File size: 10,692 Bytes
c72f0a1
81a8217
c72f0a1
 
 
 
 
 
 
 
 
81a8217
 
 
c72f0a1
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
81a8217
 
 
 
 
 
 
 
 
 
c72f0a1
 
 
 
 
 
 
 
 
 
81a8217
 
c72f0a1
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
81a8217
c72f0a1
81a8217
 
c72f0a1
 
 
 
 
 
 
 
 
81a8217
 
c72f0a1
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
"""Evaluate structured If statement definitions and compile matching DuckDB CASE expressions."""
import json
import logging
import math
import re
from copy import deepcopy

from core.user_tables import ROLLUP_CONDITION_OPS

_LOG = logging.getLogger(__name__)
_KEY = re.compile(r"^[a-zA-Z_][a-zA-Z0-9_]*$")
_DIGITS = r"[0-9](?:_?[0-9])*"
_NUMBER_PATTERN = rf"[+-]?(?:{_DIGITS}(?:\.(?:{_DIGITS})?)?|\.{_DIGITS})(?:[eE][+-]?{_DIGITS})?"
_NUMBER = re.compile(_NUMBER_PATTERN)
_EXPRESSION_KEYS = {"expr", "expression", "formula", "signal", "fn", "script", "code"}
MAX_ROWS = 100
_TRIM_CHARS = ''.join(chr(code) for code in range(0x3100) if chr(code).isspace())


class DefinitionError(ValueError):
    """A named refusal for an invalid If statement definition."""
    def __init__(self, code, detail):
        self.code = code
        super().__init__(f"{code}: {detail}")


def _deny_expressions(raw):
    if isinstance(raw, dict):
        for key, value in raw.items():
            if str(key).lower() in _EXPRESSION_KEYS:
                raise DefinitionError("if_expression_forbidden", str(key))
            _deny_expressions(value)
    elif isinstance(raw, list):
        for value in raw:
            _deny_expressions(value)


def _field(value):
    if not isinstance(value, str) or not _KEY.fullmatch(value) or len(value) > 80:
        raise DefinitionError("if_field_invalid", "use a field key")
    return value


def _scalar(value):
    if value is None or type(value) in (str, bool, int):
        return value
    if type(value) is float and math.isfinite(value):
        return value
    raise DefinitionError("if_value_invalid", "use text, a number, a boolean or null")


def _result(raw):
    if isinstance(raw, dict) and set(raw) == {"value"}:
        return {"value": _scalar(raw["value"])}
    if isinstance(raw, dict) and set(raw) == {"field"}:
        return {"field": _field(raw["field"])}
    raise DefinitionError("if_result_invalid", "use {value} or {field}")


def clean(raw):
    """Validate only the If statement bag, preserving row order and literal value types."""
    if isinstance(raw, str):
        raise DefinitionError('if_expression_forbidden', 'expression text')
    _deny_expressions(raw)
    if not isinstance(raw, dict) or set(raw) != {"rows", "default"}:
        raise DefinitionError("if_definition_invalid", "use rows and one default, without expression text")
    if not isinstance(raw["rows"], list) or len(raw["rows"]) > MAX_ROWS:
        raise DefinitionError("if_rows_invalid", f"use at most {MAX_ROWS} ordered rows")
    rows = []
    for item in raw["rows"]:
        if not isinstance(item, dict) or set(item) != {"field", "op", "value", "then"}:
            raise DefinitionError("if_row_invalid", "use field, op, value and then")
        if not isinstance(item['op'], str) or item["op"] not in ROLLUP_CONDITION_OPS:
            raise DefinitionError("if_operator_invalid", str(item["op"]))
        rows.append({"field": _field(item["field"]), "op": item["op"],
                     "value": _scalar(item["value"]), "then": _result(item["then"])})
    return {"rows": rows, "default": _result(raw["default"])}


def references(defn):
    """Return source keys from conditions, result copies and the default copy."""
    definition = clean(defn)
    keys = {item["field"] for item in definition["rows"]}
    keys.update(item["then"]["field"] for item in definition["rows"] if "field" in item["then"])
    if "field" in definition["default"]:
        keys.add(definition["default"]["field"])
    return keys


def admit_sources(defn, hidden, previous=None):
    """Require readable sources for a new definition while preserving an existing shared definition."""
    definition = clean(defn)
    unchanged = json.dumps(definition, sort_keys=True) == json.dumps(previous, sort_keys=True)
    if not unchanged and references(definition) & set(hidden):
        raise DefinitionError('if_source_forbidden',
                              'Choose source fields that you can read.')
    return definition


def _text(value):
    if value is None:
        return ""
    if type(value) is bool:
        return "true" if value else "false"
    # DuckDB uses simple Unicode lowercase, including a single i for dotted capital I.
    return ''.join('i' if char == '\u0130' else char.lower() for char in str(value).strip())


def _number(text):
    if not _NUMBER.fullmatch(text):
        return None
    try:
        number = float(text)
        return number if math.isfinite(number) else None
    except (ValueError, TypeError, OverflowError):
        return None


def _matches(item, row):
    left, right = _text(row[item["field"]]), _text(item["value"])
    op = item["op"]
    if op in ("is_empty", "is_not_empty"):
        return (left == "") if op == "is_empty" else (left != "")
    if op in ("contains", "not_contains"):
        return (right in left) if op == "contains" else (right not in left)
    ln, rn = _number(left), _number(right)
    a, b = (ln, rn) if ln is not None and rn is not None else (left, right)
    if op in ("eq", "neq"):
        return (a == b) if op == "eq" else (a != b)
    if left == "" or right == "":
        return False
    return {"gt": a > b, "gte": a >= b, "lt": a < b, "lte": a <= b}[op]


