ScriptCollection 4.2.72__py3-none-any.whl → 4.2.74__py3-none-any.whl

ScriptCollection/Executables.py

@@ -706,14 +706,13 @@ def OCRAnalysisOfFolder() -> int:
     parser.add_argument('-e', '--extensions', required=False, default="pdf,docx,jpg,png,xlsx")
     parser.add_argument('-l', '--languages', required=False, default="eng")
     parser.add_argument('-f', '--folder', required=False, default=None)
-    parser.add_argument('-d', '--datafolder', required=False, default=None)
     args = parser.parse_args()
     sc = ScriptCollectionCore()
     if args.folder is None:
         args.folder = os.getcwd()
     languages=args.languages.split(",")
     extensions=args.extensions.split(",")
-    sc.ocr_analysis_of_folder(args.folder, args.serviceaddress, extensions, languages,args.datafolder,args.folder,[])
+    sc.ocr_analysis_of_folder(args.folder, args.serviceaddress, extensions, languages,args.folder,[])
     return 0
 
 
@@ -722,11 +721,10 @@ def OCRAnalysisOfFile() -> int:
     parser.add_argument('-s', '--serviceaddress', required=False, default=None)
     parser.add_argument('-l', '--languages', required=False, default="eng")
     parser.add_argument('-f', '--file', required=True)
-    parser.add_argument('-d', '--datafolder', required=False, default=None)
     args = parser.parse_args()
     sc = ScriptCollectionCore()
     languages=args.languages.split(",")
-    sc.ocr_analysis_of_file(args.file, args.serviceaddress, languages,args.datafolder,".")
+    sc.ocr_analysis_of_file(args.file, args.serviceaddress, languages,".")
     return 0
 
 
@@ -736,14 +734,13 @@ def OCRAnalysisOfRepository() -> int:
     parser.add_argument('-e', '--extensions', required=False, default="pdf,docx,jpg,png,xlsx")
     parser.add_argument('-l', '--languages', required=False, default="eng")
     parser.add_argument('-f', '--folder', required=False, default=None)
-    parser.add_argument('-d', '--datafolder', required=False, default=None)
     args = parser.parse_args()
     sc = ScriptCollectionCore()
     if args.folder is None:
         args.folder = os.getcwd()
     languages=args.languages.split(",")
     extensions=args.extensions.split(",")
-    sc.ocr_analysis_of_repository(args.folder, args.serviceaddress, extensions, languages,args.datafolder)
+    sc.ocr_analysis_of_repository(args.folder, args.serviceaddress, extensions, languages)
     return 0
 
 

ScriptCollection/ScriptCollectionCore.py

@@ -37,7 +37,7 @@ from .ProgramRunnerBase import ProgramRunnerBase
 from .ProgramRunnerPopen import ProgramRunnerPopen
 from .SCLog import SCLog, LogLevel
 
-version = "4.2.72"
+version = "4.2.74"
 __version__ = version
 
 class VSCodeWorkspaceShellTask:
@@ -2674,7 +2674,14 @@ TXDX
         self.run_program_argsasarray("pip", arguments, folder,print_live_output=self.log.loglevel==LogLevel.Debug)
 
     @GeneralUtilities.check_arguments
-    def ocr_analysis_of_folder(self, folder: str, serviceaddress: str, extensions: list[str], languages: list[str], datafolder: str,base_folder_for_entry: str,ignore_pattern:list[str] ) -> list[str]:  # Returns a list of changed files due to ocr-analysis.
+    def ocr_analysis_of_folder_using_local_docker_image(self, folder: str, extensions: list[str], languages: list[str],base_folder_for_entry: str,ignore_pattern:list[str] ) -> list[str]:  # Returns a list of changed files due to ocr-analysis.
+        #TODO start docker server
+        serviceaddress:str=None#TODO
+        self.ocr_analysis_of_folder(folder, serviceaddress, extensions, languages, base_folder_for_entry,ignore_pattern)
+        #TODO stop docker server
+
+    @GeneralUtilities.check_arguments
+    def ocr_analysis_of_folder(self, folder: str, serviceaddress: str, extensions: list[str], languages: list[str],base_folder_for_entry: str,ignore_pattern:list[str] ) -> list[str]:  # Returns a list of changed files due to ocr-analysis.
         supported_extensions = ['png', 'jpg', 'jpeg', 'tiff', 'bmp', 'gif', 'pdf', 'docx', 'doc', 'xlsx', 'xls', 'pptx', 'ppt']
         changes_files: list[str] = []
         if base_folder_for_entry is None:
@@ -2687,13 +2694,13 @@ TXDX
             file_lower = file.lower()
             for extension in extensions:
                 if file_lower.endswith("."+extension):
-                    if self.ocr_analysis_of_file(file, serviceaddress, languages,datafolder,base_folder_for_entry):
+                    if self.ocr_analysis_of_file(file, serviceaddress, languages,base_folder_for_entry):
                         changes_files.append(file)
                     break
         for subfolder in GeneralUtilities.get_direct_folders_of_folder(folder):
             if GeneralUtilities.is_ignored_by_glob_pattern(os.path.dirname(subfolder),subfolder,ignore_pattern):
                 continue
-            for file in self.ocr_analysis_of_folder(subfolder, serviceaddress, extensions, languages,datafolder,base_folder_for_entry+"/"+os.path.basename(subfolder), ignore_pattern):
+            for file in self.ocr_analysis_of_folder(subfolder, serviceaddress, extensions, languages,base_folder_for_entry+"/"+os.path.basename(subfolder), ignore_pattern):
                 changes_files.append(file)
         return changes_files
 
@@ -2707,8 +2714,7 @@ TXDX
         return False
     
     @GeneralUtilities.check_arguments
-    def ocr_analysis_of_file(self, file: str, serviceaddress: str, languages: list[str], datafolder: str,readable_folder_entry:str ) -> bool:  # Returns true if the ocr-file was generated or updated. Returns false if the existing ocr-file was not changed.
-        GeneralUtilities.write_message_to_stdout(f"Starting OCR analysis of file {file}...")
+    def ocr_analysis_of_file(self, file: str, serviceaddress: str, languages: list[str], readable_folder_entry:str ) -> bool:  # Returns true if the ocr-file was generated or updated. Returns false if the existing ocr-file was not changed.
         supported_extensions = ['png', 'jpg', 'jpeg', 'tiff', 'bmp', 'webp', 'gif', 'pdf', 'rtf', 'docx', 'doc', 'odt', 'xlsx', 'xls', 'ods', 'pptx', 'ppt', 'odp']
         if not self.__it_supported_extension(file, supported_extensions):
             raise ValueError(f"File '{file}' is not supported due to unsupported extension. Supported extensions are: {', '.join(supported_extensions)}")
@@ -2722,7 +2728,8 @@ TXDX
                     return False
         except:
             pass
-        ocr_content = self.get_ocr_content_of_file(file, serviceaddress, languages,datafolder)
+        GeneralUtilities.write_message_to_stdout(f"Starting OCR-analysis of file \"{file}\"...")
+        ocr_content = self.get_ocr_content_of_file(file, serviceaddress, languages)
         GeneralUtilities.ensure_file_exists(target_file)
         if readable_folder_entry is None:
             readable_folder_entry="."
@@ -2734,30 +2741,52 @@ OCR-content:
         return True
 
