Merge pull request #54 from krishnasism/optimisations/small-optimisat…

…ions-refactor-krish-18012024

Small optimisations - rewrote some lines

hotpdf/memory_map.py

@@ -78,7 +78,6 @@ def __get_span_chars(self, spans: Generator[ET.Element, None, None], drop_duplic
             for char in span.iterfind(".//"):
                 char.set("span_id", span_id)
                 yield char
-        del seen_span_hashes
 
     def load_memory_map(self, page: ET.Element, drop_duplicate_spans: bool = True) -> None:
         """
@@ -102,7 +101,9 @@ def load_memory_map(self, page: ET.Element, drop_duplicate_spans: bool = True) -
             chars = self.__get_page_chars(page)
         for char in chars:
             char_bbox = char.attrib["bbox"]
-            char_x0, char_y0, char_x1, _ = map(float, char_bbox.split())
+            char_x0, char_y0, char_x1, char_y1 = map(float, char_bbox.split())
+            if char_x0 < 0 or char_y0 < 0 or char_x1 < 0 or char_y1 < 0:
+                continue
             char_c = char.attrib["c"]
             char_span_id = char.attrib.get("span_id")
             cell_x = math.floor(char_x0)
@@ -115,9 +116,6 @@ def load_memory_map(self, page: ET.Element, drop_duplicate_spans: bool = True) -
                 x_end=cell_x_end,
                 span_id=char_span_id,
             )
-            if not 0 < cell_x or not 0 < cell_y:
-                continue
-
             self.memory_map.insert(value=char_c, row_idx=cell_y, column_idx=cell_x)
             char_hot_characters.append((
                 char_c,
@@ -133,34 +131,27 @@ def load_memory_map(self, page: ET.Element, drop_duplicate_spans: bool = True) -
         self.height = self.memory_map.rows
 
     @lru_cache
-    def extract_text_from_bbox(self, x0: float, x1: float, y0: float, y1: float) -> str:
+    def extract_text_from_bbox(self, x0: int, x1: int, y0: int, y1: int) -> str:
         """
         Extract text within a specified bounding box.
 
         Args:
-            x0 (float): Left x-coordinate of the bounding box.
-            x1 (float): Right x-coordinate of the bounding box.
-            y0 (float): Bottom y-coordinate of the bounding box.
-            y1 (float): Top y-coordinate of the bounding box.
+            x0 (int): Left x-coordinate of the bounding box.
+            x1 (int): Right x-coordinate of the bounding box.
+            y0 (int): Bottom y-coordinate of the bounding box.
+            y1 (int): Top y-coordinate of the bounding box.
 
         Returns:
             str: Extracted text within the bounding box.
         """
-        cell_x0 = math.floor(x0)
-        cell_x1 = math.ceil(x1)
-        cell_y0 = math.floor(y0)
-        cell_y1 = math.ceil(y1)
-
-        extracted_text = ""
-        for row in range(cell_y0, cell_y1 + 1):
-            if 0 <= row < self.memory_map.rows:
-                row_text = ""
-                for col in range(cell_x0, cell_x1 + 1):
-                    if 0 <= col < self.memory_map.columns:
-                        row_text += self.memory_map.get(row_idx=row, column_idx=col)
-                if row_text:
-                    extracted_text += row_text
-                    extracted_text += "\n"
+        extracted_text: str = ""
+        for row in range(max(y0, 0), min(y1, self.memory_map.rows - 1) + 1):
+            row_text: str = ""
+            row_text = "".join([
+                self.memory_map.get(row_idx=row, column_idx=col) for col in range(max(x0, 0), min(x1, self.memory_map.columns - 1) + 1)
+            ])
+            if row_text:
+                extracted_text += row_text + "\n"
 
         return extracted_text
 

hotpdf/utils.py

@@ -1,5 +1,4 @@
 import math
-from copy import deepcopy
 from typing import Union
 
 from .data.classes import ElementDimension, HotCharacter, PageResult
@@ -60,7 +59,7 @@ def filter_adjacent_coords(text: list[str], page_hot_character_occurences: PageR
     for anchor_hot_character in anchor_hot_character_instances:
         neighbours = [anchor_hot_character]
         reference_hot_character = anchor_hot_character
-        for _, coords_j in enumerate(page_hot_character_occurences[1:]):
+        for coords_j in page_hot_character_occurences[1:]:
             neighbour_hot_character = find_neighbour_coord(
                 reference_character=reference_hot_character,
                 hot_characters=coords_j,
@@ -69,7 +68,8 @@ def filter_adjacent_coords(text: list[str], page_hot_character_occurences: PageR
                 neighbours.append(neighbour_hot_character)
                 reference_hot_character = neighbour_hot_character
         if len(neighbours) == max_len:
-            adjacent_groups.append(deepcopy(neighbours))
+            adjacent_groups.append(neighbours[:])
+            neighbours.clear()
         neighbours = []
     return adjacent_groups