From d76b12edd1a42c3c1bcf07c5ef458df260fb682c Mon Sep 17 00:00:00 2001
From: catus felis <phanmanhdan@gmail.com>
Date: Mon, 2 Dec 2024 12:43:53 +0700
Subject: [PATCH] Adds filter parameters to allow selective dictionary building

---
 .github/workflows/release_all.yml          |  4 ++--
 .github/workflows/release_all_external.yml |  1 +
 .vscode/launch.json                        |  5 +++--
 bin/convert_all.py                         | 17 ++++++++++++++++-
 4 files changed, 22 insertions(+), 5 deletions(-)

diff --git a/.github/workflows/release_all.yml b/.github/workflows/release_all.yml
index 15275dc..5e1023c 100644
--- a/.github/workflows/release_all.yml
+++ b/.github/workflows/release_all.yml
@@ -1,4 +1,4 @@
-name: Release all dictionaries
+name: Release sample dictionaries
 
 on:
   workflow_dispatch:
@@ -60,7 +60,7 @@ jobs:
 
     - name: Build all dictionaries
       run: |
-        python ./bin/convert_all.py --input_folder=$INPUT_DIR --output_folder=$OUTPUT_DIR --extension=tab
+        python ./bin/convert_all.py --input_folder=$INPUT_DIR --output_folder=$OUTPUT_DIR --extension=tab --filter=CCCE,ThienChuu
         
     - name: Zip all artifacts for release
       run: |
diff --git a/.github/workflows/release_all_external.yml b/.github/workflows/release_all_external.yml
index 4753f31..debbbf7 100644
--- a/.github/workflows/release_all_external.yml
+++ b/.github/workflows/release_all_external.yml
@@ -9,6 +9,7 @@ on:
       - '*.md'
       - LICENSE
       - bin/create_inflections.py      
+      - .github/workflows/release_all.yml   
 
 env:
   INPUT_DIR: ./ext-dict
diff --git a/.vscode/launch.json b/.vscode/launch.json
index c48a526..3744836 100644
--- a/.vscode/launch.json
+++ b/.vscode/launch.json
@@ -6,7 +6,7 @@
     "configurations": [
         {
             "name": "Python: Current File",
-            "type": "python",
+            "type": "debugpy",
             "request": "launch",
             "program": "./bin/convert_all.py",
             "console": "integratedTerminal",
@@ -14,7 +14,8 @@
             "args": [
                 "--input_folder=./dict",
                 "--output_folder=./output",
-                "--extension=tab"
+                "--extension=tab",
+                "--filter=CCCE,ThienChuu"
             ]
         }
     ]
diff --git a/bin/convert_all.py b/bin/convert_all.py
index 34181b5..4331af9 100755
--- a/bin/convert_all.py
+++ b/bin/convert_all.py
@@ -71,6 +71,7 @@ def main() -> None:
     parser.add_argument('-o', '--output_folder', required=True, help='Output folder containing dictionary files')
     parser.add_argument('-e', '--extension', default='tab', help='Filename extention for input dictionary files. Default is .tab')
     parser.add_argument('-m', '--metadata', default='dfo', help='Filename extention for input metadata for dictionary. Default is .dfo')
+    parser.add_argument('-f', '--filter', help='Filter only dictionary entries with matching keys (seperated by comma)')
 
     args, array = parser.parse_known_args()
 
@@ -78,6 +79,7 @@ def main() -> None:
     output_folder = args.output_folder.replace(' ', '\\ ')
     extension = args.extension
     metadata = args.metadata
+    dict_filters = args.filter.split(",") if args.filter is not None else []
 
     if input_folder:
         metafilelist = sorted(glob.glob(input_folder + f'/*.{metadata}'), reverse=True)
@@ -113,6 +115,19 @@ def main() -> None:
         datafilelist.clear()
 
         for key in common_keys:
+            if dict_filters:
+                include_dict = False
+                for filter in dict_filters:
+                    if filter in key:
+                        include_dict = True
+
+                        break
+            
+            if include_dict:
+                print(f"Including this dictionary: {key}")
+            else:
+                continue
+
             metafilelist.append(meta_dict[key])
 
             datafile = data_dict[key]
@@ -143,7 +158,7 @@ def main() -> None:
 
     # use_only_these = {'Tu-dien-ThienChuu-TranVanChanh'}
     for filepath, datafile in zip(metafilelist, datafilelist):
-        folder, filename = os.path.split(filepath)
+        _, filename = os.path.split(filepath)
         filebase, fileext = os.path.splitext(filename)
 
         # if filebase not in use_only_these: