gcd now has string similarity metrics

files/gcd-findmatch.py

@@ -1,11 +1,15 @@
 #!/usr/bin/env python
 
 import sys,os,re
+import difflib as dl
 
+# Define color codes
 Green="\033[32;1m"
 Red="\033[31;1m"
 Reset="\033[0m"
+# Available shortcut keys
 key_list="1234567890qwertyuiop"
+
 class getch:
     def get(self):
         fd = sys.stdin.fileno()
@@ -16,13 +20,30 @@ class getch:
         finally:
             termios.tcsetattr(fd, termios.TCSADRAIN, old_settings)
         return ch
+
 def printerr(s):
     sys.stderr.write(s+"\n")
     sys.stderr.flush()
+
+def colorize_matches(directories,repl):
+    return [repl.sub("%s\g<1>%s"%(Green,Reset),d) for d in directories]
+
+def search_close_match(directories, key):
+    """ Return a list of Components whose name closely matches key """
+    match=[]
+    key=key.lower()
+    for name in directories:
+        s=dl.SequenceMatcher(None, name.lower(), key)
+        s_ratio=s.ratio()
+        if s_ratio > 0:
+            match.append(( s_ratio, name ))
+    match.sort(key=lambda x:x[0], reverse=True)
+    best_matches=[x[1] for x in match[0:10]]
+    return best_matches
+
 # Print help
 if len(sys.argv)==1 or sys.argv[-1]=="-h":
     printerr("Guess cd: Find first match in folder, interactive if multiple matches\n  Arguments: [dir/]pattern_to_match_folders")
-if sys.argv[-1]=="-h":
     sys.exit(0)
 
 pattern=".".join(sys.argv[1:])
@@ -35,34 +56,46 @@ if pat_dir=="":
 if not os.path.exists(pat_dir):
     printerr("%s: %sNo such folder%s"%(pat_dir,Red,Reset))
     sys.exit(1)
+
 # list sub-folders in the folder
-current_folders=[d for d in os.listdir(pat_dir) if os.path.isdir(d)]
+current_folders=[d for d in os.listdir(pat_dir) if os.path.isdir(os.path.join(pat_dir,d))]
 current_folders=[d for d in current_folders if not d.startswith(".")]
 current_folders.sort()
 
+# no sub-folders
+if len(current_folders)==0:
+    printerr("There are no subfolders in %s"%(pat_dir,))
+    sys.exit(0)
+
 # find matches
 match_case=re.compile(".*"+pat_base+".*")
 repl_case=re.compile("("+pat_base+")")
 
 matching_folders=[d for d in current_folders if match_case.match(d)]
-matches=[repl_case.sub("%s\g<1>%s"%(Green,Reset),d) for d in matching_folders]
+matches=colorize_matches(matching_folders, repl_case)
+#[repl_case.sub("%s\g<1>%s"%(Green,Reset),d) for d in matching_folders]
 
 # no matches, try case insensitive
 if len(matching_folders)==0:
     match_nocase=re.compile(".*"+pat_base+".*",re.I)
     repl_nocase=re.compile("("+pat_base+")",re.I)
     matching_folders=[d for d in current_folders if match_nocase.match(d)]
-    matches=[repl_nocase.sub("%s\g<1>%s"%(Green,Reset),d) for d in matching_folders]
+    matches=colorize_matches(matching_folders, repl_nocase)
+    #matches=[repl_nocase.sub("%s\g<1>%s"%(Green,Reset),d) for d in matching_folders]
+
+# No matches, try close match
+if len(matching_folders)==0:    
+    matching_folders=search_close_match(current_folders, pat_base)
+    matches=[d for d in matching_folders]
+    print("close")
+    print(matches)
 
 # One match, print and return
 if len(matching_folders)==1:
     printerr(matches[0])
     print(os.path.join(pat_dir,matching_folders[0]))
     sys.exit(0)
-# No matches
+
-if len(matching_folders)==0:    
-    printerr("%sno matches%s"%(Red,Reset))
-    sys.exit(1)
 # Many matches, ask the user for folder index
 for i,m in enumerate(matches[0:len(key_list)]):
     printerr(key_list[i]+": "+m)