add CLI test and reformat

Author: wuan

pyproject.toml

@@ -7,7 +7,7 @@ license = "Apache-2.0"
 readme = "README.rst"
 
 [tool.poetry.scripts]
-scan-pdf = 'scan_pdf.console:run'
+scan-pdf = 'scan_pdf.cli:run'
 
 [tool.poetry.dependencies]
 python = "^3.9"

scan_pdf/cli.py

@@ -0,0 +1,131 @@
+#!/usr/bin/env python
+
+from __future__ import print_function
+
+import argparse
+import logging
+import os
+import shutil
+import stat
+
+from scan_pdf import Scanner, Converter, Combiner, TempDir
+
+
+def parse_args() -> argparse.Namespace:
+    parser = argparse.ArgumentParser(
+        description="Produce PDF from Scanner with document-feeder"
+    )
+    parser.add_argument(
+        "output_file_name", type=str, nargs=1, help="name of the produced output file"
+    )
+    parser.add_argument(
+        "--device",
+        dest="device",
+        type=str,
+        default=None,
+        help="optional device locator",
+    )
+    parser.add_argument(
+        "--resolution",
+        dest="resolution",
+        type=int,
+        default=300,
+        help="scan resolution DPI (default: 300)",
+    )
+    parser.add_argument(
+        "--flatbed",
+        dest="flatbed",
+        action="store_true",
+        help="scan only one page from flatbed glass",
+    )
+    parser.add_argument(
+        "-d", dest="debug", action="store_true", help="enable debug output"
+    )
+    parser.add_argument(
+        "--duplex",
+        dest="duplex",
+        action="store_true",
+        help="scan both sides of document",
+    )
+    parser.add_argument(
+        "--color-mode",
+        dest="color_mode",
+        action="store",
+        choices=["bw", "gray", "color"],
+        default="gray",
+        help="default: gray",
+    )
+    parser.add_argument(
+        "--threshold",
+        dest="threshold",
+        default=None,
+        action="store",
+        help="disabled by default",
+    )
+    parser.add_argument("--color-depth", dest="color_depth", default=4)
+    parser.add_argument(
+        "--paper-format",
+        dest="paper_format",
+        action="store",
+        choices=Scanner.paper_formats.keys(),
+        default="A4",
+        help="default: A4",
+    )
+    parser.add_argument(
+        "--paper-left", dest="paper_left", default=0, help="override left offset"
+    )
+    parser.add_argument(
+        "--paper-top", dest="paper_top", default=0, help="override top offset"
+    )
+    parser.add_argument(
+        "--paper-height", dest="paper_height", default=0, help="override paper height"
+    )
+    parser.add_argument(
+        "--paper-width", dest="paper_width", default=0, help="override paper width"
+    )
+    options = parser.parse_args()
+    return options
+
+
+def run() -> None:
+    options = parse_args()
+
+    logging.basicConfig(level=logging.DEBUG if options.debug else logging.INFO)
+    logger = logging.getLogger(__name__)
+
+    origin_dir = os.getcwd()
+
+    with TempDir() as temp_dir:
+        os.chmod(temp_dir, stat.S_IRWXU)
+        os.chdir(temp_dir)
+
+        output_file_name = options.output_file_name[0]
+
+        scanner = Scanner(options)
+
+        output_basenames = scanner.get_page_file_basenames()
+
+        if len(output_basenames) > 0:
+            converter = Converter(options)
+            for page_number, output_basename in enumerate(output_basenames, start=1):
+                converter.convert(
+                    output_basename,
+                    scanner.page_file_suffix,
+                    options.duplex and page_number % 2 == 0,
+                )
+
+            combiner = Combiner(options)
+            combiner.combine(
+                [
+                    output_basename + converter.page_file_suffix
+                    for output_basename in output_basenames
+                ]
+            )
+
+            os.chdir(origin_dir)
+            shutil.move(
+                os.path.join(temp_dir, os.path.basename(output_file_name)),
+                output_file_name,
+            )
+        else:
+            logger.warn("no pages found")

scan_pdf/console.py

@@ -8,21 +8,23 @@
 
 
 class Converter:
-    page_file_suffix = '.pdf'
+    page_file_suffix = ".pdf"
 
     def __init__(self, options: argparse.Namespace):
         self.options = options
 
-    def convert(self, source_basename: str, source_suffix: str, flip: bool = False) -> int:
-        logger.info('convert %s', source_basename)
-        args = ['convert']
+    def convert(
+        self, source_basename: str, source_suffix: str, flip: bool = False
+    ) -> int:
+        logger.info("convert %s", source_basename)
+        args = ["convert"]
 
