.. _field-examples: Fields ====== Driver supports all field types available in Swimlane platform with automatic validation and coercion to best Python representations of field values. TrackingIdField --------------- The "Tracking Id" field is always readonly, and exists on all records from all apps .. code-block:: python assert record['Tracking Id'] == 'RA-7' try: record['Tracking Id'] = 'New Tracking ID' except ValidationError: assert record['Tracking Id'] == 'RA-7' TextField --------- Supports automatic :class:`str` type coercion for non-string types Not all types are supported. Here are some examples of valid values: - Number: "24352435234", 2343 - Text: "24352435234", "Henry" - Date: "24352435234" as a DateTime ONLY - Tracking Id: "23452435234", "SW-132" Assume the example below has a record with a text field "Severity" with the value "7": .. code-block:: python # Coerce to int, add, and automatically set back to str type in field assert record['Severity'] == '7' # Update value, setting the text field to an int record['Severity'] = int(record['Severity']) + 1 # Value has been coerced to a string assert isinstance(record['Severity'], str) assert record['Severity'] == '8' NumericField ------------ Number type enforcement validation. Supports any subclass of :class:`numbers.Number` .. code-block:: python assert isinstance(record['Numeric'], numbers.Number) try: record['Numeric'] = 'Not a Number' except ValidationError: record['Numeric'] = 5 assert record['Numeric'] == 5 DatetimeField ------------- Full support for all valid Swimlane datetime subtypes: - Date & Time - Date - Time - Timespan All datetime values throughout driver are returned as Pendulum_ objects. .. _Pendulum: https://pendulum.eustace.io/ Fields can be set to appropriate :class:`~datetime.datetime` types or respective :class:`Pendulum` instances, and will always be converted to corresponding Pendulum type matching the field subtype. Datetime subtypes (not including Timespan) can be set to full datetime instances, and any extraneous information will be discarded. Timezone support is automatic, and can mostly be ignored when using Pendulum instances. All instances returned from server are set to UTC. Date & Time ^^^^^^^^^^^ Full specific date & time field Returns :class:`pendulum.DateTime` instances .. code-block:: python datetime_field_value = record['Datetime'] # Drop-in replacement and extension of Python's builtin datetime assert isinstance(datetime_field_value, datetime.datetime) assert isinstance(datetime_field_value, pendulum.Pendulum) assert datetime_field_value.year == 2017 assert datetime_field_value.month == 4 assert datetime_field_value.day == 10 assert datetime_field_value.hour == 16 # Set to a datetime/pendulum instance # See warning below record['Datetime'] = now = pendulum.now().replace(microseconds=0) # Fail on generic timestamp/string (not enough context to guarantee consistent behavior) try: record['Datetime'] = '2017-05-11 11:10:09' except ValidationError: pass assert record['Datetime'] == now .. warning:: Mongo only supports millisecond resolution, datetimes returned from Swimlane API lose nanosecond resolution, leading to potentially slightly inconsistent datetimes before and after saving a record. For consistency, nanoseconds are automatically stripped from datetimes when the field is set to a datetime with nanosecond precision. Field equality comparisons with `pendulum.now()` or other datetime instances with nanosecond resolution will not be accurate unless the nanoseconds are manually removed from the compared datetime. .. code-block:: python # 2017-09-20 12:34:56.987654 now = pendulum.now() # 2017-09-20 12:34:56.987000 record['Datetime'] = now assert record['Datetime'] != now For guaranteed equality checks, simplest solution is to remove the microsecond component entirely when setting the field value in cases where sub-second resolution isn't important. .. code-block:: python # 2017-09-20 12:34:56.000000 now = pendulum.now().replace(microsecond=0) # 2017-09-20 12:34:56.000000 record['Datetime'] = now assert record['Datetime'] == now Manual rounding or less/greater than comparisons are necessary in cases where millisecond resolution is necessary .. code-block:: python ## Rounding comparison # 2017-09-20 12:34:56.987654 now = pendulum.now() # 2017-09-20 12:34:56.987000 record['Datetime'] = now # 2017-09-20 12:34:56.987000 rounded_now = now.replace( microsecond=math.floor(now.microsecond / 1000) * 1000 ) assert record['Datetime'] == rounded_now ## Proximity comparison # 0.000654 assert abs((record['Datetime'] - now).total_seconds()) < 0.001 Date ^^^^ Date of year with no time component (2017-06-01). Returns :class:`pendulum.Date` instances .. code-block:: python date_field = record['Date'] assert isinstance(date_field, datetime.date) assert isinstance(date_field, pendulum.Date) # Set to full datetime, time portion is dropped and Date instance is always returned record['Date'] = pendulum.now() assert isinstance(record['Date'], pendulum.Date) # Set to just date record['Date'] = pendulum.now().date() assert isinstance(record['Date'], pendulum.Date) Time ^^^^ Time of day with no date component (12:34:56). Returns :class:`pendulum.Time` instances .. code-block:: python time_field = record['Time'] assert isinstance(time_field, datetime.time) assert