From 7e93ffe71f89e20836b2eab1d33de8884cad1788 Mon Sep 17 00:00:00 2001
From: Martinez <overwatch@covalent.cz>
Date: Sun, 12 Nov 2017 22:12:09 +0100
Subject: [PATCH] import

---
 .gitignore |   2 +
 dedup.py   | 129 +++++++++++++++++++++++++++++++++++++++++++++++++++++
 install.sh |  11 +++++
 version.py |   8 ++++
 4 files changed, 150 insertions(+)
 create mode 100644 .gitignore
 create mode 100755 dedup.py
 create mode 100755 install.sh
 create mode 100644 version.py

diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..653e6f1
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1,2 @@
+over
+__pycache__
diff --git a/dedup.py b/dedup.py
new file mode 100755
index 0000000..dc2e371
--- /dev/null
+++ b/dedup.py
@@ -0,0 +1,129 @@
+#! /usr/bin/env python3
+# encoding: utf-8
+
+# library imports
+import datetime
+import glob
+import hashlib
+import shutil
+import os
+import over
+
+# local imports
+import version
+
+# --------------------------------------------------
+# Exceptions
+
+class ConfigurationError(Exception):
+	pass
+
+# --------------------------------------------------
+# Functions
+
+# --------------------------------------------------
+# Classes
+
+class UniqueStore:
+	def __init__(self, path, algo):
+		self.path = path
+		self.hasher = getattr(hashlib, algo)
+		self.log = over.text.Output("store")
+		
+		
+		self.store = {} # size: {hash: path in store}
+		
+		self.log("loading storage", self.log.tl.start)
+		
+		i = 0
+		for file in os.listdir(path):
+			self.add(os.path.join(path, file), write_enabled=False)
+			i += 1
+		
+		self.log("loaded %d blocks" %(i), self.log.tl.done)
+	
+	def add(self, path, write_enabled=True, verbose=False):
+		"""
+		Adds a unique file to the storage. Returns True iff the file was new, False otherwise.
+		
+		@while adding a file to the hashed store
+		"""
+		
+		stat = os.stat(path)
+		size = stat.st_size
+		hasher = self.hasher(open(path, "rb").read())
+		hash_raw = hasher.digest()
+		hash_hr = hasher.hexdigest()[:16]
+		
+		is_new = False
+		
+		if size in self.store:
+			suspects = self.store[size]
+			
+			if hash_raw in suspects:
+				store_path = suspects[hash_raw]
+				if verbose: self.log("%s (%s) is already known as %s" %(path, hash_hr, store_path))
+			else:
+				is_new = True
+		else:
+			is_new = True
+		
+		if is_new:
+			filename_src = os.path.basename(path)
+			filename_src_name, filename_src_suffix = os.path.splitext(filename_src)
+			if "." in filename_src_suffix: filename_src_suffix = filename_src_suffix[1:]
+			mtime = datetime.datetime.fromtimestamp(stat.st_mtime)
+			filename_str = "%s %s %s.%s" %(
+				filename_src_name,
+				mtime.strftime("%Y-%m-%d %H:%M:%S"),
+				hash_hr,
+				filename_src_suffix
+			)
+			
+			if size not in self.store:
+				self.store[size] = {}
+			
+			self.store[size][hash_raw] = filename_str
+			
+			if write_enabled:
+				shutil.copy(path, os.path.join(self.path, filename_str))
+			
+			if verbose:
+				self.log("%s (%s) added" %(path, hash_hr), self.log.tl.done)
+		
+		return is_new
+
+# --------------------------------------------------
+
+if __name__ == "__main__":
+	main = over.app.Main("dedup", version.str, "AO-JSL")
+	main.add_option("store", "Directory to put unqie files into.", str, count=1, abbr="s")
+	main.add_option("algo", "Hashing operation to use, use anything supported by hashlib. Either <M>sha256<.> or <M>sha512<.> is recommended, though. The latter is significantly faster on 64bit CPUs.", str, ["sha512"], count=1, abbr="a")
+	main.add_option("verbose", "Describe the action taken with each input file.", bool, [False], abbr="v")
+	main.add_doc("Description", ["Imports files from directory <m>targets<.> into the <W>--<g>output<.> flat directory, making sure these are unique. The files are all renamed to <c>original-filename mtime hash.suffix<.>."])
+	main.setup()
+	
+	if not main.cfg.store:
+		main.print("no <W>--<g>store<.> specified", main.print.tl.fail)
+		main.exit(1)
+	
+	store_dir = os.path.realpath(main.cfg.store)
+	
+	if not os.path.exists(store_dir):
+		main.print("creating <y>%s<.>" %(store_dir), main.print.tl.exec)
+		os.mkdir(store_dir)
+	
+	if not os.path.isdir(store_dir):
+		main.print("<r>%s<.> is not a writable store directory" %(store_dir), main.print.tl.fail)
+		main.exit(1)
+	
+	store = UniqueStore(store_dir, main.cfg.algo)
+	
+	for src in main.targets:
+		if main.cfg.verbose:
+			main.print("processing %s" %(src))
+		
+		files = [f for f in glob.glob(os.path.join(os.path.realpath(src), "**")) if os.path.isfile(f)]
+		
+		for file in files:
+			store.add(file, verbose=main.cfg.verbose)
diff --git a/install.sh b/install.sh
new file mode 100755
index 0000000..7a37abe
--- /dev/null
+++ b/install.sh
@@ -0,0 +1,11 @@
+#! /bin/bash
+
+ROOT="${1}"
+LIBDIR="${ROOT}/usr/lib/over/dedup"
+BINDIR="${ROOT}/usr/bin"
+BIN="dedup"
+
+mkdir -p "${LIBDIR}"
+cp *.py "${LIBDIR}"
+mkdir -p "${BINDIR}"
+ln -s "${LIBDIR}/${BIN}.py" "${BINDIR}/${BIN}"
diff --git a/version.py b/version.py
new file mode 100644
index 0000000..5ed0a41
--- /dev/null
+++ b/version.py
@@ -0,0 +1,8 @@
+#! /usr/bin/env python3
+# encoding: utf-8
+
+major = 0 # VERSION_MAJOR_IDENTIFIER
+minor = 0 # VERSION_MINOR_IDENTIFIER
+# VERSION_LAST_MM 0.0
+patch = 0 # VERSION_PATCH_IDENTIFIER
+str = ".".join(str(v) for v in (major, minor, patch))