Update NSIS installer

- Add manual pages in HTML format and helper for Tesseract command line
- Don't remove the installation directory recursively
- Add GitHub action for Tesseract installer for Windows
- Add docbook-xml to required packages (needed for doc)
- Use unicode for NSIS installer
- Optionally sign executables
- Add more file properties to installer
- Update configuration for use with pacman
- Build Windows installer only for 64 bit Windows

Signed-off-by: Stefan Weil <sw@weilnetz.de>

.github/workflows/installer-for-windows.yml

@@ -0,0 +1,27 @@
+# GitHub actions - Create Tesseract installer for Windows
+
+name: Cross build for Windows
+
+on:
+  # Trigger workflow in GitHub web frontend or from API.
+  workflow_dispatch:
+    inputs:
+      targets:
+        description: 'Target operating system'
+        required: true
+        default: 'Windows (64 bit)'
+        type: choice
+        options:
+          - 'Windows (64 bit)'
+
+jobs:
+  build64:
+    runs-on: [ubuntu-24.04]
+    steps:
+    - uses: actions/checkout@v4
+    - name: Build Tesseract installer (64 bit)
+      run: nsis/build.sh x86_64
+    - uses: actions/upload-artifact@v4
+      with:
+        name: Tesseract Installer for Windows (64 bit)
+        path: dist

.github/workflows/pkg-config-crosswrapper

@@ -0,0 +1,54 @@
+#! /bin/sh
+# pkg-config wrapper for cross-building
+# Sets pkg-config search path to search multiarch and historical cross-compiling paths.
+
+# If the user has already set PKG_CONFIG_LIBDIR, believe it (even if empty):
+# it's documented to be an override
+if [ x"${PKG_CONFIG_LIBDIR+set}" = x ]; then
+  # GNU triplet for the compiler, e.g. i486-linux-gnu for Debian i386,
+  # i686-linux-gnu for Ubuntu i386
+  basename="$(basename "$0")"
+  triplet="${basename%-pkg-config}"
+  # Normalized multiarch path if any, e.g. i386-linux-gnu for i386
+  dpkg-architecture >/dev/null 2>&1
+  if [ "$?" != 0 ]; then
+      # dpkg-architecture is missing.
+      echo "Please install dpkg-dev to use pkg-config when cross-building" >&2
+      exit 1
+  fi
+  multiarch="$(dpkg-architecture -t"${triplet}" -qDEB_HOST_MULTIARCH 2>/dev/null)"
+  # Native multiarch path
+  native_multiarch="$(cat /usr/lib/pkg-config.multiarch)"
+
+  # This can be used for native builds as well, in that case, just exec pkg-config "$@" directly.
+  if [ "$native_multiarch" = "$multiarch" ]; then
+     exec pkg-config "$@"
+  fi
+
+  PKG_CONFIG_LIBDIR="/usr/local/${triplet}/lib/pkgconfig"
+  # For a native build we would also want to append /usr/local/lib/pkgconfig
+  # at this point; but this is a cross-building script, so don't
+  PKG_CONFIG_LIBDIR="$PKG_CONFIG_LIBDIR:/usr/local/share/pkgconfig"
+
+  if [ -n "$multiarch" ]; then
+    PKG_CONFIG_LIBDIR="/usr/local/lib/${multiarch}/pkgconfig:$PKG_CONFIG_LIBDIR"
+    PKG_CONFIG_LIBDIR="$PKG_CONFIG_LIBDIR:/usr/lib/${multiarch}/pkgconfig"
+  fi
+
+  PKG_CONFIG_LIBDIR="$PKG_CONFIG_LIBDIR:/usr/${triplet}/lib/pkgconfig"
+  # For a native build we would also want to append /usr/lib/pkgconfig
+  # at this point; but this is a cross-building script, so don't
+  # If you want to allow use of un-multiarched -dev packages for crossing
+  # (at the risk of finding build-arch stuff you didn't want, if not in a clean chroot)
+  # Uncomment the next line:
+  # PKG_CONFIG_LIBDIR="$PKG_CONFIG_LIBDIR:/usr/lib/pkgconfig"
+  # ... but on Ubuntu we rely cross-building with non-multiarch libraries:
+  if dpkg-vendor --derives-from Ubuntu; then
+      PKG_CONFIG_LIBDIR="$PKG_CONFIG_LIBDIR:/usr/lib/pkgconfig"
+  fi
+  PKG_CONFIG_LIBDIR="$PKG_CONFIG_LIBDIR:/usr/share/pkgconfig"
+
+  export PKG_CONFIG_LIBDIR
+fi
+
+exec pkg-config "$@"

nsis/Makefile.am

@@ -4,7 +4,7 @@ all:
 
 if MINGW
 
-gitrev="$(shell git --git-dir=${abs_top_srcdir}/.git --work-tree=${abs_top_srcdir} describe --always --tags)"
+gitrev="$(shell git --git-dir=${abs_top_srcdir}/.git --work-tree=${abs_top_srcdir} describe --always --tags | sed s/^v//)"
 
 .PHONY: winsetup
 
@@ -17,6 +17,6 @@ winpath.exe: winpath.cpp
 	x86_64-w64-mingw32-strip --strip-unneeded $@
 
 winsetup: Plugins/x86-unicode/INetC.dll winpath.exe
-	makensis -DCROSSBUILD -DSHARED -DSRCDIR=$(top_srcdir) -DVERSION=${gitrev} $(shell test "$(host_cpu)" = x86_64 && echo "-DW64") -NOCD $(top_srcdir)/nsis/tesseract.nsi
+	makensis -DCROSSBUILD -DSHARED -DSIGNCODE=$(SIGNCODE) -DSRCDIR=$(top_srcdir) -DVERSION=${gitrev} $(shell test "$(host_cpu)" = x86_64 && echo "-DW64") -NOCD $(top_srcdir)/nsis/tesseract.nsi
 
 endif

nsis/build.sh

@@ -0,0 +1,115 @@
+#!/bin/bash
+
+# GitHub actions - Create Tesseract installer for Windows
+
+# Author: Stefan Weil (2010-2024)
+
+set -e
+set -x
+
+LANG=C.UTF-8
+
+ARCH=$1
+
+if [ "$ARCH" = "i686" ]; then
+  MINGW=/mingw32
+else
+  ARCH=x86_64
+  MINGW=/mingw64
+fi
+
+ROOTDIR=$PWD
+DISTDIR=$ROOTDIR/dist
+HOST=$ARCH-w64-mingw32
+TAG=$(cat VERSION).$(date +%Y%m%d)
+BUILDDIR=bin/ndebug/$HOST-$TAG
+PKG_ARCH=mingw-w64-${ARCH/_/-}
+
+# Install packages.
+sudo apt-get update --quiet
+sudo apt-get install --assume-yes --no-install-recommends --quiet \
+  asciidoc xsltproc docbook-xml docbook-xsl \
+  automake dpkg-dev libtool pkg-config default-jdk-headless \
+  mingw-w64-tools nsis g++-"$PKG_ARCH" \
+  makepkg pacman-package-manager
+
+# Install pacman-package-manager and its dependencies (from Ubuntu 22.10).
+# sudo curl -Os http://de.archive.ubuntu.com/ubuntu/pool/universe/p/pacman-package-manager/pacman-package-manager_6.0.1-4_amd64.deb
+# sudo curl -Os http://de.archive.ubuntu.com/ubuntu/pool/universe/p/pacman-package-manager/libalpm13_6.0.1-4_amd64.deb
+# sudo curl -Os http://de.archive.ubuntu.com/ubuntu/pool/universe/p/pacman-package-manager/makepkg_6.0.1-4_amd64.deb
+# sudo dpkg -i *.deb || true
+# sudo apt-get install --fix-broken --assume-yes --no-install-recommends --quiet
+
+# Configure pacman.
+
+# Enable mirrorlist.
+sudo sed -Ei 's/^#.*(Include.*mirrorlist)/\1/' /etc/pacman.conf
+(
+# Add msys key for pacman.
+cd /usr/share/keyrings
+sudo curl -Os https://raw.githubusercontent.com/msys2/MSYS2-keyring/master/msys2.gpg
+sudo curl -Os https://raw.githubusercontent.com/msys2/MSYS2-keyring/master/msys2-revoked
+sudo curl -Os https://raw.githubusercontent.com/msys2/MSYS2-keyring/master/msys2-trusted
+)
+(
+# Add active environments for pacman.
+# See https://www.msys2.org/docs/repos-mirrors/.
+sudo mkdir -p /etc/pacman.d
+cd /etc/pacman.d
+cat <<eod | sudo tee mirrorlist >/dev/null
+[mingw64]
+Include = /etc/pacman.d/mirrorlist.mingw
+eod
+sudo curl -O https://raw.githubusercontent.com/msys2/MSYS2-packages/master/pacman-mirrors/mirrorlist.mingw
+# sudo curl -O https://raw.githubusercontent.com/msys2/MSYS2-packages/master/pacman-mirrors/mirrorlist.msys
+)
+
+sudo pacman-key --init
+sudo pacman-key --populate msys2
+sudo pacman -Syu --noconfirm
+
+# Install required pacman packages.
+sudo pacman -S --noconfirm \
+ mingw-w64-x86_64-curl-winssl \
+ mingw-w64-x86_64-giflib \
+ mingw-w64-x86_64-icu \
+ mingw-w64-x86_64-leptonica \
+ mingw-w64-x86_64-libarchive \
+ mingw-w64-x86_64-libidn2 \
+ mingw-w64-x86_64-openjpeg2 \
+ mingw-w64-x86_64-openssl \
+ mingw-w64-x86_64-pango \
+ mingw-w64-x86_64-libpng \
+ mingw-w64-x86_64-libtiff \
+ mingw-w64-x86_64-libwebp
+
+sudo ln -sf "$PWD/.github/workflows/pkg-config-crosswrapper" "/usr/bin/$HOST-pkg-config"
+
+git config --global user.email "sw@weilnetz.de"
+git config --global user.name "Stefan Weil"
+git tag -a "v$TAG" -m "Tesseract $TAG"
+
+# Run autogen.
+./autogen.sh
+
+# Build Tesseract installer.
+mkdir -p "$BUILDDIR" && cd "$BUILDDIR"
+
+# Run configure.
+PKG_CONFIG_PATH=$MINGW/lib/pkgconfig
+export PKG_CONFIG_PATH
+# Disable OpenMP (see https://github.com/tesseract-ocr/tesseract/issues/1662).
+../../../configure --disable-openmp --host="$HOST" --prefix="/usr/$HOST" \
+  CXX="$HOST-g++-posix" \
+  CXXFLAGS="-fno-math-errno -Wall -Wextra -Wpedantic -g -O2 -isystem $MINGW/include" \
+  LDFLAGS="-L$MINGW/lib"
+
+make all training
+MINGW_INSTALL=${PWD}${MINGW}
+make install-jars install training-install html prefix="$MINGW_INSTALL"
+mkdir -p dll
+ln -sv $($ROOTDIR/nsis/find_deps.py $MINGW_INSTALL/bin/*.exe $MINGW_INSTALL/bin/*.dll | sort | uniq) dll/
+make winsetup prefix="$MINGW_INSTALL"
+
+# Copy result for upload.
+mkdir -p "$DISTDIR" && cp nsis/tesseract-ocr-w*-setup-*.exe "$DISTDIR"

