From 7b3f268df25922c515174862da95569aea547367 Mon Sep 17 00:00:00 2001
From: Ander Punnar <ander@kvlt.ee>
Date: Tue, 22 Jun 2021 16:36:30 +0300
Subject: [PATCH] [__download] improvements

1. post download checksum verification
2. detect hashes without prefix
3. add optional --destination
4. updated man
---
 .../conf/type/__download/explorer/remote_cmd  |  19 ---
 .../type/__download/explorer/remote_cmd_get   |  16 +++
 .../type/__download/explorer/remote_cmd_sum   |  82 +++++++++++
 cdist/conf/type/__download/explorer/state     |  60 ++------
 cdist/conf/type/__download/gencode-local      | 131 +++++++++++++++---
 cdist/conf/type/__download/gencode-remote     |  46 +++++-
 cdist/conf/type/__download/man.rst            |  31 ++++-
 cdist/conf/type/__download/manifest           |   2 +-
 cdist/conf/type/__download/parameter/optional |   3 +-
 9 files changed, 292 insertions(+), 98 deletions(-)
 delete mode 100755 cdist/conf/type/__download/explorer/remote_cmd
 create mode 100755 cdist/conf/type/__download/explorer/remote_cmd_get
 create mode 100755 cdist/conf/type/__download/explorer/remote_cmd_sum

diff --git a/cdist/conf/type/__download/explorer/remote_cmd b/cdist/conf/type/__download/explorer/remote_cmd
deleted file mode 100755
index e3e35b45..00000000
--- a/cdist/conf/type/__download/explorer/remote_cmd
+++ /dev/null
@@ -1,19 +0,0 @@
-#!/bin/sh -e
-
-if [ -f "$__object/parameter/cmd-get" ]
-then
-    cmd="$( cat "$__object/parameter/cmd-get" )"
-
-elif command -v curl > /dev/null
-then
-    cmd="curl -L -o - '%s'"
-
-elif command -v fetch > /dev/null
-then
-    cmd="fetch -o - '%s'"
-
-else
-    cmd="wget -O - '%s'"
-fi
-
-echo "$cmd"
diff --git a/cdist/conf/type/__download/explorer/remote_cmd_get b/cdist/conf/type/__download/explorer/remote_cmd_get
new file mode 100755
index 00000000..9f1cd59c
--- /dev/null
+++ b/cdist/conf/type/__download/explorer/remote_cmd_get
@@ -0,0 +1,16 @@
+#!/bin/sh -e
+
+if [ -f "$__object/parameter/cmd-get" ]
+then
+    cat "$__object/parameter/cmd-get"
+elif
+    command -v curl > /dev/null
+then
+    echo "curl -sSL -o - '%s'"
+elif
+    command -v fetch > /dev/null
+then
+    echo "fetch -o - '%s'"
+else
+    echo "wget -O - '%s'"
+fi
diff --git a/cdist/conf/type/__download/explorer/remote_cmd_sum b/cdist/conf/type/__download/explorer/remote_cmd_sum
new file mode 100755
index 00000000..84df663c
--- /dev/null
+++ b/cdist/conf/type/__download/explorer/remote_cmd_sum
@@ -0,0 +1,82 @@
+#!/bin/sh -e
+
+if [ ! -f "$__object/parameter/sum" ]
+then
+    exit 0
+fi
+
+if [ -f "$__object/parameter/cmd-sum" ]
+then
+    cat "$__object/parameter/cmd-sum"
+    exit 0
+fi
+
+sum_should="$( cat "$__object/parameter/sum" )"
+
+if echo "$sum_should" | grep -Fq ':'
+then
+    sum_hash="$( echo "$sum_should" | cut -d : -f 1 )"
+else
+    if echo "$sum_should" | grep -Eq '^[0-9]+\s[0-9]+$'
+    then
+        sum_hash='cksum'
+    elif
+        echo "$sum_should" | grep -Eiq '^[a-f0-9]{32}$'
+    then
+        sum_hash='md5'
+    elif
+        echo "$sum_should" | grep -Eiq '^[a-f0-9]{40}$'
+    then
+        sum_hash='sha1'
+    elif
+        echo "$sum_should" | grep -Eiq '^[a-f0-9]{64}$'
+    then
+        sum_hash='sha256'
+    else
+        echo 'hash format detection failed' >&2
+        exit 1
+    fi
+fi
+
+os="$( "$__explorer/os" )"
+
+case "$sum_hash" in
+    cksum)
+        echo "cksum %s | awk '{print \$1\" \"\$2}'"
+    ;;
+    md5)
+        case "$os" in
+            freebsd)
+                echo "md5 -q %s"
+            ;;
+            *)
+                echo "md5sum %s | awk '{print \$1}'"
+            ;;
+        esac
+    ;;
+    sha1)
+        case "$os" in
+            freebsd)
+                echo "sha1 -q %s"
+            ;;
+            *)
+                echo "sha1sum %s | awk '{print \$1}'"
+            ;;
+        esac
+    ;;
+    sha256)
+        case "$os" in
+            freebsd)
+                echo "sha256 -q %s"
+            ;;
+            *)
+                echo "sha256sum %s | awk '{print \$1}'"
+            ;;
+        esac
+    ;;
+    *)
+        # we arrive here only if --sum is given with unknown format prefix
+        echo "unknown hash format: $sum_hash" >&2
+        exit 1
+    ;;
+esac
diff --git a/cdist/conf/type/__download/explorer/state b/cdist/conf/type/__download/explorer/state
index 68b517c5..881a1c09 100755
--- a/cdist/conf/type/__download/explorer/state
+++ b/cdist/conf/type/__download/explorer/state
@@ -1,6 +1,11 @@
 #!/bin/sh -e
 
