Merge pull request #89 from nulib/xparse-fix

Xparse fix
nulib · Nov 20, 2023 · f8c7058 · f8c7058
2 parents 5a35d7c + ac7e1b1
commit f8c7058
Show file tree

Hide file tree

Showing 4 changed files with 33 additions and 48 deletions.
diff --git a/nulrdcscripts/ingest/README.md b/nulrdcscripts/ingest/README.md
@@ -20,19 +20,19 @@ poetry run ingest -i INPUT_PATH -o OUTPUT_FILEPATH
 poetry run ingest -i INPUT_PATH -l INVENTORY_PATH
 ```
 
-### Skip ".md5" files
+### Parse x files by filename
 ```
-poetry run ingest -i INPUT_PATH -s .md5
+poetry run ingest -i INPUT_PATH -x parse
 ```
 
-### Custom description
+### Skip ".md5" files
 ```
-poetry run ingest -i INPUT_PATH -d "Date/Time" "Barcode"
+poetry run ingest -i INPUT_PATH -s .md5
 ```
 
-### Parse x files by filename
+### Custom description
 ```
-poetry run ingest -i INPUT_PATH -x parse
+poetry run ingest -i INPUT_PATH -d "date/time" "barcode"
 ```
 
 ### Example File Structure
@@ -64,15 +64,15 @@ input_folder
 |A|P|S|X|  
 |:-------:|:-------:|:-------:|:-------:|  
 |-a or _a|-p or _p|-s or _s|-x or _x|
-|-am or _am|-pm or _pm|spectrogram|-Asset or _Asset|
-|-am- or _am\_|-pm- or _pm\_|.json|Back.|
-|||.log|-Can or _Can|
-|||.pdf|-Ephemera or _Ephemera|
-|||.xml|Front.|
-|||.xml.gz|
-|||.framemd5|
-|||.qctool.mkv|
-|||dpx.txt|
+|-am or _am|-pm or _pm|spectrogram|-asset or _asset|
+|-am- or _am\_|-pm- or _pm\_|.json|-Asset or _Asset|
+|||.log|back or Back|
+|||.pdf|front or Front|
+|||.xml|-can or _can|
+|||.xml.gz|-Can or _Can|
+|||.framemd5|-ephemera or _ephemera|
+|||.qctool.mkv|-Ephemera or _Ephemera|
+|||dpx.txt|.jpg|
 
 Anything that can't be identified will be set to 'S'
 

diff --git a/nulrdcscripts/ingest/ingest.py b/nulrdcscripts/ingest/ingest.py
@@ -157,40 +157,47 @@ def main():
         elif "parse" in args.aux_parse:
             aux_dict = {
                 "front": {
-                    "identifiers": ["Front."],
+                    "identifiers": ["front", "Front"],
                     "type": "pattern",
                     "role": "X",
                     "label": "asset front",
                     "file_builder": "_auxiliary_",
                 },
                 "back": {
-                    "identifiers": ["Back."],
+                    "identifiers": ["back", "Back"],
                     "type": "pattern",
                     "role": "X",
                     "label": "asset back",
                     "file_builder": "_auxiliary_",
                 },
                 "asset": {
-                    "identifiers": ["_Asset", "-Asset"],
+                    "identifiers": ["_asset", "-asset", "_Asset", "-Asset"],
                     "type": "pattern",
                     "role": "X",
                     "label": "asset",
                     "file_builder": "_auxiliary_",
                 },
                 "can": {
-                    "identifiers": ["_Can", "-Can"],
+                    "identifiers": ["_can", "-can", "_Can", "-Can"],
                     "type": "pattern",
                     "role": "X",
                     "label": "can",
                     "file_builder": "_auxiliary_",
                 },
                 "ephemera": {
-                    "identifiers": ["_Ephemera", "-Ephemera"],
+                    "identifiers": ["_ephemera", "-ephemera", "_Ephemera", "-Ephemera"],
                     "type": "pattern",
                     "role": "X",
                     "label": "ephemera",
                     "file_builder": "_auxiliary_",
                 },
+                "auxiliary": {
+                    "identifiers": ["_x", "-x"],
+                    "type": "pattern",
