key in existing_metrics and key != "YEAR": if key not in ["SLA Metrics", "Incident Open Duration"]: existing_metrics[key] = UpdateDict(existing_metrics[key], val, mode) else: existing_metrics[key] = UpdateDict(existing_metrics[key], val, "initialize") else: existing_metrics[key] = val SaveJsonList(listname, existing_metrics) def UpdateDict(existing_dict: dict, new_dict: dict, mode: str) -> dict: for newkey, newvalue in new_dict.items(): if newkey in existing_dict: for sub_key, sub_value in newvalue.items(): if sub_key in existing_dict[newkey]: if mode == "increment": existing_dict[newkey][sub_key] = str(int(existing_dict[newkey][sub_key]) + int(sub_value)) elif mode == "initialize": existing_dict[newkey][sub_key] = sub_value else: existing_dict[newkey][sub_key] = sub_value else: existing_dict[newkey] = newvalue return existing_dict def LoadJsonList(list_name: str) -> dict: results = demisto.executeCommand("getList", {"listName": list_name})[0]["Contents"] if "Item not found" not in results: return json.loads(results) return {} def SaveJsonList(list_name: str, json_data: dict): res = demisto.executeCommand( "core-api-post", {"uri": "/lists/save", "body": {"name": list_name, "data": json.dumps(json_data), "type": "json"}} )[0]["Contents"] # If error, existing list, so set the list contents if "Script failed to run" in res: demisto.executeCommand("setList", {"listName": list_name, "listData": json.dumps(json_data)}) def NormalDate(date_str: str, first_day=True) -> str: if len(date_str.split("-")) == 3: return date_str year, month = map(int, date_str.split("-")) if first_day: return f"{year}-{month:02d}-01" else: _, last_day = monthrange(year, month) return f"{year}-{month:02d}-{last_day:02d}" def FoundIncidents(res: List): if res and isinstance(res, list) and isinstance(res[0].get("Contents"), dict): if "data" not in res[0]["Contents"]: raise DemistoException(res[0].get("Contents")) elif res[0]["Contents"]["data"] is None: return False return True return None def main(): try: XLOG = "\n" XLOG += LogMessage("Starting Incident Search") inccount = 0 page = 0 period: dict = {} monthly: dict = {} arguments = demisto.args() firstday = NormalDate(arguments["firstday"]) lastday = NormalDate(arguments["lastday"], first_day=False) esflag = arguments["esflag"] thisyear_list = arguments["thisyearlist"] lastyear_list = arguments["lastyearlist"] windowstart = arguments.get("windowstart", "") windowend = arguments.get("windowend", "") computeduration = arguments.get("computeduration", "no") mode = arguments["mode"] query = arguments.get("query", "") filters = BuildFilters([item.strip().lower() for item in arguments.get("filters", "").split(",")]) timers = arguments.get("slatimers") if timers: slatimers = [item.strip().lower() for item in timers.split(",")] else: slatimers = [] windows = BuildWindows(firstday, lastday) for w in windows: XLOG += LogMessage(f"Start Two Day Window: {w[0]} | End: {w[1]} | {inccount}, {page}") page = 0 while True: if esflag == "false": response: List = GetIncLargeWindow(w, page, filters, query) if not FoundIncidents(response): break inccount, monthly, period = ProcessResponse( w, response, monthly, period, inccount, slatimers, windowstart, windowend, computeduration ) page += 1 # Switch to 4 hour window if the ES flag is set since it thows error next page if # 10000 or more incidents were found even while paging through a smaller size page else: curday = 0 curhour = 4 page = 0 # Process all the 4 hour windows in the two days while True: response = GetIncSmallWindow(w, page, curday, curhour, filters, query) if FoundIncidents(response): inccount, monthly, period = ProcessResponse( w, response, monthly, period, inccount, slatimers, windowstart, windowend, computeduration ) page += 1 # If no incidents found, step to the next 4 hour window else: curhour += 4 page = 0 # Are we done with the current day if curhour > 24: curhour = 4 # If on the second day of two day window, reset and start a new 2 day window if curday == 1: curday = 0 break # On the first day of the 2 day window, step to the second day else: # noqa: RET508 curday = 1 XLOG += LogMessage(f"Total Found Incident Count {inccount}") # Limit the results to the top twenty incident types removeKeys = TopTwenty(period) for k, m in monthly.items(): monthly[k] = MonthlyIncidents(removeKeys, m) # Create the list of records from the collected values records = [] for m in monthly.values(): for r in m.values(): records.extend(r) XLOG += LogMessage(f"Top Twenty Incident Count: {len(records)}") metrics, json_metrics, metrics2, json_metrics2 = GenerateTables(firstday, lastday, records, slatimers) if mode != "noupdate": if mode != "initialize": RollYearList(thisyear_list, lastyear_list, json_metrics) if "Incidents" not in json_metrics2: UpdateMetricsList(thisyear_list, json_metrics, mode) else: UpdateMetricsList(lastyear_list, json_metrics, mode) UpdateMetricsList(thisyear_list, json_metrics2, mode) return_results(fileResult("xsoar_value_metrics.csv", metrics)) if "Incidents" in json_metrics2: return_results(fileResult("xsoar_value_metrics2.csv", metrics2)) return_results(XLOG) except Exception as ex: demisto.error(traceback.format_exc()) return_error(f"XSOARValueMetrics: Exception failed to execute. Error: {ex!s}\n{XLOG}\n") if __name__ in ("__main__", "__builtin__", "builtins"): main()