-dst="/$__object_id"
+if [ -f "$__object/parameter/destination" ]
+then
+    dst="$( cat "$__object/parameter/destination" )"
+else
+    dst="/$__object_id"
+fi
 
 if [ ! -f "$dst" ]
 then
@@ -16,57 +21,18 @@ fi
 
 sum_should="$( cat "$__object/parameter/sum" )"
 
-if [ -f "$__object/parameter/cmd-sum" ]
+if echo "$sum_should" | grep -Fq ':'
 then
-    # shellcheck disable=SC2059
-    sum_is="$( eval "$( printf \
-        "$( cat "$__object/parameter/cmd-sum" )" \
-        "$dst" )" )"
-else
-    os="$( "$__explorer/os" )"
-
-    if echo "$sum_should" | grep -Eq '^[0-9]+\s[0-9]+$'
-    then
-        sum_is="$( cksum "$dst" | awk '{print $1" "$2}' )"
-
-    elif echo "$sum_should" | grep -Eiq '^md5:[a-f0-9]{32}$'
-    then
-        case "$os" in
-            freebsd)
-                sum_is="md5:$( md5 -q "$dst" )"
-            ;;
-            *)
-                sum_is="md5:$( md5sum "$dst" | awk '{print $1}' )"
-            ;;
-        esac
-
-    elif echo "$sum_should" | grep -Eiq '^sha1:[a-f0-9]{40}$'
-    then
-        case "$os" in
-            freebsd)
-                sum_is="sha1:$( sha1 -q "$dst" )"
-            ;;
-            *)
-                sum_is="sha1:$( sha1sum "$dst" | awk '{print $1}' )"
-            ;;
-        esac
-
-    elif echo "$sum_should" | grep -Eiq '^sha256:[a-f0-9]{64}$'
-    then
-        case "$os" in
-            freebsd)
-                sum_is="sha256:$( sha256 -q "$dst" )"
-            ;;
-            *)
-                sum_is="sha256:$( sha256sum "$dst" | awk '{print $1}' )"
-            ;;
-        esac
-    fi
+    sum_should="$( echo "$sum_should" | cut -d : -f 2 )"
 fi
 
