BUG: Add font stack to q/Q operations in layout mode (#3225)

Closes #3212.

Author: hackowitz-af

CONTRIBUTORS.md

@@ -25,6 +25,7 @@ history and [GitHub's 'Contributors' feature](https://github.com/py-pdf/pypdf/gr
 * [Gutteridge, David H.](https://github.com/dhgutteridge)
 * [Hale, Joseph](https://github.com/thehale)
 * [harshhes](https://github.com/harshhes)
+* [Jackowitz, Noah](https://github.com/hackowitz-af) | [LinkedIn](https://www.linkedin.com/in/noah-jackowitz/)
 * [JianzhengLuo](https://github.com/JianzhengLuo)
 * [Karvonen, Harry](https://github.com/Hatell/)
 * [King, Hunter](https://github.com/neversphere)

pypdf/_text_extraction/_layout_mode/_text_state_manager.py

@@ -1,7 +1,7 @@
 """manage the PDF transform stack during "layout" mode text extraction"""
 
 from collections import ChainMap, Counter
-from typing import Any, Dict, List, MutableMapping, Union
+from typing import Any, Dict, List, MutableMapping, Tuple, Union
 from typing import ChainMap as ChainMapType
 from typing import Counter as CounterType
 
@@ -43,6 +43,7 @@ def __init__(self) -> None:
         self.Tz: float = 100.0
         self.TL: float = 0.0
         self.Ts: float = 0.0
+        self.font_stack: List[Tuple[Union[Font, None], Union[int, float]]] = []
         self.font: Union[Font, None] = None
         self.font_size: Union[int, float] = 0
 
@@ -167,6 +168,7 @@ def reset_trm(self) -> TextStateManagerChainMapType:
 
     def remove_q(self) -> TextStateManagerChainMapType:
         """Rewind to stack prior state after closing a 'q' with internal 'cm' ops"""
+        self.font, self.font_size = self.font_stack.pop(-1)
         self.transform_stack = self.reset_tm()
         self.transform_stack.maps = self.transform_stack.maps[
             self.q_queue.pop(self.q_depth.pop(), 0) :
@@ -175,6 +177,7 @@ def remove_q(self) -> TextStateManagerChainMapType:
 
     def add_q(self) -> None:
         """Add another level to q_queue"""
+        self.font_stack.append((self.font, self.font_size))
         self.q_depth.append(len(self.q_depth))
 
     def add_cm(self, *args: Any) -> TextStateManagerChainMapType:

tests/test_text_extraction.py

@@ -338,3 +338,18 @@ def test_iss3074():
     # pypdf.errors.PdfReadError: ZeroDivisionError: float division by zero
     txt = reader.pages[0].extract_text(extraction_mode="layout")
     assert txt.strip().startswith("AAAAAA")
+
+
+@pytest.mark.enable_socket
+def test_layout_mode_text_state():
+    """Ensure the text state is stored and reset with q/Q operators."""
+    # Get the PDF from issue #3212
+    url = "https://github.com/user-attachments/files/19396790/garbled.pdf"
+    name = "garbled-font.pdf"
+    reader = PdfReader(BytesIO(get_data_from_url(url, name=name)))
+    # Get the txt from issue #3212 and normalize line endings
+    txt_url = "https://github.com/user-attachments/files/19510731/garbled-font.layout.txt"
+    txt_name = "garbled-font.layout.txt"
+    expected = get_data_from_url(txt_url, name=txt_name).decode("utf-8").replace("\r\n", "\n")
+
+    assert expected == reader.pages[0].extract_text(extraction_mode="layout")