Merge pull request #16 from simonflueckiger/simonflueckiger-update-to…

…-tesserocr-2.6.0

Update to tesserocr 2.6.0

appveyor.yml

@@ -1,34 +1,37 @@
-version: 2.5.2.{build}
+version: 2.6.0.{build}
 image: Visual Studio 2022
 
 environment:
-  # https://github.com/microsoft/vcpkg/blob/master/ports/leptonica/vcpkg.json
-  VCPKG_COMMIT_ID: ddea2bb # Leptonica 1.83.1
   PYTHON_VERSIONS: 36, 37, 38, 39, 310, 311
   # 2023-01-18 conda-build currently not supported for python 3.11 https://github.com/conda/conda/issues/11170
-  CONDA_SKIP_PYTHON_VERSIONS: 311
+  # they added support for conda >= v23.5.0 but 32bit version is stuck at v4.x
+  #CONDA_SKIP_PYTHON_VERSIONS_64: 311
+  CONDA_SKIP_PYTHON_VERSIONS_32: 311
 
   matrix:
     - job_name: Build Wheels - Tesseract 3.x
       job_group: Build Wheels
+      VCPKG_COMMIT_ID: 1e9facc # Leptonica 1.82.0 (https://github.com/microsoft/vcpkg/blob/master/ports/leptonica/vcpkg.json)
       TESSERACT_COMMIT_ID: 7e5f0d6 # Tesseract 3.05.02
-      
+
     - job_name: Build Conda Packages - Tesseract 3.x
       job_group: Build Conda Packages
       job_depends_on: Build Wheels - Tesseract 3.x
 
     - job_name: Build Wheels - Tesseract 4.x
       job_group: Build Wheels
+      VCPKG_COMMIT_ID: 1e9facc # Leptonica 1.82.0 (https://github.com/microsoft/vcpkg/blob/master/ports/leptonica/vcpkg.json)
       TESSERACT_COMMIT_ID: f38e7a7 # Tesseract 4.1.3
 
     - job_name: Build Conda Packages - Tesseract 4.x
       job_group: Build Conda Packages
       job_depends_on: Build Wheels - Tesseract 4.x
-    
+
     - job_name: Build Wheels - Tesseract 5.x
       job_group: Build Wheels
+      VCPKG_COMMIT_ID: ddea2bb # Leptonica 1.83.1 (https://github.com/microsoft/vcpkg/blob/master/ports/leptonica/vcpkg.json)
       TESSERACT_COMMIT_ID: 71af454 # Tesseract 5.3.1
-      
+
     - job_name: Build Conda Packages - Tesseract 5.x
       job_group: Build Conda Packages
       job_depends_on: Build Wheels - Tesseract 5.x
@@ -143,6 +146,9 @@ for:
           # ------ Build tesserocr ----------------------------------------
 
           cd tesserocr
+          
+          # dumpbin.exe required for finding dependencies in setup.py
+          $env:Path = "C:\Program Files (x86)\Microsoft Visual Studio 14.0\VC\bin;" + $env:Path
 
           # for each {64bit and 32bit} x {Python Versions}
           ForEach ($platform_x64 in @($true, $false)) {
@@ -214,7 +220,8 @@ for:
 
     build_script:
       - pwsh: |
-          $conda_skip_python_versions = $env:CONDA_SKIP_PYTHON_VERSIONS -split ", "
+          $conda_skip_python_versions_64 = $env:CONDA_SKIP_PYTHON_VERSIONS_64 -split ", "
+          $conda_skip_python_versions_32 = $env:CONDA_SKIP_PYTHON_VERSIONS_32 -split ", "
           