nsis/tesseract.nsi

@@ -22,10 +22,18 @@
 SetCompressor /FINAL /SOLID lzma
 SetCompressorDictSize 32
 
+Unicode true
+
 ; Settings which normally should be passed as command line arguments.
 ;define CROSSBUILD
 ;define SHARED
 ;define W64
+!ifndef COMMENTS
+!define COMMENTS "GitHub CI build"
+!endif
+!ifndef COMPANYNAME
+!define COMPANYNAME "Open Source Community"
+!endif
 !ifndef SRCDIR
 !define SRCDIR .
 !endif
@@ -40,7 +48,7 @@ SetCompressorDictSize 32
 !define PRODUCT_WEB_SITE "https://github.com/tesseract-ocr/tesseract"
 !endif
 !define GITHUB_RAW_FILE_URL \
-  "https://raw.githubusercontent.com/tesseract-ocr/tessdata_fast/master"
+  "https://raw.githubusercontent.com/tesseract-ocr/tessdata_fast/main"
 
 !ifdef CROSSBUILD
 !addincludedir ${SRCDIR}\nsis\include
@@ -56,13 +64,17 @@ SetCompressorDictSize 32
 !endif
 
 # Name of program and file
-!ifdef VERSION
+!define OUTFILE "${SETUP}-${VERSION}.exe"
-OutFile ${SETUP}-${VERSION}.exe
+OutFile ${OUTFILE}
-!else
+
-OutFile ${SETUP}.exe
+!ifdef SIGNCODE
+!finalize "${SIGNCODE} %1"
+!uninstfinalize "${SIGNCODE} %1"
 !endif
 
-!define PREFIX "../usr/${ARCH}-w64-mingw32"
+!ifndef PREFIX
+!define PREFIX "../mingw64"
+!endif
 !define TRAININGDIR "${PREFIX}/bin"
 
 # General Definitions
@@ -72,6 +84,21 @@ Caption "${PRODUCT_NAME} ${VERSION}"
 BrandingText /TRIMCENTER "(c) 2010-2019 ${PRODUCT_NAME}"
 !endif
 
+; File properties.
+!define /date DATEVERSION "%Y%m%d%H%M%S"
+VIProductVersion "${VERSION}"
+VIAddVersionKey "ProductName" "${PRODUCT_NAME}"
+VIAddVersionKey "Comments" "${COMMENTS}"
+VIAddVersionKey "CompanyName" "${COMPANYNAME}"
+VIAddVersionKey "FileDescription" "Tesseract OCR"
+!define /date DATETIME "%Y-%m-%d-%H-%M-%S"
+VIAddVersionKey "FileVersion" "${DATETIME}"
+VIAddVersionKey "InternalName" "Tesseract"
+VIAddVersionKey "LegalCopyright" "Apache-2.0"
+#VIAddVersionKey "LegalTrademarks" ""
+VIAddVersionKey "OriginalFilename" "${OUTFILE}"
+VIAddVersionKey "ProductVersion" "${VERSION}"
+
 !define REGKEY "SOFTWARE\${PRODUCT_NAME}"
 ; HKLM (all users) vs HKCU (current user) defines
 !define env_hklm 'HKLM "SYSTEM\CurrentControlSet\Control\Session Manager\Environment"'
@@ -100,10 +127,12 @@ BrandingText /TRIMCENTER "(c) 2010-2019 ${PRODUCT_NAME}"
 !define MUI_FINISHPAGE_LINK "View Tesseract on GitHub"
 !define MUI_FINISHPAGE_LINK_LOCATION "https://github.com/tesseract-ocr/tesseract"
 !define MUI_FINISHPAGE_NOAUTOCLOSE
+!ifdef SHOW_README
 ; Showing the README does not work.
-;!define MUI_FINISHPAGE_SHOWREADME "$INSTDIR\doc\README.md"
+!define MUI_FINISHPAGE_SHOWREADME "$INSTDIR\doc\README.md"
-;!define MUI_FINISHPAGE_SHOWREADME_FUNCTION ShowReadme
+!define MUI_FINISHPAGE_SHOWREADME_FUNCTION ShowReadme
-;!define MUI_FINISHPAGE_SHOWREADME_TEXT "Show README"
+!define MUI_FINISHPAGE_SHOWREADME_TEXT "Show README"
+!endif
 !define MUI_STARTMENUPAGE_REGISTRY_ROOT HKLM
 !define MUI_STARTMENUPAGE_REGISTRY_KEY ${REGKEY}
 !define MUI_STARTMENUPAGE_REGISTRY_VALUENAME StartMenuGroup
@@ -128,9 +157,7 @@ Var OLD_KEY
 !insertmacro MUI_PAGE_WELCOME
 !insertmacro MUI_PAGE_LICENSE "${SRCDIR}\LICENSE"
 !insertmacro MULTIUSER_PAGE_INSTALLMODE
-!ifdef VERSION
   Page custom PageReinstall PageLeaveReinstall
-!endif
 !insertmacro MUI_PAGE_COMPONENTS
 !insertmacro MUI_PAGE_DIRECTORY
 !insertmacro MUI_PAGE_STARTMENU Application $StartMenuGroup
@@ -178,8 +205,10 @@ Section -Main SEC0000
   File ${PREFIX}/bin/tesseract.exe
   File ${PREFIX}/bin/libtesseract-*.dll
 !ifdef CROSSBUILD
-  File ${SRCDIR}\dll\${ARCH}-w64-mingw32\*.dll
+  File ../dll/*.dll
 !endif
+  File winpath.exe
+  File ../doc/*.html
   CreateDirectory "$INSTDIR\tessdata"
   SetOutPath "$INSTDIR\tessdata"
   File ${PREFIX}/share/tessdata/pdf.ttf
@@ -245,7 +274,8 @@ SectionEnd
 Section "Shortcuts creation" SecCS
   SetOutPath $INSTDIR
   CreateDirectory "$SMPROGRAMS\${PRODUCT_NAME}"
-  CreateShortCut "$SMPROGRAMS\${PRODUCT_NAME}\Console.lnk" $WINDIR\system32\CMD.EXE
+  CreateShortCut "$SMPROGRAMS\${PRODUCT_NAME}\Console.lnk" "$INSTDIR\winpath.exe" "cmd"
+  CreateShortCut "$SMPROGRAMS\${PRODUCT_NAME}\Dokumentation.lnk" "$INSTDIR\tesseract.1.html"
   CreateShortCut "$SMPROGRAMS\${PRODUCT_NAME}\Homepage.lnk" "${PRODUCT_WEB_SITE}"
   CreateShortCut "$SMPROGRAMS\${PRODUCT_NAME}\ReadMe.lnk" "${PRODUCT_WEB_SITE}/wiki/ReadMe"
   CreateShortCut "$SMPROGRAMS\${PRODUCT_NAME}\FAQ.lnk" "${PRODUCT_WEB_SITE}/wiki/FAQ"
@@ -258,14 +288,12 @@ SectionEnd
 SectionGroup "Language data" SecGrp_LD
     Section "English" SecLang_eng
     SectionIn RO
-      SetOutPath "$INSTDIR\tessdata"
+      !insertmacro Download_Lang_Data eng
-      File ${SRCDIR}\tessdata\eng.*
     SectionEnd
 
     Section "Orientation and script detection" SecLang_osd
     SectionIn 1
-      SetOutPath "$INSTDIR\tessdata"
+      !insertmacro Download_Lang_Data osd
-      File ${SRCDIR}\tessdata\osd.*
     SectionEnd
 SectionGroupEnd
 
@@ -466,7 +494,7 @@ SectionGroup "Additional language data (download)" SecGrp_ALD
   SectionEnd
 
   ; The language names are documented here:
-  ; https://github.com/tesseract-ocr/tesseract/blob/master/doc/tesseract.1.asc#languages
+  ; https://github.com/tesseract-ocr/tesseract/blob/main/doc/tesseract.1.asc#languages
 
   Section /o "Afrikaans" SecLang_afr
     AddSize 2530
@@ -588,6 +616,11 @@ SectionGroup "Additional language data (download)" SecGrp_ALD
     !insertmacro Download_Lang_Data deu
   SectionEnd
 