+sum_cmd="$( "$__type_explorer/remote_cmd_sum" )"
+
+sum_is="$( eval "$( printf "$sum_cmd" "'$dst'" )" )"
+
 if [ -z "$sum_is" ]
 then
-    echo 'no checksum from target' >&2
+    echo 'existing destination checksum failed' >&2
     exit 1
 fi
 
diff --git a/cdist/conf/type/__download/gencode-local b/cdist/conf/type/__download/gencode-local
index 571d2c3c..d1b0d0d5 100755
--- a/cdist/conf/type/__download/gencode-local
+++ b/cdist/conf/type/__download/gencode-local
@@ -11,34 +11,133 @@ fi
 
 url="$( cat "$__object/parameter/url" )"
 
-tmp="$( mktemp )"
-
-dst="/$__object_id"
+if [ -f "$__object/parameter/destination" ]
+then
+    dst="$( cat "$__object/parameter/destination" )"
+else
+    dst="/$__object_id"
+fi
 
 if [ -f "$__object/parameter/cmd-get" ]
 then
     cmd="$( cat "$__object/parameter/cmd-get" )"
 
-elif command -v wget > /dev/null
-then
-    cmd="wget -O - '%s'"
-
 elif command -v curl > /dev/null
 then
-    cmd="curl -L -o - '%s'"
+    cmd="curl -sSL -o - '%s'"
 
 elif command -v fetch > /dev/null
 then
     cmd="fetch -o - '%s'"
 
+elif command -v wget > /dev/null
+then
+    cmd="wget -O - '%s'"
+
 else
-    echo 'no usable locally installed utility for downloading' >&2
+    echo 'local download failed, no usable utility' >&2
     exit 1
 fi
 
-printf "$cmd > %s\n" \
-    "$url" \
-    "$tmp"
+echo "download_tmp=\"\$( mktemp )\""
+
+# shellcheck disable=SC2059
+printf "$cmd > \"\$download_tmp\"\n" "$url"
+
+if [ -f "$__object/parameter/sum" ]
+then
+    sum_should="$( cat "$__object/parameter/sum" )"
+
+    if [ -f "$__object/parameter/cmd-sum" ]
+    then
+        local_cmd_sum="$( cat "$__object/parameter/cmd-sum" )"
+    else
+        if echo "$sum_should" | grep -Fq ':'
+        then
+            sum_hash="$( echo "$sum_should" | cut -d : -f 1 )"
+
+            sum_should="$( echo "$sum_should" | cut -d : -f 2 )"
+        else
+            if echo "$sum_should" | grep -Eq '^[0-9]+\s[0-9]+$'
+            then
+                sum_hash='cksum'
+            elif
+                echo "$sum_should" | grep -Eiq '^[a-f0-9]{32}$'
+            then
+                sum_hash='md5'
+            elif
+                echo "$sum_should" | grep -Eiq '^[a-f0-9]{40}$'
+            then
+                sum_hash='sha1'
+            elif
+                echo "$sum_should" | grep -Eiq '^[a-f0-9]{64}$'
+            then
+                sum_hash='sha256'
+            else
+                echo 'hash format detection failed' >&2
+                exit 1
+            fi
+        fi
+
+        case "$sum_hash" in
+            cksum)
+                local_cmd_sum="cksum %s | awk '{print \$1\" \"\$2}'"
+            ;;
+            md5)
+                if command -v md5 > /dev/null
+                then
+                    local_cmd_sum="md5 -q %s"
+                elif
+                    command -v md5sum > /dev/null
+                then
+                    local_cmd_sum="md5sum %s | awk '{print \$1}'"
+                fi
+            ;;
+            sha1)
+                if command -v sha1 > /dev/null
+                then
+                    local_cmd_sum="sha1 -q %s"
+                elif
+                    command -v sha1sum > /dev/null
+                then
+                    local_cmd_sum="sha1sum %s | awk '{print \$1}'"
+                fi
+            ;;
+            sha256)
+                if command -v sha256 > /dev/null
+                then
+                    local_cmd_sum="sha256 -q %s"
+                elif
+                    command -v sha256sum > /dev/null
+                then
+                    local_cmd_sum="sha256sum %s | awk '{print \$1}'"
+                fi
+            ;;
+            *)
+                # we arrive here only if --sum is given with unknown format prefix
+                echo "unknown hash format: $sum_hash" >&2
+                exit 1
+            ;;
+        esac
+
+        if [ -z "$local_cmd_sum" ]
+        then
+            echo 'local checksum verification failed, no usable utility' >&2
+            exit 1
+        fi
+    fi
+
+    # shellcheck disable=SC2059
+    echo "sum_is=\"\$( $( printf "$local_cmd_sum" "\"\$download_tmp\"" ) )\""
+
+    echo "if [ \"\$sum_is\" != '$sum_should' ]; then"
+
+    echo "echo 'local download checksum mismatch' >&2"
+
+    echo "rm -f \"\$download_tmp\""
+
+    echo 'exit 1; fi'
+fi
 
