exam-segment-django/segment/ocr/group_pictures.py

import re
import shutil

import glob
from pprint import pprint

import segment.ocr.luo_ocr.ocr as luo_ocr
# from pypinyin import lazy_pinyin
from segment.ocr.split_topic_en import topic_type_line

# def to_pinyin_camel(s):
#     '''文件123.txt'''
#     py_ls = lazy_pinyin(s)
#     py_camel = [py.capitalize() for py in py_ls]
#     return "".join(py_camel)
#
#
# def rename_filename(filename):
#     "将文件名转变为拼音"
#     filename_en = to_pinyin_camel(filename)
#     try:
#         shutil.copy(filename, filename_en)
#     except shutil.SameFileError:
#         pass
#     return filename_en


# def request_ocr(filename):
#     '''中文无法上传需要修改成英文'''
#     url = "http://117.50.17.141/ocr"
#     data = {}
#     filename = rename_filename(filename)
#     files = {"mydata": open(filename, "rb")}
#     r = requests.post(url, data, files=files)
#     print(filename)
#     print(r.json())
#     return r.json()['text']


topic_start = re.compile("^\s*(\d+)\s*[\.、:：,，]")
topic_start2 = re.compile("^\s*[(<〈《]?(\d+)\)\s*[\.、:：,，]?")


def is_topic_start(s, subject):
    """开始节点"""
    if subject in ['math', 'english', 'chinese', 'physics', 'chemistry', 'biology']:
        if topic_start.match(s):
            return True
        elif subject == 'math':
            if topic_start2.match(s):
                return True
        return False
    else:
        raise ValueError("subject={} is not supported!".format(subject))


# -------------------------符合下列条件的则为结束-------------------------
topic_end = re.compile("D\s*[\.、:：]")
topic_end2 = re.compile("^\s*G\s*[\.、:：]")


def is_topic_end(s, subject):
    """结束节点"""
    if subject in ['math', 'english', 'chinese', 'physics', 'chemistry', 'biology']:
        if topic_end.search(s):
            return True
        elif subject == 'english':
            if topic_end2.match(s):
                return True
        return False
    else:
        raise ValueError("subject={} is not supported!".format(subject))


# -------------------------符合下列条件的则为跳过舍去-------------------------
topic_filter = re.compile("^\s*[一二三四五六七八九十]+\s*[\.、:：]")
general_filter = ['选择题', '单选题', '多选题',
                  '填空题', '单空题', '多空题',
                  '解答题', '简答题', '证明题',
                  '选做题', '实验题', '第II卷',
                  '第I卷', '第二卷', ]

english_filter = [["听", "材料"], ["听", "对话"], ["听", "独白"],
                  ['第一节'], ['第二节'], ['语言知识运用'], ['第II卷'],
                  ['第二部分'], ['第三部分'], ['第四部分']]

chinese_filter = re.compile("^\s*[(（]\s*[一二三四五六七八九十]\s*[）)]\s*[\.、:：]?")


def contains_all(s, words):
    for word in words:
        if all([w in s for w in word]):
            return True
    return False


def is_topic_skip(s, subject):
    """判断该行是否可以去掉，跳跃节点"""
    if subject == 'english':
        return contains_all(s, english_filter)

    elif subject in ['math', 'chinese', 'physics', 'chemistry', 'biology']:
        if topic_filter.match(s):
            return True
        if subject == 'chinese':
            if chinese_filter.match(s):
                return True
        for topic_type in general_filter:
            if topic_type in s:
                return True
        return False
    else:
        raise ValueError("subject={} is not supported!".format(subject))


# ----------------------action----------------------
def group_pictures(pictures, subject=''):
    """Assume pictures are row based"""
    # texts = []
    # for picture in pictures:
    #     t = luo_ocr.ocr_py(picture)
    #     t = t.replace("\r", "").replace("\n", "")
    #     # print(str(t))
    #     texts.append(str(t))
    # # texts = bd_ocr(pictures)
    # # print(texts)

    texts = [luo_ocr.ocr_py(picture).replace("\r", "").replace("\n", "") for picture in pictures]

    groups = []
    start = 0
    for i, t in enumerate(texts):
        if is_topic_start(t, subject):
            groups.append([start, i])
            start = i
        elif is_topic_end(t, subject):
            groups.append([start, i + 1])
            start = i + 1
        elif is_topic_skip(t, subject):
            if i > start:
                groups.append([start, i])
            start = i + 1

    len_text = len(texts)
    if len_text > start:
        groups.append([start, len_text])
    return texts, groups