     @GeneralUtilities.check_arguments
-    def get_ocr_content_of_file(self, file: str, serviceaddress: str, languages: list[str], datafolder: str) -> str:  # serviceaddress = None means local executable
+    def get_ocr_content_of_file(self, file: str, serviceaddress: str, languages: list[str]) -> str:
         result: str = None
-        extension = Path(file).suffix
+        extension = Path(file).suffix[1:]
+        mime_types = {
+            "pdf": "application/pdf",
+            "png": "image/png",
+            "jpg": "image/jpeg",
+            "jpeg": "image/jpeg",
+            "txt": "text/plain",
+            "json": "application/json",
+            "doc": "application/msword",
+            "docx": "application/vnd.openxmlformats-officedocument.wordprocessingml.document",
+            "xls": "application/vnd.ms-excel",
+            "xlsx": "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
+        }
         if serviceaddress is None:
-            arguments= ["OCRAnalysis", "--File", file, "--Languages", "+".join(languages)] 
-            if datafolder is not None:
-                arguments.append("--OCRDataFolder")
-                arguments.append(datafolder)
-            program_result = self.run_program_argsasarray("simpleocrcli",arguments)
-            result = program_result[1]
-        else:
-            languages_for_url = '%2B'.join(languages)
-            package_url: str = f"https://{serviceaddress}/GetOCRContent?languages={languages_for_url}&fileType={extension}"
-            headers = {'Cache-Control': 'no-cache'}
-            r = requests.put(package_url, timeout=5, headers=headers, data=GeneralUtilities.read_binary_from_file(file))
+            server_url_file:str= GeneralUtilities.normalize_path(f"{str(Path.home())}/.ScriptCollection/OCR/ServiceURL.txt")
+            if os.path.isfile(server_url_file):
+                for line in GeneralUtilities.read_nonempty_lines_from_file(server_url_file):
+                    if not line.startswith("#"):
+                        serviceaddress = line.strip()
+                        break
+        GeneralUtilities.assert_not_null(serviceaddress, "ocr-service-address must not be null.")
+        mime_type = mime_types.get(extension.lower(), "application/octet-stream")
+        service_url: str = f"{serviceaddress}/API/v1/SimpleOCR/GetOCRContent?mimeType={mime_type}"
+        for language in languages:
+            service_url = service_url + f"&languages={language}"
+        headers = {'Cache-Control': 'no-cache'}
+        with open(file, "rb") as f:
+            files_to_analyse = {
+                "fileContent": (os.path.basename(file), f, mime_type)
+            }
+            r = requests.put(service_url, timeout=3600, headers=headers, files=files_to_analyse,verify=True)
             if r.status_code != 200:
-                raise ValueError(f"Checking for latest tor package resulted in HTTP-response-code {r.status_code}.")
+                if r.status_code == 400:
+                    return f"Could not calculate ocr-content for file \"{file}\". File may be broken."
+                else:
+                    raise ValueError(f"Retrieving ocr-content for file \"{file}\" resulted in HTTP-response-code {r.status_code}.")
+
             result = GeneralUtilities.bytes_to_string(r.content)
         return result
 
     @GeneralUtilities.check_arguments
-    def ocr_analysis_of_repository(self, folder: str, serviceaddress: str, extensions: list[str], languages: list[str], datafolder: str) -> None:
+    def ocr_analysis_of_repository(self, folder: str, serviceaddress: str, extensions: list[str], languages: list[str]) -> None:
         self.assert_is_git_repository(folder)
-        self.ocr_analysis_of_folder(folder, serviceaddress, extensions, languages, datafolder,".",[".git"])
+        self.ocr_analysis_of_folder(folder, serviceaddress, extensions, languages,".",[".git"])
 
     @GeneralUtilities.check_arguments
     def update_timestamp_in_file(self, target_file: str) -> None:
@@ -3299,7 +3328,7 @@ OCR-content:
         if os.path.isabs(target_file):
             target_file=GeneralUtilities.resolve_relative_path(target_file,repository_folder)
         target_file=GeneralUtilities.normalize_path(target_file)
-        files=self.get_all_files_in_git_repository(repository_folder,ignore_ignored_files,include_submodules)
+        files=[path.replace("\\","/") for path in self.get_all_files_in_git_repository(repository_folder,ignore_ignored_files,include_submodules)]
         GeneralUtilities.ensure_file_exists(target_file)
         GeneralUtilities.write_lines_to_file(target_file, files)
 
@@ -3322,7 +3351,7 @@ OCR-content:
         GeneralUtilities.ensure_file_exists(target_file)
         GeneralUtilities.write_lines_to_file(target_file, commits)
 
-
     @GeneralUtilities.check_arguments
     def is_runnning_in_container(self) ->bool:
+        """this function is based on a convention and does not do a real check."""
         return os.environ.get("ISRUNNINGINCONTAINER") == "true"

{scriptcollection-4.2.72.dist-info → scriptcollection-4.2.74.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ScriptCollection
-Version: 4.2.72
+Version: 4.2.74
 Summary: The ScriptCollection is the place for reusable scripts.
 Home-page: https://github.com/anionDev/ScriptCollection
 Author: Marius Göcke