isinstance(time_field, pendulum.Time) # Set to full datetime, date portion is dropped and Time instance is always returned record['Time'] = pendulum.now() assert isinstance(record['Time'], pendulum.Time) # Set to just time record['Time'] = pendulum.now().time() assert isinstance(record['Time'], pendulum.Time) .. warning:: Time instances do not respect timezone information, and should always be provided in UTC. Recommend using full Pendulum datetime instances when working with Time fields. When using full datetimes, the timezone is respected before dropping the date portion. Timespan ^^^^^^^^ Time period (2 hours, 4 minutes, 15 seconds). Returns :class:`pendulum.Duration` instances .. code-block:: python timespan_field = record['Timespan'] assert isinstance(timespan_field, datetime.timedelta) assert isinstance(timespan_field, pendulum.Duration) .. note:: Only subtype that cannot handle datetime/Pendulum instances. Must use datetime.timedelta or pendulum.Duration instances instead. ValuesListField --------------- Enforces valid selection options available in UI. Single Select ^^^^^^^^^^^^^ Single-select mode values are accessed and set directly .. code-block:: python # Valid option enforcement record['Status'] = 'Open' try: record['Status'] = 'Not a valid option' except ValidationError: record['Status'] = 'Closed' assert record['Status'] == 'Closed' Multi Select ^^^^^^^^^^^^ Uses a cursor that behaves similar to a standard list to provide selection functionality and value enforcement. .. code-block:: python # Uses cursor for multi-select support with support for select, deselect, iteration, etc. vl_cursor = record['Values List'] assert len(vl_cursor) == 2 # Adding the same value multiple times is ignored vl_cursor.select('Option 3') assert len(vl_cursor) == 3 vl_cursor.select('Option 3') assert len(vl_cursor) == 3 # Remove element raises exception if not already added vl_cursor.deselect('Option 3') assert len(vl_cursor) == 2 try: vl_cursor.deselect('Option 3') except KeyError: assert len(vl_cursor) == 2 # Respects field's valid options and types, raising ValidationError for invalid values try: vl_cursor.select('Not a valid option') except ValidationError: assert len(vl_cursor) == 2 Field can be set directly to any iterable, overwriting current selection entirely .. code-block:: python vl_original_values = list(record['Values List']) record['Values List'] = [] assert len(record['Values List']) == 0 # All elements must pass validation, or entire set operation fails try: record['Values List'] = ['Option 1', 'Not a valid option'] except ValidationError: assert len(record['Values List']) == 0 record['Values List'] = vl_original_values assert len(record['Values List']) == 2 ListField --------- Text and numeric list field. Uses a :class:`TextListFieldCursor` or a :class:`NumericListFieldCursor` depending on the field type to enforce min/max item count restrictions, min/max character/word limits, and numeric range restrictions. Cursor works exactly like a normal primitive Python :class:`list` with added validation around any methods modifying the list or its items, and when overriding the field value entirely. .. code-block:: python # Cursor behaving like a list text_list_cursor = record['Text List Field'] # Iteration for value in text_list_cursor: print(value) # Modification text_list_cursor.reverse() text_list_cursor.insert(0, 'new value') # Index/slice assert text_list_cursor[0] == 'new value' # Contains assert 'new value' in text_list_cursor # Type validation # Failing validation will not modify the field value original_values = list(text_list_cursor) try: text_list_cursor.append(123) except ValidationError: assert len(original_values) == len(text_list_cursor) # Replacement # Can be set directly to a new list of values record['Text List Field'] = ['new', 'values'] # Any invalid values will abort the entire operation try: record['Text List Field'] = ['text', 456] except ValidationError: assert list(record['Text List Field']) == ['new', 'values'] UserGroupField -------------- Returns UserGroup instances (current API limitation) .. code-block:: python usergroup = record['Group'] assert isinstance(usergroup, UserGroup) assert usergroup.id == '58de1d1c07637a0264c0ca71' assert usergroup.name == 'Everyone' # UserGroup comparisons with specific User/Group instances assert usergroup == swimlane.groups.get(name='Everyone') Set User, Group, or UserGroup .. code-block:: python assert isinstance(swimlane.user, User) record['User'] = swimlane.user assert record['User'] == swimlane.user Value must be a UserGroup instance or extension; Usernames, IDs, display names, etc. are all ambiguous .. code-block:: python record['UserGroup'] = swimlane.user try: record['UserGroup'] = 'Everyone' except ValidationError: # Will not work, string is ambiguous and not a valid value pass assert record['UserGroup'] == swimlane.user .. note:: Field support both single-select and multi-select modes like values lists. Uses similar cursor as values list for multi-select, works exactly the same but for UserGroup objects instead of strings. AttachmentsField ---------------- Returns a cursor managing iteration existing attachments. .. code-block:: python attachments = record['Attachment'] assert isinstance(attachments, AttachmentCursor) for