mbox2csv 0.1.2 → 0.2.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 4075cf789bd3819e1642195c3787a2406ce5f04c3cea31345bb2a76a2783a960
-  data.tar.gz: dd347f1302b8b90363b1b975753c2eda79a52d09ce90bf620df8946e37ae3363
+  metadata.gz: d01ca30a89acfe78cea4db6c1b61ee76a10ee9f268ffc51c80ed15136c98b279
+  data.tar.gz: 295a46bbcee3b62796febe837c145a002df6ad03dcbf1f2e9d3b3a455a46f3c7
 SHA512:
-  metadata.gz: 93d4bf15e4932ab1bd47c6553b1cd3136d5763dee587ce3bcf06a607910770707be5d4c8c7a21017c0469acb7725fccff037376bf0bdadd42e0ee8e06ba84bb5
-  data.tar.gz: 0fc729d86a2d57552a98eab459a78bd7d0472853ede8aeb3488f9f110daed88a38c9161ff497262b38da3f5558897993335b6bbe8d331c7848415f488e308353
+  metadata.gz: e17144ea47a735f13d00070bca37c2e7fbbecd8e91685abb943628fc08f2e1b68e845864b69492fc333b3e72c8c07e98e78d662951d6aa4d8ecba54b6fe1dfe1
+  data.tar.gz: 833c1c6e20dbb6faa17d27e65db529877b08810eca35288277dc78b54f13f6034ba94318ae5376764613bd1ba0024b54a8705dbf6dcda06f1580c55dfd318132

data/README.md

@@ -34,6 +34,10 @@ parser = Mbox2CSV::MboxParser.new(mbox_file, all_emails, sender_stats_all_emails
 
 # Parse the MBOX file, save email data, and generate statistics
 parser.parse
+
+# Extract attachments
+parser.extract_attachments(extract: true, filetypes: %w[pdf jpg], output_folder: "exports")
+
 ```
 
 ## 🔑 License

data/lib/mbox2csv.rb

@@ -1,12 +1,15 @@
 require 'base64'
 require 'csv'
 require 'mail'
+require 'fileutils'
+require 'ruby-progressbar'
 
 module Mbox2CSV
-    # Main class
+    # Main class for parsing MBOX files, saving email data/statistics to CSV,
+    # and (optionally) extracting selected attachment types to disk.
     class MboxParser
         # Initializes the MboxParser with file paths for the MBOX file, output CSV file,
-        # and statistics CSV files for sender statistics.
+        # and statistics CSV files for sender and recipient statistics.
         #
         # @param [String] mbox_file Path to the MBOX file to be parsed.
         # @param [String] csv_file Path to the output CSV file where parsed email data will be saved.
@@ -18,40 +21,84 @@ module Mbox2CSV
             @statistics = EmailStatistics.new
             @stats_csv_file = stats_csv_file
             @recipient_stats_csv_file = recipient_stats_csv_file
+            @senders_folder = 'senders/'
+            FileUtils.mkdir_p(@senders_folder) # Create the senders folder if it doesn't exist
         end
 
         # Parses the MBOX file and writes the email data to the specified CSV file.
         # It also saves sender and recipient statistics to separate CSV files.
+        # A progress bar is displayed during the parsing process.
         def parse
+            total_lines = File.foreach(@mbox_file).inject(0) { |c, _line| c + 1 }
+            progressbar = ProgressBar.create(title: "Parsing Emails", total: total_lines, format: "%t: |%B| %p%%")
+
             CSV.open(@csv_file, 'w') do |csv|
-                # Write CSV header
                 csv << ['From', 'To', 'Subject', 'Date', 'Body']
 
                 File.open(@mbox_file, 'r') do |mbox|
                     buffer = ""
                     mbox.each_line do |line|
+                        progressbar.increment
                         if line.start_with?("From ")
                             process_email_block(buffer, csv) unless buffer.empty?
-                            buffer = "" # Reset buffer
+                            buffer = ""
                         end
-                        buffer << line # Append line to buffer
+                        buffer << line
                     end
-                    process_email_block(buffer, csv) unless buffer.empty? # Process last email block
+                    process_email_block(buffer, csv) unless buffer.empty?
                 end
             end
             puts "Parsing completed. Data saved to #{@csv_file}"
 
-            # Save and print statistics after parsing
             @statistics.save_sender_statistics(@stats_csv_file)
             @statistics.save_recipient_statistics(@recipient_stats_csv_file)
         rescue => e
             puts "Error processing MBOX file: #{e.message}"
         end
 
+        # Extract selected attachment file types from the MBOX into a folder.
+        #
+        # @param [Boolean] extract         Flag to enable/disable extraction.
+        # @param [Array<String>] filetypes Array of extensions to extract (e.g., %w[pdf jpg png]).
+        # @param [String] output_folder    Directory to write attachments into.
+        # @return [Integer]                Number of files successfully written.
+        def extract_attachments(extract: true, filetypes: [], output_folder: 'attachments')
+            return 0 unless extract
+
+            wanted_exts = Array(filetypes).map { |e| e.to_s.downcase.sub(/\A\./, '') }.uniq
+            raise ArgumentError, "filetypes must not be empty when extract: true" if wanted_exts.empty?
+
+            FileUtils.mkdir_p(output_folder)
+            total_written = 0
+
+            total_lines = File.foreach(@mbox_file).inject(0) { |c, _| c + 1 }
+            progressbar = ProgressBar.create(title: "Extracting Attachments", total: total_lines, format: "%t: |%B| %p%%")
+
+            File.open(@mbox_file, 'r') do |mbox|
+                buffer = ""
+                mbox.each_line do |line|
+                    progressbar.increment
+                    if line.start_with?("From ")
+                        total_written += process_attachment_block(buffer, wanted_exts, output_folder) unless buffer.empty?
+                        buffer = ""
+                    end
+                    buffer << line
+                end
+                total_written += process_attachment_block(buffer, wanted_exts, output_folder) unless buffer.empty?
+            end
+
+            puts "Attachment extraction completed. #{total_written} file(s) saved to #{output_folder}"
+            total_written
+        rescue => e
+            puts "Error extracting attachments: #{e.message}"
+            0
+        end
+
         private
 
         # Processes an individual email block from the MBOX file, extracts the email fields,
-        # and writes them to the CSV. Also records email statistics for analysis.
+        # and writes them to the CSV. Also records email statistics for analysis and creates
+        # sender-specific CSV files.
         #
         # @param [String] buffer The email block from the MBOX file.
         # @param [CSV] csv The CSV object where email data is written.
@@ -62,14 +109,13 @@ module Mbox2CSV
             to = ensure_utf8(mail.to ? mail.to.join(", ") : '', 'UTF-8')
             subject = ensure_utf8(mail.subject ? mail.subject : '', 'UTF-8')
             date = ensure_utf8(mail.date ? mail.date.to_s : '', 'UTF-8')
-
             body = decode_body(mail)
 
-            # Write to CSV
             csv << [from, to, subject, date, body]
 
-            # Record email for statistics
             @statistics.record_email(from, to, body.length)
+
+            save_email_to_sender_csv(from, to, subject, date, body)
         rescue => e
             puts "Error processing email block: #{e.message}"
         end
@@ -79,7 +125,8 @@ module Mbox2CSV
         # @param [Mail] mail The mail object to decode.
         # @return [String] The decoded email body.
         def decode_body(mail)
-            body = if mail.multipart?
+            body =
+                    if mail.multipart?
             part = mail.text_part || mail.html_part
             part&.body&.decoded || ''
         else
@@ -108,15 +155,125 @@ module Mbox2CSV
         text = text.force_encoding(charset) if charset
         text.encode('UTF-8', invalid: :replace, undef: :replace, replace: '?')
     end
+
+    # Saves an email to a sender-specific CSV file.
+    #
+    # @param [String] from The sender of the email.
+    # @param [String] to The recipient(s) of the email.
+    # @param [String] subject The subject of the email.
+    # @param [String] date The date of the email.
+    # @param [String] body The body of the email.
+    def save_email_to_sender_csv(from, to, subject, date, body)
+        return if from.empty?
+
+        sender_file = File.join(@senders_folder, "#{sanitize_filename(from)}.csv")
+
+        CSV.open(sender_file, 'a') do |csv|
+            if File.size?(sender_file).nil? || File.size(sender_file).zero?
+                csv << ['From', 'To', 'Subject', 'Date', 'Body'] # Add header if file is new
+            end
+            csv << [from, to, subject, date, body]
+        end
+    rescue => e
+        puts "Error writing to sender CSV file for #{from}: #{e.message}"
+    end
+
+    # Sanitizes filenames by replacing invalid characters with underscores.
+    #
+    # @param [String] filename The input filename.
+    # @return [String] A sanitized version of the filename.
+    def sanitize_filename(filename)
+        filename.gsub(/[^0-9A-Za-z\-]/, '_')
+    end
+
+    # --- Helpers for attachment extraction ---
+
+    # Process a single email block to extract wanted attachments.
+    def process_attachment_block(buffer, wanted_exts, output_folder)
+        return 0 if buffer.nil? || buffer.empty?
