davidkt wrote:

Why do we need Kurt 2.0? It's simple:

import kurt
import sys
p = kurt.Project()
for path in sys.argv[1:-1]:
    costume = kurt.Costume.load(path)
    p.stage.costumes.append(costume)
p.save(sys.argv[-1])

import hashlib
import json
import os
import sys
import time
import zipfile
import PIL
project = {
    'children': [],
    'costumes': [],
    'currentCostumeIndex': 0,
    'info': {
        'author': u'',
        'comment': u'',
        'flashVersion': '',
        'hasCloudData': False,
        'projectID': '',
        'scriptCount': 0,
        'spriteCount': 0,
        'userAgent': '',
        'videoOn': False
    },
    'lists': [],
    'objName': 'Stage',
    'penLayerMD5': '279467d0d49e152706ed66539b577c00.png',
    'scriptComments': [],
    'scripts': [],
    'sounds': [],
    'tempoBPM': 60,
    'variables': [],
    'videoAlpha': 0.5,
}
zip_file = zipfile.ZipFile(sys.argv[-1], "w")
for (i, path) in enumerate(sys.argv[1:-1]):
    contents = open(path, 'rb').read()
    (folder, filename) = os.path.split(path)
    (name, extension) = os.path.splitext(filename)
    assert extension.lower() in (".png", ".jpg")
    zi = zipfile.ZipInfo("%i.png" % i)
    zi.date_time = time.localtime(time.time())[:6]
    zi.compress_type = zipfile.ZIP_DEFLATED
    zi.external_attr = 0777 << 16L
    zip_file.writestr(zi, contents)
    project['costumes'].append({
        'baseLayerID': i,
        'baseLayerMD5': hashlib.md5(contents).hexdigest() + extension,
        'bitmapResolution': 1,
        'costumeName': u'image1',
        'rotationCenterX': 240,
        'rotationCenterY': 180,
    })
zi = zipfile.ZipInfo("project.json")
zi.date_time = time.localtime(time.time())[:6]
zi.compress_type = zipfile.ZIP_DEFLATED
zi.external_attr = 0777 << 16L
zip_file.writestr(zi, json.dumps(project))

blob8108 wrote:

davidkt wrote:

Why do we need Kurt 2.0? It's simple:

Good question!

Sure, it's easy to read the ZIP file – that's what's so great about the 2.0 format – but it's a pain to actually manipulate the JSON. Compare the following code to import images into a project file.

With Kurt:

import kurt
import sys
p = kurt.Project()
for path in sys.argv[1:-1]:
    costume = kurt.Costume.load(path)
    p.stage.costumes.append(costume)
p.save(sys.argv[-1])

Without Kurt:

import hashlib
import json
import os
import sys
import time
import zipfile
import PIL
project = {
    'children': [],
    'costumes': [],
    'currentCostumeIndex': 0,
    'info': {
        'author': u'',
        'comment': u'',
        'flashVersion': '',
        'hasCloudData': False,
        'projectID': '',
        'scriptCount': 0,
        'spriteCount': 0,
        'userAgent': '',
        'videoOn': False
    },
    'lists': [],
    'objName': 'Stage',
    'penLayerMD5': '279467d0d49e152706ed66539b577c00.png',
    'scriptComments': [],
    'scripts': [],
    'sounds': [],
    'tempoBPM': 60,
    'variables': [],
    'videoAlpha': 0.5,
}
zip_file = zipfile.ZipFile(sys.argv[-1], "w")
for (i, path) in enumerate(sys.argv[1:-1]):
    contents = open(path, 'rb').read()
    (folder, filename) = os.path.split(path)
    (name, extension) = os.path.splitext(filename)
    assert extension.lower() in (".png", ".jpg")
    zi = zipfile.ZipInfo("%i.png" % i)
    zi.date_time = time.localtime(time.time())[:6]
    zi.compress_type = zipfile.ZIP_DEFLATED
    zi.external_attr = 0777 << 16L
    zip_file.writestr(zi, contents)
    project['costumes'].append({
        'baseLayerID': i,
        'baseLayerMD5': hashlib.md5(contents).hexdigest() + extension,
        'bitmapResolution': 1,
        'costumeName': u'image1',
        'rotationCenterX': 240,
        'rotationCenterY': 180,
    })
zi = zipfile.ZipInfo("project.json")
zi.date_time = time.localtime(time.time())[:6]
zi.compress_type = zipfile.ZIP_DEFLATED
zi.external_attr = 0777 << 16L
zip_file.writestr(zi, json.dumps(project))

—and this is a simple example. Kurt provides a nice interface for working with the data in the Scratch files, and it's the same interface whether you're using Scratch 1.4 or Scratch 2.0; so code you write for working with Scratch 1.4 files will also work with Scratch 2.0 files, with no extra effort on your part. Which also means you can convert between formats very easily, as Hardmath points out.

It also makes working with blocks much easier: it has all the information for the blocks, can search them by text or command, and can even parse them from a special text syntax.

Does that answer your question?

davidkt wrote:

It just must have been a lot easier to make Kurt 2.0, as I can't even see how you made kurt 1.4.

Why can't it just install everything at once?

nXIII wrote:

Most of the keys in info are optional, but Kurt outputs invalid values like empty strings for the user agent/project ID fields. It might be better not to include them in kurt-generated files at all.

blob8108 wrote:

davidkt wrote:

It just must have been a lot easier to make Kurt 2.0, as I can't even see how you made kurt 1.4.

The 2.0-specific bits were much easier, yes. But Kurt 2.0 overall was much harder, as it supports both and I put a lot of thought into the interface.

Pillow is the Python Imaging Library, used for loading / converting / saving images.

Why can't it just install everything at once?

If you use `pip install kurt`, it should fetch and install all the dependencies for you. But you may have trouble with pillow because you need binaries or to compile it from source.

davidkt wrote:

blob8108 wrote:

Pillow is the Python Imaging Library, used for loading / converting / saving images.

That's PIL.

blob8108 wrote:

davidkt wrote:

blob8108 wrote:

Pillow is the Python Imaging Library, used for loading / converting / saving images.

That's PIL.

Mmm. Pillow is the friendly PIL fork.

blob8108 wrote:

nXIII wrote:

Most of the keys in info are optional, but Kurt outputs invalid values like empty strings for the user agent/project ID fields. It might be better not to include them in kurt-generated files at all.

Mmkay. Does it matter outputting empty ones? Which ones precisely should it not include? After all, your validator accepts it…

Travellingcritic wrote:

This helped me download the pymunk library as well