rolfedh-doc-utils 0.1.3__py3-none-any.whl → 0.1.5__py3-none-any.whl

archive_unused_files.py

@@ -1,7 +1,8 @@
 """
 Archive Unused AsciiDoc Files
 
-Scans './modules' and './assemblies' for AsciiDoc files not referenced by any other AsciiDoc file in the project. Optionally archives and deletes them.
+Automatically discovers and scans 'modules' and 'assemblies' directories for AsciiDoc files 
+not referenced by any other AsciiDoc file in the project. Optionally archives and deletes them.
 
 For full documentation and usage examples, see archive_unused_files.md in this directory.
 """
@@ -11,14 +12,19 @@ from doc_utils.unused_adoc import find_unused_adoc
 from doc_utils.file_utils import parse_exclude_list_file
 
 def main():
-    parser = argparse.ArgumentParser(description='Archive unused AsciiDoc files.')
+    parser = argparse.ArgumentParser(
+        description='Archive unused AsciiDoc files.',
+        epilog='By default, automatically discovers all modules and assemblies directories in the repository.'
+    )
     parser.add_argument('--archive', action='store_true', help='Move the files to a dated zip in the archive directory.')
+    parser.add_argument('--scan-dir', action='append', default=[], help='Specific directory to scan (can be used multiple times). If not specified, auto-discovers directories.')
     parser.add_argument('--exclude-dir', action='append', default=[], help='Directory to exclude (can be used multiple times).')
     parser.add_argument('--exclude-file', action='append', default=[], help='File to exclude (can be used multiple times).')
     parser.add_argument('--exclude-list', type=str, help='Path to a file containing directories or files to exclude, one per line.')
     args = parser.parse_args()
 
-    scan_dirs = ['./modules', './modules/rn', './assemblies']
+    # Use provided scan directories or None for auto-discovery
+    scan_dirs = args.scan_dir if args.scan_dir else None
     archive_dir = './archive'
 
     exclude_dirs = list(args.exclude_dir)

doc_utils/topic_map_parser.py

@@ -22,8 +22,16 @@ def detect_repo_type(base_path='.'):
         if yml_files:
             return 'topic_map'
     
-    # Check for master.adoc files
-    master_files = glob.glob(os.path.join(base_path, '**/master.adoc'), recursive=True)
+    # Check for master.adoc files using os.walk to avoid symlink issues
+    master_files = []
+    for root, dirs, files in os.walk(base_path):
+        # Skip symbolic link directories to prevent infinite recursion
+        dirs[:] = [d for d in dirs if not os.path.islink(os.path.join(root, d))]
+        
+        # Check for master.adoc in this directory
+        if 'master.adoc' in files:
+            master_files.append(os.path.join(root, 'master.adoc'))
+            
     if master_files:
         return 'master_adoc'
     

doc_utils/unused_adoc.py

@@ -5,7 +5,77 @@ import re
 from .file_utils import collect_files, write_manifest_and_archive
 from .topic_map_parser import detect_repo_type, get_all_topic_map_references
 
