From c7c8bfb3055a4e8193a544abf56b3ee62e10e3de Mon Sep 17 00:00:00 2001
From: SWHL <liekkaskono@163.com>
Date: Wed, 4 Dec 2024 08:12:16 +0800
Subject: [PATCH] feat(rapidocr_paddle): support en char rec

---
 python/rapidocr_paddle/cal_rec_boxes/main.py | 59 +++++++++++---------
 python/setup_paddle.py                       |  2 +-
 python/tests/test_paddle.py                  | 17 ++----
 3 files changed, 40 insertions(+), 38 deletions(-)

diff --git a/python/rapidocr_paddle/cal_rec_boxes/main.py b/python/rapidocr_paddle/cal_rec_boxes/main.py
index 5e0d71b..cbc5294 100644
--- a/python/rapidocr_paddle/cal_rec_boxes/main.py
+++ b/python/rapidocr_paddle/cal_rec_boxes/main.py
@@ -77,34 +77,26 @@ class CalRecBoxes:
         word_box_list = []
         word_box_content_list = []
         cn_width_list = []
+        en_width_list = []
         cn_col_list = []
-        for word, word_col, state in zip(word_list, word_col_list, state_list):
-            if state == "cn":
-                if len(word_col) != 1:
-                    char_seq_length = (word_col[-1] - word_col[0] + 1) * cell_width
-                    char_width = char_seq_length / (len(word_col) - 1)
-                    cn_width_list.append(char_width)
-                cn_col_list += word_col
-                word_box_content_list += word
-            else:
-                cell_x_start = bbox_x_start + int(word_col[0] * cell_width)
-                cell_x_end = bbox_x_start + int((word_col[-1] + 1) * cell_width)
-                cell = [
-                    [cell_x_start, bbox_y_start],
-                    [cell_x_end, bbox_y_start],
-                    [cell_x_end, bbox_y_end],
-                    [cell_x_start, bbox_y_end],
-                ]
-                word_box_list.append(cell)
-                word_box_content_list.append("".join(word))
+        en_col_list = []
 
-        if len(cn_col_list) != 0:
-            if len(cn_width_list) != 0:
-                avg_char_width = np.mean(cn_width_list)
+        def cal_char_width(width_list, word_col_):
+            if len(word_col_) == 1:
+                return
+            char_total_length = (word_col_[-1] - word_col_[0] + 1) * cell_width
+            char_width = char_total_length / (len(word_col_) - 1)
+            width_list.append(char_width)
+
+        def cal_box(col_list, width_list, word_box_list_):
+            if len(col_list) == 0:
+                return
+            if len(width_list) != 0:
+                avg_char_width = np.mean(width_list)
             else:
                 avg_char_width = (bbox_x_end - bbox_x_start) / len(rec_txt)
 
-            for center_idx in cn_col_list:
+            for center_idx in col_list:
                 center_x = (center_idx + 0.5) * cell_width
                 cell_x_start = max(int(center_x - avg_char_width / 2), 0) + bbox_x_start
                 cell_x_end = (
@@ -117,7 +109,20 @@ class CalRecBoxes:
                     [cell_x_end, bbox_y_end],
                     [cell_x_start, bbox_y_end],
                 ]
-                word_box_list.append(cell)
+                word_box_list_.append(cell)
+
+        for word, word_col, state in zip(word_list, word_col_list, state_list):
+            if state == "cn":
+                cal_char_width(cn_width_list, word_col)
+                cn_col_list += word_col
+                word_box_content_list += word
+            else:
+                cal_char_width(en_width_list, word_col)
+                en_col_list += word_col
+                word_box_content_list += word
+
+        cal_box(cn_col_list, cn_width_list, word_box_list)
+        cal_box(en_col_list, en_width_list, word_box_list)
         sorted_word_box_list = sorted(word_box_list, key=lambda box: box[0][0])
         return word_box_content_list, sorted_word_box_list
 
@@ -256,5 +261,9 @@ class CalRecBoxes:
                 p23[np.where(p23[:, 1] == np.min(p23[:, 1]))],
                 p23[np.where(p23[:, 1] == np.max(p23[:, 1]))],
             )
-
+        # 解决单字矩形框重叠导致多个相同框的情况
+        p1 = p1[:1, :]
+        p2 = p2[:1, :]
+        p3 = p3[:1, :]
+        p4 = p4[:1, :]
         return np.array([p1, p2, p3, p4]).reshape((-1, 2)).tolist()
diff --git a/python/setup_paddle.py b/python/setup_paddle.py
index aa1e0ce..2398efb 100644
--- a/python/setup_paddle.py
+++ b/python/setup_paddle.py
@@ -29,7 +29,7 @@ MODULE_NAME = "rapidocr_paddle"
 obtainer = GetPyPiLatestVersion()
 try:
     latest_version = obtainer(MODULE_NAME)
-    VERSION_NUM = obtainer.version_add_one(latest_version)
+    VERSION_NUM = obtainer.version_add_one(latest_version, add_patch=True)
 except:
     VERSION_NUM = "0.0.1"
 
diff --git a/python/tests/test_paddle.py b/python/tests/test_paddle.py
index 4783525..f75d846 100644
--- a/python/tests/test_paddle.py
+++ b/python/tests/test_paddle.py
@@ -212,18 +212,11 @@ def test_input_three_ndim_one_channel():
         ),
         (
             "text_vertical_words.png",
-            [
-                "已",
-                "取",
-                "之",
-                "時",
-                "不",
-                "參",
-                "一",
-                "人",
-                "见",
-                "而",
-            ],  # rapidocr_onnxruntime: 見, rapidocr_paddle: 见
+            ["已", "取", "之", "時", "不", "參", "一", "人", "见", "而"],
+        ),
+        (
+            "issue_170.png",
+            ["T", "E", "S", "T"],
         ),
     ],
 )