Implementing validation feature

ptulsconv/__init__.py

@@ -2,6 +2,6 @@ from .ptuls_grammar import protools_text_export_grammar
 from .ptuls_parser_visitor import DictionaryParserVisitor
 from .transformations import TimecodeInterpreter
 
-__version__ = '0.5.3'
+__version__ = '0.6.0'
 __author__  = 'Jamie Hardt'
 __license__ = 'MIT'

ptulsconv/__main__.py

@@ -29,6 +29,15 @@ def main():
 
     parser.add_option_group(filter_opts)
 
+    warn_options = OptionGroup(title="Warning and Validation Options", parser=parser)
+    warn_options.add_option('-W', action='store_true', dest='warnings',
+                            help='Generate warnings for common errors (missing code numbers etc.)')
+
+    warn_options.add_option('-S', action='store_true', dest='spelling',
+                            help='Check spelling and warn on misspellings.')
+
+    parser.add_option_group(warn_options)
+
     output_opts = OptionGroup(title="Output Options", parser=parser)
     output_opts.add_option('--json', default=False, action='store_true', dest='write_json',
                            help='Output a JSON document instead of XML. If this option is enabled, --xform will have '
@@ -90,7 +99,8 @@ def main():
 
         convert(input_file=args[1], output_format=output_format, start=options.in_time, end=options.out_time,
                 include_muted=options.include_muted, xsl=options.xslt, select_reel=options.select_reel,
-                progress=False, output=sys.stdout, log_output=sys.stderr)
+                progress=False, output=sys.stdout, log_output=sys.stderr,
+                warnings=options.warnings, spelling=options.spelling)
     except FileNotFoundError as e:
         print_fatal_error("Error trying to read input file")
         raise e

ptulsconv/commands.py

@@ -7,8 +7,10 @@ from xml.etree.ElementTree import TreeBuilder, tostring
 import subprocess
 import pathlib
 import ptulsconv
+from itertools import chain
 
-from .reporting import print_section_header_style, print_status_style
+from .reporting import print_section_header_style, print_status_style, print_warning
+from .validations import *
 
 # field_map maps tags in the text export to fields in FMPXMLRESULT
 #  - tuple field 0 is a list of tags, the first tag with contents will be used as source
@@ -148,7 +150,7 @@ def fmp_transformed_dump(data, input_file, xsl_name, output):
 
 def convert(input_file, output_format='fmpxml', start=None, end=None, select_reel=None,
             progress=False, include_muted=False, xsl=None,
-            output=sys.stdout, log_output=sys.stderr):
+            output=sys.stdout, log_output=sys.stderr, warnings=False, spelling=False):
     with open(input_file, 'r') as file:
         print_section_header_style('Parsing')
         ast = ptulsconv.protools_text_export_grammar.parse(file.read())
@@ -184,6 +186,16 @@ def convert(input_file, output_format='fmpxml', start=None, end=None, select_ree
             parsed = reel_xform.transform(parsed)
 
 
+        if warnings:
+            for warning in chain(validate_unique_field(parsed, field='QN'),
+                                 validate_non_empty_field(parsed, field='QN'),
+                                 validate_non_empty_field(parsed, field='CN'),
+                                 validate_non_empty_field(parsed, field='Char'),
+                                 validate_dependent_value(parsed, key_field='CN', dependent_field='Char'),
+                                 validate_dependent_value(parsed, key_field='CN', dependent_field='Actor'),):
+
+                print_warning(warning.report_message())
+
         if output_format == 'json':
             json.dump(parsed, output)
         elif output_format == 'fmpxml':

ptulsconv/movie_export.py

@@ -0,0 +1,13 @@
+import ffmpeg # ffmpeg-python
+
+
+def create_movie(event):
+    start = event['Movie.Start_Offset_Seconds']
+    duration = event['PT.Clip.Finish_Seconds'] - event['PT.Clip.Start_Seconds']
+    input_movie = event['Movie.Filename']
+    print("Will make movie starting at {}, dur {} from movie {}".format(start, duration, input_movie))
+
+
+def export_movies(events):
+    for event in events:
+        create_movie(event)

ptulsconv/reporting.py

@@ -12,12 +12,21 @@ def print_section_header_style(str):
     else:
         sys.stderr.write("%s\n\n" % str)
 
+
 def print_status_style(str):
     if sys.stderr.isatty():
         sys.stderr.write("\033[3m - %s\033[0m\n" % str)
     else:
         sys.stderr.write(" - %s\n" % str)
 
+
+def print_warning(warning_string):
+    if sys.stderr.isatty():
+        sys.stderr.write("\033[3m - %s\033[0m\n" % warning_string)
+    else:
+        sys.stderr.write(" - %s\n" % warning_string)
+
+
 def print_advisory_tagging_error(failed_string, position, parent_track_name=None, clip_time=None):
     if sys.stderr.isatty():
         sys.stderr.write("\n")

ptulsconv/validations.py

@@ -0,0 +1,54 @@
+from dataclasses import dataclass
+from sys import stderr
+
+
+@dataclass
+class ValidationError:
+    message: str
+    event: dict
+
+    def report_message(self):
+        return f"{self.message}: event at {self.event['PT.Clip.Start']} on track {self.event['PT.Track.Name']}"
+
+def validate_value(input_dict, key_field, predicate):
+    for event in input_dict['events']:
+        val = event[key_field]
+        if not predicate(val):
+            yield ValidationError(message='Field {} not in range'.format(val),
+                                  event=event)
+
+
+def validate_unique_field(input_dict, field='QN'):
+    values = set()
+    for event in input_dict['events']:
+        if event[field] in values:
+            yield ValidationError(message='Re-used {}'.format(field), event=event)
+
+
+def validate_non_empty_field(input_dict, field='QN'):
+    for event in input_dict['events']:
+        if field not in event.keys() or len(event[field]) == 0:
+            yield ValidationError(message='Empty field {}'.format(field), event=event)
+
+
+def validate_dependent_value(input_dict, key_field, dependent_field):
+    """
+    Validates that two events with the same value in `key_field` always have the
+    same value in `dependent_field`
+    """
+    value_map = dict()
+    for event in input_dict['events']:
+        if key_field not in event.keys():
+            continue
+
+        if event[key_field] not in value_map.keys():
+            value_map[event[key_field]] = event.get(dependent_field, None)
+        else:
+            if value_map[event[key_field]] != event.get(dependent_field, None):
+                yield ValidationError(message='Field {} depends on key field {} (value={}), expected {}, was {}'
+                                      .format(dependent_field, key_field, event[key_field], value_map[key_field],
+                                              event.get(dependent_field, None)), event=event)
+
+
+
+