fix 添加对空文本的检查和对txt_len为0的保护

[narc1ssus1]

如果为空文本或者检测模型检测框到空区域,会报错ZeroDivisionError: float division by zero,添加对空文本的检查和对txt_len为0的保护

[SWHL]

感谢哈，晚点我review一下

…

---- 回复的原邮件 ---- | 发件人 | ***@***.***> | | 发送日期 | 2025年06月05日 16:55 | | 收件人 | RapidAI/RapidOCR ***@***.***> | | 抄送人 | Subscribed ***@***.***> | | 主题 | Re: [RapidAI/RapidOCR] fix 添加对空文本的检查和对txt_len为0的保护 (PR #443) | Reopened #443. — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[Copilot]

Pull Request Overview

This PR addresses a ZeroDivisionError by adding checks for empty text and a txt_len of 0.

• Prevents errors in OCR word box calculations by returning empty lists when no valid text is provided.
• Protects average character width calculation by returning 0.0 when the text length is zero.

Comments suppressed due to low confidence (2)

python/rapidocr/cal_rec_boxes/main.py:80

• [nitpick] Consider adding an inline comment explaining why empty text or a line_txt_len of 0 should result in returning empty lists to improve code clarity.

if not rec_txt or word_info.line_txt_len == 0:

python/rapidocr/cal_rec_boxes/main.py:167

• [nitpick] Add a comment to clarify that returning 0.0 is intended to prevent division by zero when no text is present, aiding future maintainability.

if txt_len == 0:

[SWHL]

LGTM