---
title: Query CSV data sources
description: Use csv.from() and experimental csv.from() to query CSV data with Flux. Query a CSV string, CSV file, or CSV data from a URL.
url: https://docs.influxdata.com/flux/v0/query-data/csv/
estimated_tokens: 6423
product: Flux
version: v0
---

# Query CSV data sources

Use [`csv.from()`](/flux/v0/stdlib/csv/from/) and [experimental `csv.from()`](/flux/v0/stdlib/experimental/csv/from/) to query CSV data with Flux. Query a CSV string, CSV file, or CSV data from a URL. Import the `csv` or `experimental/csv` package.

-   [CSV parsing modes](#csv-parsing-modes)
-   [Results structure](#results-structure)
-   [Examples](#examples)
    -   [Query an annotated CSV string](#query-an-annotated-csv-string)
    -   [Query a raw CSV string](#query-a-raw-csv-string)
    -   [Query CSV data from a file](#query-csv-data-from-a-file)
    -   [Query CSV data from a URL](#query-csv-data-from-a-url)

## CSV parsing modes

`csv.from()` supports two CSV parsing modes:

-   **annotations**: *(Default)* Use CSV annotations to determine column data types and table grouping.
-   **raw**: Parse all columns as strings and use the first row as the **header row** and all subsequent rows as data.

When using the **annotations** parsing mode, CSV data must include all annotation rows (`#datatype`, `#group`, and `#default`).

## Results structure

The structure of results returned by `csv.from()` depends on the [parsing mode](#csv-parsing-modes) used.

-   **annotations**: `csv.from()` returns a stream of tables grouped by columns defined as `true` in the `#group` annotation row.
-   **raw**: `csv.from()`returns a stream of tables with no grouping (all rows are in a single table). All data is formatted as strings.

## Examples

If just getting started, use the [Flux REPL](/influxdb/cloud/tools/flux-repl/) or the [InfluxDB Data Explorer](/influxdb/cloud/query-data/execute-queries/data-explorer/) to execute Flux queries.

-   [Query an annotated CSV string](#query-an-annotated-csv-string)
-   [Query a raw CSV string](#query-a-raw-csv-string)
-   [Query CSV data from a file](#query-csv-data-from-a-file)
-   [Query CSV data from a URL](#query-csv-data-from-a-url)

### Query an annotated CSV string

1. Import the [`csv` package](/flux/v0/stdlib/csv/).
2. Use [`csv.from()`](/flux/v0/stdlib/csv/from/) and the `csv` parameter to specify the annotated CSV string to query.

#### Query

```js
import "csv"

csvData =
    "
#group,false,false,true,true,true,false,false
#datatype,string,long,string,string,string,long,double
#default,_result,,,,,,
,result,table,dataset,metric,sensorID,timestamp,value
,,0,air-sensors,humidity,TLM0100,1627049400000000000,34.79
,,0,air-sensors,humidity,TLM0100,1627049700000000000,34.65
,,1,air-sensors,humidity,TLM0200,1627049400000000000,35.64
,,1,air-sensors,humidity,TLM0200,1627049700000000000,35.67
,,2,air-sensors,temperature,TLM0100,1627049400000000000,71.84
,,2,air-sensors,temperature,TLM0100,1627049700000000000,71.87
,,3,air-sensors,temperature,TLM0200,1627049400000000000,74.10
,,3,air-sensors,temperature,TLM0200,1627049700000000000,74.17
"

csv.from(csv: csvData)
```

#### Results

| dataset | metric | sensorID | timestamp | value |
| --- | --- | --- | --- | --- |
| air-sensors | humidity | TLM0100 | 1627049400000000000 | 34.79 |
| air-sensors | humidity | TLM0100 | 1627049700000000000 | 34.65 |

| dataset | metric | sensorID | timestamp | value |
| --- | --- | --- | --- | --- |
| air-sensors | humidity | TLM0200 | 1627049400000000000 | 35.64 |
| air-sensors | humidity | TLM0200 | 1627049700000000000 | 35.67 |

| dataset | metric | sensorID | timestamp | value |
| --- | --- | --- | --- | --- |
| air-sensors | temperature | TLM0100 | 1627049400000000000 | 71.84 |
| air-sensors | temperature | TLM0100 | 1627049700000000000 | 71.87 |

| dataset | metric | sensorID | timestamp | value |
| --- | --- | --- | --- | --- |
| air-sensors | temperature | TLM0200 | 1627049400000000000 | 74.10 |
| air-sensors | temperature | TLM0200 | 1627049700000000000 | 74.17 |

### Query a raw CSV string

1. Import the [`csv` package](/flux/v0/stdlib/csv/).
    
2. Use [`csv.from()`](/flux/v0/stdlib/csv/from/) and provide the following parameters:
    
    -   **csv**: CSV string to query
    -   **mode**: raw

#### Query

```js
import "csv"

csvData =
    "
dataset,metric,sensorID,timestamp,value
air-sensors,humidity,TLM0100,1627049400000000000,34.79
air-sensors,humidity,TLM0100,1627049700000000000,34.65
air-sensors,humidity,TLM0200,1627049400000000000,35.64
air-sensors,humidity,TLM0200,1627049700000000000,35.67
air-sensors,temperature,TLM0100,1627049400000000000,71.84
air-sensors,temperature,TLM0100,1627049700000000000,71.87
air-sensors,temperature,TLM0200,1627049400000000000,74.10
air-sensors,temperature,TLM0200,1627049700000000000,74.17
"

csv.from(csv: csvData, mode: "raw")
```

#### Results

When using the **raw** CSV parsing mode, all columns values are strings.

| dataset | metric | sensorID | timestamp | value |
| --- | --- | --- | --- | --- |
| air-sensors | humidity | TLM0100 | 1627049400000000000 | 34.79 |
| air-sensors | humidity | TLM0100 | 1627049700000000000 | 34.65 |
| air-sensors | humidity | TLM0200 | 1627049400000000000 | 35.64 |
| air-sensors | humidity | TLM0200 | 1627049700000000000 | 35.67 |
| air-sensors | temperature | TLM0100 | 1627049400000000000 | 71.84 |
| air-sensors | temperature | TLM0100 | 1627049700000000000 | 71.87 |
| air-sensors | temperature | TLM0200 | 1627049400000000000 | 74.10 |
| air-sensors | temperature | TLM0200 | 1627049700000000000 | 74.17 |

### Query CSV data from a file

1. Import the [`csv` package](/flux/v0/stdlib/csv/).
2. Use [`csv.from()`](/flux/v0/stdlib/csv/from/) and the `file` parameter to query CSV data from a file.

#### Flux must have access to the file system

To query CSV data from a file, Flux must have access to the filesystem. If Flux does not have access to the file system, the query will return an error similar to:

```
failed to read file: filesystem service is uninitialized
```

If using InfluxDB Cloud or InfluxDB OSS, the Flux process **does not** have access to the filesystem.

#### Query

```js
import "csv"

csv.from(file: "/path/to/example.csv")
```

#### /path/to/example.csv

```csv
#group,false,false,true,true,true,false,false
#datatype,string,long,string,string,string,long,double
#default,_result,,,,,,
,result,table,dataset,metric,sensorID,timestamp,value
,,0,air-sensors,humidity,TLM0100,1627049400000000000,34.79
,,0,air-sensors,humidity,TLM0100,1627049700000000000,34.65
,,1,air-sensors,humidity,TLM0200,1627049400000000000,35.64
,,1,air-sensors,humidity,TLM0200,1627049700000000000,35.67
,,2,air-sensors,temperature,TLM0100,1627049400000000000,71.84
,,2,air-sensors,temperature,TLM0100,1627049700000000000,71.87
,,3,air-sensors,temperature,TLM0200,1627049400000000000,74.10
,,3,air-sensors,temperature,TLM0200,1627049700000000000,74.17
```

#### Results

| dataset | metric | sensorID | timestamp | value |
| --- | --- | --- | --- | --- |
| air-sensors | humidity | TLM0100 | 1627049400000000000 | 34.79 |
| air-sensors | humidity | TLM0100 | 1627049700000000000 | 34.65 |

| dataset | metric | sensorID | timestamp | value |
| --- | --- | --- | --- | --- |
| air-sensors | humidity | TLM0200 | 1627049400000000000 | 35.64 |
| air-sensors | humidity | TLM0200 | 1627049700000000000 | 35.67 |

| dataset | metric | sensorID | timestamp | value |
| --- | --- | --- | --- | --- |
| air-sensors | temperature | TLM0100 | 1627049400000000000 | 71.84 |
| air-sensors | temperature | TLM0100 | 1627049700000000000 | 71.87 |

| dataset | metric | sensorID | timestamp | value |
| --- | --- | --- | --- | --- |
| air-sensors | temperature | TLM0200 | 1627049400000000000 | 74.10 |
| air-sensors | temperature | TLM0200 | 1627049700000000000 | 74.17 |

### Query CSV data from a URL

1. Import the [`experimental/csv` package](/flux/v0/stdlib/experimental/csv/).
2. Use the [experimental `csv.from()` function](/flux/v0/stdlib/experimental/csv/from/) and `url` parameter to specify the URL to query.

The **experimental `csv.from()`** function does not support multiple parsing modes and only works with [annotated CSV](/influxdb/cloud/reference/syntax/annotated-csv/).

```js
import "experimental/csv"

csv.from(url: "https://example.com/example.csv")
```

**To use the parsing modes available in `csv.from()`:**

1. Import the [`csv`](/flux/v0/stdlib/csv/) and [`experimental/http`](/flux/v0/stdlib/experimental/http/) packages.
2. Use [`http.get()`](/flux/v0/stdlib/experimental/http/) to fetch the CSV data.
3. Use [`string()`](/flux/v0/stdlib/universe/string/) to convert the response body to a string.
4. Use [`csv.from()`](/flux/v0/stdlib/csv/from/) to parse the CSV data and return results.

```js
import "csv"
import "experimental/http"

url = "https://example.com/example.csv"
csvData = string(v: http.get(url: url).body)

csv.from(csv: csvData, mode: "raw")
```

#### Related

-   [csv.from() function](/flux/v0/stdlib/csv/from/)
-   [Experimental csv.from() function](/flux/v0/stdlib/experimental/csv/from/)
-   [Annotated CSV](/influxdb/cloud/reference/syntax/annotated-csv/)