{scriptcollection-4.2.72.dist-info → scriptcollection-4.2.74.dist-info}/RECORD

@@ -1,6 +1,6 @@
 ScriptCollection/AnionBuildPlatform.py,sha256=K-PHarX802A0PU8uRu0GNcEZiXujFoXHACe-X9YJsAQ,11711
 ScriptCollection/CertificateUpdater.py,sha256=Pa6eyjQSx7IIvj4PQVMI0IwMs01KQrNSB7Qa-7lRfBs,9375
-ScriptCollection/Executables.py,sha256=qpo0g5peWdlK5uLIUCyLDB9c3JCk0ETbtmOJXZwuHh4,44510
+ScriptCollection/Executables.py,sha256=SsA3zeDL8QEsh7GxjDjatv5P4eFeDBPqB0F8pNXoYzA,44234
 ScriptCollection/GeneralUtilities.py,sha256=3Fgp0fAXF-rfcohy6k1RsRcMXEVRF15fHl8QJnViKIg,65497
 ScriptCollection/HTTPMaintenanceOverheadHelper.py,sha256=TToNtyO1XzsMbBsTBf3o0xgOK0v4Jf03qw2Z0xb2nCk,2007
 ScriptCollection/ProcessesRunner.py,sha256=o5raxIt3lknNPoPrjNzJ2bprRPJ3SnL0rrR7crraD7E,1523
@@ -9,7 +9,7 @@ ScriptCollection/ProgramRunnerMock.py,sha256=uTu-aFle1W_oKjeQEmuPsFPQpvo0kRf2FrR
 ScriptCollection/ProgramRunnerPopen.py,sha256=BPY7-ZMIlqT7JOKz8qlB5c0laF2Js-ijzqk09GxZC48,3821
 ScriptCollection/ProgramRunnerSudo.py,sha256=_khC3xuTdrPoLluBJZWfldltmmuKltABJPcbjZSFW-4,4835
 ScriptCollection/SCLog.py,sha256=8TRy1LeYMsPOIuWUcnUNNbO5pd-cNBS-3cn-kdzP8FU,4768
-ScriptCollection/ScriptCollectionCore.py,sha256=qU8XbRln8Ozaqe_Z4XnU7--XQfmlB0z4RoSupVB01_s,180421
+ScriptCollection/ScriptCollectionCore.py,sha256=5RXTdMUUCb1XvzPEj8JTqQosPQ4YgRufsbXpY6ZEHa8,181964
 ScriptCollection/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 ScriptCollection/OCIImages/AbstractImageHandler.py,sha256=83qDMILwxhH9DbC0sb358Vu8PXEysmJJyap_6gECZqs,1627
 ScriptCollection/OCIImages/OCIImageManager.py,sha256=aBogkSXNDyi8NO11N-s03nuFJEv7PyJ-wjHuYYeZfvs,6662
@@ -47,8 +47,8 @@ ScriptCollection/TFCPS/NodeJS/TFCPS_CodeUnitSpecific_NodeJS.py,sha256=GQLE6FeR-X
 ScriptCollection/TFCPS/NodeJS/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 ScriptCollection/TFCPS/Python/TFCPS_CodeUnitSpecific_Python.py,sha256=9XK7XnbeOnq_4siVoWovogStoKFiZLhGh3C_f2YaznI,13621
 ScriptCollection/TFCPS/Python/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-scriptcollection-4.2.72.dist-info/METADATA,sha256=eVq5U2dxHmaGEZnjzCZg-9DYI1S6s5sHx3Zbofx7Xfk,7691
-scriptcollection-4.2.72.dist-info/WHEEL,sha256=aeYiig01lYGDzBgS8HxWXOg3uV61G9ijOsup-k9o1sk,91
-scriptcollection-4.2.72.dist-info/entry_points.txt,sha256=27XwAJEcaMEc1be0Ec1vKHCbiU4Ziu8jKL-SqsrYOIQ,4680
-scriptcollection-4.2.72.dist-info/top_level.txt,sha256=hY2hOVH0V0Ce51WB76zKkIWTUNwMUdHo4XDkR2vYVwg,17
-scriptcollection-4.2.72.dist-info/RECORD,,
+scriptcollection-4.2.74.dist-info/METADATA,sha256=X82BuSGCRx0eK2pnFKs2wEhh7_xViAovRmeiZpqitfc,7691
+scriptcollection-4.2.74.dist-info/WHEEL,sha256=aeYiig01lYGDzBgS8HxWXOg3uV61G9ijOsup-k9o1sk,91
+scriptcollection-4.2.74.dist-info/entry_points.txt,sha256=27XwAJEcaMEc1be0Ec1vKHCbiU4Ziu8jKL-SqsrYOIQ,4680
+scriptcollection-4.2.74.dist-info/top_level.txt,sha256=hY2hOVH0V0Ce51WB76zKkIWTUNwMUdHo4XDkR2vYVwg,17
+scriptcollection-4.2.74.dist-info/RECORD,,