Compare PDFs

[pmkochie]

I recently had the need to compare 2 PDFs. I found a website that does it fairly well https://draftable.com/compare. The site looks like it OCRs the 2 PDFs and then compares the contents. The site also compares 2 Word documents.
It would be nice to be able to compare 2 documents based on the contents after an OCR scan.

[sdottaka]

WinMerge can compare text within PDF files using plugins. Two plugins available are xdocdiffPlugin and Apache Tika plugin.

[xdocdiffPlugin]
http://freemind.s57.xrea.com/xdocdiffPlugin/en/index.html

[Apache Tika]

Additionally, PDF files can be compared as images by using the following method. In image comparison, Windows' OCR function can be used to extract text from within the image.

[elesbb]

I see!
Thank you. I originally downloaded the 3.0.0 Jar from Apache Tika website, then was trying to install it manually. Just letting WinMerge do it worked.

I did notice it downloaded 2.6.0, any way to update to the 3.0.0 jar?

[sdottaka]

I did notice it downloaded 2.6.0, any way to update to the 3.0.0 jar?

You need to edit C:\Program Files\WinMerge\Command\Apache-Tika\tika.bat. I think it's better to wait for the next version.

[elesbb]

Thanks! Will do

[qingrenquan]

Thanks! Will do  谢谢你，谢谢将尽

修改tika.bat就可以使用tika3.0了，因为只是调用了一下jar位置。
java8无法使用tika3.0，不过程序默认现有java环境为java19，是可以使用tika3.0的，也可以改java.bat，选择某个java版本。但是默认java19就可以了，大于java11+即可使用tika3.0。

这是问ai的tika.bat脚本，去访问d:\jar\tika-x.jar
@echo off
setlocal EnableDelayedExpansion
set TikaVer=2.6.0
set TikaJar=tika-app-%TikaVer%.jar
set DOWNLOAD_URL=https://repo1.maven.org/maven2/org/apache/tika/tika-app/%TikaVer%/%TikaJar%
set TIKA_PATH=D:\jar%TikaJar%
set MESSAGE='Apache Tika is not installed. Do you want to download it from %DOWNLOAD_URL%'
set TITLE='Apache Tika Plugin'
set TIKA_SHA256=fa289b58a5c1bb531ace78324625512a9448aa8472b5eb88b65988964048815a

cd "%APPDATA%\WinMerge"
if not exist "D:\jar%TikaJar%" (
cd "%~dp0...."
if not exist "D:\jar%TikaJar%" (
mkdir "D:\jar" 2> NUL
cd "D:\jar"
for %%i in (%TIKA_PATH%) do mkdir %%~pi 2> NUL
powershell "if ((New-Object -com WScript.Shell).Popup(%MESSAGE%,0,%TITLE%,1) -ne 1) { throw }" > NUL
if errorlevel 1 (
echo "download is canceled" 1>&2
) else (
start "Downloading..." /WAIT powershell -command "[Net.ServicePointManager]::SecurityProtocol = [Net.SecurityProtocolType]::Tls12; Invoke-WebRequest -Uri %DOWNLOAD_URL% -UseBasicParsing -Outfile %TIKA_PATH%"
powershell -command "$(CertUtil -hashfile '%TIKA_PATH%' SHA256)[1] -replace ' ','' -eq '%TIKA_SHA256%'" | findstr True > NUL
if errorlevel 1 (
echo %TIKA_PATH%: download failed 1>&2
del %TIKA_PATH% 2> NUL
)
)
)
)

"%~dp0..\Java\java.bat" -jar "D:\jar%TikaJar%" %3 %4 %5 %6 %7 %8 %9 "%~1" > "%~2"

我遇到了默认下载2.6，jar包损坏的问题。所以就修改这个tika.bat了找原因了（在下面你可以看到我的贴子留言）。

[qingrenquan]

Thanks! Will do  谢谢你，谢谢将尽

请注意，如果epub带有图片，tika3.0默认会有一个ocr图片过程，winmerge可能会卡死崩溃。

可以修改tika.bat脚本，去禁用默认的ocr功能。
✅ 解决方案：禁用 Tesseract OCR
方法 1：使用环境变量临时禁用 OCR
在 PowerShell 运行：
powershell复制编辑$env:TESSDATA_PREFIX="DISABLED"
java -jar d:\jar\tika-app-3.1.0.jar -t 2.epub > extracted_text1.txt

默认的ocr效果是不好的，我执行没有ocr，而是一堆乱码其他东西（图片ocr），似乎没有正确配置tesseract ocr。
使用ocr，需要单独在java项目里去用这个tesseract ocr这个识别图片文字能力。

我现在可以正常从epub通过tika提取文本（包括图片名称，但是是文本非图片），默认tika -extract 是只提取图片。

不过由于epub是有层级的zip，其实完全可以用python改名为zip或者解压为文件夹，然后直接对比文件夹就可以

[pmkochie]

I can't download it because my company firewall blockers content from Russia and China.

…

On Wed, Apr 5, 2023 at 8:33 AM Takashi Sawanaka ***@***.***> wrote: WinMerge can compare text within PDF files using plugins. Two plugins available are xdocdiffPlugin and Apache Tika plugin. [xdocdiffPlugin] http://freemind.s57.xrea.com/xdocdiffPlugin/en/index.html [Apache Tika] [image: image] <https://user-images.githubusercontent.com/98126/230129792-7f15a9e4-ddd7-459b-88b9-38750f5a708f.png> Additionally, PDF files can be compared as images by using the following method. In image comparison, Windows' OCR function can be used to extract text from within the image. [image: image] <https://user-images.githubusercontent.com/98126/230129581-bb701956-098f-4e92-9fc4-c359f91585c8.png> [image: image] <https://user-images.githubusercontent.com/98126/230127765-db87a691-8c14-434c-ba9a-55b60f151e0f.png> — Reply to this email directly, view it on GitHub <#1792 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AGVYPKI4BHVQ2TCTZSTRCZ3W7WGE3ANCNFSM6AAAAAAWUFTAUU> . You are receiving this because you authored the thread.Message ID: ***@***.***>

-- -- Paul Kochie ***@***.***

________oooo |___|__|__q\___\,____ L.---.\__|__|.---.(OlllllllO) (o)_) (o)_)---(o)_)