-def find_unused_adoc(scan_dirs, archive_dir, archive=False, exclude_dirs=None, exclude_files=None):
+def find_scan_directories(base_path='.', exclude_dirs=None):
+    """
+    Automatically find all 'modules' and 'assemblies' directories in the repository.
+    
+    Returns a list of paths to scan.
+    """
+    scan_dirs = []
+    exclude_dirs = exclude_dirs or []
+    
+    for root, dirs, files in os.walk(base_path):
+        # Skip symbolic links to prevent issues
+        dirs[:] = [d for d in dirs if not os.path.islink(os.path.join(root, d))]
+        
+        # Skip excluded directories
+        for exclude_dir in exclude_dirs:
+            abs_exclude = os.path.abspath(exclude_dir)
+            if os.path.abspath(root).startswith(abs_exclude):
+                dirs[:] = []  # Don't descend into excluded directories
+                break
+        
+        # Skip hidden directories and common non-content directories
+        dirs[:] = [d for d in dirs if not d.startswith('.') and d not in ['node_modules', 'build', 'dist', 'target']]
+        
+        # Look for modules and assemblies directories
+        for d in dirs:
+            if d in ['modules', 'assemblies']:
+                dir_path = os.path.join(root, d)
+                # Check if this directory or any subdirectory contains .adoc files
+                has_adoc = False
+                for subroot, subdirs, subfiles in os.walk(dir_path):
+                    # Skip symbolic links
+                    subdirs[:] = [sd for sd in subdirs if not os.path.islink(os.path.join(subroot, sd))]
+                    if any(f.endswith('.adoc') for f in subfiles):
+                        has_adoc = True
+                        break
+                if has_adoc:
+                    scan_dirs.append(dir_path)
+    
+    # Also check for modules/rn pattern if modules exists
+    modules_dirs = [d for d in scan_dirs if os.path.basename(d) == 'modules']
+    for modules_dir in modules_dirs:
+        rn_dir = os.path.join(modules_dir, 'rn')
+        if os.path.isdir(rn_dir):
+            # Check if rn directory or subdirectories contain .adoc files
+            has_adoc = False
+            for subroot, subdirs, subfiles in os.walk(rn_dir):
+                subdirs[:] = [sd for sd in subdirs if not os.path.islink(os.path.join(subroot, sd))]
+                if any(f.endswith('.adoc') for f in subfiles):
+                    has_adoc = True
+                    break
+            if has_adoc:
+                scan_dirs.append(rn_dir)
+    
+    return scan_dirs
+
+def find_unused_adoc(scan_dirs=None, archive_dir='./archive', archive=False, exclude_dirs=None, exclude_files=None):
+    # Print safety warning
+    print("\n⚠️  SAFETY: Work in a git branch! Run without --archive first to preview.\n")
+    
+    # If no scan_dirs provided, auto-discover them
+    if not scan_dirs:
+        scan_dirs = find_scan_directories(exclude_dirs=exclude_dirs)
+        if scan_dirs:
+            print(f"Auto-discovered directories to scan:")
+            for dir_path in sorted(scan_dirs):
+                print(f"  - {dir_path}")
+        else:
+            print("No 'modules' or 'assemblies' directories found containing .adoc files.")
+            print("Please run this tool from your documentation repository root.")
+            return
+    
     # Detect repository type
     repo_type = detect_repo_type()
     print(f"Detected repository type: {repo_type}")

doc_utils/unused_images.py

@@ -7,6 +7,9 @@ from .file_utils import collect_files, write_manifest_and_archive
 IMAGE_EXTENSIONS = {'.png', '.jpg', '.jpeg', '.gif', '.svg'}
 
 def find_unused_images(scan_dirs, archive_dir, archive=False, exclude_dirs=None, exclude_files=None):
+    # Print safety warning
+    print("\n⚠️  SAFETY: Work in a git branch! Run without --archive first to preview.\n")
+    
     image_files = collect_files(scan_dirs, IMAGE_EXTENSIONS, exclude_dirs, exclude_files)
     adoc_files = collect_files(['.'], {'.adoc'}, exclude_dirs, exclude_files)
     referenced_images = set()

{rolfedh_doc_utils-0.1.3.dist-info → rolfedh_doc_utils-0.1.5.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: rolfedh-doc-utils
-Version: 0.1.3
+Version: 0.1.5
 Summary: CLI tools for AsciiDoc documentation projects
 Author: Rolfe Dlugy-Hegwer
 License: MIT License

