Add strict2_parse to increment and decrement tags

[aswamy]

Note

Stacked on Shopify/liquid#2065 (assign/capture strict2_parse)

Summary

Adds strict2_parse to Increment and Decrement tags. Both tags previously accepted any string as a variable name via markup.strip — no Parser validation was performed in any mode.

Changes:

• Both tags now include ParserSwitching and use parse_with_selected_parser
• lax_parse: preserves the original markup.strip behavior
• strict_parse: delegates to lax_parse
• strict2_parse: validates the variable name via p.consume(:id) + p.consume(:end_of_string)

This means {% increment foo bar %} and {% decrement 11aa %} are now rejected in strict2 mode.

Tophat

bundle install
bundle exec irb -r liquid

# Rejected in strict2
Liquid::Template.parse('{% increment foo bar %}', error_mode: :strict2)
# => Liquid::SyntaxError

Liquid::Template.parse('{% decrement 11aa %}', error_mode: :strict2)
# => Liquid::SyntaxError

# Accepted
Liquid::Template.parse('{% increment my-var %}', error_mode: :strict2)
# => OK

# Lax mode unchanged
Liquid::Template.parse('{% increment foo bar %}', error_mode: :lax)
# => OK (preserves backward compatibility)

🤖 Generated with Claude Code

[aswamy]

i thought of merging the code between increment and decrement since they share so much, but didn't see it with other similar tags like case and if so i just kept it verbose

[EvilGenius13]

Seems fair to leave them separate. I do have a question.

The guard return @variable_name = stripped if stripped.empty? - I think we want to remove it because in other tags like capture if we don't find a variable name we raise a syntax error through p.consume(:id)

[aswamy]

Hmm yeah i think that's fair. I looked at how lax mode handles {% increment %} (without a variable name) and apparently it's allowed. Internally, it looks like it's creating an empty-string variable that can be accessed via self[""]. Ill rewrite this into a valid variable name. Good catch!