I want parse my text line and build syntax tree.
How do this in Fasm code ?

Txt db 'print: 4*var+1',13,10,0

Somthing like this.

Have you read Dragon book?

Thanks.
Now reading.

Some people think the first step is tokenization. This is where the source is broken into larger abstractions. To do this you will need to construct a model of your language. For example, maybe you limit input to ASCII, or just numbers and math symbols. Maybe you have predefined symbols or more complex constructions. Find a model that can be decoded - it's possible for a language model to have built-in problems.

There are languages to describe language models: BNF, EBNF, etc ...

If you want to build a very small langauge for learning - I would say to just jump in and make mistakes. Like a simple calculator (with variables), etc ... There are tools to build parsers from a language model: YACC, etc ... It can get really complex fast.

The dragon book is getting quite dated - as are the tools I mention above. So much research has happened since then, and modern languages have also evolved. Yet, it's a great overview.

FYI: there is a section of the board specifically for language design.

The disadvantage of jumping directly to parser generators skipping the lexical analysis step is that one has to deal with whitespace chunks and comments (that are allowed nearly everywhere) in the grammar which at least doubles the complexity for any decent language.

Note also that the generators tend to be quite bad at providing good diagnostic error messages except for common cases.

Still, depends on the language. For some specific tasks the disadvantages are not a problem.