rolfedh_doc_utils-0.1.5.dist-info/RECORD

@@ -0,0 +1,17 @@
+archive_unused_files.py,sha256=KMC5a1WL3rZ5owoVnncvfpT1YeMKbVXq9giHvadDgbM,1936
+archive_unused_images.py,sha256=PG2o3haovYckgfhoPhl6KRG_a9czyZuqlLkzkupKTCY,1526
+check_scannability.py,sha256=gcM-vFXKHGP_yFBz7-V5xbXWhIMmtMzBYIGwP9CFbzI,5140
+find_unused_attributes.py,sha256=fk-K32eoCVHxoj7RiBNgSmX1arBLuwYfdSAOMc-wIx0,1677
+doc_utils/__init__.py,sha256=qqZR3lohzkP63soymrEZPBGzzk6-nFzi4_tSffjmu_0,74
+doc_utils/file_utils.py,sha256=fpTh3xx759sF8sNocdn_arsP3KAv8XA6cTQTAVIZiZg,4247
+doc_utils/scannability.py,sha256=XwlmHqDs69p_V36X7DLjPTy0DUoLszSGqYjJ9wE-3hg,982
+doc_utils/topic_map_parser.py,sha256=tKcIO1m9r2K6dvPRGue58zqMr0O2zKU1gnZMzEE3U6o,4571
+doc_utils/unused_adoc.py,sha256=2cbqcYr1os2EhETUU928BlPRlsZVSdI00qaMhqjSIqQ,5263
+doc_utils/unused_attributes.py,sha256=HBgmHelqearfWl3TTC2bZGiJytjLADIgiGQUNKqXXPg,1847
+doc_utils/unused_images.py,sha256=nqn36Bbrmon2KlGlcaruNjJJvTQ8_9H0WU9GvCW7rW8,1456
+rolfedh_doc_utils-0.1.5.dist-info/licenses/LICENSE,sha256=vLxtwMVOJA_hEy8b77niTkdmQI9kNJskXHq0dBS36e0,1075
+rolfedh_doc_utils-0.1.5.dist-info/METADATA,sha256=pSqhon_Hxstw72EnvJ-4UusaepxhdOvxUI0OqTF0toQ,8152
+rolfedh_doc_utils-0.1.5.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+rolfedh_doc_utils-0.1.5.dist-info/entry_points.txt,sha256=i8LqEsp0KD4YyVI_7wQ1TMgCuag32D7gQes6bLufmtM,216
+rolfedh_doc_utils-0.1.5.dist-info/top_level.txt,sha256=BkaYN3KbtNvLQjs-QGBKCJb5UAtjEbC_IqxSSIN9P-w,95
+rolfedh_doc_utils-0.1.5.dist-info/RECORD,,

rolfedh_doc_utils-0.1.3.dist-info/RECORD

@@ -1,17 +0,0 @@
-archive_unused_files.py,sha256=MFdAtE8CDthida5H5FhvtEHZWrQAZZoAB-DQkTOc6gs,1537
-archive_unused_images.py,sha256=PG2o3haovYckgfhoPhl6KRG_a9czyZuqlLkzkupKTCY,1526
-check_scannability.py,sha256=gcM-vFXKHGP_yFBz7-V5xbXWhIMmtMzBYIGwP9CFbzI,5140
-find_unused_attributes.py,sha256=fk-K32eoCVHxoj7RiBNgSmX1arBLuwYfdSAOMc-wIx0,1677
-doc_utils/__init__.py,sha256=qqZR3lohzkP63soymrEZPBGzzk6-nFzi4_tSffjmu_0,74
-doc_utils/file_utils.py,sha256=fpTh3xx759sF8sNocdn_arsP3KAv8XA6cTQTAVIZiZg,4247
-doc_utils/scannability.py,sha256=XwlmHqDs69p_V36X7DLjPTy0DUoLszSGqYjJ9wE-3hg,982
-doc_utils/topic_map_parser.py,sha256=4vmOPInoKAFO54qOz4EuK24hHTngI4GXzrw0D_QZzSc,4224
-doc_utils/unused_adoc.py,sha256=aPhq4KQG5VJGCg_cnqyhNWuADGEQA_dtjmvxtf7ylGA,2185
-doc_utils/unused_attributes.py,sha256=HBgmHelqearfWl3TTC2bZGiJytjLADIgiGQUNKqXXPg,1847
-doc_utils/unused_images.py,sha256=P9vcm00BidrLmxhjeczBtiFU-1wgfN5nCYdZjeCH1kM,1329
-rolfedh_doc_utils-0.1.3.dist-info/licenses/LICENSE,sha256=vLxtwMVOJA_hEy8b77niTkdmQI9kNJskXHq0dBS36e0,1075
-rolfedh_doc_utils-0.1.3.dist-info/METADATA,sha256=nTGGOIULiuqFcveMw4kGsHF-NVbe-YK3sbElusT9ku0,8152
-rolfedh_doc_utils-0.1.3.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-rolfedh_doc_utils-0.1.3.dist-info/entry_points.txt,sha256=i8LqEsp0KD4YyVI_7wQ1TMgCuag32D7gQes6bLufmtM,216
-rolfedh_doc_utils-0.1.3.dist-info/top_level.txt,sha256=BkaYN3KbtNvLQjs-QGBKCJb5UAtjEbC_IqxSSIN9P-w,95
-rolfedh_doc_utils-0.1.3.dist-info/RECORD,,