 if echo "$__target_host" | grep -Eq '^[0-9a-fA-F:]+$'
 then
@@ -47,12 +146,10 @@ else
     target_host="$__target_host"
 fi
 
-printf '%s %s %s:%s\n' \
+# shellcheck disable=SC2016
+printf '%s "$download_tmp" %s:%s\n' \
     "$__remote_copy" \
-    "$tmp" \
     "$target_host" \
     "$dst"
 
-echo "rm -f '$tmp'"
-
-echo 'downloaded' > "$__messages_out"
+echo "rm -f \"\$download_tmp\""
diff --git a/cdist/conf/type/__download/gencode-remote b/cdist/conf/type/__download/gencode-remote
index 029a0801..e49bcec3 100755
--- a/cdist/conf/type/__download/gencode-remote
+++ b/cdist/conf/type/__download/gencode-remote
@@ -6,17 +6,51 @@ state_is="$( cat "$__object/explorer/state" )"
 
 if [ "$download" = 'remote' ] && [ "$state_is" != 'present' ]
 then
-    cmd="$( cat "$__object/explorer/remote_cmd" )"
+    cmd_get="$( cat "$__object/explorer/remote_cmd_get" )"
 
     url="$( cat "$__object/parameter/url" )"
 
-    dst="/$__object_id"
+    if [ -f "$__object/parameter/destination" ]
+    then
+        dst="$( cat "$__object/parameter/destination" )"
+    else
+        dst="/$__object_id"
+    fi
 
-    printf "$cmd > %s\n" \
-        "$url" \
-        "$dst"
+    echo "download_tmp=\"\$( mktemp )\""
 
-    echo 'downloaded' > "$__messages_out"
+    # shellcheck disable=SC2059
+    printf "$cmd_get > \"\$download_tmp\"\n" "$url"
+
+    if [ -f "$__object/parameter/sum" ]
+    then
+        sum_should="$( cat "$__object/parameter/sum" )"
+
+        if [ -f "$__object/parameter/cmd-sum" ]
+        then
+            remote_cmd_sum="$( cat "$__object/parameter/cmd-sum" )"
+        else
+            remote_cmd_sum="$( cat "$__object/explorer/remote_cmd_sum" )"
+
+            if echo "$sum_should" | grep -Fq ':'
+            then
+                sum_should="$( echo "$sum_should" | cut -d : -f 2 )"
+            fi
+        fi
+
+        # shellcheck disable=SC2059
+        echo "sum_is=\"\$( $( printf "$remote_cmd_sum" "\"\$download_tmp\"" ) )\""
+
+        echo "if [ \"\$sum_is\" != '$sum_should' ]; then"
+
+        echo "echo 'remote download checksum mismatch' >&2"
+
+        echo "rm -f \"\$download_tmp\""
+
+        echo 'exit 1; fi'
+    fi
+
+    echo "mv \"\$download_tmp\" '$dst'"
 fi
 
 if [ -f "$__object/parameter/onchange" ] && [ "$state_is" != "present" ]
diff --git a/cdist/conf/type/__download/man.rst b/cdist/conf/type/__download/man.rst
index a1278cfb..c16510a9 100644
--- a/cdist/conf/type/__download/man.rst
+++ b/cdist/conf/type/__download/man.rst
@@ -8,7 +8,7 @@ cdist-type__download - Download a file
 
