PDFテキスト変換

pdfminer

コマンド

!pip install pdfminer.six
!python /usr/local/bin/pdf2txt.py -o data.txt data.pdf

Python

from pdfminer.high_level import extract_text

text = extract_text("data.pdf")

pdfbox

コマンド

# 最新バージョンでURLが変わる
!wget https://www-eu.apache.org/dist/pdfbox/2.0.21/pdfbox-app-2.0.21.jar -O pdfbox-app.jar

# テキスト変換
!java -jar pdfbox-app.jar ExtractText -sort -encoding UTF-8 data.pdf

# 画像
!java -jar pdfbox-app.jar PDFToImage -imageType png -dpi 300 data.pdf

Python

!pip install python-pdfbox

import pdfbox

p = pdfbox.PDFBox()
p.extract_text("data.pdf", sort=True)

poppler

コマンド

!apt install poppler-utils poppler-data

# テキスト変換
!pdftotext -layout data.pdf

!pdfinfo data.pdf

# リペア
!pdftocairo -pdf data.pdf data_repaired.pdf

PDFの表変換

tabula

コマンド

!wget https://github.com/tabulapdf/tabula-java/releases/download/v1.0.4/tabula-1.0.4-jar-with-dependencies.jar -O tabula.jar

# lattice
!java -jar tabula.jar -o data.csv -p all -l data.pdf

# stream
!java -jar tabula.jar -o data.csv -p all -t data.pdf

Python

!pip install tabula-py

import pandas as pd
from tabula import read_pdf

dfs = read_pdf("data.pdf", pages="all", lattice=True)

dfs = read_pdf("data.pdf", pages="all", lattice=True, pandas_options={"header": None})

Camelot

コマンド

!apt install python3-tk ghostscript
!pip install camelot-py[cv]

# !pip install camelot-py[plot]

# !camelot --help

!camelot -p all -o data.csv -f csv lattice data.pdf

!camelot -p all -o data.csv -f csv -strip ' .\n' -split lattice -scale 40 data.pdf

Python

import camelot

tables = camelot.read_pdf("data.pdf", pages="all", split_text=True, strip_text=" \n", line_scale=40)

pdfplumber

!pip install pdfplumber
!apt install libmagickwand-dev ghostscript

# PDFを画像変換できるように/etc/ImageMagick-6/policy.xmlを上書き
%%writefile /etc/ImageMagick-6/policy.xml
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE policymap>

<policymap>

  <policy domain="resource" name="memory" value="256MiB"/>
  <policy domain="resource" name="map" value="512MiB"/>
  <policy domain="resource" name="width" value="16KP"/>
  <policy domain="resource" name="height" value="16KP"/>
  <policy domain="resource" name="area" value="128MB"/>
  <policy domain="resource" name="disk" value="1GiB"/>

  <policy domain="delegate" rights="none" pattern="URL"/>
  <policy domain="delegate" rights="none" pattern="HTTPS"/>
  <policy domain="delegate" rights="none" pattern="HTTP"/>

  <policy domain="path" rights="none" pattern="@*"/>
  <policy domain="cache" name="shared-secret" value="passphrase" stealth="true"/>

  <policy domain="coder" rights="none" pattern="PS"/>
  <policy domain="coder" rights="none" pattern="PS2"/>
  <policy domain="coder" rights="none" pattern="PS3"/>
  <policy domain="coder" rights="none" pattern="EPS"/>
  <policy domain="coder" rights="read|write" pattern="PDF" />
  <policy domain="coder" rights="none" pattern="XPS"/>
</policymap>

!pdfplumber < data.pdf > data.csv

import pdfplumber
import pandas as pd

pdf = pdfplumber.open("data.pdf")

page = pdf.pages[0]

page.find_tables()[0]

# 文字の位置確認
page.chars

# cropでテキスト取得
crop = page.within_bbox((0, 90, page.width, 105))

s = crop.extract_text()
s

# PDF確認
im = page.to_image()
im

table_settings = {

    # 垂直基準
    "vertical_strategy": "lines",
    # 垂直区切を数値指定（リスト）
    "explicit_vertical_lines": [],

    # 水平基準
    "horizontal_strategy": "lines",
    # 水平区切を数値指定（リスト）
    "explicit_horizontal_lines": [],

    # 許容範囲内の場合、同じ水平位置または垂直位置に調整　※上に調整？
    "snap_tolerance": 3,

    # 許容範囲内の場合、結合
    "join_tolerance": 3,

    # テーブルの再構築を試みる前に破棄されるよりも短いエッジ？
    "edge_min_length": 3,

    # 最小の文字の高さ
    "min_words_vertical": 3,

    # 最小の文字の高さ
    "min_words_horizontal": 1,

    # 空白文字を単語の一部と認識し、区切り文字にしない
    "keep_blank_chars": False,

    # 文字の間隔が以下の場合、単語と認識
    "text_tolerance": 3,
    "text_x_tolerance": None,
    "text_y_tolerance": None,
    
    # テキストの左端と右端が垂直線と完全に一致していない場合の許容値？
    "intersection_tolerance": 3,
    "intersection_x_tolerance": None,
    "intersection_y_tolerance": None,
}

# 文字確認
im.reset().draw_rects(page.extract_words())

# テーブル確認
im.reset().debug_tablefinder()

with pdfplumber.open("data.pdf") as pdf:

    dfs = []

    for page in pdf.pages:

        table = page.extract_table(table_settings)

        df_tmp = pd.DataFrame(table[1:], columns=table[0])

        dfs.append(df_tmp)

df = pd.concat(dfs)

OCR

tesseract-ocr

!add-apt-repository ppa:alex-p/tesseract-ocr -y
!apt update
!apt install tesseract-ocr
!apt install libtesseract-dev
!tesseract -v

!apt install tesseract-ocr-jpn  tesseract-ocr-jpn-vert
!apt install tesseract-ocr-script-jpan tesseract-ocr-script-jpan-vert
!tesseract --list-langs
!pip install pytesseract

try:
    from PIL import Image
except ImportError:
    import Image
    
import pytesseract

import cv2
import numpy as np

from google.colab.patches import cv2_imshow

img = cv2.imread("test.jpg")

# 白黒反転
img_gray, _ = cv2.decolor(img)

cv2_imshow(img_gray)