attachment in attachments: # Yields Attachment instances assert isinstance(attachment, Attachment) assert attachment.filename == '5f09afe50064b2bd718e77818b565df1.pcap' assert attachment.file_id == '58ebb22907637a0b488b7b17' assert isinstance(attachment.upload_date, datetime) # Retrieve file bytes as BytesIO stream (file-like object) stream = attachment.download() assert isinstance(stream, BytesIO) content = stream.read() assert len(content) > 0 Upload new attachment with a given filename and a file-like object .. code-block:: python # Read file from disk and add as new attachment with open('/path/to/file', 'rb') as file_handle: record['Attachment'].add('filename.txt', file_handle) # Create new attachment from data already loaded into a file-like object # Useful when attaching data already read from disk or when that file data is used multiple times from io import BytesIO with open('/path/to/file', 'rb') as file_handle: data = file_handle.read() record['Attachment'].add('filename.txt', BytesIO(data)) Example showing adding a request response body as an attachment .. code-block:: python from io import BytesIO import requests response = requests.get('http://httpbin.org/json') record['Attachment'].add('example.json', BytesIO(response.content)) record.save() .. note:: Attachment is uploaded, and associated with record locally, immediately. Association with attachment on server is not persisted until calling :meth:`record.save`. Clear all attachments .. code-block:: python assert len(record['Attachment']) == 1 del record['Attachment'] assert len(record['Attachment']) == 0 # Not cleared on server until saved record.save() ReferenceField -------------- Returns ReferenceCursor with lazy retrieval of target app definition and referenced records as accessed. Yields (and caches) Record instances when iterated over. .. note:: Orphaned referenced records (records deleted but referenced not yet removed) are ignored, and automatically removed during iteration. Saving a record after iterating over a reference field will remove those orphaned references on the server. .. code-block:: python reference = record['Reference'] assert isinstance(reference, ReferenceCursor) assert len(reference) == 3 for referenced_record in reference: assert isinstance(referenced_record, Record) assert referenced_record._app != app assert referenced_record._app == reference.target_app Add or remove references to Records .. code-block:: python other_app = swimlane.apps.get(name='Reference App') ref_target_record = other_app.records.get(id='58e24e8607637a0b488849d4') # Records added multiple times are ignored record['Reference'].add(ref_target_record) assert len(record['Reference']) == 4 record['Reference'].add(ref_target_record) assert len(record['Reference']) == 4 # Remove reference. Raises exception if not already referenced record['Reference'].remove(ref_target_record) assert len(record['Reference']) == 3 Target app validation .. code-block:: python # Cannot reference a record from an app that is not the reference field's target app try: record['Reference'].add(record) except ValidationError: assert len(record['Reference']) == 3 Override all references .. code-block:: python # Can be set to a list of records directly # Acts similar to values list, any invalid records cause the entire operation to fail record['Reference'] = [ref_target_record] assert len(record['Reference']) == 1 try: record['Reference'] = [ref_id, ref_target_record] except ValidationError: assert len(record['Reference']) == 1 CommentsField ------------- Cursor managing iteration and addition of comments .. code-block:: python comments = record['Comments'] assert isinstance(comments, CommentCursor) assert len(comments) == 1 for comment in comments: # Yields Comment instances assert isinstance(comment, Comment) assert isinstance(comment.message, str) assert isinstance(comment.user, UserGroup) assert isinstance(comment.created_date, datetime) assert isinstance(comment.is_rich_text, boolean) # Add new comment comments.comment('New comment message') # Add new rich text comment comments.comment('

New Comment

', rich_text=True) # Not persisted until saved, but still listed on local record assert len(comments) == 2 assert comments[1].message == str(comments[1]) == 'New comment message' .. note:: Like attachments, comments are associated with a record only locally until calling :meth:`record.save`. HistoryField ------------ Returns a readonly RevisionCursor object that abstracts out retrieval of record history. Each item in the RevisionCursor is a RecordRevision object, which performs additional requests to history API endpoints as accessed. See the "Resources" section of the documentation for more information about the RecordRevision object. .. code-block:: python history = record['History'] assert isinstance(history, RevisionCursor) # Get number of revisions num_revisions = len(history) # Iterate backwards over revisions for idx, revision in enumerate(history): assert isinstance(revision, Revision) assert isinstance(revision.modified_date, datetime) assert isinstance(revision.user, UserGroup) assert num_revisions - revision.revision_number == idx # revision.version is a full Record instance, and fields can be accessed like a normal Record assert revision.version.id == record.id