move names into separate file

papermage/magelib/__init__.py

@@ -6,18 +6,20 @@
 
 """
 
-
 from .box import Box
+from .document import Document, Prediction
+from .entity import Entity
+from .image import Image
+from .indexer import EntityBoxIndexer, EntitySpanIndexer
 from .layer import Layer
-from .document import (
+from .metadata import Metadata
+from .names import (
  AbstractsFieldName,
  AlgorithmsFieldName,
  AuthorsFieldName,
  BibliographiesFieldName,
  BlocksFieldName,
  CaptionsFieldName,
- Document,
- Prediction,
  EntitiesFieldName,
  EquationsFieldName,
  FiguresFieldName,
@@ -40,10 +42,6 @@
  TokensFieldName,
  WordsFieldName,
 )
-from .entity import Entity
-from .image import Image
-from .indexer import EntityBoxIndexer, EntitySpanIndexer
-from .metadata import Metadata
 from .span import Span
 
 __all__ = [

papermage/magelib/document.py

@@ -14,40 +14,16 @@
 from .image import Image
 from .layer import Layer
 from .metadata import Metadata
+from .names import (
+ EntitiesFieldName,
+ ImagesFieldName,
+ MetadataFieldName,
+ RelationsFieldName,
+ SymbolsFieldName,
+ TokensFieldName,
+)
 from .span import Span
 
-# document field names
-SymbolsFieldName = "symbols"
-ImagesFieldName = "images"
-MetadataFieldName = "metadata"
-EntitiesFieldName = "entities"
-RelationsFieldName = "relations"
-
-PagesFieldName = "pages"
-TokensFieldName = "tokens"
-RowsFieldName = "rows"
-BlocksFieldName = "blocks"
-WordsFieldName = "words"
-SentencesFieldName = "sentences"
-ParagraphsFieldName = "paragraphs"
-
-# these come from vila
-TitlesFieldName = "titles"
-AuthorsFieldName = "authors"
-AbstractsFieldName = "abstracts"
-KeywordsFieldName = "keywords"
-SectionsFieldName = "sections"
-ListsFieldName = "lists"
-BibliographiesFieldName = "bibliographies"
-EquationsFieldName = "equations"
-AlgorithmsFieldName = "algorithms"
-FiguresFieldName = "figures"
-TablesFieldName = "tables"
-CaptionsFieldName = "captions"
-HeadersFieldName = "headers"
-FootersFieldName = "footers"
-FootnotesFieldName = "footnotes"
-
 
 class Prediction(NamedTuple):
  name: str

papermage/magelib/entity.py

@@ -10,10 +10,10 @@
 from .box import Box
 from .image import Image
 from .metadata import Metadata
+from .names import TokensFieldName
 from .span import Span
 
 if TYPE_CHECKING:
- from .document import TokensFieldName
  from .layer import Layer
 
 
@@ -160,8 +160,7 @@ def symbols_from_boxes(self) -> List[str]:
  if self.layer.doc.symbols is None:
  raise ValueError("This Entity's Document is missing symbols")
 
- # TODO: maybe just an import error for TokensFieldName?
- matched_tokens = self.intersect_by_box(name="tokens")
+ matched_tokens = self.intersect_by_box(name=TokensFieldName)
  return [self.layer.doc.symbols[span.start : span.end] for t in matched_tokens for span in t.spans]
 
  @property

papermage/magelib/names.py

@@ -0,0 +1,37 @@
+"""
+
+@kylel
+
+"""
+
+# document field names
+SymbolsFieldName = "symbols"
+ImagesFieldName = "images"
+MetadataFieldName = "metadata"
+EntitiesFieldName = "entities"
+RelationsFieldName = "relations"
+
+PagesFieldName = "pages"
+TokensFieldName = "tokens"
+RowsFieldName = "rows"
+BlocksFieldName = "blocks"
+WordsFieldName = "words"
+SentencesFieldName = "sentences"
+ParagraphsFieldName = "paragraphs"
+
+# these come from vila
+TitlesFieldName = "titles"
+AuthorsFieldName = "authors"
+AbstractsFieldName = "abstracts"
+KeywordsFieldName = "keywords"
+SectionsFieldName = "sections"
+ListsFieldName = "lists"
+BibliographiesFieldName = "bibliographies"
+EquationsFieldName = "equations"
+AlgorithmsFieldName = "algorithms"
+FiguresFieldName = "figures"
+TablesFieldName = "tables"
+CaptionsFieldName = "captions"
+HeadersFieldName = "headers"
+FootersFieldName = "footers"
+FootnotesFieldName = "footnotes"