def _available(item, keys):
    return item["field"] in keys and ("field" not in item["then"] or item["then"]["field"] in keys)


def _copy(result, row):
    return deepcopy(result["value"] if "value" in result else row.get(result["field"]))


def evaluate(defn, row):
    """Return the first matching result, skipping and reporting deleted source references."""
    definition = clean(defn)
    row = row if isinstance(row, dict) else {}
    missing = references(definition) - row.keys()
    if missing:
        _LOG.warning("if_source_missing: %s", ", ".join(sorted(missing)))
    for item in definition["rows"]:
        if _available(item, row) and _matches(item, row):
            return _copy(item["then"], row)
    return _copy(definition["default"], row)


def _literal(value):
    if value is None:
        return "NULL"
    if type(value) is bool:
        return "TRUE" if value else "FALSE"
    if type(value) is int:
        return repr(value)
    if type(value) is float:
        return f"CAST('{value!r}' AS DOUBLE)"
    return "'" + value.replace("'", "''") + "'"


def _sql_text(expression):
    return f"lower(trim(coalesce(CAST(({expression}) AS VARCHAR), ''), {_literal(_TRIM_CHARS)}))"


def _sql_number(expression):
    cast = f"TRY_CAST(NULLIF({expression}, '') AS DOUBLE)"
    return (f"CASE WHEN regexp_full_match({expression}, {_literal(_NUMBER_PATTERN)}) "
            f"AND isfinite({cast}) THEN {cast} ELSE NULL END")


def _sql_match(item, expression):
    left, right = _sql_text(expression), _literal(_text(item["value"]))
    op = item["op"]
    if op in ("is_empty", "is_not_empty"):
        return f"{left} {'=' if op == 'is_empty' else '<>'} ''"
    if op in ("contains", "not_contains"):
        return f"strpos({left}, {right}) {'>' if op == 'contains' else '='} 0"
    ln, rn = _sql_number(left), _sql_number(right)
    operator = {"eq": "=", "neq": "<>", "gt": ">", "gte": ">=", "lt": "<", "lte": "<="}[op]
    test = (f"CASE WHEN ({ln}) IS NOT NULL AND ({rn}) IS NOT NULL "
            f"THEN ({ln}) {operator} ({rn}) ELSE {left} {operator} {right} END")
    if op in ("gt", "gte", "lt", "lte"):
        test = f"{left} <> '' AND {right} <> '' AND ({test})"
    return test


def sql_case(defn, column_sql):
    """Compile a DuckDB CASE with typed results and the same first-match and missing-ref rules."""
    definition = clean(defn)
    missing = references(definition) - column_sql.keys()
    if missing:
        _LOG.warning("if_source_missing: %s", ", ".join(sorted(missing)))

    def result_sql(result):
        expression = (_literal(result["value"]) if "value" in result
                      else column_sql.get(result["field"], "NULL"))
        # VARIANT preserves mixed scalar types instead of coercing every result to text.
        return f"CAST(({expression}) AS VARIANT)"

    branches = [f"WHEN {_sql_match(item, column_sql[item['field']])} THEN {result_sql(item['then'])}"
                for item in definition["rows"] if _available(item, column_sql)]
    default = result_sql(definition["default"])
    return "CASE " + " ".join(branches) + f" ELSE {default} END" if branches else default


def materialize(fields, rows):
    """Return computed cells and diagnostics without changing input rows or field definitions."""
    by_key = {str(f["key"]): f for f in fields if isinstance(f, dict) and f.get("key")}
    definitions = {key: clean(f["if"]) for key, f in by_key.items() if f.get("type") == "if"}
    if not definitions:
        return {}, []
    diagnostics = [{"field": key, "missing": sorted(references(defn) - by_key.keys())}
                   for key, defn in definitions.items() if references(defn) - by_key.keys()]
    for key, definition in definitions.items():
        definition['rows'] = [item for item in definition['rows'] if _available(item, by_key)]
        if 'field' in definition['default'] and definition['default']['field'] not in by_key:
            definition['default'] = {'value': None}
    changes = {}
    for rid, row in rows.items():
        context = {key: row.get(key) for key in by_key}
        done, active = set(), set()

        def resolve(key):
            if key in done:
                return
            if key in active:
                raise DefinitionError("if_cycle", key)
            active.add(key)
            for dependency in references(definitions[key]) & definitions.keys():
                resolve(dependency)
            context[key] = evaluate(definitions[key], context)
            active.remove(key)
            done.add(key)

        for key in definitions:
            try:
                resolve(key)
            except DefinitionError as exc:
                context[key] = None
                diagnostic = {"field": key, "code": exc.code}
                if diagnostic not in diagnostics:
                    diagnostics.append(diagnostic)
        changes[str(rid)] = {key: context[key] for key in definitions}
    return changes, diagnostics


def project_fields(fields, hidden):
    """Omit the definition when it could disclose a hidden source key or comparison value."""
    projected = []
    for field in fields:
        item = dict(field)
        if item.get("type") == "if" and isinstance(item.get("if"), dict):
            if references(item["if"]) & set(hidden):
                item.pop("if")
                item["ifConfigured"] = True
        projected.append(item)
    return projected