feat: add new element types (#332)

This PR adds some new element types that can be used especially by
pdf/image parition.
The new element types correspond to the ones added in unstructured in
this PR:
Unstructured-IO/unstructured#2700

CHANGELOG.md

@@ -1,3 +1,6 @@
+## 0.7.26-dev0
+* feat: add a set of new `ElementType`s to extend future element types recognition
+
 ## 0.7.25
 
 * fix: replace `Rectangle.is_in()` with `Rectangle.is_almost_subregion_of()` when filling in an inferred element with embedded text

unstructured_inference/__version__.py

@@ -1 +1 @@
-__version__ = "0.7.25"  # pragma: no cover
+__version__ = "0.7.26-dev0"  # pragma: no cover

unstructured_inference/constants.py

@@ -26,11 +26,20 @@ class Source(Enum):
 
 
 class ElementType:
+    PARAGRAPH = "Paragraph"
     IMAGE = "Image"
+    PARAGRAPH_IN_IMAGE = "ParagraphInImage"
     FIGURE = "Figure"
     PICTURE = "Picture"
     TABLE = "Table"
+    PARAGRAPH_IN_TABLE = "ParagraphInTable"
     LIST = "List"
+    FORM = "Form"
+    PARAGRAPH_IN_FORM = "ParagraphInForm"
+    CHECK_BOX_CHECKED = "CheckBoxChecked"
+    CHECK_BOX_UNCHECKED = "CheckBoxUnchecked"
+    RADIO_BUTTON_CHECKED = "RadioButtonChecked"
+    RADIO_BUTTON_UNCHECKED = "RadioButtonUnchecked"
     LIST_ITEM = "List-item"
     FORMULA = "Formula"
     CAPTION = "Caption"
@@ -42,6 +51,9 @@ class ElementType:
     TEXT = "Text"
     UNCATEGORIZED_TEXT = "UncategorizedText"
     PAGE_BREAK = "PageBreak"
+    CODE_SNIPPET = "CodeSnippet"
+    PAGE_NUMBER = "PageNumber"
+    OTHER = "Other"
 
 
 FULL_PAGE_REGION_THRESHOLD = 0.99