-        color_depth = self.options.color_depth if self.options.color_mode != 'bw' else 1
-        args += ['-depth', str(color_depth)]
-        args += ['-density', str(self.options.resolution)]
-        args += ['-compress', 'zip']
+        color_depth = self.options.color_depth if self.options.color_mode != "bw" else 1
+        args += ["-depth", str(color_depth)]
+        args += ["-density", str(self.options.resolution)]
+        args += ["-compress", "zip"]
         if flip:
-            args += ['-rotate', '180']
+            args += ["-rotate", "180"]
         target_file = source_basename + self.page_file_suffix
         args += [source_basename + source_suffix, target_file]
 

scan_pdf/convert.py

@@ -19,50 +19,71 @@ def __init__(self, left: int, width: int, top: int, height: int):
 
 class Scanner:
     paper_formats = {
-        'A3': PaperFormat(0, 297, 0, 420),
-        'A4': PaperFormat(0, 210, 0, 297),
-        'A5': PaperFormat(0, 149, 0, 218),
+        "A3": PaperFormat(0, 297, 0, 420),
+        "A4": PaperFormat(0, 210, 0, 297),
+        "A5": PaperFormat(0, 149, 0, 218),
     }
-    page_file_suffix = '.pnm'
+    page_file_suffix = ".pnm"
 
     def __init__(self, options: argparse.Namespace):
-
-        args = ['scanimage', '-b', '--format=pnm']
+        args = ["scanimage", "-b", "--format=pnm"]
 
         if options.device:
             args += ["-d", options.device]
 
-        color_mode = 'Gray'
-        if options.color_mode == 'bw':
-            color_mode = 'Lineart'
-        elif options.color_mode == 'color':
-            color_mode = 'Color'
-        args += ['--mode', color_mode]
+        color_mode = "Gray"
+        if options.color_mode == "bw":
+            color_mode = "Lineart"
+        elif options.color_mode == "color":
+            color_mode = "Color"
+        args += ["--mode", color_mode]
 
         if options.flatbed:
-            args += ['--source', 'Flatbed', '--batch-count', '1']
+            args += ["--source", "Flatbed", "--batch-count", "1"]
         else:
-            args += ['--source', 'Automatic Document Feeder']
-            args += ['--adf-mode', 'Duplex' if options.duplex else 'Simplex']
+            args += ["--source", "Automatic Document Feeder"]
+            args += ["--adf-mode", "Duplex" if options.duplex else "Simplex"]
 
-        args += ['--resolution', str(options.resolution)]
+        args += ["--resolution", str(options.resolution)]
 
         if options.threshold:
-            args += ['--halftoning', 'None', '--threshold', str(options.threshold)]
-
-        paper_format = self.paper_formats.get(options.paper_format, self.paper_formats['A4'])
-
-        args += ['-l', str(int(options.paper_left if options.paper_left else paper_format.left))]
-        args += ['-x', str(int(options.paper_width if options.paper_width else paper_format.width))]
-        args += ['-t', str(int(options.paper_top if options.paper_top else paper_format.top))]
-        args += ['-y', str(int(options.paper_height if options.paper_height else paper_format.height))]
+            args += ["--halftoning", "None", "--threshold", str(options.threshold)]
+
+        paper_format = self.paper_formats.get(
+            options.paper_format, self.paper_formats["A4"]
+        )
+
+        args += [
+            "-l",
+            str(int(options.paper_left if options.paper_left else paper_format.left)),
+        ]
+        args += [
+            "-x",
+            str(
+                int(options.paper_width if options.paper_width else paper_format.width)
+            ),
+        ]
+        args += [
+            "-t",
+            str(int(options.paper_top if options.paper_top else paper_format.top)),
+        ]
+        args += [
+            "-y",
+            str(
+                int(
+                    options.paper_height
+                    if options.paper_height
+                    else paper_format.height
+                )
+            ),
+        ]
 
         logger.debug("call %s", " ".join(args))
         retval = subprocess.call(args)
         logger.debug("call retured %d", retval)
 
     def get_page_file_basenames(self) -> List[str]:
-        output_files = glob.glob('out*' + self.page_file_suffix)
+        output_files = glob.glob("out*" + self.page_file_suffix)
         output_files = sorted(output_files, key=cmp_to_key(self.compare_output_names))
 
         return [os.path.splitext(output_file)[0] for output_file in output_files]

scan_pdf/scan.py

@@ -7,13 +7,18 @@
 
 class TempDir:
     def __init__(self) -> None:
-        self.temp_dir :Optional[str] = None
+        self.temp_dir: Optional[str] = None
 
     def __enter__(self) -> str:
-        self.temp_dir = tempfile.mkdtemp(prefix='scanpdf_')
+        self.temp_dir = tempfile.mkdtemp(prefix="scanpdf_")
         return self.temp_dir
 
     # noinspection PyUnusedLocal
-    def __exit__(self, type_value: type[BaseException], value: BaseException, traceback: TracebackType) -> None:
+    def __exit__(
+        self,
+        type_value: type[BaseException],
+        value: BaseException,
+        traceback: TracebackType,
+    ) -> None:
         if self.temp_dir is not None:
             shutil.rmtree(self.temp_dir, ignore_errors=True)

scan_pdf/util.py

@@ -1,10 +1,11 @@
 import os
+from argparse import Namespace
 
 
-class Options(object):
+class Options(Namespace):
     pass
 
 
 def touch(file_name: str):
-    with open(file_name, 'a'):
+    with open(file_name, "a"):
         os.utime(file_name, None)