FFProbe Data Extraction Tool

nulrdcscripts/tools/ffprobeExtraction/main.py

@@ -0,0 +1,38 @@
+import os
+import progressbar
+import subprocess
+import time
+from nulrdcscripts.tools.ffprobeExtraction.params import args
+
+
+def main():
+    # Check if the input file exists
+    if not os.path.isfile(args.input_path):
+        print(f"Error: The input file '{args.input_path}' does not exist.")
+        return
+
+    if args.output_format == "JSON" or args.output_format == "json":
+        output_format = "json"
+    elif args.output_format == "XML" or args.output_format == "xml":
+        output_format = "xml"
+    else:
+        raise ValueError(
+            "This file format is not supported by this script for ffprobe data extraction."
+        )
+
+    output_path = os.path.splitext(args.input_path)[0] + {output_format}
+
+    input_path = os.path.abspath(args.input_path)
+    output_path = os.path.abspath(output_path)
+
+    command = f'ffprobe -f lavfi -i "movie={input_path},signalstats" -show_frames -show_streams -of {output_format} > {output_path}'
+    bar = progressbar.ProgressBar(max_value=progressbar.UnknownLength)
+    for i in range(100):
+        time.sleep(0.1)
+        bar.update(i)
+
+    subprocess.run(command)
+
+
+if __name__ == "__main__":
+    main()

nulrdcscripts/tools/ffprobeExtraction/params.py

@@ -0,0 +1,23 @@
+import argparse
+
+parser = argparse.ArgumentParser()
+
+parser.add_argument(
+    "--input",
+    "-i",
+    action="store",
+    dest="input_path",
+    type=str,
+    help="full path to input file",
+)
+
+parser.add_argument(
+    "-of",
+    "--outputformat",
+    action="store",
+    dest="outputformat",
+    default="json",
+    type=str,
+    help="format that you want the ffprobe report to be in. JSON is the default and the other accepted value is XML",
+)
+args = parser.parse_args()

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "nul-rdc-scripts"
-version = "0.5.2"
+version = "0.6.0"
 description = "Scripts for NUL RDC Digitization Team"
 authors = [
     "Northwestern University Libraries <[email protected]>",
@@ -29,6 +29,7 @@ md5 = 'nulrdcscripts.tools.md5generation.main:main'
 trim = 'nulrdcscripts.tools.videotrimmer.main:main'
 micro = 'nulrdcscripts.ingestMicro.ingest:main'
 meta = 'nulrdcscripts.tools.generatemetadataTEMP.main:main'
+ffprobedata = 'nulrdcscripts.tools.ffprobeExtract.main:main'
 [build-system]
 requires = ["poetry-core>=1.0.0"]
 build-backend = "poetry.core.masonry.api"