+                    "role": "X",
+                    "label": "image",
+                    "file_builder": "_auxiliary_",
+                }
             }
         # add the aux_dict to the beginning of the role_dict
         # this will catch X files that also have a/p identifiers in the filename
@@ -439,9 +446,9 @@ def import_inventories(source_inventories):
                         # work type is assumed by the presence of format-specific column headers
                         if "Width (cm.)" in reader.fieldnames:
                             work_type = "IMAGE"
-                        elif "Speed IPS" in reader.fieldnames:
+                        elif "speed IPS" or "Speed IPS" in reader.fieldnames:
                             work_type = "AUDIO"
-                        elif "Region" or "Stock" in reader.fieldnames:
+                        elif "region" or "Region" or "stock" or "Stock" in reader.fieldnames:
                             work_type = "VIDEO"
                         else:
                             print(

diff --git a/nulrdcscripts/vproc/data/mediaconch_policies/AJA_NTSC_VHS-2SAS-MKV.xml b/nulrdcscripts/vproc/data/mediaconch_policies/AJA_NTSC_VHS-2SAS-MKV.xml
@@ -17,8 +17,6 @@
   <rule name="Video/CodecID is V_MS/VFW/FOURCC / FFV1" value="CodecID" tracktype="Video" occurrence="*" operator="=">V_MS/VFW/FOURCC / FFV1</rule>
   <rule name="Video/Width is 720" value="Width" tracktype="Video" occurrence="*" operator="=">720</rule>
   <rule name="Video/Height is 486" value="Height" tracktype="Video" occurrence="*" operator="=">486</rule>
-  <rule name="Video/PixelAspectRatio is 0.909" value="PixelAspectRatio" tracktype="Video" occurrence="*" operator="=">0.909</rule>
-  <rule name="Video/DisplayAspectRatio is 1.347" value="DisplayAspectRatio" tracktype="Video" occurrence="*" operator="=">1.347</rule>
   <rule name="Video/FrameRate_Mode is CFR" value="FrameRate_Mode" tracktype="Video" occurrence="*" operator="=">CFR</rule>
   <rule name="Video/FrameRate is 29.970" value="FrameRate" tracktype="Video" occurrence="*" operator="=">29.970</rule>
   <rule name="Video/Standard is NTSC" value="Standard" tracktype="Video" occurrence="*" operator="=">NTSC</rule>
@@ -28,7 +26,6 @@
   <rule name="Video/ScanType is Interlaced" value="ScanType" tracktype="Video" occurrence="*" operator="=">Interlaced</rule>
   <rule name="Video/ScanOrder is BFF" value="ScanOrder" tracktype="Video" occurrence="*" operator="=">BFF</rule>
   <rule name="Video/Compression_Mode is Lossless" value="Compression_Mode" tracktype="Video" occurrence="*" operator="=">Lossless</rule>
-  <rule name="Video/TimeCode_Source is Matroska tags" value="TimeCode_Source" tracktype="Video" occurrence="*" operator="=">Matroska tags</rule>
   <rule name="Video/Default is Yes" value="Default" tracktype="Video" occurrence="*" operator="=">Yes</rule>
   <rule name="Video/Forced is No" value="Forced" tracktype="Video" occurrence="*" operator="=">No</rule>
   <rule name="Video/colour_description_present is Yes" value="colour_description_present" tracktype="Video" occurrence="*" operator="=">Yes</rule>

diff --git a/nulrdcscripts/vproc/data/mediaconch_policies/AJA_NTSC_VHS-2SAS-MOV.xml b/nulrdcscripts/vproc/data/mediaconch_policies/AJA_NTSC_VHS-2SAS-MOV.xml
@@ -2,22 +2,17 @@
 <policy type="and" name="P0371-GRNG-F566_v01.mov">
   <rule name="General/VideoCount is 1" value="VideoCount" tracktype="General" occurrence="*" operator="=">1</rule>
   <rule name="General/AudioCount is 2" value="AudioCount" tracktype="General" occurrence="*" operator="=">2</rule>
-  <rule name="General/OtherCount is 1" value="OtherCount" tracktype="General" occurrence="*" operator="=">1</rule>
   <rule name="General/FileExtension is mov" value="FileExtension" tracktype="General" occurrence="*" operator="=">mov</rule>
-  <rule name="General/Format is QuickTime" value="Format" tracktype="General" occurrence="*" operator="=">QuickTime</rule>
+  <rule name="General/Format_Profile is QuickTime" value="Format_Profile" tracktype="General" occurrence="*" operator="=">QuickTime</rule>
   <rule name="General/FrameRate is 29.970" value="FrameRate" tracktype="General" occurrence="*" operator="=">29.970</rule>
-  <rule name="General/Encoded_Library is Apple QuickTime" value="Encoded_Library" tracktype="General" occurrence="*" operator="=">Apple QuickTime</rule>
-  <rule name="General/Encoded_Library_Name is Apple QuickTime" value="Encoded_Library_Name" tracktype="General" occurrence="*" operator="=">Apple QuickTime</rule>
   <rule name="Video/StreamOrder is 0" value="StreamOrder" tracktype="Video" occurrence="*" operator="=">0</rule>
   <rule name="Video/ID is 1" value="ID" tracktype="Video" occurrence="*" operator="=">1</rule>
   <rule name="Video/Format is YUV" value="Format" tracktype="Video" occurrence="*" operator="=">YUV</rule>
   <rule name="Video/CodecID is v210" value="CodecID" tracktype="Video" occurrence="*" operator="=">v210</rule>
   <rule name="Video/BitRate_Mode is CBR" value="BitRate_Mode" tracktype="Video" occurrence="*" operator="=">CBR</rule>
   <rule name="Video/Width is 720" value="Width" tracktype="Video" occurrence="*" operator="=">720</rule>
-  <rule name="Video/Width_CleanAperture is 704" value="Width_CleanAperture" tracktype="Video" occurrence="*" operator="=">704</rule>
+  <rule name="Video/Width_CleanAperture is 704" value="Width_CleanAperture" tracktype="Video" occurrence="*" operator="=">720</rule>
   <rule name="Video/Height is 486" value="Height" tracktype="Video" occurrence="*" operator="=">486</rule>
-  <rule name="Video/PixelAspectRatio is 0.909" value="PixelAspectRatio" tracktype="Video" occurrence="*" operator="=">0.909</rule>
-  <rule name="Video/DisplayAspectRatio is 1.347" value="DisplayAspectRatio" tracktype="Video" occurrence="*" operator="=">1.347</rule>
   <rule name="Video/Rotation is 0.000" value="Rotation" tracktype="Video" occurrence="*" operator="=">0.000</rule>
   <rule name="Video/FrameRate_Mode is CFR" value="FrameRate_Mode" tracktype="Video" occurrence="*" operator="=">CFR</rule>
   <rule name="Video/FrameRate is 29.970" value="FrameRate" tracktype="Video" occurrence="*" operator="=">29.970</rule>
@@ -28,13 +23,12 @@
   <rule name="Video/ScanType is Interlaced" value="ScanType" tracktype="Video" occurrence="*" operator="=">Interlaced</rule>
   <rule name="Video/ScanOrder is BFF" value="ScanOrder" tracktype="Video" occurrence="*" operator="=">BFF</rule>
   <rule name="Video/Compression_Mode is Lossless" value="Compression_Mode" tracktype="Video" occurrence="*" operator="=">Lossless</rule>
-  <rule name="Video/Language is en" value="Language" tracktype="Video" occurrence="*" operator="=">en</rule>
   <rule name="Video/colour_description_present is Yes" value="colour_description_present" tracktype="Video" occurrence="*" operator="=">Yes</rule>
   <rule name="Video/colour_primaries is BT.601 NTSC" value="colour_primaries" tracktype="Video" occurrence="*" operator="=">BT.601 NTSC</rule>
   <rule name="Video/transfer_characteristics is BT.709" value="transfer_characteristics" tracktype="Video" occurrence="*" operator="=">BT.709</rule>
   <rule name="Video/matrix_coefficients is BT.601" value="matrix_coefficients" tracktype="Video" occurrence="*" operator="=">BT.601</rule>
