我使用非结构化 io 来解析 pdf 中的表数据。我编写了以下代码来从我的 pdf 中提取表格，其中主要是银行对账单。将 PDF 直接转换为 csv 不起作用，因此我首先继续转换为 json，因为这样可以更好地显示表格、标题和叙述文本。 python代码如下所示：
from unstructured.staging.base import elements_to_json, elements_from_json
from unstructured.staging.base import convert_to_dict
from unstructured.documents.elements import Title, NarrativeText, Table
from unstructured.staging.base import convert_to_csv
import json
import csv
import pandas as pd
input = "a1.pdf"
elements = partition_pdf(filename=input, infer_table_structure=True)
tables = [el for el in elements if el.category == "Table"]
print(tables[0].text)
print(tables[0].metadata.text_as_html)
convert_to_dict(elements)
output = "outputs.json"
elements_to_json(elements, filename=output)
elements = elements_from_json(filename=output)
#df = pd.read_json(output)
#df.to_csv('finale.csv')
# Read JSON file
with open('outputs.json', 'r') as file:
data = json.load(file)
# Extract Titles and Tables
titles = []
tables = []
for element in data:
if element['type'] == 'Title':
titles.append(element)
elif element['type'] == 'Table':
tables.append(element)
print(titles)
print(tables)
isd_csv = convert_to_csv(tables)
with open('output_tables.csv', 'w', newline='') as csvfile:
writer = csv.writer(csvfile)
writer.writerows(isd_csv.splitlines()) # Split lines and write as rows```
该表提取得相当准确，但它是 json 格式，如下所示。 csv 函数不起作用。是否可以在 json 文件上运行一些脚本并获取 csv 形式或 HTML 形式的表格、标题和叙述文本，这会提供更高的准确性（最好使用 csv）。首选脚本是 python，因为我项目的所有其他部分都是基于 python 的。
[
{
"element_id": "c1eb5f7eb3b8a8ebd25ddcc63e305006",
"metadata": {
"coordinates": {
"layout_height": 2200,
"layout_width": 1700,
"points": [
[
144.7,
309.2
],
[
144.7,
353.6
],
[
675.2,
353.6
],
[
675.2,
309.2
]
],
"system": "PixelSpace"
},
"detection_class_prob": 0.55205,
"filename": "a1.pdf",
"filetype": "application/pdf",
"languages": [
"eng"
],
"last_modified": "2024-01-09T13:02:21",
"page_number": 1
},
"text": "CIBC Account Statement",
"type": "Title"
},
{
"element_id": "bfe2311a88365d06d632e2bc1df2e5bd",
"metadata": {
"coordinates": {
"layout_height": 2200,
"layout_width": 1700,
"points": [
[
144.1,
430.4
],
[
144.1,
458.1
],
[
419.5,
458.1
],
[
419.5,
430.4
]
],
"system": "PixelSpace"
},
"detection_class_prob": 0.63581,
"filename": "a1.pdf",
"filetype": "application/pdf",
"languages": [
"eng"
],
"last_modified": "2024-01-09T13:02:21",
"page_number": 1,
"parent_id": "c1eb5f7eb3b8a8ebd25ddcc63e305006"
},
"text": "2431536 ONTARIO INC.",
"type": "NarrativeText"
},
{
"element_id": "045c030ba3939ff17e7c1f26bc38175f",
"metadata": {
"coordinates": {
"layout_height": 2200,
"layout_width": 1700,
"points": [
[
1110.8,
433.0
],
[
1110.8,
458.0
],
[
1423.9,
458.0
],
[
1423.9,
433.0
]
],
"system": "PixelSpace"
},
"detection_class_prob": 0.34244,
"filename": "a1.pdf",
"filetype": "application/pdf",
"languages": [
"eng"
],
"last_modified": "2024-01-09T13:02:21",
"page_number": 1
},
"text": "For Aug 1 to Aug 31, 2021",
"type": "Title"
},
{
"element_id": "146dab2a99fac51981d5b136459f7be7",
"metadata": {
"coordinates": {
"layout_height": 2200,
"layout_width": 1700,
"points": [
[
1114.1,
496.8
],
[
1114.1,
552.4
],
[
1318.1,
552.4
],
[
1318.1,
496.8
]
],
"system": "PixelSpace"
},
"detection_class_prob": 0.30347,
"filename": "a1.pdf",
"filetype": "application/pdf",
"languages": [
"eng"
],
"last_modified": "2024-01-09T13:02:21",
"page_number": 1,
"parent_id": "045c030ba3939ff17e7c1f26bc38175f"
},
"text": "Account number 70-10117",
"type": "NarrativeText"
},
{
"element_id": "6b525f6ae0b734f9a6869972775aa16e",
"metadata": {
"coordinates": {
"layout_height": 2200,
"layout_width": 1700,
"points": [
[
143.5,
574.1
],
[
143.5,
659.1
],
[
920.8,
659.1
],
[
920.8,
574.1
]
],
"system": "PixelSpace"
},
"detection_class_prob": 0.86173,
"filename": "a1.pdf",
"filetype": "application/pdf",
"languages": [
"eng"
],
"last_modified": "2024-01-09T13:02:21",
"page_number": 1,
"parent_id": "045c030ba3939ff17e7c1f26bc38175f"
},
"text": "The names shown are based on our current records, as of March 2, 2022. This statement does not reflect any changes in account holders and account holder names that may have occurred prior to this date.",
"type": "NarrativeText"
},
{
"element_id": "cbd860785f81791a1c73b9c75ba6407a",
"metadata": {
"coordinates": {
"layout_height": 2200,
"layout_width": 1700,
"points": [
[
1114.1,
591.3
],
[
1114.1,
646.8
],
[
1383.9,
646.8
],
[
1383.9,
591.3
]
],
"system": "PixelSpace"
},
"detection_class_prob": 0.42099,
"filename": "a1.pdf",
"filetype": "application/pdf",
"languages": [
"eng"
],
"last_modified": "2024-01-09T13:02:21",
"page_number": 1,
"parent_id": "045c030ba3939ff17e7c1f26bc38175f"
},
"text": "Branch transit number 07342",
"type": "NarrativeText"
},
{
"element_id": "7aa85348022f61c9fc791f63346fd68f",
"metadata": {
"coordinates": {
"layout_height": 2200,
"layout_width": 1700,
"points": [
[
1116.2,
803.6
],
[
1116.2,
837.5
],
[
1437.6,
837.5
],
[
1437.6,
803.6
]
],
"system": "PixelSpace"
},
"detection_class_prob": 0.69574,
"filename": "a1.pdf",
"filetype": "application/pdf",
"languages": [
"eng"
],
"last_modified": "2024-01-09T13:02:21",
"page_number": 1
},
"text": "Contact information",
"type": "Title"
},
{
"element_id": "048adaba40bd8c5501ca9a5219891e32",
"metadata": {
"coordinates": {
"layout_height": 2200,
"layout_width": 1700,
"points": [
[
126.0,
813.0
],
[
126.0,
1050.9
],
[
1579.5,
1050.9
],
[
1579.5,
813.0
]
],
"system": "PixelSpace"
},
"detection_class_prob": 0.53915,
"filename": "a1.pdf",
"filetype": "application/pdf",
"languages": [
"eng"
],
"last_modified": "2024-01-09T13:02:21",
"page_number": 1,
"parent_id": "7aa85348022f61c9fc791f63346fd68f",
"text_as_html": "<table><tr><td>Opening balance on Aug 1, 2021</td><td>$43,719.76</td><td>@</td><td>1800 465 CIBC (2422)</td></tr><tr><td>Withdrawals</td><td>14,316.36</td><td></td><td>Contact us by phone for questions this of</td></tr><tr><td>Deposits</td><td>20,801.01</td><td></td><td>on update, change personal information, and general inquiries,</td></tr><tr><td>Closing balance on Aug 31, 2021</td><td>$50,204.41</td><td></td><td>24 hours a day, 7 days a week.</td></tr></table>"
},
"text": "Opening balance on Aug 1, 2021 $43,719.76 Withdrawals - 14,316.36 Deposits + 20,801.01 Closing balance on Aug 31, 2021 = $50,204.41 1 800 465 CIBC (2422) Contact us by phone for questions on this update, change of personal information, and general inquiries, 24 hours a day, 7 days a week.",
"type": "Table"
},
{
"element_id": "0e6ab9db7f9a8c1b8cefa24c9ccff170",
"metadata": {
"coordinates": {
"layout_height": 2200,
"layout_width": 1700,
"points": [
[
1159.8,
1030.2
],
[
1159.8,
1085.7
],
[
1415.5,
1085.7
],
[
1415.5,
1030.2
]
],
"system": "PixelSpace"
},
"filename": "a1.pdf",
"filetype": "application/pdf",
"languages": [
"eng"
],
"last_modified": "2024-01-09T13:02:21",
"page_number": 1,
"parent_id": "7aa85348022f61c9fc791f63346fd68f"
},
"text": "TTY hearing impaired 1 800 465 7401",
"type": "NarrativeText"
},
{
"element_id": "43ea2d37a7b84d6e13db58e27ead0f01",
"metadata": {
"coordinates": {
"layout_height": 2200,
"layout_width": 1700,
"points": [
[
1159.8,
1099.6
],
[
1159.8,
1194.6
],
[
1501.6,
1194.6
],
[
1501.6,
1099.6
]
],
"system": "PixelSpace"
},
"filename": "a1.pdf",
"filetype": "application/pdf",
"languages": [
"eng"
],
"last_modified": "2024-01-09T13:02:21",
"page_number": 1
},
"text": "Outside Canada and the U.S. 1 902 420 CIBC (2422) www.cibc.com",
"type": "Title"
},
{
"element_id": "15b618d7fc57f900d275074b3651a1fb",
"metadata": {
"coordinates": {
"layout_height": 2200,
"layout_width": 1700,
"points": [
[
147.2,
1306.2
],
[
147.2,
1340.2
],
[
451.8,
1340.2
],
[
451.8,
1306.2
]
],
"system": "PixelSpace"
},
"detection_class_prob": 0.64498,
"filename": "a1.pdf",
"filetype": "application/pdf",
"languages": [
"eng"
],
"last_modified": "2024-01-09T13:02:21",
"page_number": 1
},
"text": "Transaction details",
"type": "Title"
},
{
"element_id": "0c9edae94cc9cec1eb6c559448a54385",
"metadata": {
"coordinates": {
"layout_height": 2200,
"layout_width": 1700,
"points": [
[
122.4,
1348.7
],
[
122.4,
1998.8
],
[
1593.5,
1998.8
],
[
1593.5,
1348.7
]
],
"system": "PixelSpace"
},
"detection_class_prob": 0.90372,
"filename": "a1.pdf",
"filetype": "application/pdf",
"languages": [
"eng"
],
"last_modified": "2024-01-09T13:02:21",
"page_number": 1,
"parent_id": "15b618d7fc57f900d275074b3651a1fb",
"text_as_html": "<table><thead><th rowspan=\"2\">Date Aug1</th><th>Description</th><th>Withdrawals ($) Deposits</th><th rowspan=\"2\">($) Balance ($) $43,719.76</th></thead><thead><th></th><th>_ Opening balance</th><th></th><th></th></thead><tr><td rowspan=\"5\">Aug 3</td><td>INTERNET BILL PMT000000214174</td><td>113.70</td><td>43,606.06</td></tr><tr><td></td><td>FIDO</td><td></td><td></td></tr><tr><td></td><td>A506*#ERRRKT QD</td><td></td><td></td></tr><tr><td></td><td>E-TRANSFER103853228862 Erhan Bakirci OD</td><td>2,000.00</td><td>41,606.06</td></tr><tr><td></td><td>ABO GEERT</td><td></td><td></td></tr><tr><td rowspan=\"5\">Aug 5</td><td>INTERNET BILL PMT000000235565 MASTERCARD, MBNA CANADA B A506*#ERRRKT</td><td>80.00</td><td>41,526.06</td></tr><tr><td></td><td>QD</td><td></td><td></td></tr><tr><td></td><td>INTERNET BILL PMT000000216308</td><td>54.50</td><td>41,471.56</td></tr><tr><td></td><td>PROVIDENT ENERGY MANAGEME A506*#ERRRKT</td><td></td><td></td></tr><tr><td></td><td>QD</td><td></td><td></td></tr><tr><td rowspan=\"2\">Aug 23.</td><td>INTERNET BILL PMT000000233266 ROGERS (9 DIGIT ACCOUNT N A506*#ERRRHET QD</td><td>142.30</td><td>41,329.26.</td></tr><tr><td></td><td></td><td></td><td></td></tr></table>"
},
"text": "Date Aug 1 Aug 3 Aug 5 Aug 23 Description Opening balance INTERNET BILL PMT000000214174 FIDO 4506*********792 E-TRANSFER103853228862 Erhan Bakirci 4506*********792 INTERNET BILL PMT000000235565 MASTERCARD, MBNA CANADA B 4506*********792 INTERNET BILL PMT000000216308 PROVIDENT ENERGY MANAGEME 4506*********792 INTERNET BILL PMT000000233266 ROGERS (9 DIGIT ACCOUNT N 4506*********792 Withdrawals ($) 113.70 2,000.00 80.00 54.50 142.30 Deposits ($) Balance ($) $43,719.76 43,606.06 41,606.06 41,526.06 41,471.56 41,329.26 (continued on next page)",
"type": "Table"
},
{
"element_id": "6c8bfe0f2340b6f05354176802ea337c",
"metadata": {
"coordinates": {
"layout_height": 2200,
"layout_width": 1700,
"points": [
[
138.9,
2118.5
],
[
138.9,
2140.1
],
[
328.2,
2140.1
],
[
328.2,
2118.5
]
],
"system": "PixelSpace"
},
"detection_class_prob": 0.59505,
"filename": "a1.pdf",
"filetype": "application/pdf",
"languages": [
"eng"
],
"last_modified": "2024-01-09T13:02:21",
"page_number": 1,
"parent_id": "15b618d7fc57f900d275074b3651a1fb"
},
"text": "10774E BUS-2018/09",
"type": "NarrativeText"
},
{
"element_id": "33ed8c9803ae2dbbee4c676a19442128",
"metadata": {
"coordinates": {
"layout_height": 2200,
"layout_width": 1700,
"points": [
[
1435.3,
2119.9
],
[
1435.3,
2143.5
],
[
1553.4,
2143.5
],
[
1553.4,
2119.9
]
],
"system": "PixelSpace"
},
"detection_class_prob": 0.44888,
"filename": "a1.pdf",
"filetype": "application/pdf",
"languages": [
"eng"
],
"last_modified": "2024-01-09T13:02:21",
"page_number": 1,
"parent_id": "15b618d7fc57f900d275074b3651a1fb"
},
"text": "Page 1 of 2",
"type": "NarrativeText"
},
{
"element_id": "c1eb5f7eb3b8a8ebd25ddcc63e305006",
"metadata": {
"coordinates": {
"layout_height": 2200,
"layout_width": 1700,
"points": [
[
153.8,
106.9
],
[
153.8,
140.2
],
[
553.5,
140.2
],
[
553.5,
106.9
]
],
"system": "PixelSpace"
},
"detection_class_prob": 0.49285,
"filename": "a1.pdf",
"filetype": "application/pdf",
"languages": [
"eng"
],
"last_modified": "2024-01-09T13:02:21",
"page_number": 2
},
"text": "CIBC Account Statement",
"type": "Title"
},
{
"element_id": "369b08001627723a6f47814a95a5405b",
"metadata": {
"coordinates": {
"layout_height": 2200,
"layout_width": 1700,
"points": [
[
1110.6,
103.4
],
[
1110.6,
211.6
],
[
1467.0,
211.6
],
[
1467.0,
103.4
]
],
"system": "PixelSpace"
},
"detection_class_prob": 0.36904,
"filename": "a1.pdf",
"filetype": "application/pdf",
"languages": [
"eng"
],
"last_modified": "2024-01-09T13:02:21",
"page_number": 2,
"parent_id": "c1eb5f7eb3b8a8ebd25ddcc63e305006"
},
"text": "Aug 1 to Aug 31, 2021 Account number: 70-10117 Branch transit number: 07342",
"type": "NarrativeText"
},
{
"element_id": "2babb7fc184bfcfac14a46fded4eaef4",
"metadata": {
"coordinates": {
"layout_height": 2200,
"layout_width": 1700,
"points": [
[
147.2,
286.2
],
[
147.2,
320.8
],
[
611.2,
320.8
],
[
611.2,
286.2
]
],
"system": "PixelSpace"
},
"detection_class_prob": 0.76239,
"filename": "a1.pdf",
"filetype": "application/pdf",
"languages": [
"eng"
],
"last_modified": "2024-01-09T13:02:21",
"page_number": 2
},
"text": "Transaction details (continued)",
"type": "Title"
},
{
"element_id": "73d209c6bf0ebc45f357118b41f43977",
"metadata": {
"coordinates": {
"layout_height": 2200,
"layout_width": 1700,
"points": [
[
137.0,
329.6
],
[
137.0,
1122.5
],
[
1588.8,
1122.5
],
[
1588.8,
329.6
]
],
"system": "PixelSpace"
},
"detection_class_prob": 0.92363,
"filename": "a1.pdf",
"filetype": "application/pdf",
"languages": [
"eng"
],
"last_modified": "2024-01-09T13:02:21",
"page_number": 2,
"parent_id": "2babb7fc184bfcfac14a46fded4eaef4",
