ABX/pkg/abx/file_context.py
filmfreedom-org 5e24e06796 Replaced parsers sub-package with parsers.py
After discovering that sub-packages in Blender add-ons don't work (this
seems to be very broken!), I replaced the parsers subpackage with a
parsers.py module (almost as simple as concatenating the source files in
the package). I may be removing this name-parsing system from ABX
entirely, soon, possibly moving it into KitCAT.
2021-06-29 13:53:48 -05:00

593 lines
22 KiB
Python

# file_context.py
"""
Contextual metadata acquired from the file system, file name, directory structure, and
sidecar data files.
Data is acquired from file and directory names and also from yaml files in the tree.
The yaml files are loaded in increasing priority from metadata.yaml, abx.yaml, <dirname>.yaml.
They are also loaded from the top of the tree to the bottom, with the most local Values
overriding the top-level ones.
@author: Terry Hancock
@copyright: 2019 Anansi Spaceworks.
@license: GNU General Public License, version 2.0 or later. (Python code)
Creative Commons Attribution-ShareAlike, version 3.0 or later. (Website Templates).
@contact: digitante@gmail.com
"""
import os, re, copy, string, collections
import yaml
DEFAULT_YAML = {}
with open(os.path.join(os.path.dirname(__file__), 'abx.yaml')) as def_yaml_file:
DEFAULT_YAML.update(yaml.safe_load(def_yaml_file))
TESTPATH = os.path.abspath(os.path.join(os.path.dirname(__file__), '..', 'testdata', 'myproject', 'Episodes', 'A.001-Pilot', 'Seq', 'LP-LastPoint', 'A.001-LP-1-BeginningOfEnd-anim.txt'))
from . import accumulate
from .accumulate import RecursiveDict
from .enum import Enum
from .ranks import RankNotFound
from abx.parsers import NameParsers
log_level = Enum('DEBUG', 'INFO', 'WARNING', 'ERROR')
from .name_schema import FieldSchema
from .name_context import NameContext
#from .render_profile import RenderProfileMap
class FileContext(NameContext):
"""
Collected information about a file's storage location on disk.
Collects name and path information from a filepath, used to identify
the file's role in a project. In order to do this correctly, the
FileContext object needs a schema defined for the project, which
explains how to read and parse project file names, to determine what
unit, name, or role they might have in the project.
For this, you will need to have a <project>.yaml file which defines
the 'project_schema' (a list of dictionaries used to initialize a list
of FieldSchema objects). Examples of <project>.yaml are provided in the
'myproject.yaml' file in the test data in the source distribution of
ABX, and you can also see a "live" example in the "Lunatics!" project.
Subclass from NameContext, so please read more information there.
Attributes:
root (filepath):
The root directory of the project as an absolute operating system
filepath. This should be used for finding the root where it is
currently, not stored for permanent use, as it will be wrong if
the project is relocated.
render_root (filepath):
The root directory for rendering. We often have this symlinked to
a large drive to avoid congestion. Usually just <root>/Renders.
filetype (str):
Filetype code or extension for this file. Usually identifies what
sort of file it is and may imply how it is used in some cases.
role (str):
Explicit definition of file's role in the project, according to
roles specified in <project>.yaml. For a default, see 'abx.yaml'
in the ABX source code. Derived from the file name.
title (str):
Title derived from the filename.
The relationship between this and the NameContext title is unclear
at present -- probably we should be setting the NameContext.title
property from here (?)
comment (str):
Comment field from the filename. This is a free field generally
occurring after the role, using a special delimiter and meant to
be readable by humans. It may indicate an informal backup or
saved version of the file outside of the VCS, as opposed to
a main, VCS-tracked copy. Or it may indicate some variant version
of the file.
name_contexts (list[NameContext]):
A list of NameContext objects contained in this file, typically
one-per-scene in a Blender file.
filepath (str):
O/S and location dependent absolute path to the file.
filename (str):
Unaltered filename from disk.
file_exists (bool):
Does the file exist on disk (yet)?
This may be false if the filename has been determined inside
the application, but the file has not been written to disk yet.
folder_exists (bool):
Does the containing folder exist (yet)?
folders (list(str)):
List of folder names from the project root to the current file,
forming a relative path from the root to this file.
omit_ranks (dict[str:int]):
How many ranks are omitted from the beginning of filename
fields? (Implementation).
provided_data (RecursiveDict):
The pile of data from project YAML files. This is a special
dictionary object that does "deep updates" in which sub-dictionaries
and sub-lists are updated recursively rather than simply being
replaced at the top level. This allows the provided_data to
accumulate information as it looks up the project tree to the
project root. It is not recommended to directly access this data.
(Implementation)
abx_fields (RecursiveDict):
A pile of 'abx.yaml' file with directives affecting how ABX should
behave with this file. This can be used to set custom behavior in
different project units. For example, we use it to define different
render profiles for different project units.
notes (list(str)):
A primitive logging facility. This stores warning and information
messages about the discovery process to aid the production designer
in setting up the project correctly.
NOTE that the clear method does not clear the notes! There is a
separate clear_notes() method.
parsers (list):
A list of registered parser implementations for analyzing file
names. FileContext tries them all, and picks the parser which
reports the best score -- that is, parser score themselves on
how likely their parse is to be correct. So if a parser hits a
problem, it demerits its score, allowing another parser to take
over.
Currently there are only three parsers provided: a custom one,
originally written to be specific to "Lunatics!" episodes
('abx_episode', now obsolete?), a parser using the project_schema
system ('abx_schema', now the preferred choice), and a "dumb"
parser design to fallback on if no schema is provided, which reads
only the filetype and possible role, title, and comment fields,
guessing from common usage with no explicit schema
('abx_fallback').
This implementation could probably benefit from some more application of
computer science and artificial intelligence, but I've settled on a
"good enough" solution and the assumption that production designers would
probably rather just learn how to use the YAML schemas correctly, than
to try to second-guess a sloppy AI system.
As of v0.2.6, FileContext does NOT support getting any information
directly from the operating system path for the file (i.e. by reading
directory names), although this would seem to be a good idea.
Therefore, project units have to be specified by additional unit-level
YAML documents (these can be quite small), explicitly setting the
unit-level information for directories above the current object, and
by inference from the project schema and the filename (which on "Lunatics!"
conveys all the necessary information for shot files, but perhaps not
for library asset files).
"""
# IMMUTABLE DEFAULTS:
filepath = None
root = None
folders = ()
#ranks = ()
project_units = ()
filename = None
fields = None
#subunits = ()
code = '_'
# defaults = {
# 'filetype': None, 'role': None, 'hierarchy': None, 'project': None,
# 'series': None, 'episode': None, 'seq': None, 'block': None,
# 'camera': None, 'shot': None, 'title': None }
def __init__(self, path=None):
"""
Collect path context information from a given filepath.
(Searches the filesystem for context information).
"""
NameContext.__init__(self, None, {})
self.clear()
self.clear_notes()
if path:
self.update(path)
def clear(self):
"""
Clear the contents of the FileContext object.
Nearly the same as reinitializing, but the notes
attribute is left alone, to preserve the log history.
"""
NameContext.clear(self)
# Identity
self.root = os.path.abspath(os.environ['HOME'])
self.render_root = os.path.join(self.root, 'Renders')
self.filetype = ''
self.role = ''
self.title = ''
self.comment = ''
# Containers
#self.notes = []
self.name_contexts = {}
# Status / Settings
self.filepath = None
self.filename = None
self.file_exists = False
self.folder_exists = False
self.omit_ranks = {
'edit': 0,
'render': 0,
'filename': 0,
'scene': 0}
# Defaults
self.provided_data = RecursiveDict(DEFAULT_YAML, source='default')
self.abx_fields = DEFAULT_YAML['abx']
self.render_profiles = {} #RenderProfileMap()
def clear_notes(self):
"""
Clear the log history in the notes attribute.
"""
# We use this for logging, so it doesn't get cleared by the
# normal clear process.
self.notes = []
def update(self, path):
"""
Update the FileContext based on a new file path.
