From adc92ee610f06452696bc02cf8027a3832de79ba Mon Sep 17 00:00:00 2001
From: Aleteoryx <alyx@aleteoryx.me>
Date: Fri, 21 Feb 2025 02:02:05 -0500
Subject: [PATCH] storage

---
 .gitignore    |  3 ++
 PLAN          | 22 +++++++++++++
 backupper.tcl | 85 +++++++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 110 insertions(+)
 create mode 100644 .gitignore
 create mode 100644 PLAN
 create mode 100755 backupper.tcl

diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000000000000000000000000000000000000..0a11896adf299181be31eb4e82830cd821b7b5c1
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1,3 @@
+*~
+store/
+*.png
diff --git a/PLAN b/PLAN
new file mode 100644
index 0000000000000000000000000000000000000000..d3c6f25b8f0c4ef90c796f4216459978bd590e1c
--- /dev/null
+++ b/PLAN
@@ -0,0 +1,22 @@
+store consists of md5 content-addressed files
+
+store file contents:
+"AMEBAK" [version byte = 0] [compression mode byte] [file contents]
+
+compression mode:
+  'r' - uncompressed
+  'F' - deFlate
+
+compression should be attempted for all files, algorithms tested with
+the first 10MB of data. if the reduction is <5%, do not compress. this
+should make large high-entropy files have less overhead.
+
+directory contents:
+  store/ - see above
+  index.db - list of backup files
+
+backup files are stored in the store, and indexed in index.db. they are
+archive files, with the contents of each file replaced with a binary
+hash.
+
+
diff --git a/backupper.tcl b/backupper.tcl
new file mode 100755
index 0000000000000000000000000000000000000000..82f38db5e91c0f0387b22e85953fb74453245e1a
--- /dev/null
+++ b/backupper.tcl
@@ -0,0 +1,85 @@
+#!/bin/env tclsh
+
+package require sha256
+package require sqlite3
+
+proc readbak {path} {
+  set fd [open $path rb]
+  set magic [read $fd 6]
+  if {$magic != "AMEBAK"} {
+    close $fd
+    return -code error "invalid magic string: \"$magic\""
+  }
+
+  set ver [read $fd 1]
+  binary scan $ver c ver
+  if {$ver != 0} {
+    close $fd
+    return -code error "archive file too new: $ver (self: 0)"
+  }
+
+  set mode [read $fd 1]
+  switch -- $mode {
+    r {}
+    F {
+      zlib push inflate $fd -level 9
+    }
+    default {
+      close $fd
+      return -code error "unknown compression mode: $mode"
+    }
+  }
+
+  set ret [read $fd]
+  close $fd
+  return $ret
+}
+
+proc readstore {store hash} {
+  set hexhash [binary encode hex $hash]
+  set path $store
+  append path "/[string range $hexhash 0 1]"
+  append path "/[string range $hexhash 2 3]"
+  append path "/[string range $hexhash 4 5]/"
+  append path "/[string range $hexhash 6 end]"
+  if {[file exists $path]} {
+    return [readbak $path]
+  } else {
+    return -code error "missing file"
+  }
+}
+
+proc writebak {path data} {
+  set bench [string range $data 0 10000000]
+  binary scan $bench c _
+  set Fbench [zlib deflate $bench 9]
+
+  set fd [open $path wb]
+  puts -nonewline $fd "AMEBAK\0"
+
+  if {[string length $Fbench] < ([string length $bench] * 95 / 100)} {
+    puts -nonewline $fd "F"
+    zlib push deflate $fd -level 9
+  } else { puts -nonewline $fd "r" }
+
+  puts -nonewline $fd $data
+  close $fd
+}
+
+proc writestore {store data} {
+  set hash [::sha2::sha256 -bin -- $data]
+  set hexhash [binary encode hex $hash]
+  set path $store
+  append path "/[string range $hexhash 0 1]"
+  append path "/[string range $hexhash 2 3]"
+  append path "/[string range $hexhash 4 5]/"
+  file mkdir $path
+  set path "$path/[string range $hexhash 6 end]"
+
+  if {![file exists $path]} {
+    writebak $path $data
+  }
+
+  return $hash
+}
+