   <rule name="Audio/StreamOrder is 2" value="StreamOrder" tracktype="Audio" occurrence="*" operator="=">2</rule>
-  <rule name="Audio/ID is 3" value="ID" tracktype="Audio" occurrence="*" operator="=">3</rule>
+  <rule name="Audio/ID is 2" value="ID" tracktype="Audio" occurrence="*" operator="=">3</rule>
   <rule name="Audio/Format is PCM" value="Format" tracktype="Audio" occurrence="*" operator="=">PCM</rule>
   <rule name="Audio/Format_Settings_Endianness is Little" value="Format_Settings_Endianness" tracktype="Audio" occurrence="*" operator="=">Little</rule>
   <rule name="Audio/Format_Settings_Sign is Signed" value="Format_Settings_Sign" tracktype="Audio" occurrence="*" operator="=">Signed</rule>
@@ -43,10 +37,7 @@
   <rule name="Audio/Channels is 2" value="Channels" tracktype="Audio" occurrence="*" operator="=">2</rule>
   <rule name="Audio/SamplingRate is 48000" value="SamplingRate" tracktype="Audio" occurrence="*" operator="=">48000</rule>
   <rule name="Audio/BitDepth is 24" value="BitDepth" tracktype="Audio" occurrence="*" operator="=">24</rule>
-  <rule name="Audio/Delay_DropFrame is No" value="Delay_DropFrame" tracktype="Audio" occurrence="*" operator="=">No</rule>
-  <rule name="Audio/Delay_Source is Container" value="Delay_Source" tracktype="Audio" occurrence="*" operator="=">Container</rule>
-  <rule name="Audio/StreamOrder is 3" value="StreamOrder" tracktype="Audio" occurrence="*" operator="=">3</rule>
-  <rule name="Audio/ID is 4" value="ID" tracktype="Audio" occurrence="*" operator="=">4</rule>
+  <rule name="Audio/ID is 3" value="ID" tracktype="Audio" occurrence="*" operator="=">3</rule>
   <rule name="Audio/Format is PCM" value="Format" tracktype="Audio" occurrence="*" operator="=">PCM</rule>
   <rule name="Audio/Format_Settings_Endianness is Little" value="Format_Settings_Endianness" tracktype="Audio" occurrence="*" operator="=">Little</rule>
   <rule name="Audio/Format_Settings_Sign is Signed" value="Format_Settings_Sign" tracktype="Audio" occurrence="*" operator="=">Signed</rule>
@@ -55,14 +46,4 @@
   <rule name="Audio/Channels is 2" value="Channels" tracktype="Audio" occurrence="*" operator="=">2</rule>
   <rule name="Audio/SamplingRate is 48000" value="SamplingRate" tracktype="Audio" occurrence="*" operator="=">48000</rule>
   <rule name="Audio/BitDepth is 24" value="BitDepth" tracktype="Audio" occurrence="*" operator="=">24</rule>
-  <rule name="Audio/Delay_DropFrame is No" value="Delay_DropFrame" tracktype="Audio" occurrence="*" operator="=">No</rule>
-  <rule name="Audio/Delay_Source is Container" value="Delay_Source" tracktype="Audio" occurrence="*" operator="=">Container</rule>
-  <rule name="Other/StreamOrder is 1" value="StreamOrder" tracktype="Other" occurrence="*" operator="=">1</rule>
-  <rule name="Other/ID is 2" value="ID" tracktype="Other" occurrence="*" operator="=">2</rule>
-  <rule name="Other/Type is Time code" value="Type" tracktype="Other" occurrence="*" operator="=">Time code</rule>
-  <rule name="Other/Format is QuickTime TC" value="Format" tracktype="Other" occurrence="*" operator="=">QuickTime TC</rule>
-  <rule name="Other/TimeCode_FirstFrame is 00:00:00:00" value="TimeCode_FirstFrame" tracktype="Other" occurrence="*" operator="=">00:00:00:00</rule>
-  <rule name="Other/TimeCode_Striped is Yes" value="TimeCode_Striped" tracktype="Other" occurrence="*" operator="=">Yes</rule>
-  <rule name="Other/Title is Untitled" value="Title" tracktype="Other" occurrence="*" operator="=">Untitled</rule>
-  <rule name="Other/Language is en" value="Language" tracktype="Other" occurrence="*" operator="=">en</rule>
 </policy>