README
Collects metrics for a time window and the top 20 incident types in that period. Creates a small CSV with four tables 1) All Incidents 2) Closed Incidents 3) Open Duration 4) SLA timer durations. The time window is expected to be a complete month specified by the “firstday” and “lastday” arguments. If partial months are used, the open durations and SLA metrics is the average of the last set of incidents found, while incident counts are incremented.
The “slatimers” argument is a CSV list of custom SLA timer fields to include in the metrics.
Example: slatimers="customsla1,customsla2,customsla3"
The “filters” argument is a CSV list thats support the following field names to filter incidents on: status, notstatus, type, severity, owner:
"severity" values are: unknown, information, low, medium, high, critical
"status or notstatus" values are: pending, active, done, archive
"type" is the name of a single incident type
"owner" is the name of a single incident owner
Example: filters ="type=typea,status=done,severity=high"
If the “query” parameter is passed, the “filters” argument is ignored. The “query” parameter is a Lucene/Bleve search string in the incidents search box. The “query” string is used to select which incidents - do not specify any dates. These are controlled by the “firstday” and “lastday” parameters.
If the “windowstart” and “windowend” parameters are passed with the name of timer fields, the duration is calculated from windowend.endDate - windowstart.startDate for the “UserWindow” SLA metric.
The “mode” argument controls saving monthly statistics in this year’s XSOAR list (a JSON object) as specified in the “thisyearlist” argument. The default mode is “increment” and expects the time windows for each query to be contiguous with no gaps or overlaps in the time window specified by the “firstday” and “lastday” arguments. If the time windows overlap, then incidents will be double counted. If there are gaps between time windows, then incidents may be missed. If the query needs to run and not update the saved statistics, use “mode=noupdate”. In the event a month in the saved statistics becomes corrupted, this is corrected by using “mode=initialize” with the first day and the last day of the month to reset the values.
The “computeduration” argument allows the overall incident open duration to be computed from the created and closed dates versus using the openDuration field. Set to “yes” to use computed duration. This is helpful when incidents do not have valid data in the openDuration field.
Script Data
| Name | Description |
|---|---|
| Script Type | python3 |
Inputs
| Argument Name | Description |
|---|---|
| firstday | First day to find incident occurrences. Example: “2023-03-01”. “2023-01” defaults to “2023-01-01”. |
| lastday | Last day to find incident occurrences. Does not support a different year than the first day.Example: “2023-06-30”. “2023-01” defaults to “2023-01-31”. |
| esflag | If using Elasticsearch, set this to “true”. Elasticsearch has a 10000 incident search limit and this flag reduces the search windows from 2 days to to 4 hours. |
| slatimers | CSV list of additional SLA timer fields to include in the metrics. Example: slatimers=”customsla1,customsla2,customsla3”. |
| filters | CSV list of incident filters. Example: filters =”type=typea,status=done”. |
| thisyearlist | List name to store this year’s monthly result. |
| lastyearlist | List name to store last year’s results. |
| mode | Controls the how statistics are saved to this years data list. |
| query | Query string for incidents using Lucene/Bleve. Do not include any date related fields - those are controlled by the “firstday” and “lastday” arguments. Ignores the “filter” argument. |
| windowstart | Timer field name to use as the start time for a user defined duration. |
| windowend | Timer field name to use as the end time for a user defined duration. |
| computeduration | Compute the incident open duration from the created and closed dates. |
Outputs
There are no outputs for this script.