Question

我有一个脚本，可以根据输入文件预测产品名称。代码如下：

output_dir = "C:\\Users\\Lenovo\\.spyder-py3\\NER_training"
DIR = 'C:\\Users\\Lenovo\\.spyder-py3\\Testing\\'
print("Loading from", output_dir)
nlp2 = spacy.load(output_dir)
with open('eng_productnames.csv', newline='') as myFile:
    reader = csv.reader(myFile)
    for rowz in reader:
        try:
            filenamez = rowz[1]
            file = open(DIR+filenamez, "r", encoding ='utf-8')
            filecontentszz = file.read()
            for s in filecontentszz:
                filecontentszz = re.sub(r'\s+', ' ', filecontentszz)
                #filecontents = filecontents.encode().decode('unicode-escape')
                filecontentszz = ''.join([line.lower() for line in filecontentszz]) 
                doc2 = nlp2(filecontentszz)
                for ent in doc2.ents:
                    print(filenamez, ent.label_, ent.text)

                break

        except Exception as e:`

以字符串形式给我输出：

07-09-18 N021024s16PASBUNDLEACK - Acknowledgement P.txt PRODUCT ABC1
06-22-18 Letter from Supl.txt PRODUCT ABC2
06-22-18 Letter from Req to Change .txt PRODUCT ABC3

现在，我想将所有这些详细信息导出到具有2列的CSV中，其中一列为FILENAME，另一列为PRODUCT，其中PRODUCT在相应的列名称下具有所有文件名和产品名称。所有产品名称均以PRODUCT开头，然后为字符串中的名称。我该如何解决：

输出csv应该如下：

Filename                                                             PRODUCT
  07-09-18 Acknowledgement P.txt                                 ABC1
  06-22-18 Letter Req to Change.txt                              ABC2

Answer 1

您可以使用csv.writer，而不是打印到屏幕，使writerow写入每一行到输出文件。

output_dir = "C:\\Users\\Lenovo\\.spyder-py3\\NER_training"
DIR = 'C:\\Users\\Lenovo\\.spyder-py3\\Testing\\'
print("Loading from", output_dir)
nlp2 = spacy.load(output_dir)
with open('eng_productnames.csv', newline='') as input_file, \
        open('output.csv', 'w') as output_file:
    reader = csv.reader(input_file)
    writer = csv.writer(output_file)
    writer.writerow(["Filename", "Product"])  # this is the header row
    for rowz in reader:
        try:
            filenamez = rowz[1]
            file = open(DIR+filenamez, "r", encoding ='utf-8')
            filecontentszz = file.read()
            for s in filecontentszz:
                filecontentszz = re.sub(r'\s+', ' ', filecontentszz)
                #filecontents = filecontents.encode().decode('unicode-escape')
                filecontentszz = ''.join([line.lower() for line in filecontentszz]) 
                doc2 = nlp2(filecontentszz)
                for ent in doc2.ents:
                    writer.writerow([filenamez, ent.text])

                break

我在这里假设filenamez和ent.text在每一列中包含您想要的信息。如果不是这种情况，则可以在写入CSV之前操纵它们以获取所需的内容。

Answer 2

有很多方法可以实现这一目标。我更喜欢使用的是Pandas，这是一个功能强大的库，可以处理CSV文件。您可以创建字典：

predicted_products = {'FILENAME': [], 'PRODUCT': []}

并将文件名和产品迭代地附加到相应的列表中。

完成此操作后，将预测产品转换为DataFrame，然后调用to_csv函数：

import Pandas as pd
predicted_products_df = pd.DataFrame.from_dict(predicted_products)
predicted_products_df.to_csv('your_path/file_name.csv')

我更喜欢这种方式，因为在保存文件之前，您可以更轻松地编辑数据。

对于您现有的代码，我想print(filenamez, ent.label_, ent.text)打印输出。如果是这样，则：

import Pandas as pd
output_dir = "C:\\Users\\Lenovo\\.spyder-py3\\NER_training"
DIR = 'C:\\Users\\Lenovo\\.spyder-py3\\Testing\\'
print("Loading from", output_dir)
nlp2 = spacy.load(output_dir)
predicted_products = {'FILENAME': [], 'PRODUCT': []}
with open('eng_productnames.csv', newline='') as myFile:
    reader = csv.reader(myFile)
    for rowz in reader:
        try:
            filenamez = rowz[1]
            file = open(DIR+filenamez, "r", encoding ='utf-8')
            filecontentszz = file.read()
            for s in filecontentszz:
                filecontentszz = re.sub(r'\s+', ' ', filecontentszz)
                #filecontents = filecontents.encode().decode('unicode-escape')
                filecontentszz = ''.join([line.lower() for line in filecontentszz]) 
                doc2 = nlp2(filecontentszz)
                for ent in doc2.ents:
                    print(filenamez, ent.label_, ent.text)
                    predicted_products['FILENAME'].append(filenamez + ' ' + ent.label_)
                    predicted_products['PRODUCT'].append(ent.text)
                break

        except Exception as e:

predicted_products_df = pd.DataFrame.from_dict(predicted_products)
predicted_products_df.to_csv('your_path/file_name.csv')

将输出从循环写入csv

2 个答案: