def create_entity_localization_command_sequence(
entity_name: str,
EntityExtractorType: Type[EntityExtractor],
extract_contexts: bool = False,
DetectedEntityType: Optional[Type[SerializableEntity]] = None,
upload_func: Optional[EntityUploadCallable] = None,
colorize_options: ColorizeOptions = ColorizeOptions(),
colorize_func: Optional[ColorizeFunc] = None,
) -> List[Type[Command]]: # type: ignore
"""
Create a set of commands that can be used to locate a new type of entity. In the simplest case,
all you have to provide is and 'entity_name' to be used for naming output files, and
'entity_type' that can be used to filter which commands are being run when you the full
pipeline is run, and an 'EntityExtractorType' that locates all instances of that entity in the
TeX. This function creates the commands necessary to colorize the entities, compile the
LaTeX, raster the pages, and locate the colors in the pages. You may define additional
paramters (e.g., 'colorize_options') to fine-tune the commands.
To extract the contexts for an entity (i.e., the sentences in which the entities appear),
set 'extract_contexts' to True.
If you are trying to find the locations of a new type of entity, it is highly recommended that
you use this convenience methods instead of creating new commands yourself.
"""
commands: CommandList = []
directories.register(f"detected-{entity_name}")
commands.append(
make_detect_entities_command(entity_name, EntityExtractorType))
if extract_contexts:
directories.register(f"contexts-for-{entity_name}")
commands.append(make_extract_contexts_command(entity_name))
directories.register(f"sources-with-colorized-{entity_name}")
directories.register(f"compiled-sources-with-colorized-{entity_name}")
directories.register(f"paper-images-with-colorized-{entity_name}")
directories.register(f"diffed-images-with-colorized-{entity_name}")
directories.register(f"{entity_name}-locations")
commands.append(
make_locate_entities_command(entity_name, None, DetectedEntityType,
colorize_options, colorize_func))
if upload_func is not None:
upload_command = make_upload_entities_command(
entity_name, upload_func, DetectedEntityType=DetectedEntityType)
commands.append(upload_command)
return commands
def entity_key_for_contexts(entity: SerializableEntity) -> Any:
"""
When constructing snippets for the contexts symbols appear in, determine whether
two symbols should be highlighted as the 'same symbol' using their MathML.
"""
symbol = cast(SerializableSymbol, entity)
return symbol.mathml
commands = [
ExtractSymbols,
FindSymbolMatches,
make_extract_contexts_command(
"symbols",
EntityType=SerializableSymbol,
entity_key=entity_key_for_contexts,
tex_wrapper=TexWrapper(before=r"\htmlClass{match-highlight}{",
after="}",
braces=True),
),
make_locate_entities_command(
"equation-tokens",
DetectedEntityType=SerializableToken,
colorize_options=ColorizeOptions(
adjust_color_positions=adjust_color_positions,
braces=True,
when=filter_atom_tokens,
),
),
make_locate_entities_command(
"symbols-with-affixes",
input_entity_name="symbols",
@dataclass(frozen=True)
class EntityWithType(SerializableEntity):
type_: str
def exclude_symbols(entity: SerializableEntity) -> bool:
if entity.id_.startswith("definiendum") or entity.id_.startswith("term"):
return cast(EntityWithType, entity).type_ != "symbol"
return True
commands: CommandList = [
TokenizeSentences,
CreateAnnotationFiles,
DetectDefinitions,
make_extract_contexts_command(entity_name="definitions"),
make_locate_entities_command(
"definitions",
DetectedEntityType=EntityWithType,
# Do not locate terms that are symbols because these will already be detect more
# robustly in dedicated commands for symbol localization.
colorize_options=ColorizeOptions(when=exclude_symbols),
),
upload_command,
]
definitions_pipeline = EntityPipeline(
"definitions",
commands,
depends_on=["symbols", "sentences"],
)
from .colorize import adjust_color_positions
from .extractor import GlossaryTermExtractor
from .upload import upload_terms
commands = create_entity_localization_command_sequence(
"glossary-terms",
GlossaryTermExtractor,
DetectedEntityType=Term,
colorize_options=ColorizeOptions(adjust_color_positions=adjust_color_positions),
upload_func=upload_terms,
)
# Before uploading entities, extract contexts that each term appeared in.
upload_command_index = len(commands)
for i, command in enumerate(commands):
if command.get_name() == "upload-glossary-terms":
upload_command_index = i
directories.register("contexts-for-glossary-terms")
commands.insert(
upload_command_index,
make_extract_contexts_command(
"glossary-terms",
EntityType=Term,
tex_wrapper=TexWrapper(before="**", after="**"),
),
)
terms_pipeline = EntityPipeline("glossary-terms", commands)
register_entity_pipeline(terms_pipeline)