"""
# Basic File Path Info
self.filepath = os.path.abspath(path)
self.filename = os.path.basename(path)
# Does the file path exist?
if os.path.exists(path):
self.file_exists = True
self.folder_exists = True
else:
self.file_exists = False
if os.path.exists(os.path.dirname(path)):
self.folder_exists = True
else:
self.folder_exists = False
# - Should we create it? / Are we creating it?
# We should add a default YAML file in the ABX software to guarantee
# necessary fields are in place, and to document the configuration for
# project developers.
# Data from YAML Files
#self._collect_yaml_data()
self.provided_data = RecursiveDict(DEFAULT_YAML, source='default')
kitcat_root, kitcat_data, abx_data = accumulate.get_project_data(self.filepath)
self.root = kitcat_root
self.provided_data.update(kitcat_data)
path = os.path.abspath(os.path.normpath(self.filepath))
root = os.path.abspath(self.root)
self.folders = [os.path.basename(self.root)]
self.folders.extend(os.path.normpath(os.path.relpath(path, root)).split(os.sep)[:-1])
self.abx_fields = abx_data
# Did we find the YAML data for the project?
# Did we find the project root?
self.render_profiles = self.abx_fields['render_profiles']
#self.render_profiles = RenderProfileMap(self.abx_fields['render_profiles'])
# TODO: Bug?
# Note that 'project_schema' might not be correct if overrides are given.
# As things are, I think it will simply append the overrides, and this
# may lead to odd results. We'd need to actively compress the list by
# overwriting according to rank
#
try:
self._load_schemas(self.provided_data['project_schema'])
self.namepath_segment = [d['code'] for d in self.provided_data['project_unit']]
self.code = self.namepath[-1]
except:
print("Can't find Name Path. Missing <project>.yaml file?")
pass
# print("\n(899) filename = ", self.filename)
# if 'project_schema' in self.provided_data:
# print("(899) project_schema: ", self.provided_data['project_schema'])
# else:
# print("(899) project schema NOT DEFINED")
#
# print("(904) self.namepath_segment = ", self.namepath_segment)
# Was there a "project_schema" section?
# - if not, do we fall back to a default?
# Was there a "project_unit" section?
# - if not, can we construct what we need from project_root & folders?
# Is there a definitions section?
# Do we provide defaults?
try:
self.render_root = os.path.join(self.root,
self.provided_data['definitions']['render_root'])
except KeyError:
self.render_root = os.path.join(self.root, 'Renders')
self.omit_ranks = {}
try:
for key, val in self.provided_data['definitions']['omit_ranks'].items():
self.omit_ranks[key] = int(val)
except KeyError:
self.omit_ranks.update({
'edit': 0,
'render': 1,
'filename': 1,
'scene': 3})
# Data from Parsing the File Name
if ( 'parser' in self.provided_data['definitions'] and
self.provided_data['definitions']['parser'] in NameParsers):
# If project has defined what parser it wants (and it is registered),
# then restrict to that parser:
parser_selection = [self.provided_data['definitions']['parser']]
else:
parser_selection = NameParsers.keys()
if 'parser_options' in self.provided_data['definitions']:
parser_options = self.provided_data['definitions']['parser_options']
else:
parser_options = {}
# TESTING:
# Previous code locked-in the schema parser, so I'm starting with that:
parser_selection = ['abx_schema']
self.parsers = [NameParsers[p](
schemas = self.schemas,
definitions = self.provided_data['definitions'],
**parser_options)
for p in parser_selection]
parser_chosen, parser_score = self._parse_filename()
self.log(log_level.INFO, "Parsed with %s, score: %d" %
(parser_chosen, parser_score))
# TODO:
# We don't currently consider the information from the folder names,
# though we could get some additional information this way
def __repr__(self):
s = '{0}(data='.format(self.__class__.__name__)
#s = s + super().__repr__()
s = s + str(self.code) + '(' + str(self.rank) + ')'
s = s + ')'
return s
def log(self, level, msg):
"""
Log a message to the notes attribute.
This is a simple facility for tracking issues with the production
source tree layout, schemas, and file contexts.
"""
if type(level) is str:
level = log_level.index(level)
self.notes.append((level, msg))
def get_log_text(self, level=log_level.INFO):
"""
Returns the notes attribute as a block of text.
"""
level = log_level.number(level)
return '\n'.join([
': '.join((log_level.name(note[0]), note[1]))
for note in self.notes
if log_level.number(note[0]) >= level])
def _parse_filename(self):
"""
Try available fuzzy data parsers on the filename, and pick the one
that returns the best score.
"""
fields = {}
best_score = 0.0
best_parser_name = None
for parser in self.parsers:
score, fielddata = parser(self.filename, self.namepath)
if score > best_score:
fields = fielddata
best_parser_name = parser.name
best_score = score
self.fields.update(fields)
self._pull_up_last_rank_fields()
return best_parser_name, best_score
def _pull_up_last_rank_fields(self):
if ( 'rank' in self.fields and
self.fields['rank'] in self.fields and
isinstance(self.fields[self.fields['rank']], collections.Mapping) ):
for key, val in self.fields[self.fields['rank']].items():
self.fields[key] = val
# def _collect_yaml_data(self):
@property
def filetype(self):
"""
Filetype suffix for the file (usually identifies format).
"""
if 'filetype' in self.fields:
return self.fields['filetype']
else:
return ''
@filetype.setter
def filetype(self, filetype):
self.fields['filetype'] = filetype
@property
def role(self):
"""
Role field from the filename, or guessed from filetype.
"""
if 'role' in self.fields:
return self.fields['role']
else:
return ''
@role.setter
def role(self, role):
self.fields['role'] = role
@property
def title(self):
"""
Title field parsed from the file name.
"""
if 'title' in self.fields:
return self.fields['title']
else:
return ''
@title.setter
def title(self, title):
self.fields['title'] = title
@property
def comment(self):
"""
Comment field parsed from the filename.
Meant to be a human-readable extension to the filename, often used to
represent an informal version, date, or variation on the file.
"""
if 'comment' in self.fields:
return self.fields['comment']
else:
return ''
@comment.setter
def comment(self, comment):
self.fields['comment'] = comment
@classmethod
def deref_implications(cls, values, matchfields):
"""
NOT USED: Interpret information from reading folder names.
"""
subvalues = {}
for key in values:
# TODO: is it safe to use type tests here instead of duck tests?
if type(values[key])==int and values[key] < len(matchfields):
subvalues[key]=matchfields[values[key]]
elif type(values[key]==dict):
subvalues[key]=cls.deref_implications(values[key], matchfields)
elif type(values[key]==list):
vals = []
for val in values[key]:
vals.append(cls.deref_implications(val, matchfields))
return subvalues
def get_path_implications(self, path):
"""
NOT USED: Extract information from folder names.
"""
data = {}
prefix = r'(?:.*/)?'
suffix = r'(?:/.*)?'
for implication in self.schema['path_implications']:
matched = re.compile(prefix + implication['match'] + suffix).match(path)
if matched and matched.groups:
data.update(self.deref_implications(implication['values'], matched.groups()))
return data
def new_name_context(self, rank=None, **kwargs):
"""
Get a NameContext object representing a portion of this file.
In Blender, generally in a 1:1 relationship with locally-defined
scenes.
"""
fields = {}
fields.update(self.fields)
namepath_segment = []
ranks = [s.rank for s in self.schemas]
i_rank = len(self.namepath)
if i_rank == 0:
old_rank = None
else:
old_rank = ranks[i_rank -1]
# The new rank will be the highest rank mentioned, or the
# explicitly requested rank or
# one rank past the namepath
#
new_rank = self.schemas[i_rank].rank
for schema in self.schemas[i_rank:]:
if schema.rank in kwargs:
fields[schema.rank] = {'code':kwargs[schema.rank]}
new_rank = schema.rank
namepath_segment.append(kwargs[schema.rank])
elif rank is not None:
namepath_segment.append(schema.default)
if ranks.index(schema.rank) <= ranks.index(rank):
new_rank = schema.rank
if old_rank:
delta_rank = ranks.index(new_rank) - ranks.index(old_rank)
else:
# I think in this case, it's as if the old_rank number is -1?
delta_rank = ranks.index(new_rank) + 1
# Truncate to the new rank:
namepath_segment = namepath_segment[:delta_rank]
fields['rank'] = new_rank
fields['code'] = namepath_segment[-1]
name_context = NameContext(self, fields,
namepath_segment=namepath_segment)
self.name_contexts[str(id(name_context))] = name_context
return name_context