+ Section /o "German Fraktur" SecLang_deu_latf
+    AddSize 6130
+    !insertmacro Download_Lang_Data deu_latf
+  SectionEnd
+
   Section /o "Divehi" SecLang_div
     AddSize 1690
     !insertmacro Download_Lang_Data div
@@ -648,11 +681,6 @@ SectionGroup "Additional language data (download)" SecGrp_ALD
     !insertmacro Download_Lang_Data fra
   SectionEnd
 
- Section /o "German Fraktur" SecLang_frk
-    AddSize 6130
-    !insertmacro Download_Lang_Data frk
-  SectionEnd
-
  Section /o "French - Middle (ca. 1400-1600)" SecLang_frm
     AddSize 1930
     !insertmacro Download_Lang_Data frm
@@ -1130,7 +1158,7 @@ Section -un.Main UNSEC0000
   RMDir  "$SMPROGRAMS\${PRODUCT_NAME}"
   DetailPrint "Removing registry info"
   DeleteRegKey HKLM "Software\Tesseract-OCR"
-  SendMessage ${HWND_BROADCAST} ${WM_WININICHANGE} 0 "STR:Environment" /TIMEOUT=5000
+  SendMessage ${HWND_BROADCAST} ${WM_WININICHANGE} 0 "STR:Environment" /TIMEOUT=1000
 
   # remove the Add/Remove information
   DeleteRegKey HKLM "${UNINST_KEY}"
@@ -1138,7 +1166,15 @@ Section -un.Main UNSEC0000
   DeleteRegValue HKLM "${REGKEY}" Path
   DeleteRegKey /IfEmpty HKLM "${REGKEY}\Components"
   DeleteRegKey /IfEmpty HKLM "${REGKEY}"
-  RMDir /r "$INSTDIR"
+  Delete "$INSTDIR\*.dll"
+  Delete "$INSTDIR\*.exe"
+  Delete "$INSTDIR\*.html"
+  Delete "$INSTDIR\doc\AUTHORS"
+  Delete "$INSTDIR\doc\LICENSE"
+  Delete "$INSTDIR\doc\README.md"
+  RMDir "$INSTDIR\doc"
+  RMDir /r "$INSTDIR\tessdata"
+  RMDir "$INSTDIR"
 SectionEnd
 
 Function PageReinstall
@@ -1389,10 +1425,12 @@ Function .onInstFailed
   MessageBox MB_OK "Installation failed."
 FunctionEnd
 
+!ifdef SHOW_README
 Function ShowReadme
   Exec '"wordpad" "doc\README.md"'
   ;BringToFront
 FunctionEnd
+!endif
 
 ; Prevent running multiple instances of the installer
 Function PreventMultipleInstances

nsis/winpath.cpp

@@ -0,0 +1,39 @@
+// Copyright (C) 2024 Stefan Weil
+//
+// SPDX-License-Identifier: Apache-2.0
+//
+// winpath - run a Windows program with extended PATH
+//
+// Usage:
+//
+//     winpath [CMD [ARGUMENT ...]]
+//
+// Example:
+//
+//     winpath cmd
+//
+// This will start a Windows command line with PATH extended by
+// the location of the winpath executable.
+
+#include <process.h>    // _spawnvp
+#include <stdlib.h>     // _putenv_s
+#include <string.h>     // strcpy, strcat
+
+static char path[4096];
+
+int main(int argc, char *argv[]) {
+  if (argc > 1) {
+    char *dir = argv[0];
+    char *last = strrchr(dir, '\\');
+    if (last != nullptr) {
+      *last = '\0';
+    }
+    strcpy(path, dir);
+    strcat(path, ";");
+    strcat(path, getenv("PATH"));
+    _putenv_s("PATH", path);
+    _spawnvp(_P_WAIT, argv[1], argv + 1);
+    //~ _spawnvp(_P_OVERLAY, argv[1], argv + 1);
+  }
+  return 0;
+}