Python Google Images Downloader

This tool lets you download results from Google Images pretty fast. It uses aiohttp under the hood. It is essentially a Python 3.7 rewrite of google-images-downloader without the restriction of not using external dependencies.

Installation

The following should work to install the utility and check that it will run:

pip install pygidl
pygidl -h

Basic Usage

From the command line:

basic usage is something like pygidl "cats and dogs". This will create an output directory in your current working directory named for the timestamp that the command was run. Underneath that directory will be another directory with the slugified version of your query string. Underneath that directory will be the downloaded images, named by their sha256 hashes and file type extensions:

pygidl "cats and dogs"
tree .
.
└── 2020-01-21-17-57-34
    └── cats-and-dogs
        ├── 01d2dde343a45e3a1fcc5e7cd3cace33398c9b06a97e494d4329f264e57d5f57.jpg
        ├── 026cef34db26cbd5fa246bc720c1234b39ffa07737e43523b160390c13d5d3e6.jpeg
        ├── 03a0f2ebeed5d91acaed73ad303bd724767e101688e33d1a5557cca9139972d7.webp
        ├── 0affcf3198b40063e9302c4515380d6796946098c7c1c3c043072815e29e2770.jpeg
        ...

Advanced Usage

The command can be configured to support several more complex query scenarios:

Verbose Output

Use the -v/--verbose flag to change the log level to show more messages. The log level is "WARNING" by default. Supplying -v once sets it to "INFO", and two or more times sets it to "DEBUG".

Prefixes and Suffixes

The -p/--prefix and -s/--suffix flags can be used to run multiple copies of the same Google Images query with extra prefix or suffix strings. For example:

pygidl -p Andorra -p Angola -s "on a ship" -s "on a plane" flag
tree -d .
.
└── 2020-01-21-18-15-59
    ├── andorra-flag-on-a-plane
    ├── andorra-flag-on-a-ship
    ├── angola-flag-on-a-plane
    └── angola-flag-on-a-ship

5 directories

Output Groups

You can override the name of the directory that contains the results of each query with the -g/--group flag. For example:

pygidl -g "Cute Animals" -p fluffy -p adorable -s dog -s cat "" -v
tree . -d
.
└── cute-animals
    ├── adorable-cat
    ├── adorable-dog
    ├── fluffy-cat
    └── fluffy-dog

5 directories

Face Search

You can tell Google Images to find faces with the -f/--face flag. For example:

pygidl -g "Tom Hanks" -p "" -p young "Tom Hanks" -s "" -s "Oscars" -f -v
tree -d .
.
└── tom-hanks
    ├── tom-hanks
    ├── tom-hanks-oscars
    ├── young-tom-hanks
    └── young-tom-hanks-oscars

5 directories

Programmatic Usage

Something like the following should work (assuming you have opencv-python installed in your environment):

import asyncio
import os

import cv2

from pygidl import scrape_google_images


downloaded_image_paths = asyncio.run(
    scrape_google_images(
        base_query="cats and dogs",
        prefixes=["cute", "adorable"],
        suffixes=["playing", "running"],
        group="cute-animals",
        output_dir=os.getcwd(),
        face=False,
    )
)
for path in downloaded_image_paths:
    image = cv2.imread(path)
    if image is None:
        print(f"could not load image {path}")
        continue
    height, width = image.shape[:2]
    print(f"image {path} has size {width}x{height}")

Known Issues and Limitations

• Only returns max of 100 results per query
• Doesn't support full range of advanced search options
• No tests
• No retries
• No report on results/metadata output option
• Sometimes Google returns results from a different template without easily-parseable metadata

Contributing

I don't think anyone will ever get this far, but if you want to open a pull request (or even better, take over ownership of the project for me!), go for it. At a minimum, new code should have type hints, docstrings, and be auto-formatted with black with an 80-character max line length. Even better would be some tests and Sphinx documentation!