           # create necessary directories
           New-Item conda_build -ItemType Directory
@@ -248,15 +255,31 @@ for:
               # (which would require closing and re-opening of powershell session)
               # this is necessary for creating and activating new environments
               # any Miniconda version installed on AppVeyor could be used as long as we update it
-              if ($platform_x64) { 
-                  (& "C:\Miniconda37-x64\Scripts\conda.exe" "shell.powershell" "hook") | Out-String | Invoke-Expression
+              if ($platform_x64) {
+                  # download and install miniconda 64bit
+                  $filename = "Miniconda3-py39_23.3.1-0-Windows-x86_64.exe"
+                  Invoke-WebRequest https://repo.anaconda.com/miniconda/$filename -OutFile $filename
+                  Start-Process -FilePath $filename -ArgumentList "/InstallationType=JustMe", "/RegisterPython=0", "/S", "/D=$env:UserProfile\Miniconda3-x64" -Wait -NoNewWindow
+                  (& "$env:UserProfile\Miniconda3-x64\Scripts\conda.exe" "shell.powershell" "hook") | Out-String | Invoke-Expression
+                  
+                  # we want the latest conda to build for the most recent python version 
+                  # (since more recent Miniconda versions like Miniconda310 are not yet installed on this AppVeyor image)
+                  # conda version >= 23.5.0 is required (only attainable with miniconda >= 3.8) for conda-build on python 3.11,
+                  # using "conda update -y -n base -c defaults conda" doesn't work if installed version too far behind
+                  conda install -y -n base -c defaults conda=23.5.0
               } else { 
-                  (& "C:\Miniconda37\Scripts\conda.exe" "shell.powershell" "hook") | Out-String | Invoke-Expression
+                  # download and install miniconda 32bit
+                  $filename = "Miniconda3-py39_4.12.0-Windows-x86.exe"
+                  Invoke-WebRequest https://repo.anaconda.com/miniconda/$filename -OutFile $filename
+                  Start-Process -FilePath $filename -ArgumentList "/InstallationType=JustMe", "/RegisterPython=0", "/S", "/D=$env:UserProfile\Miniconda3" -Wait -NoNewWindow
+                  (& "$env:UserProfile\Miniconda3\Scripts\conda.exe" "shell.powershell" "hook") | Out-String | Invoke-Expression
+                  
+                  # we want the latest conda to build for the most recent python version
+                  # 32bit version is stuck at v4.x
+                  conda update -y -n base -c defaults conda
               }
               
-              # we want the latest conda to build for the most recent python version 
-              # (since more recent Miniconda versions like Miniconda310 are not yet installed on this AppVeyor image)
-              conda update -y -n base -c defaults conda
+              $conda_skip_python_versions = if ($platform_x64) { $conda_skip_python_versions_64 } else { $conda_skip_python_versions_32 }
 
               ForEach ($package in $packages) {
                   if ($conda_skip_python_versions.Contains($package.py_ver_str)) { Continue }

res/setup.py

@@ -4,6 +4,7 @@
 import shutil
 import codecs
 import re
+import subprocess
 from distutils.util import strtobool
 import setuptools  # needed for bdist_wheel
 from distutils.core import setup
@@ -79,6 +80,27 @@ def version_to_int(version):
     return int(version_str, 16)
 
 
+def find_dll_dependencies_recursively(dll_path, search_paths):
+    dumpbin = subprocess.run(['dumpbin.exe', '/dependents', dll_path], universal_newlines=True, stdout=subprocess.PIPE, stderr=subprocess.PIPE)
+    dependency_names = re.findall(r'^\s{4}(\S*\.dll)$', dumpbin.stdout, re.MULTILINE)
+
+    dependencies = []
+    for dependency_name in dependency_names:
+        for search_path in search_paths:
+            dependency_direct_full_path = os.path.join(search_path, dependency_name)
+            if os.path.isfile(dependency_direct_full_path):
+                dependencies.append(dependency_direct_full_path)
+                break
+
+    dependencies_recursive = []
+    for dependency_direct_full_path in dependencies:
+        dependencies_recursive.extend(find_dll_dependencies_recursively(dependency_direct_full_path, search_paths))
+
+    dependencies.extend(dependencies_recursive)
+
+    return list(set(dependencies))
+
+
 def find_libraries(library_stems, search_paths, extension):
     library_paths = []
 
@@ -163,23 +185,8 @@ def __init__(self, name, sources, *args, **kw):
     "leptonica"
 ]
 
-# indentations with respect to dependency
-runtime_libraries = [
-    "tesseract",
-        "leptonica",
-            "gif",
-            "jpeg",
-            "openjp2",
-            "png",
-            "zlib",
-            "tiff",
-                "lzma",
-            "webpmux",
-            "webp"
-]
-
-
-runtime_library_paths = find_libraries(runtime_libraries, [vcpkg_bin, tesseract_bin], "dll")
+runtime_library_paths = find_libraries(["tesseract"], [tesseract_bin], "dll")
+runtime_library_paths.extend(find_dll_dependencies_recursively(runtime_library_paths[0], [vcpkg_bin, tesseract_bin]))
 _LOGGER.info("runtime libraries found:\n\t{}".format("\n\t".join(runtime_library_paths)))
 
 build_dependency_paths = find_libraries(build_dependencies, [vcpkg_lib, tesseract_lib], "lib")