david.reindl/firefly-import-preprocessor

Fork 0

Reindl David (IT-PTR-CEN2-SL10) ec95ea26c3 additional, improved transformers / skipIf to eliminate records

2026-05-06 23:17:54 +02:00

26 KiB

Raw Permalink Blame History

Firefly Import Preprocessor

Version: 1.0.0
Date: 03 May 2026
Status: Production Ready

🌐 Deutsch

Overview
Installation & Setup
Quick Start
Configuration
Transformation Types
CLI Reference
Debug Mode
Firefly III Integration
Architecture
Error Handling

Overview

The Firefly Import Preprocessor is a production-ready PHP preprocessor for bank CSV export files. It transforms bank data into a standardised format and can optionally import it into Firefly III.

Core Features

✅ Full CSV transformation with complex pipelines
✅ Metadata extraction via regex (IBAN, currency, account name)
✅ 14 transformation types for flexible data processing
✅ Firefly III integration — CLI, Docker, and HTTP upload
✅ Debug mode for full processing transparency
✅ Production ready with complete error handling
✅ Zero dependencies for core functionality

Workflow

Input CSV
    ↓
Extract metadata (regex)
    ↓
Transform data rows (pipeline)
    ↓
Write output CSV
    ↓
[Optional] Import into Firefly III

Installation & Setup

Requirements

PHP 8.1+
Composer (recommended)
[Optional] Docker for Firefly III integration

Installation

# 1. Clone / copy the repository
cd ff-imp-preprocessor

# 2. Install dependencies (optional, dev tools only)
composer install

# 3. Create configuration
cp config/config.example.json config/config.json
# Edit config/config.json with your settings

# 4. Create directories
mkdir -p config/import/{source,output,archive,error}
chmod 755 config/import/{source,output,archive,error}

# 5. Run a test
php bin/transformer.php validate config/config.json input.csv

Quick Start

1. Adjust configuration

Edit config/config.json and make sure the extraction rules match your CSV format:

{
  "metadata": {
    "extractionRules": [
      {
        "name": "account_iban",
        "lineNumber": 2,
        "regex": "IBAN:\\s*([A-Z0-9 ]+)",
        "captureGroup": 1
      }
    ]
  },
  "csvStructure": {
    "headerLine": 5,
    "delimiter": ";",
    "encoding": "UTF-8"
  }
}

2. Validate CSV

php bin/transformer.php validate config/config.json input.csv

3. Run transformation

php bin/transformer.php transform input.csv config/config.json

# With debug mode for troubleshooting
php bin/transformer.php transform input.csv config/config.json --debug

4. Inspect output

php bin/transformer.php test input.csv config/config.json --debug
# Shows up to 10 transformed rows and debug logs

Configuration

config.json structure

`metadata` — Metadata extraction

{
  "metadata": {
    "extractionRules": [
      {
        "name": "account_iban",
        "lineNumber": 2,
        "regex": "IBAN:\\s*([A-Z0-9 ]+)",
        "captureGroup": 1
      },
      {
        "name": "currency_code",
        "lineNumber": 3,
        "regex": "Currency:\\s*([A-Z]{3})",
        "captureGroup": 1
      }
    ]
  }
}

Field	Type	Description
`name`	string	Name of the metadata variable (used in `constantvalue`)
`lineNumber`	int	Line number in CSV (1-based, human-readable)
`regex`	string	Regex pattern for extraction (without delimiters)
`captureGroup`	int	Capture group index (0 = full match, 1 = first group, etc.)

Regex example:

Pattern: IBAN:\s*([A-Z0-9 ]+)
Input: IBAN: CH93 0077 2020 6262 5252 7
Capture group 1: CH93 0077 2020 6262 5252 7

`csvStructure` — CSV format

{
  "csvStructure": {
    "headerLine": 5,
    "delimiter": ";",
    "encoding": "UTF-8",
    "hasBom": false
  }
}

Field	Type	Default	Description
`headerLine`	int	5	Line number of the header row (1-based)
`delimiter`	string	`;`	CSV delimiter
`encoding`	string	`UTF-8`	Character encoding (UTF-8, ISO-8859-1, CP1252)
`hasBom`	bool	false	Whether the file has a BOM (Byte Order Mark)

