Logo extraction with no slash

[dgarijo]

Detected by @tpronk
I think there might be an issue with extracting a logo when there is no slash (/) in the path to the logo. For illustration, below is a snippet of the README.md of the somef-demo-repo, followed by a snippet of the JSON output of SOMEF. Note that logo1.png is not recognized as a logo, but logo_directory/logo2.png is. Same result if I use logo.png and if I don't have the logo_directory/logo2.png in the README.md

README.md

# Image
Images used to illustrate the software component.
![logo1.png](logo1.png)

# Logo
Main logo used to represent the target software component.
![logo2.png](logo_directory/logo2.png)

SOMEF Output

"logo": [
  {
    "result": {
      "type": "Url",
      "value": "https://raw.githubusercontent.com/tpronk/somef-demo-repo/main/logo_directory/logo2.png"
    },
    "confidence": 1,
    "technique": "regular_expression",
    "source": "https://raw.githubusercontent.com/tpronk/somef-demo-repo/main/README.md"
  }
],
"image": [
  {
    "result": {
      "type": "Url",
      "value": "https://raw.githubusercontent.com/tpronk/somef-demo-repo/main/logo1.png"
    },
    "confidence": 1,
    "technique": "regular_expression",
    "source": "https://raw.githubusercontent.com/tpronk/somef-demo-repo/main/README.md"
  }
]

[tpronk]

As I kept adding fields for SOMEF to extract, I discovered this one might be a bit trickier.

1. This version of the somef-demo-repo yields the same output as above.
2. In a later version, I added some more fields to extract, including invocation. I also moved logo2 to below the invocation section. This version does not yield any image nor logo. Instead, the markdown containing logo2 is made part of the description field, and logo1 does not appear in the output.
3. In a yet later version, I added more fields, including package_distribution, related_documentation, and related_papers. Here, logo1 is again extracted into image, and logo2 is again extracted into logo.

[dgarijo]

The invocation bit is an error by the classifier (need to improve training corpus). In theory if there is something called "logo" we should consider it a logo. But we may only allow one, hence one being classified as image. Will have to look into it.