+
+        mail = Mail.read_from_string(buffer)
+        return 0 unless mail
+
+        written = 0
+        date = (mail.date rescue nil)
+        date_str = date ? date.strftime("%Y-%m-%d") : "unknown_date"
+        time_str = date ? date.strftime("%H-%M-%S") : "unknown_time"
+
+        Array(mail.attachments).each do |att|
+            begin
+                original_name = att.filename || att.name || "attachment"
+                base = File.basename(original_name, ".*")
+                ext  = File.extname(original_name).downcase.sub(/\A\./, '')
+
+                # If no ext present, try to infer from MIME type
+                ext = mime_to_ext(att.mime_type) if ext.empty? && att.mime_type
+
+                # Skip if extension not desired
+                next unless wanted_exts.include?(ext.downcase)
+
+                safe_base = sanitize_filename(base)
+                fname = "#{safe_base}_#{date_str}_#{time_str}.#{ext}"
+                path  = File.join(output_folder, fname)
+
+                # Ensure uniqueness if file already exists
+                path = uniquify_path(path)
+
+                # Write decoded content
+                File.open(path, "wb") { |f| f.write(att.body.decoded) }
+                written += 1
+            rescue => e
+                puts "Failed to save attachment '#{att&.filename}': #{e.message}"
+            end
+        end
+
+        written
+    rescue => e
+        puts "Error processing attachment block: #{e.message}"
+        0
+    end
+
+    # Minimal MIME→extension mapping; extend as needed.
+    def mime_to_ext(mime)
+        map = {
+            'application/pdf' => 'pdf',
+            'image/jpeg'      => 'jpg',
+            'image/jpg'       => 'jpg',
+            'image/png'       => 'png',
+            'image/gif'       => 'gif',
+            'text/plain'      => 'txt',
+            'application/zip' => 'zip',
+            'application/vnd.openxmlformats-officedocument.wordprocessingml.document' => 'docx',
+            'application/vnd.openxmlformats-officedocument.spreadsheetml.sheet' => 'xlsx',
+            'application/msword' => 'doc',
+            'application/vnd.ms-excel' => 'xls'
+        }
+        map[mime] || 'bin'
+    end
+
+    # If a path exists, append an incrementing suffix before the extension.
+    def uniquify_path(path)
+        return path unless File.exist?(path)
+        dir  = File.dirname(path)
+        base = File.basename(path, ".*")
+        ext  = File.extname(path)
+        i = 1
+        new_path = File.join(dir, "#{base}_#{i}#{ext}")
+        while File.exist?(new_path)
+            i += 1
+            new_path = File.join(dir, "#{base}_#{i}#{ext}")
+        end
+        new_path
+    end
 end
 
 # The EmailStatistics class is responsible for gathering and writing statistics related to emails.
 # It tracks sender frequency, recipient frequency, and calculates the average email body length per sender.
 class EmailStatistics
     def initialize
-        @sender_counts = Hash.new(0) # Keeps count of emails per sender
-        @recipient_counts = Hash.new(0) # Keeps count of emails per recipient
-        @body_lengths = Hash.new { |hash, key| hash[key] = [] } # Stores body lengths per sender
+        @sender_counts = Hash.new(0)
+        @recipient_counts = Hash.new(0)
+        @body_lengths = Hash.new { |hash, key| hash[key] = [] }
     end
 
     # Records an email's sender, recipients, and body length for statistical purposes.
@@ -177,3 +334,8 @@ class EmailStatistics
     end
 end
 end
+
+# --- Usage example ---
+# parser = Mbox2CSV::MboxParser.new("inbox.mbox", "emails.csv", "sender_stats.csv", "recipient_stats.csv")
+# parser.parse
+# parser.extract_attachments(extract: true, filetypes: %w[pdf jpg], output_folder: "exports")

metadata

@@ -1,14 +1,13 @@
 --- !ruby/object:Gem::Specification
 name: mbox2csv
 version: !ruby/object:Gem::Version
-  version: 0.1.2
+  version: 0.2.1
 platform: ruby
 authors:
 - firefly-cpp
-autorequire:
 bindir: bin
 cert_chain: []
-date: 2025-01-10 00:00:00.000000000 Z
+date: 1980-01-02 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: base64
@@ -52,7 +51,20 @@ dependencies:
     - - "~>"
       - !ruby/object:Gem::Version
         version: 2.8.1
-description:
+- !ruby/object:Gem::Dependency
+  name: ruby-progressbar
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '1.11'
+  type: :runtime
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '1.11'
 email:
 - iztok@iztok-jr-fister.eu
 executables: []
@@ -69,7 +81,6 @@ metadata:
   homepage_uri: https://github.com/firefly-cpp/mbox2csv
   source_code_uri: https://github.com/firefly-cpp/mbox2csv
   changelog_uri: https://github.com/firefly-cpp/mbox2csv
-post_install_message:
 rdoc_options: []
 require_paths:
 - lib
@@ -84,8 +95,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.5.22
-signing_key:
+rubygems_version: 3.6.7
 specification_version: 4
 summary: Parse MBOX files and export email data into CSV format
 test_files: []