 DESCRIPTION
 -----------
-By default type will try to use ``wget``, ``curl`` or ``fetch``.
+By default type will try to use ``curl``, ``fetch`` or ``wget``.
 If download happens in target (see ``--download``) then type will
 fallback to (and install) ``wget``.
 
@@ -16,6 +16,8 @@ If download happens in local machine, then environment variables like
 ``{http,https,ftp}_proxy`` etc can be used on cdist execution
 (``http_proxy=foo cdist config ...``).
 
+To change downloaded file's owner, group or permissions, use ``require='__download/path/to/file' __file ...``.
+
 
 REQUIRED PARAMETERS
 -------------------
@@ -25,14 +27,29 @@ url
 
 OPTIONAL PARAMETERS
 -------------------
+destination
+   Downloaded file's destination in target. If unset, ``$__object_id`` is used.
+
 sum
-   Checksum is used to decide if existing destination file must be redownloaded.
-   By default output of ``cksum`` without filename is expected.
-   Other hash formats supported with prefixes: ``md5:``, ``sha1:`` and ``sha256:``.
+   Supported formats: ``cksum`` output without file name, MD5, SHA1 and SHA256.
+
+   Type tries to detect hash format with regexes, but prefixes
+   ``cksum:``, ``md5:``, ``sha1:`` and ``sha256:`` are also supported.
+
+   Checksum have two purposes - state check and post-download verification.
+   In state check, if destination checksum mismatches, then content of URL
+   will be downloaded to temporary file. If downloaded temporary file's
+   checksum matches, then it will be moved to destination (overwritten).
+
+   For local downloads it is expected that usable utilities for checksum
+   calculation exist in the system.
 
 download
-   If ``local`` (default), then download file to local storage and copy
-   it to target host. If ``remote``, then download happens in target.
+   If ``local`` (default), then file is downloaded to local storage and copied
+   to target host. If ``remote``, then download happens in target.
+
+   For local downloads it is expected that usable utilities for downloading
+   exist in the system. Type will try to use ``curl``, ``fetch`` or ``wget``.
 
 cmd-get
    Command used for downloading.
@@ -62,7 +79,7 @@ EXAMPLES
     require='__directory/opt/cpma' \
         __download /opt/cpma/cnq3.zip \
             --url https://cdn.playmorepromode.com/files/cnq3/cnq3-1.51.zip \
-            --sum md5:46da3021ca9eace277115ec9106c5b46
+            --sum 46da3021ca9eace277115ec9106c5b46
 
     require='__download/opt/cpma/cnq3.zip' \
         __unpack /opt/cpma/cnq3.zip \
diff --git a/cdist/conf/type/__download/manifest b/cdist/conf/type/__download/manifest
index 7ec8d86d..3d4c498b 100755
--- a/cdist/conf/type/__download/manifest
+++ b/cdist/conf/type/__download/manifest
@@ -1,6 +1,6 @@
 #!/bin/sh -e
 
-if grep -Eq '^wget' "$__object/explorer/remote_cmd"
+if grep -Eq '^wget' "$__object/explorer/remote_cmd_get"
 then
     __package wget
 fi
diff --git a/cdist/conf/type/__download/parameter/optional b/cdist/conf/type/__download/parameter/optional
index d69e083e..e809ef78 100644
--- a/cdist/conf/type/__download/parameter/optional
+++ b/cdist/conf/type/__download/parameter/optional
@@ -1,5 +1,6 @@
-sum
 cmd-get
 cmd-sum
+destination
 download
 onchange
+sum