Add playground tools read-parquet-file command #5682

vdesabou · Jun 6, 2024 · 047259d · 047259d
1 parent d8f54af
commit 047259d
Show file tree

Hide file tree

Showing 7 changed files with 495 additions and 264 deletions.
diff --git a/scripts/cli/completions.bash b/scripts/cli/completions.bash
diff --git a/scripts/cli/playground b/scripts/cli/playground
@@ -92,6 +92,7 @@ playground_usage() {
   printf "%s\n" "$(bold "Tools commands:")"
   printf "  %s   🧰 Tools commands\n" "$(green "tools")                                    "
   printf "  %s   🔐 Read provided avro file\n" "$(green "tools read-avro-file")                     "
+  printf "  %s   🔐 Read provided parquet file\n" "$(green "tools read-parquet-file")                  "
   echo
   printf "%s\n" "$(bold "Debug commands:")"
   printf "  %s   🐞 Debug commands\n" "$(green "debug")                                    "
@@ -3288,7 +3289,8 @@ playground_tools_usage() {
   echo
   # :command.usage_commands
   printf "%s\n" "$(bold "== Commands ==")"
-  printf "  %s   🔐 Read provided avro file\n" "$(green "read-avro-file")"
+  printf "  %s   🔐 Read provided avro file\n" "$(green "read-avro-file")   "
+  printf "  %s   🔐 Read provided parquet file\n" "$(green "read-parquet-file")"
   echo
 
   # :command.long_usage
@@ -3338,6 +3340,41 @@ playground_tools_read_avro_file_usage() {
   fi
 }
 
+# :command.usage
+playground_tools_read_parquet_file_usage() {
+  if [[ -n $long_usage ]]; then
+    printf "playground tools read-parquet-file - 🔐 Read provided parquet file\n"
+    echo
+
+  else
+    printf "playground tools read-parquet-file - 🔐 Read provided parquet file\n"
+    echo
+
+  fi
+
+  printf "%s\n" "$(bold "== Usage ==")"
+  printf "  playground tools read-parquet-file [OPTIONS]\n"
+  printf "  playground tools read-parquet-file --help | -h\n"
+  echo
+
+  # :command.long_usage
+  if [[ -n $long_usage ]]; then
+    printf "%s\n" "$(bold "== Options ==")"
+
+    # :command.usage_flags
+    # :flag.usage
+    printf "  %s\n" "$(magenta "--file, -f FILE")"
+    printf "    🔖 Parquet file to read\n    \n    ❕ It must be absolute full path\n    \n    🎓 Tip: use <tab> completion to trigger fzf completion\n"
+    echo
+
+    # :command.usage_fixed_flags
+    printf "  %s\n" "$(magenta "--help, -h")"
+    printf "    Show this help\n"
+    echo
+
+  fi
+}
+
 # :command.usage
 playground_debug_usage() {
   if [[ -n $long_usage ]]; then
@@ -12211,6 +12248,20 @@ validate_file_exists_and_avro() {
   [[ "$real_file" == *.avro ]] || logerror "<$real_file> is not an Avro file. Please provide a file with .avro extension."
 }
 
+# src/lib/validations/validate_file_exists_and_parquet.sh
+validate_file_exists_and_parquet() {
+  file="$1"
+
+  real_file=$file
+  if [[ $file == *"@"* ]]
+  then
+    real_file=$(echo "$file" | cut -d "@" -f 2)
+  fi
+
+  [[ -f "$real_file" ]] || logerror "<$real_file> does not correspond to the path of an existing file, please make sure to use absolute full path or correct relative path !"
+  [[ "$real_file" == *.parquet ]] || logerror "<$real_file> is not an Parquet file. Please provide a file with .parquet extension."
+}
+
 # src/lib/validations/validate_file_exists_with_trick.sh
 validate_file_exists_with_trick() {
   file="$1"
@@ -18443,6 +18494,26 @@ playground_tools_read_avro_file_command() {
 
 }
 
+# :command.function
+playground_tools_read_parquet_file_command() {
+  # src/commands/tools/read-parquet-file.sh
+  file="${args[--file]}"
+
+  if [[ $file == *"@"* ]]
+  then
+    file=$(echo "$file" | cut -d "@" -f 2)
+  fi
+
+  filename=$(basename $file)
+
+  log "🔖 ${filename}.parquet schema"
+  docker run --quiet --rm -v ${file}:/tmp/${filename} nathanhowell/parquet-tools schema /tmp/${filename}
+
+  log "🔖 ${filename}.parquet content"
+  docker run --quiet --rm -v ${file}:/tmp/${filename} nathanhowell/parquet-tools cat /tmp/${filename}
+
+}
+
 # :command.function
 playground_debug_enable_remote_debugging_command() {
   # src/commands/debug/enable-remote-debugging.sh
@@ -30303,6 +30374,13 @@ playground_tools_parse_requirements() {
       shift $#
       ;;
 
+    read-parquet-file)
+      action="read-parquet-file"
+      shift
+      playground_tools_read_parquet_file_parse_requirements "$@"
+      shift $#
+      ;;
+
     # :command.command_fallback
     "")
       playground_tools_usage >&2
@@ -30403,6 +30481,70 @@ playground_tools_read_avro_file_parse_requirements() {
 
 }
 
+# :command.parse_requirements
+playground_tools_read_parquet_file_parse_requirements() {
+  # :command.fixed_flags_filter
+  while [[ $# -gt 0 ]]; do
+    case "${1:-}" in
+      --help | -h)
+        long_usage=yes
+        playground_tools_read_parquet_file_usage
+        exit
+        ;;
+
+      *)
+        break
+        ;;
+
+    esac
+  done
+
+  # :command.command_filter
+  action="tools read-parquet-file"
+
+  # :command.parse_requirements_while
+  while [[ $# -gt 0 ]]; do
+    key="$1"
+    case "$key" in
+      # :flag.case
+      --file | -f)
+
+        # :flag.case_arg
+        if [[ -n ${2+x} ]]; then
+          args['--file']="$2"
+          shift
+          shift
+        else
+          printf "%s\n" "--file requires an argument: --file, -f FILE" >&2
+          exit 1
+        fi
+        ;;
+
+      -?*)
+        printf "invalid option: %s\n" "$key" >&2
+        exit 1
+        ;;
+
+      *)
+        # :command.parse_requirements_case
+        # :command.parse_requirements_case_simple
+        printf "invalid argument: %s\n" "$key" >&2
+        exit 1
+
+        ;;
+
+    esac
+  done
+
+  # :command.validations
+  # :flag.validations
+  if [[ -v args['--file'] && -n $(validate_file_exists_and_parquet "${args['--file']:-}") ]]; then
+    printf "validation error in %s:\n%s\n" "--file, -f FILE" "$(validate_file_exists_and_parquet "${args['--file']:-}")" >&2
+    exit 1
+  fi
+
+}
+
 # :command.parse_requirements
 playground_debug_parse_requirements() {
   # :command.fixed_flags_filter
@@ -36288,6 +36430,7 @@ run() {
     "tcp-proxy toggle-writes-service") playground_tcp_proxy_toggle_writes_service_command ;;
     "tools") playground_tools_command ;;
     "tools read-avro-file") playground_tools_read_avro_file_command ;;
+    "tools read-parquet-file") playground_tools_read_parquet_file_command ;;
     "debug") playground_debug_command ;;
     "debug enable-remote-debugging") playground_debug_enable_remote_debugging_command ;;
     "debug testssl") playground_debug_testssl_command ;;

diff --git a/scripts/cli/playground.json b/scripts/cli/playground.json
@@ -959,6 +959,21 @@
               "description": "🔖 Avro file to read\n\n❕ It must be absolute full path\n\n🎓 Tip: use <tab> completion to trigger fzf completion\n"
             }
           ]
+        },
+        {
+          "name": "read-parquet-file",
+          "description": "🔐 Read provided parquet file\n",
+          "usage": "playground tools read-parquet-file [OPTIONS]",
+          "options": [
+            {
+              "names": [
+                "--file",
+                "-f"
+              ],
+              "argument": "FILE",
+              "description": "🔖 Parquet file to read\n\n❕ It must be absolute full path\n\n🎓 Tip: use <tab> completion to trigger fzf completion\n"
+            }
+          ]
         }
       ]
     },

diff --git a/scripts/cli/playground.yaml b/scripts/cli/playground.yaml
@@ -1086,6 +1086,22 @@ subcommands:
             
             🎓 Tip: use <tab> completion to trigger fzf completion
             
+      - name: read-parquet-file
+        description: | 
+          🔐 Read provided parquet file
+        usage: playground tools read-parquet-file [OPTIONS]
+        options:
+        - names:
+          - --file
+          - -f
+          argument: FILE
+          description: | 
+            🔖 Parquet file to read
+            
+            ❕ It must be absolute full path
+            
+            🎓 Tip: use <tab> completion to trigger fzf completion
+            
   - name: debug
     description: | 
         🐞 Debug commands

diff --git a/scripts/cli/src/bashly.yml b/scripts/cli/src/bashly.yml
@@ -1459,6 +1459,24 @@ commands:
 
         🎓 Tip: use <tab> completion to trigger fzf completion
 
+  - name: read-parquet-file
+    help: |-
+      🔐 Read provided parquet file
+    flags:
+    - long: --file
+      short: -f
+      required: false
+      completions:
+        - $(playground get-specific-file-extension "$cur" --extension "parquet")
+      validate: file_exists_and_parquet
+      arg: file
+      help: |-
+        🔖 Parquet file to read
+
+        ❕ It must be absolute full path
+
+        🎓 Tip: use <tab> completion to trigger fzf completion
+
 - name: debug
   expose: always
   group: Debug

diff --git a/scripts/cli/src/commands/tools/read-parquet-file.sh b/scripts/cli/src/commands/tools/read-parquet-file.sh
@@ -0,0 +1,14 @@
+file="${args[--file]}"
+
+if [[ $file == *"@"* ]]
+then
+  file=$(echo "$file" | cut -d "@" -f 2)
+fi
+
+filename=$(basename $file)
+
+log "🔖 ${filename}.parquet schema"
+docker run --quiet --rm -v ${file}:/tmp/${filename} nathanhowell/parquet-tools schema /tmp/${filename}
+
+log "🔖 ${filename}.parquet content"
+docker run --quiet --rm -v ${file}:/tmp/${filename} nathanhowell/parquet-tools cat /tmp/${filename}
diff --git a/scripts/cli/src/lib/validations/validate_file_exists_and_parquet.sh b/scripts/cli/src/lib/validations/validate_file_exists_and_parquet.sh
@@ -0,0 +1,13 @@
+## [@bashly-upgrade validations]
+validate_file_exists_and_parquet() {
+  file="$1"
+
+  real_file=$file
+  if [[ $file == *"@"* ]]
+  then
+    real_file=$(echo "$file" | cut -d "@" -f 2)
+  fi
+
+  [[ -f "$real_file" ]] || logerror "<$real_file> does not correspond to the path of an existing file, please make sure to use absolute full path or correct relative path !"
+  [[ "$real_file" == *.parquet ]] || logerror "<$real_file> is not an Parquet file. Please provide a file with .parquet extension."
+}