dzuelke/grammatista

A utility for finding, managing, transforming and processing translatable information in software projects.

grammatista's Issues

The PHP parser should work on an AST instead of tokens internally

given the pattern t(declare(singular_message), declare(parameters), declare(domain)) and a code like t("msg", [], "master") the token matching breaks down completely since in the declare(parameters) part it looks for a single token and then fails to detect the domain.

We could solve this by changing the parsing to look for the next token in the pattern when at the declare() position (in this case the ,) while skipping all parts in brackets/parenthesis. But I think this would be rather brittle.

Additionally having a generic way to parse an AST would also allow us implementing other languages pretty easily. Python supports getting the AST from code, with Javascript we could use http://lisperator.net/uglifyjs/ast.

The PHP parser should skip parsed tokens on successful match

Assuming a pattern definition like

'->t(declare(singular_message))' => [...],
't(declare(singular_message))' => [...],

both pattern will match for a line like $this->t(...), since

grammatista/lib/Parser/Php.php

Line 373 in dd1e835

if(($pattern = $this->compareToken($tokens, $i)) !== false) {

doesn't explicitly skip the matched pattern count, but will simply try parsing with the next token.

One Problem with skipping the amount of matched tokens could be code like $this->t($this->t(...)) which would then only find the outer t() call. That seems to me to be a rather edge case though, so we could add an option to a pattern which allows the old behaviour?