`columnTransformations` — Column transformations

{
  "columnTransformations": [
    {
      "sourceColumn": "BookingDate",
      "transformations": [
        {
          "type": "dateformat",
          "fromFormat": "d.m.Y",
          "toFormat": "Y-m-d"
        }
      ],
      "outputColumn": "date",
      "outputAction": "overwrite"
    }
  ]
}

outputAction:

Value	Behaviour
`overwrite`	Replace the target column with the transformation result (default)
`create`	Write the result into a new output column
`append`	Concatenate the result to the end of the existing column value. Add `"appendDelimiter": " "` (any string) to insert a separator between the existing and new value — the delimiter is omitted when the target column is still empty
`append-if-not-empty`	Same as `append` (including optional `appendDelimiter`) but skips entirely when the transformation result is empty — safe for optional values such as tags or notes lines
`append-line`	Same as `append` but the separator is always a newline `\n`; no leading newline when the target is empty
`overwrite-if-empty`	Only write the result if the target column is currently empty
`overwrite-if-not-empty`	Only write the result if the transformation result is not empty

`directories` — File system

{
  "directories": {
    "source": "/opt/ff-imp-preprocessor/import/source",
    "output": "/opt/ff-imp-preprocessor/import/output",
    "archive": "/opt/ff-imp-preprocessor/import/archive",
    "error": "/opt/ff-imp-preprocessor/import/error"
  }
}

Field	Description
`source`	Input directory
`output`	Output directory
`archive`	Archive for processed files
`error`	Error directory for invalid files

`fireflyImport` — Firefly III integration

Optional. When present, passing --do-import to the transform command (or using auto-import) will call the Firefly III Data Importer after the output CSV is written.

See Firefly III Integration for the full field reference and mode-specific examples.

Transformation Types

There are 14 supported transformation types that can be combined as a pipeline:

1. trim — Remove whitespace

Removes leading and trailing whitespace.

{ "type": "trim" }

Input: Coop Pronto → Output: Coop Pronto

2. lowercase — Convert to lowercase

Converts to lowercase (UTF-8 safe).

{ "type": "lowercase" }

Input: COOP PRONTO CHUR → Output: coop pronto chur

3. uppercase — Convert to uppercase

Converts to uppercase (UTF-8 safe).

{ "type": "uppercase" }

Input: Coop Pronto Chur → Output: COOP PRONTO CHUR

4. ucwordsfirst — Capitalise after word separators

Capitalises the first letter after each word separator.

{ "type": "ucwordsfirst" }

COOP PRONTO CHUR → Coop Pronto Chur
migros-rail city → Migros-Rail City
O'NEILL STORE → O'Neill Store
SAINT-JEAN-DE-MAURIENNE → Saint-Jean-De-Maurienne

Separators: space, hyphen, apostrophe, slash, period, comma, semicolon, colon, parentheses.

Guard: If the input already contains both uppercase and lowercase letters (mixed-case), it is returned unchanged. This prevents accidentally re-casing intentionally formatted strings such as "Coop pronto chur". Fully uppercase or fully lowercase inputs are always processed.

5. replace — String replacement

Replaces a substring with another string (case-sensitive).

{ "type": "replace", "search": "  ", "replace": " " }

Input: Coop Pronto (two spaces) → Output: Coop Pronto (one space)

6. split — Split column

Splits a value at a delimiter and keeps a defined part.

{ "type": "split", "delimiter": ";", "part": 0 }

Input: Coop Pronto Chur;7007 Chur → Output: Coop Pronto Chur

7. regex — Regex replacement

Replaces parts of a string using a regular expression. Uses PHP preg_replace.

{ "type": "regex", "pattern": "^(.*?);.*$", "replace": "$1" }

No match → original value is passed through unchanged (pipeline-safe).

Use capture groups as $1, $2, … in the replace field. A pattern without ^/$ anchors replaces only the matched portion, not the whole value.

8. regexextract — Regex extraction

Extracts a capture group and returns only that. Uses PHP preg_match.

{ "type": "regexextract", "pattern": "(\\d{4,} [^;]+)" }

Input: Coop Pronto Chur, 7007 Chur → Output: 7007 Chur
No match → empty string

⚠ Not pipeline-safe: A no-match discards all previous pipeline results. Use regex instead if you want to preserve the current value on no-match.

9. dateformat — Date reformatting

Converts between date formats.

{ "type": "dateformat", "fromFormat": "d.m.Y", "toFormat": "Y-m-d" }

Input: 10.12.2025 → Output: 2025-12-10

Supports all PHP DateTime format characters.

10. truncate — Truncate string

Truncates a string to a maximum length.

{ "type": "truncate", "maxLength": 100 }

11. constantvalue — Constant value from metadata

Injects an extracted metadata value as a constant for every row.

{
  "sourceColumn": "_constant_",
  "transformations": [
    { "type": "constantvalue", "metadataKey": "account_iban" }
  ],
  "outputColumn": "account_iban",
  "outputAction": "create"
}

Every row receives the extracted account_iban value (e.g. CH9300777222666888999) in a new column.

12. map — Copy / rename column

Copies a column value as-is (optionally to a new name).

{ "type": "map" }

13. pipeline — Nested pipeline

Runs a sub-pipeline as a single transformation step.

{
  "type": "pipeline",
  "steps": [
    { "type": "trim" },
    { "type": "lowercase" },
    { "type": "ucwordsfirst" }
  ]
}

Useful for grouping steps as a logical unit within a transformations array.

14. timeperiod — Map time to a period label

Parses a time string and returns the label of the matching period range. Supports midnight-spanning ranges (e.g. 22:00–03:59). Returns default (empty string by default) when no range matches or the input is invalid.

{
  "type": "timeperiod",
  "timeFormat": "H:i:s",
  "periods": [
    { "from": "04:00:00", "to": "08:59:59", "label": "Morgen" },
    { "from": "09:00:00", "to": "10:59:59", "label": "Vormittag" },
    { "from": "11:00:00", "to": "13:59:59", "label": "Mittag" },
    { "from": "14:00:00", "to": "17:59:59", "label": "Nachmittag" },
    { "from": "18:00:00", "to": "21:59:59", "label": "Abend" },
    { "from": "22:00:00", "to": "03:59:59", "label": "Nacht" }
  ],
  "default": ""
}

"09:30:00" → "Vormittag"
"23:00:00" → "Nacht" (midnight-spanning range)
"02:00:00" → "Nacht" (midnight-spanning range)
"" or unparseable input → ""

timeFormat follows PHP's DateTime::createFromFormat syntax (default H:i:s).

Row filtering — `skipIf`

Rows can be excluded from the output by adding a top-level skipIf key to the configuration. The value is a filter node — either a bare condition or a nested and/or group.

Bare condition:

"skipIf": { "column": "Buchungstext", "operator": "equals", "value": "Saldovortrag" }

AND group:

"skipIf": {
  "and": [
    { "column": "Beschreibung1", "operator": "empty" },
    { "column": "Beschreibung2", "operator": "empty" }
  ]
}

Nested AND / OR:

"skipIf": {
  "or": [
    { "column": "Amount", "operator": "gt", "value": "10000" },
    {
      "and": [
        { "column": "Type", "operator": "equals", "value": "Saldo" },
        { "column": "Notes", "operator": "empty" }
      ]
    }
  ]
}

Supported operators:

Operator	Matches when…
`empty`	column value is empty string
`not-empty`	column value is not empty
`equals`	column value equals `"value"`
`not-equals`	column value does not equal `"value"`
`contains`	column value contains `"value"`
`not-contains`	column value does not contain `"value"`
`matches`	column value matches regex `"pattern"`
`not-matches`	column value does not match regex `"pattern"`
`gt`	`(float) column > (float) value`
`gte`	`(float) column >= (float) value`
`lt`	`(float) column < (float) value`
`lte`	`(float) column <= (float) value`

Pipeline example

Multiple transformations chained:

{
  "sourceColumn": "BookingText",
  "transformations": [
    { "type": "trim" },
    { "type": "replace", "search": "  ", "replace": " " },
    { "type": "lowercase" },
    { "type": "ucwordsfirst" }
  ],
  "outputColumn": "description",
  "outputAction": "overwrite"
}

Processing:

" COOP PRONTO " → trim → "COOP PRONTO"
"COOP PRONTO" → replace → "COOP PRONTO"
"COOP PRONTO" → lowercase → "coop pronto"
"coop pronto" → ucwordsfirst → "Coop Pronto"

CLI Reference

php bin/transformer.php <command> [input] [config] [options]

Commands

Command	Description
`test`	Test run (up to 10 rows)
`transform`	Full transformation
`validate`	Validate configuration
`auto-import`	Directory monitoring
`help`	Show help

Options

Option	Description
`--debug`, `-d`	Enable debug mode
`--rows=N`	Max. N rows (`test` command)
`--output=FILE`, `-o`	Output path
`--do-import`	Import into Firefly III after transformation (`transform` only)
`--strict`	Strict validation
`--watch`	Continuous monitoring
`--interval=SEC`	Check interval in seconds (default: 60)
`--dry-run`	Simulation mode, no real operations

Debug Mode

php bin/transformer.php test input.csv config/config.json --debug

Log categories

Category	When
`transformer`	Start/end of transformation
`csv_reader`	While reading CSV
`metadata`	During metadata extraction
`metadata_warning`	On extraction problems
`transformation`	For each transformation step
`csv_writer`	While writing output CSV

Debug log output (JSON)

{
  "success": true,
  "debug_logs": [
    {
      "timestamp": 1702200120.5432,
      "category": "transformer",
      "message": "Transformation started",
      "data": { "inputFile": "input.csv", "maxRows": 0 }
    },
    {
      "timestamp": 1702200120.5445,
      "category": "metadata",
      "message": "Extraction rule applied",
      "data": { "rule_name": "account_iban", "value": "CH93..." }
    }
  ]
}

Firefly III Integration

The transformer can automatically import transformed files into Firefly III. Three operating modes cover all typical deployment scenarios.

Prerequisites (all modes)

1. Create a Firefly III Data Importer JSON configuration file

This file maps transformed CSV columns to Firefly III transaction fields (format v3).

Recommended approach: upload a sample CSV once in the Firefly III Data Importer Web UI, configure the column mapping there, then download the finished configuration. Alternatively, use config/firefly-import-config.example.json as a template and adjust default_account to your asset account ID.

2. Choose an operating mode — see sections below.

`fireflyImport` field reference

Field	Type	Description
`mode`	string	Operating mode: `cli` \| `docker` \| `http` (default: `cli`)
`jsonConfig`	string	Path to the Firefly III Data Importer JSON config file (format v3). For `cli` and `http` modes the file must exist locally; relative paths are resolved from the working directory where `php bin/transformer.php` is invoked (typically the project root). For `docker` mode the path is inside the container — local existence is not checked.
`importerCommand`	string	Full CLI command (modes: cli, docker)
`importerUrl`	string	URL of the Data Importer (mode: http)
`personalSecret`	string	The `AUTO_IMPORT_SECRET` set on the importer server (min. 16 chars). Sent as `?secret=` URL query parameter. (mode: http)
`accessToken`	string	Firefly III Personal Access Token. Sent as `Authorization: Bearer` header. Required if not already set as `FIREFLY_III_ACCESS_TOKEN` in the importer environment. (mode: http)
`deleteAfterImport`	boolean	Delete transformed CSV after successful import
`chunkSize`	integer	Split the CSV into chunks of at most N data rows and import each chunk as a separate request. Prevents server-side timeouts on large files (rule of thumb: ~3–4 s/transaction for HTTP mode). `0` or absent = no chunking (default). Applies to all modes.
`chunkRetries`	integer	Number of additional import attempts per chunk after the first. On failure the importer retries up to this many times before aborting. `0` or absent = no retry (default). Only effective when `chunkSize > 0`.
`chunkRetryDelay`	integer	Pause in seconds before each chunk request after the first, and between retry attempts for the same failed chunk. Addresses both inter-chunk cooldown and retry back-off. `0` or absent = no pause (default). Only effective when `chunkSize > 0`.
`connectionTimeout`	integer	Maximum seconds to wait for the TCP connection to the importer to be established. Distinct from `timeout` (full transfer duration). Default: `10`. (mode: http only)
`timeout`	integer	Timeout in seconds per request (default: 300). For chunked imports this applies per chunk, not for the total run.
`environment`	object	Additional environment variables (modes: cli, docker)

Mode `cli` — Transformer and Firefly on the same server

Both the transformer and the Firefly III Data Importer run on the same server. The transformer calls the importer directly as a local command.

"fireflyImport": {
  "mode": "cli",
  "jsonConfig": "/opt/firefly-data-importer/storage/configurations/ubs-import.json",
  "importerCommand": "php /opt/firefly-data-importer/artisan importer:import",
  "deleteAfterImport": false,
  "chunkSize": 50,
  "chunkRetries": 3,
  "chunkRetryDelay": 10,
  "timeout": 300,
  "environment": {
    "FIREFLY_III_URL": "https://localhost",
    "FIREFLY_III_ACCESS_TOKEN": "your-token-here"
  }
}

Mode `docker` — Transformer local, Firefly in Docker

The transformer runs locally or in its own container; the Firefly III Data Importer runs in a Docker container. The transformer calls the importer via docker exec.

Important: The transformer's output directory must be mounted as a volume in the importer container. jsonConfig is the path inside the container (not a local path). Do not use the -it flag (no TTY).

Example docker-compose.yml for the importer:

services:
  firefly-importer:
    image: fireflyiii/data-importer:latest
    volumes:
      - /opt/ff-imp-preprocessor/import:/import
    environment:
      - FIREFLY_III_URL=https://your-firefly.com
      - FIREFLY_III_ACCESS_TOKEN=your-token-here
      - CAN_POST_FILES=false

"fireflyImport": {
  "mode": "docker",
  "jsonConfig": "/import/configs/ubs-import.json",
  "importerCommand": "docker exec firefly-importer php artisan importer:import",
  "deleteAfterImport": false,
  "chunkSize": 50,
  "chunkRetries": 3,
  "chunkRetryDelay": 10,
  "timeout": 300
}

The JSON config file must be available inside the container — either via a volume mount or docker cp:

docker cp ubs-import.json firefly-importer:/import/configs/ubs-import.json

Mode `http` — Transformer local, Firefly importer on a remote server

The transformer runs locally; the Firefly III Data Importer is reachable over HTTP(S). The CSV and JSON configuration are sent as a multipart HTTP upload to the importer.

Requirements on the importer server:

CAN_POST_FILES=true
AUTO_IMPORT_SECRET=<secret>  # at least 16 characters — set this as personalSecret in your config

Local requirement: PHP extension ext-curl

"fireflyImport": {
  "mode": "http",
  "importerUrl": "https://importer.your-server.com",
  "personalSecret": "your-auto-import-secret-min-16-chars",
  "accessToken": "your-firefly-iii-personal-access-token",
  "jsonConfig": "config/ubs-import.json",
  "deleteAfterImport": false,
  "chunkSize": 50,
  "chunkRetries": 3,
  "chunkRetryDelay": 10,
  "connectionTimeout": 10,
  "timeout": 300
}

The transformer sends a POST request to {importerUrl}/autoupload?secret={personalSecret} with the CSV and JSON config as multipart form fields. The accessToken is sent as Authorization: Bearer. If FIREFLY_III_ACCESS_TOKEN is already set in the importer's environment, accessToken can be omitted.

Server-side tuning

For large imports the bottleneck is usually the Firefly III Data Importer server, not the transformer. The settings below belong in the importer's environment (.env or docker-compose.yml):

Setting	Recommended value	Notes
`PHP_MEMORY_LIMIT`	`512M` – `2048M`	Docker env var. Raise when PHP crashes with "Allowed memory size exhausted".
`CONNECTION_TIMEOUT`	`60`	Seconds to wait for TCP connect to Firefly III API. Default is ~31 s (π × 10).
`IGNORE_DUPLICATE_ERRORS`	`true`	Suppress duplicate-transaction warnings on repeated imports.

nginx reverse proxy (if applicable):

proxy_read_timeout  600s;   # must exceed the longest single-chunk import time
client_max_body_size 64M;   # must accommodate your largest chunk CSV

Docker Compose example:

services:
  firefly-importer:
    environment:
      - PHP_MEMORY_LIMIT=1024M
      - CONNECTION_TIMEOUT=60
      - IGNORE_DUPLICATE_ERRORS=true

Usage

# Transform only (no import)
php bin/transformer.php transform input.csv config/config.json

# Transform and import into Firefly III
php bin/transformer.php transform input.csv config/config.json --do-import

# Watch mode: transform and import automatically for each new CSV in source directory
php bin/transformer.php auto-import config/config.json --watch

Architecture

Components

bin/transformer.php (CLI entry point)
  ↓
TransformerEngine (orchestration)
  ├─ ConfigurationLoader (load / validate config)
  ├─ CsvReader (read CSV)
  ├─ MetadataExtractor (metadata via regex)
  ├─ ColumnTransformer (apply transformations)
  ├─ CsvWriter (write CSV)
  ├─ FireflyImporter (Firefly III integration)
  └─ DebugLogger (debug logs)

Data flow

Input CSV
  ↓
CsvReader::readMetadataLines() → array of lines
  ↓
MetadataExtractor::extract() → {iban: "...", currency: "..."}
  ↓
CsvReader::readCsvData() → array of rows
  ↓
ColumnTransformer::transformRow() → transformed row (pipeline)
  ↓
CsvWriter::write() → output CSV

Classes

Class	Responsibility
`TransformerEngine`	Orchestrates the entire workflow
`ConfigurationLoader`	Loads and validates JSON configuration
`CsvReader`	Reads CSV with metadata support
`MetadataExtractor`	Extracts metadata via regex
`ColumnTransformer`	Transforms columns (pipeline)
`CsvWriter`	Writes output CSV
`FireflyImporter`	Imports into Firefly III
`DebugLogger`	Static logger for debug output

Error Handling

Common errors

"Input file not found"

# Check the file path
ls -la input.csv

# Use an absolute path if relative paths do not work
php bin/transformer.php transform /absolute/path/input.csv config.json

"Missing metadata: account_iban"

The IBAN could not be extracted — wrong regex or wrong line number.

# Inspect the first lines of the CSV
head -5 input.csv

# Validate with debug output
php bin/transformer.php validate config.json input.csv --debug

"Invalid JSON: …"

Syntax error in config.json.

php -r "json_decode(file_get_contents('config/config.json'), true) or die('JSON invalid');"

"Configuration: 'csvStructure.headerLine' required"

A required configuration field is missing.

diff config/config.json config/config.example.json

Exception handling

try {
    $result = $engine->transform($inputFile);
    if (!$result['success']) {
        echo "Error: " . $result['error'];
    }
} catch (Exception $e) {
    echo "Fatal error: " . $e->getMessage();
}

Tips

UTF-8 handling

The transformer uses UTF-8 safe functions throughout:

mb_strtolower() instead of strtolower()
mb_strtoupper() instead of strtoupper()
mb_strlen() for correct character counting

Supported encodings: UTF-8, ISO-8859-1, CP1252.

Regex tips

Pattern without delimiters (auto-wrapped):

"pattern": "IBAN:\\s*([A-Z0-9 ]+)"
// becomes: /IBAN:\s*([A-Z0-9 ]+)/u

With explicit flags:

"pattern": "/IBAN:\\s*([A-Z0-9 ]+)/iu"
// case-insensitive

Performance

Optimised for: up to 1 million rows
Typical file size: 10–100 k rows

Batch processing

#!/bin/bash
for file in import/source/*.csv; do
    php bin/transformer.php transform "$file" config/config.json
    if [ $? -eq 0 ]; then
        mv "$file" import/archive/
    else
        mv "$file" import/error/
    fi
done

Version History

v1.0.0 (03 May 2026)

✅ Initial release
✅ 14 transformation types
✅ Metadata extraction via regex
✅ Debug mode
✅ Firefly III integration (cli / docker / http)
✅ Full documentation

License: GPL-3.0
Author: PHP CSV Transformer Project
Repository: git.andare.ch/david.reindl/ff-imp-preprocessor

26 KiB Raw Permalink Blame History Unescape Escape