"text_as_html": "<table><thead><th rowspan=\"2\">Date Aug 23__</th><th rowspan=\"2\">Description Balance forward</th><th>Withdrawals ($)</th><th>Deposits ($)</th><th>Balance ($)</th></thead><thead><th></th><th></th><th></th><th></th><th>$41,329.26</th></thead><tr><td></td><td>NTERNET BILL PMT000000233444</td><td>70.00</td><td></td><td>41,259.26</td></tr><tr><td></td><td>AMERICAN EXPRESS REGULAR</td><td></td><td></td><td></td></tr><tr><td></td><td>A506 HHH TOD</td><td></td><td></td><td></td></tr><tr><td></td><td>nnn</td><td></td><td>Beeeoo</td><td>650836\"</td></tr><tr><td></td><td>DIXIE & MEYERSIDE BANKING CENT</td><td></td><td></td><td></td></tr><tr><td></td><td>DEPOSIT 06122</td><td></td><td>5,650.00</td><td>. 52,559.26</td></tr><tr><td></td><td>DIXIE CENT</td><td></td><td></td><td></td></tr><tr><td></td><td>& MEYERSIDE BANKING</td><td></td><td>eee</td><td>een</td></tr><tr><td></td><td>DEPOSIT 06122</td><td></td><td>3,826.01</td><td>56,385.27</td></tr><tr><td>co!</td><td>DIXIE & MEYERSIDE BANKING CENT</td><td></td><td>enn</td><td>nnn</td></tr><tr><td>Aug 24</td><td>CHEQUE 76179759 | OO</td><td>1888.00</td><td></td><td>54,502.27 |</td></tr><tr><td>25</td><td>DEPOSIT 06122</td><td></td><td></td><td></td></tr><tr><td>Aug</td><td></td><td></td><td>5,650.00</td><td>60,152.27</td></tr><tr><td>co!</td><td>DIXIE & MEYERSIDE BANKING CENT</td><td></td><td>eee</td><td>neeneeeeneeneneneee</td></tr><tr><td>Aug 27</td><td>NTERNET TRANSFERO00000249234 A500 HHH ZOD</td><td>9,857.86</td><td></td><td>50,294.41</td></tr><tr><td rowspan=\"3\">\u2018Aug 30</td><td>INTERNET BILLPMTO00000254696</td><td>9.00.</td><td></td><td>50,204.41</td></tr><tr><td></td><td>BELL CANADA - ONE BILL</td><td></td><td></td><td></td></tr><tr><td></td><td>A506 HHH TOD</td><td></td><td></td><td></td></tr><tr><td>Aug 31</td><td>ACC FEE- FULL SERV</td><td></td><td></td><td></td></tr></table>"
},
"text": "Date Aug 23 Balance forward Description Withdrawals ($) Deposits ($) Aug 24 Aug 25 Aug 27 Aug 30 INTERNET BILL PMT000000233444 AMERICAN EXPRESS REGULAR 4506*********792 DEPOSIT 06122 DIXIE & MEYERSIDE BANKING CENT DEPOSIT 06122 DIXIE & MEYERSIDE BANKING CENT DEPOSIT 06122 DIXIE & MEYERSIDE BANKING CENT CHEQUE 78179759 6 DEPOSIT 06122 DIXIE & MEYERSIDE BANKING CENT INTERNET TRANSFER000000249234 4500*********302 INTERNET BILL PMT000000254696 BELL CANADA - ONE BILL 4506*********792 ACC FEE- FULL SERV BALANCE FEE WAIVER 70.00 1,883.00 9,857.86 90.00 5,650.00 5,650.00 3,826.01 5,650.00 Aug 31 25.00 25.00 Balance ($) $41,329.26 41,259.26 46,909.26 52,559.26 56,385.27 54,502.27 60,152.27 50,294.41 50,204.41 50,179.41 50,204.41",
"type": "Table"
},
{
"element_id": "0d88225a095dfbc5407269264a9c59be",
"metadata": {
"coordinates": {
"layout_height": 2200,
"layout_width": 1700,
"points": [
[
263.0,
1117.1
],
[
263.0,
1142.1
],
[
454.7,
1142.1
],
[
454.7,
1117.1
]
...
我们可以制作另一个 Json 到 csv 的代码文件，将其单独保存在一个文件中，并使您编写的第一个代码调用第二个文件中的第二个代码，这会将 json 转换为 csv
代码可以写成
import json
import csv
with open('data.json') as json_file:
data = json.load(json_file)
employee_data = data['emp_details']
data_file = open('data_file.csv', 'w')
csv_writer = csv.writer(data_file)
count = 0
for emp in employee_data:
if count == 0:
header = emp.keys()
csv_writer.writerow(header)
count += 1
csv_writer.writerow(emp.values())
data_file.close()
更深入的细节请参见 https://www.geeksforgeeks.org/convert-json-to-csv-in-python/
第二种方法可能是使用 pandas
import pandas as pd
df = pd.read_json('sample.json')
df.to_csv('file.csv')
更深入的细节正在上 https://datagy.io/python-json-to-csv/