Merge pull request #108 from VH-Lab/claude/did-hasmember-quoting

hasmember: one rule in field_search, as in DID-matlab (#236)

Author: stevevanhooser

src/did/datastructures.py

@@ -1,4 +1,5 @@
 import json
+import numbers
 import re
 
 import numpy as np
@@ -190,6 +191,37 @@ def or_operands(search_struct):
     return operands
 
 
+def _has_member(value, member):
+    """Is *member* a member of *value*, for field_search's 'hasmember'?
+
+    Text is a member of text that is the whole value or an item of its
+    comma-separated list (white space after a comma, and before one, is
+    ignored); anything is a member of a list that holds an equal element; a
+    number is a member of an equal number. Text is never a member of a
+    number, nor a number of text. This is what DID-matlab's fieldsearch and
+    its SQL search match (DID-matlab#236).
+    """
+
+    def is_number(x):
+        return isinstance(x, numbers.Real) and not isinstance(x, str)
+
+    if isinstance(value, np.ndarray):
+        value = value.tolist()
+    if isinstance(member, str):
+        if isinstance(value, str):
+            pattern = r"(^|,\s*)" + re.escape(member) + r"\s*(,|$)"
+            return re.search(pattern, value) is not None
+        if isinstance(value, (list, tuple)):
+            return any(isinstance(v, str) and v == member for v in value)
+        return False
+    if is_number(member):
+        if is_number(value):
+            return value == member
+        if isinstance(value, (list, tuple)):
+            return any(is_number(v) and v == member for v in value)
+    return False
+
+
 def field_search(a, search_struct):
     """
     Searches a dictionary to determine if it matches a search structure.
@@ -260,10 +292,7 @@ def field_search(a, search_struct):
             b = eq_len(np.array(value).shape, param1)
     elif op_lower == "hasmember":
         if is_there:
-            try:
-                b = param1 in value
-            except TypeError:
-                pass
+            b = _has_member(value, param1)
     elif op_lower == "hasfield":
         b = is_there
     elif op_lower == "partial_struct":

src/did/did_matlab_python_bridge.yaml

@@ -40,7 +40,7 @@ classes:
   - name: database
     type: class
     matlab_path: "+did/database.m"
-    matlab_last_sync_hash: "2c9d8fa7"
+    matlab_last_sync_hash: "74ae15f1"
     status: ported
     python_path: "did/database.py"
     python_class: "Database"
@@ -124,6 +124,25 @@ classes:
       did/database.py nor did/implementations/sqlitedb.py changes.
       tests/test_negated_or.py pins the behaviour with the same cases as
       MATLAB's did.unittest.TestNegatedOr.
+
+      DID-matlab c8adf741..74ae15f1 (DID-matlab#236, 2026-10-24):
+      query_struct_to_sql_str put a text 'hasmember' value into the SQL
+      unquoted (and as a double-quoted token, which SQLite reads as a column
+      when it names one), so 'value' matched every document and '1 OR 1=1'
+      was SQL injection; and the SQL could not see the members of an array
+      or cell, which doc2sql stores only as '[1 3] double'. MATLAB now
+      searches hasmember by brute force with fieldsearch, as Python always
+      has: the new protected static isBruteForceTerm (hasmember, and the
+      depends_on queries of isBruteForceDependsOn) sends a term to
+      brute_force_search, searched last in an AND and kept out of the OR
+      batches, and query_struct_to_sql_str's 'hasmember' case is an error.
+      That is what SQLiteDB already does (_query_struct_to_sql_str returns
+      None for hasmember; _SCAN_OPERATIONS ranks it last), so did/database.py
+      and did/implementations/sqlitedb.py do not change; the rule itself
+      changed in fieldsearch and field_search -- see the fieldsearch entry.
+      2c08472c merges DID-matlab main (2c9d8fa7, #237, above) into that
+      branch. tests/test_hasmember_quoting.py mirrors DID-matlab's
+      TestHasMemberQuoting.
     sync_notes: >
       Schema validation was ported on 2026-08-28, closing what had been the
       largest capability gap between the two languages. Python previously had

src/did/did_matlab_python_bridge_util.yaml

@@ -317,7 +317,7 @@ functions:
   - name: fieldsearch
     type: function
     matlab_path: "+did/+datastructures/fieldsearch.m"
-    matlab_last_sync_hash: "051f535"
+    matlab_last_sync_hash: "f4ed409e"
     status: ported
     python_path: "did/datastructures.py"
     python_name: "field_search"
@@ -330,6 +330,20 @@ functions:
       holds a non-struct value (depends_on = [] read back from a database)
       used to return early, skipping the negation; it now matches. Python's
       field_search already negated it; nothing to port.
+      DID-matlab f4ed409e (DID-matlab#236, 2026-10-24): 'hasmember' had a
+      different rule in each place. MATLAB's fieldsearch used
+      ismember(param1, value), which compares two texts character by
+      character; its SQL search matched an item of a comma-separated list;
+      field_search used `in` (a substring test, and a TypeError -- no match
+      -- on a scalar number). All three now match one rule, MATLAB's in the
+      local function fieldsearch/hasMember and Python's in _has_member: text
+      is a member of text that is the whole value or an item of its
+      comma-separated list (pattern (^|,\s*)M\s*(,|$)); anything is a
+      member of a list (MATLAB cell or numeric array) that holds an equal
+      element, and a number of an equal number; text is never a member of a
+      number, nor a number of text; a dict (MATLAB struct) has no members.
+      tests/test_hasmember_quoting.py test_rule runs the same table of
+      cases as DID-matlab's TestHasMemberQuoting/testRule.
 
   - name: orOperands
     type: function