Detect PHP but I don't have any PHP or HTML code

[kochol]

My repo is mainly C++ but github detects it as PHP recently.

Preliminary Steps

Please confirm you have...

• reviewed How Linguist Works,
• reviewed the Troubleshooting docs,
• considered implementing an override,
• verified an issue has not already been logged for your issue (linguist issues).

URL of the affected repository:

https://github.com/kochol/ariyana

Last modified on:

1/22/2021

Expected language:

C++

Detected language:

PHP

[lildude]

This analysis is happening because of all the .inc files in https://github.com/kochol/ariyana/tree/master/src/gfx/private/transcoder

The .inc extension is a very generic extension associated with multiple languages so it need to be processed by the heuristics at https://github.com/github/linguist/blob/9eb9472be957108fc48b6c9f725f2100c18b7a5e/lib/linguist/heuristics.yml#L235-L247

None of those files are matching any of the heuristics so the analysis is falling through to the classifier which is making a guess based on the content of the samples Linguist has and it's made as best a guess as it can, and picked PHP, for example:

$ LINGUIST_DEBUG=1 github-linguist src/gfx/private/transcoder/basisu_transcoder_tables_bc7_m5_alpha.inc
basisu_transcoder_tables_bc7_m5_alpha.inc: 49 lines (49 sloc)
  type:      Text
  mime type: text/plain
  Assembly = -20906.329 +  -5.592 = -20911.921
       C++ = -11378.941 +  -3.877 = -11382.819
      HTML = -17147.685 +  -5.844 = -17153.528
Motorola 68K Assembly = -44299.009 +  -5.998 = -44305.007
      NASL = -12526.644 +  -5.487 = -12532.131
       PHP =  -9516.069 +  -4.845 =  -9520.914
POV-Ray SDL = -12481.660 +  -5.305 = -12486.964
    Pascal = -15231.504 +  -5.305 = -15236.808
      Pawn = -13178.207 +  -5.998 = -13184.205
       SQL = -26224.446 +  -5.150 = -26229.597
SourcePawn = -10823.038 +  -7.096 = -10830.134
  language:  PHP
$

Looking at the the content, I don't think the classifier or heuristics would ever have detected these files correctly 😄 so you'll need to implement an override to force them to your preferred language or tell Linguist to ignore them entirely.

As an aside, once I can, I'll be moving this issue across to Discussions which I enabled today.

[kochol]

Thank you

[lildude]

🎉 Moved, and I've marked my response as the answer.

[ecm-pushbx]

I noticed today that https://github.com/Baron-von-Riedesel/DOS-debug is misdetected as 14.3 % NASL. There are some .inc files, but they are MASM/TASM/JWASM x86 assembly source files.

[lildude]

@ecm-pushbx please start a new discussion. You might want to also consider an override before you do as .inc is a very generic extension associated with quite a few language.