Add '--algo' option for algorithm selection.

The interaction with '--brute' is a bit messy, but that's
a fix for another day.

Author: eloj

README.md

@@ -43,7 +43,7 @@ This tool is meant for small problems. It is written in Python, after all.
 ### Usage
 
 ```console
-usage: superstring [-h] [-q | -v] [-s | -S | -L LOOPS | -B] [-C STR] [-F STR] [-j] [-i] [-l] [-R] [-G] [-V] [infile]
+usage: superstring [-h] [-q | -v] [-s | -S | -L LOOPS | -B | -A ALGO] [-C STR] [-F STR] [-j] [-i] [-l] [-R] [-G] [-V] [infile]
 
 Approximate Shortest Superstring Generator -- https://github.com/eloj/superstrings
 
@@ -58,6 +58,7 @@ options:
   -S, --sort               Sort input by entry frequency
   -L LOOPS, --loops LOOPS  Shuffle and regenerate until min-length doesn't improve
   -B, --brute              Use brute-force. Warning: Only for tiny inputs!
+  -A ALGO, --algo ALGO     Algorithm selection
   -C STR, --comment STR    String(s) that start a comment in the input
   -F STR, --mtf STR        Input element(s) to move-to-front
   -j, --join-only          Only join input, don't generate superstring
@@ -118,6 +119,9 @@ $ ./superstring -q --brute data/greedy-hard.txt
 cababababc
 ```
 
+The `--algo` option can be used to pick the search algorithm. The default is 'greedy', unless `--brute` was specified, in which
+case it's 'brutedp'. If you specify an invalid name here, you should get a list of valid options back.
+
 ## Python Superstring Library API
 
 `generate_superstring(list) -> str`
@@ -132,6 +136,9 @@ cababababc
 `brutedp(list) -> str`
 : Given a substring-free list of strings, returns an optimal superstring as generated by brute-force dynamic-programming approach.
 
+`brutedijkstra(list) -> str`
+: Given a substring-free list of strings, returns an optimal superstring as generated by brute-force graph search using [Dijkstra's algorithm](https://en.wikipedia.org/wiki/Dijkstra%27s_algorithm).
+
 `make_substring_free(list) -> list`
 : For Greedy to work as originally specified, its input must be _substring-free_ (aka _factor-free_), i.e contain no elements that are substrings of one another. This function will process a list to ensure this pre-condition is true.
 
@@ -151,7 +158,6 @@ res = ssp.generate_superstring(arr)
 ## TODO
 
 * Verify that the type-spec in ssp.py is actually correct.
-* Algorithm selection.
 * Implement encoding into bitstrings.
 * Built-in benchmark.
 

ssp.py

@@ -272,7 +272,7 @@ def generate_superstring(strings: list[str], func=greedy) -> str:
     return func(make_substring_free(strings))
 
 
-def basic_test():
+def basic_tests() -> None:
     """ Basic functionality test """
     # Hard input for GREEDY. Kaplan 2005.
     # { c(ab)^k, (ba)^k, (ab)^kc }
@@ -296,4 +296,4 @@ def basic_test():
     print(f"brutedijkstra(): {res}, len={len(res)}")
 
 if __name__ == "__main__":
-    basic_test()
+    basic_tests()

superstring

@@ -3,13 +3,14 @@
 	Approximate Shortest Superstring Generator
 	https://github.com/eloj/superstrings/
 
-	Copyright © 2024 Eddy Jansson
+	Copyright © 2024, 2025 Eddy Jansson
 	Licensed under the MIT License. See LICENSE file for details.
 """
 from math import log2, ceil, gcd
 import sys
 import argparse
 import random
+import inspect
 import ssp
 
 # TODO: bits/bytes helper print function
@@ -22,6 +23,15 @@ import ssp
 
 PROGRAM_VERSION = "1.0.0"
 
+def get_algos(module='ssp'):
+	return { n : f for (n,f) in inspect.getmembers(sys.modules[module], inspect.isfunction) if not '_' in n }
+
+def get_algo_func_by_name(name):
+	for algo, func in get_algos().items():
+		if name == algo:
+			return func
+	return None
+
 def bits(n):
 	""" Return the number of bits needed to encode the number _n_ """
 	if n == 0:
@@ -35,7 +45,7 @@ def bits(n):
 #	# print(a, bin(a))
 #	return bits(a)
 
-def data_loader(args):
+def data_loader(args) -> list[str]:
 	""" Load data into array """
 	res = []
 	lines = []
@@ -87,6 +97,7 @@ if __name__ == "__main__":
 	order_group.add_argument("-S", "--sort", help="Sort input by entry frequency", action="store_true", default=False)
 	order_group.add_argument("-L", "--loops", help="Shuffle and regenerate until min-length doesn't improve", action="store", type=int, default=0)
 	order_group.add_argument("-B", "--brute", help="Use brute-force. Warning: Only for tiny inputs!", action="store_true")
+	order_group.add_argument("-A", "--algo", metavar="ALGO", help="Algorithm selection", action="store", type=str)
 
 	argp.add_argument("-C", "--comment", metavar="STR", help="String(s) that start a comment in the input", action="extend", nargs=1, type=str)
 	argp.add_argument("-F", "--mtf", metavar="STR", help="Input element(s) to move-to-front", action="extend", nargs=1, type=str)
@@ -109,6 +120,27 @@ if __name__ == "__main__":
 	if args.verbose > 1:
 		print(f"Comment prefix(es): {args.comment}")
 
+	# Algorithm selection
+	algo_func = None
+	if args.algo:
+		algo_func = get_algo_func_by_name(args.algo)
+		if algo_func is None:
+			sys.exit(f"Invalid algorithm name {args.algo}, try one of {[k for k in get_algos().keys()]}")
+		args.brute = args.algo.startswith("brute")
+
+	if args.brute:
+		if not algo_func:
+			algo_func = ssp.brutedp
+		if not args.quiet:
+			if args.verbose > 0:
+				print("WARNING: Using brute-force is extremely slow. Use only on very small inputs.")
+
+	if not algo_func:
+		algo_func = ssp.greedy
+
+	if args.verbose > 0:
+		print(f"Using algorithm {algo_func}")
+
 	original_input = data_loader(args)
 	input_bps = 0
 	transformed_input = []
@@ -160,12 +192,8 @@ if __name__ == "__main__":
 	for v in unique_input:
 		unique_input_len += len(v)
 
-	if args.brute:
-		if not args.quiet:
-			if args.verbose > 0:
-				print("WARNING: Using brute-force is extremely slow. Use only on very small inputs.")
-			if len(unique_input) >= 30:
-				print(f"WARNING: This run is unlikely to return a result in reasonable time (N={len(unique_input)}).")
+	if args.brute and len(unique_input) >= 30:
+		print(f"WARNING: This run is unlikely to return a result in reasonable time (N={len(unique_input)}).")
 
 	# Calculate fixed bits-per-symbol from input.
 	## if input_bps == 0:
@@ -193,10 +221,7 @@ if __name__ == "__main__":
 	if args.join_only:
 		output = "".join(transformed_input)
 	else:
-		if args.brute:
-			min_output = ssp.generate_superstring(transformed_input, ssp.brutedp)
-		else:
-			min_output = ssp.generate_superstring(transformed_input)
+		min_output = ssp.generate_superstring(transformed_input, algo_func)
 		if args.loops:
 			# Shuffle input and regenerate superstring until minimum length doesn't improve.
 			if not args.quiet: