Download the PHP package voilab/csv without Composer

CSV parser

This class uses fgetcsv to parse a file or a string, extract columns and provide per-column methods to manipulate data.

It can parse large files, HTTP streams, any types of resources, or strings.

It comes with a basic error handling, so it is possible to collect all errors in the CSV resource and, then, do something with this array of errors.

It is extendable, so you can parse your own type of resource/stream, if you have very special needs.

Table of content

• Install
  • Install PHP5 compatible version
• Usage
  • Available methods
  • Simple example
  • Full example
• Documentation
  • Options
  • Column function parameters
  • Headers auto-sanitization
  • On before column parse function parameters
  • On row parsed function parameters
  • Aliasing columns
  • Required columns
  • No header
  • Shuffling columns when defining them
  • Seek in big files
  • Close the resource
  • Line endings problems
• Error management
  • Initialization errors
  • Error and internationalization (i18n)
• Working with database, column optimization
  • Parse function
  • Reduce function
  • Absent function
  • Example
  • Chunks
• Guessers : auto detect line ending, delimiter and encoding
  • Guess line ending
  • Guess delimiter
  • Guess encoding
• Known issues
• Testing
• Security
• License

Install

Via Composer

Create a composer.json file in your project root:

Install PHP5 compatible version

This PHP5 version can't parse streams nor iterables.

Usage

Available methods

Simple example

Full example

Documentation

Options

These are the options you can provide at constructor level or when calling from* methods. Details for fgetcsv options can be found here: https://php.net/fgetcsv and https://php.net/str_getcsv

Column function parameters

When defining a function for a column, you have access to these parameters:

Headers auto-sanitization

Note that headers are automatically trimmed and their carriage returns are removed. Also, all spaces following a space are removed. This is only for the headers. Cells content are not manipulated, except if autotrim is true.

If the column you defined in your code doesn't exist in CSV resource and doesn't appear in required array, the $meta argument will have a flag phantom set to true. This is the way to know if the column exists or not in the CSV resource during parsing.

On before column parse function parameters

Just before any CSV column data is parsed, a standard method is called so you can operate the same way on every rows and columns data. You can use that to manage encoding, for example.

Be aware of type declaration in your columns functions if you want to return other types from here.

On row parsed function parameters

When a row is completed, you can do something with all that data.

Aliasing columns

You can define aliases for columns to ease data manipulation. Just write as to activate this functionality, like CSV column name as alias.

Alias must not itself contain as string. But in the CSV resource, the header can have such a string.

Note that if you have as in a CSV resource header, you must alias it in the columns definitions. Otherwise, the parser will not find this column.

Required columns

If you have aliased a column, and it is a required column, you must use the alias inside the required option.

No header

If you have no header in you CSV resource, you need to define the parser like this.

Shuffling columns when defining them

You can define your columns in any order you want. You don't need to provide them in the order they appear in the CSV. You just have to match your keys with a header in the CSV resource.

Note that the execution order of the columns are aligned with your code. In the example below, the function A() is called after B(), even if column A appears first in CSV resource.

Seek in big files

You can use the seek mechanism to accelerate parsing big files.

Yon can specify the start index. But it is not mandatory. It is used in the error managment, to know which line bugs, or in the other methods calls, where [$index] is given.

You are responsible for keeping [seek] and [start] snychronized. If you don't, and you have errors, the indexes would be irrelevant.

Close the resource

Using fromString() and fromFile() methods, the resource will be closed automatically. With other from*() methods, you can close the resource by giving the 'close' => true option.

Line endings problems

Just as stated in official documentation, if you have problems with recognition in line endings, you can use the option below to activate auto detect.

$parser->parse($resource, [ 'autoDetectLn' => true ]);

Note that auto detect PHP ini param is not reseted to initial value after the parsing has finished.

When parsing streams (like HTTP response message body), line ending must be specified in the array options.

Error management

You can use the onError option to collect all errors, so you can give a message to the user with all errors in the file you found, in one shot.

You can stop the process of a row by checking the $meta argument. It has a key type which can be row or column. If it's column, you can throw the error and it will call onError again, but with type row. Other columns will be skipped for this row.

If you use an optimizer, you can call an Exception from there too. The key type will then have the value optimizer.

Initialization errors

Some errors are thrown before any line is parsed. You have to take this into account.

Error and internationalization (i18n)

If you want to translate error messages, you can use the onError function with meta['type'] === 'init' to throw the translated message.

Working with database, column optimization

When parsing large set of data, if one column is, for example, a user ID, it's a bad idea to call a find($id) method for each CSV row iteration. It's better to take all column values, and call for a findByIds($ids).

The build-in class Optimizer allows you to define a column this way. It takes three arguments. The first is the function needed to parse value from CSV. The second is a reduce function. It recieves all data of the column, and must return an indexed array.

For example, if you have 2 rows with values a and b, the indexed result of the reduce function would be Array ( a => something, b => something else ).

The third argument is a function called when a value is not found in the reduced function.

Parse function

Same as Column function (see above)

Reduce function

Returns an indexed array. If there's no correspondance between CSV column values and the result of the reduce function, you should not return the missing value. For example, if values are [10, 22], they are used in database query to find users by id, and user ID 22 doesn't exist, the result should be Array ( 10 => User(id=10) )

Absent function

When a value is not found in the reduced result, the default behaviour is to set the value (like there wasn't any reduce function for this row). You can override this by defining the absent function, and do what you want with the value.

If you have defined an error function, it will be called with a type of optimizer (check error management above) if you throw an error from here.

Example

Chunks

Optimizers are good in certain cases, but sometimes you want to parse your data by chunk, maniuplate it, store it, and do it again with the next chunk. You can achieve this with chunks options:

If you use optimizers, $rows will be the resultset optimized.

You don't need to use the array returned by fromString (or alike) because what you did in onChunkParsed is enough.

Guessers : auto detect line ending, delimiter and encoding

Guessing how CSV data is structured (line ending, delimiter or encoding) is a very hasardous task, with so many use-cases it's impossible to rule them all..

This package still provides a way to guess these elements, but if it doesn't fit your needs, you can easily extend or create a new class and manage your specific use-case.

If you want to implement guessing your own way, please read the code base for each guessing interfaces.

Guessing is useless with some CsvInteface implementations. For example, iterables are ignored, since data is already arranged in cells and rows. Be sure it's useful for you before you use guessing features.

Guess line ending

First thing the parser does is to detect which are the line endings. The provided implementation tries to detect line endings among \n, \r and \r\n.

Guess delimiter

Then, the parser tries to detect delimiter. In the provided implementation, an exception is thrown if delimiter is not found or if it's too ambiguous.

Guess encoding

For each cell, encoding auto detection is called. The provided implementation tries to find the current cell encoding, and encode it to the other one given in the constructor.

It is also called for the header row. If you want to encode differently between headers and datas, you can check on $meta['type'] === 'init' in your encode function (check code base).

This guesser is called BEFORE onBeforeColumnParse

Known issues

• with PSR streams, carriage returns are not supported in headers and in cells content
• guessing processes are unlikely to fit your specific needs immediately. Before creating an issue or a PR, try to extends the guess classes and make your own specific adaptations

Testing

Security

If you discover any security related issues, please use the issue tracker.

License

The MIT License (MIT). Please see License File for more information.