[FEA] Add case-insensitive options and additional string predicates to GFQL

[lmeyerov]

---

name: Feature request
about: Suggest an idea for this project
title: "[FEA] Add case-insensitive options and additional string predicates to GFQL"
labels: enhancement, gfql, help wanted, good-first-issue
assignees: ''

---

Is your feature request related to a problem? Please describe.

GFQL's string predicates have inconsistent support for case sensitivity and are missing some common string operations that users expect from a query language. Currently:

• startswith() and endswith() are always case-sensitive with no option to change this
• Several useful string predicates available in pandas are not exposed in GFQL
• The documentation doesn't clearly indicate which predicates support case-insensitive matching

Describe the solution you'd like

1. Add case parameter to startswith() and endswith()

   # Current (case-sensitive only)
   n({"name": startswith("John")})
   
   # Proposed (with optional case parameter)
   n({"name": startswith("john", case=False)})  # Would match "John", "JOHN", etc.
   n({"email": endswith(".COM", case=False)})   # Would match ".com", ".Com", etc.

2. Add commonly needed string predicates

   # Length checking
   n({"name": len_eq(5)})         # Exactly 5 characters
   n({"name": len_gt(10)})        # More than 10 characters
   n({"name": len_between(3, 20)}) # Between 3 and 20 characters
   
   # Strip/trim operations  
   n({"name": strip_eq("John")})   # Strip whitespace before comparing
   
   # Not/inverse operations
   n({"name": not_contains("test")})  # Doesn't contain pattern
   n({"name": not_match("^[0-9]+$")}) # Doesn't match regex

3. Consider adding fuzzy/similarity matching (stretch goal)

   n({"name": similar_to("Jon", threshold=0.8)})  # Fuzzy string matching

Describe alternatives you've considered

1. Query strings - Users can use pandas query syntax, but this is less discoverable and not type-safe
2. Custom predicates - Users can implement their own, but built-in support would be better
3. Post-processing - Filter results after the query, but this is less efficient

Additional context

• The implementation should follow the existing pattern in graphistry/compute/predicates/str.py
• All new predicates should inherit from ASTPredicate and delegate to pandas string methods
• Tests should be added to ensure compatibility with both pandas and cuDF backends
• Documentation should be updated in docs/source/gfql/spec/language.md

Implementation hints

For case-insensitive startswith/endswith, the implementation could use pandas' case conversion:

def startswith(self, pat, case=True, na=None):
    if not case:
        return self._series.str.lower().str.startswith(pat.lower(), na=na)
    return self._series.str.startswith(pat, na=na)

This would be a great first issue for someone familiar with Python and pandas who wants to contribute to GFQL!