Merge pull request #439 from OpenPecha/feat-get_span_text

feat: added get_span_text

Author: ta4tsering

src/openpecha/pecha/__init__.py

@@ -1,9 +1,11 @@
 import json
+from multiprocessing import reduction
 import shutil
 from pathlib import Path
 from typing import Dict, List, Optional
 
 from stam import AnnotationStore, Offset, Selector
+from openpecha.pecha.annotations import span
 
 from openpecha.exceptions import StamAddAnnotationError, FileNotFoundError, MetaDataValidationError
 from openpecha.ids import (
@@ -27,6 +29,7 @@ def __init__(self, pecha_id: str, pecha_path: Path) -> None:
         self.bases = self.load_bases()
         self.annotations = []
 
+
     @classmethod
     def from_path(cls, pecha_path: Path) -> "Pecha":
         # Validate that the path exists
@@ -252,6 +255,20 @@ def update_annotation(self, annotation_id: str, annotation:List[BaseAnnotation],
         self.add(annotation_id, annotation)
         return self
 
+    def get_base_text(self) -> str:
+        base_dir = self.pecha_path / "base"
+        for base_file in base_dir.glob("*.txt"):
+            base_text = base_file.read_text(encoding="utf-8")
+        return base_text
+
+    def get_span_text(self, span: Optional[span] = None) -> str:
+        base_text = self.get_base_text()
+        if span is None:
+            return base_text
+        else:
+            return base_text[span.start:span.end]
+
+
 
 def get_anns(ann_store: AnnotationStore, include_span: bool = False):
     anns = []

tests/pecha/test_get_span_text.py

@@ -0,0 +1,18 @@
+from openpecha.pecha import Pecha
+from openpecha.pecha.annotations import span
+from pathlib import Path
+
+def test_get_span_text():
+    pecha = Pecha.from_path(Path("tests/pecha/data/I5003D420"))
+    text = pecha.get_span_text(span(start=0, end=12))
+    print(text)
+    assert text == "In Sanskrit:"
+
+def test_get_span_text_without_span():
+    pecha = Pecha.from_path(Path("tests/pecha/data/I5003D420"))
+    text = pecha.get_span_text()
+    assert text == pecha.get_base_text()
+
+if __name__ == "__main__":
+    test_get_span_text()
+    test_get_span_text_without_span()