|
539
|
1 """
|
|
|
2 Parsing utilities for gene rules (GPRs).
|
|
|
3
|
|
|
4 This module provides:
|
|
|
5 - RuleErr: structured errors for malformed rules
|
|
|
6 - RuleOp: valid logical operators (AND/OR)
|
|
|
7 - OpList: nested list structure representing parsed rules with explicit operator
|
|
|
8 - RuleStack: helper stack to build nested OpLists during parsing
|
|
|
9 - parseRuleToNestedList: main entry to parse a rule string into an OpList
|
|
|
10 """
|
|
|
11 from enum import Enum
|
|
|
12 from typing import List, Union, Optional
|
|
|
13
|
|
542
|
14 try:
|
|
|
15 from . import general_utils as utils
|
|
|
16 except:
|
|
|
17 import general_utils as utils
|
|
|
18
|
|
539
|
19 class RuleErr(utils.CustomErr):
|
|
|
20 """
|
|
|
21 Error type for rule syntax errors.
|
|
|
22 """
|
|
|
23 errName = "Rule Syntax Error"
|
|
|
24 def __init__(self, rule :str, msg = "no further details provided") -> None:
|
|
|
25 super().__init__(
|
|
|
26 f"rule \"{rule}\" is malformed, {msg}",
|
|
|
27 "please verify your input follows the validity guidelines")
|
|
|
28
|
|
|
29 class RuleOp(Enum):
|
|
|
30 """
|
|
|
31 Valid logical operators for gene rules.
|
|
|
32 """
|
|
|
33 OR = "or"
|
|
|
34 AND = "and"
|
|
|
35
|
|
|
36 @classmethod
|
|
|
37 def isOperator(cls, op :str) -> bool:
|
|
|
38 return op.upper() in cls.__members__
|
|
|
39
|
|
|
40 def __str__(self) -> str: return self.value
|
|
|
41
|
|
|
42 class OpList(List[Union[str, "OpList"]]):
|
|
|
43 """
|
|
|
44 Parsed rule structure: a list with an associated operator for that level.
|
|
|
45 """
|
|
|
46 def __init__(self, op :Optional[RuleOp] = None) -> None:
|
|
|
47 """
|
|
|
48 (Private) Initializes an instance of OpList.
|
|
|
49
|
|
|
50 Args:
|
|
|
51 op (str): Operator to be assigned to the OpList. Defaults to "".
|
|
|
52
|
|
|
53 Returns:
|
|
|
54 None : practically, an OpList instance.
|
|
|
55 """
|
|
|
56 self.op = op
|
|
|
57
|
|
|
58 def setOpIfMissing(self, op :RuleOp) -> None:
|
|
|
59 """
|
|
|
60 Sets the operator of the OpList if it's missing.
|
|
|
61
|
|
|
62 Args:
|
|
|
63 op (str): Operator to be assigned to the OpList.
|
|
|
64
|
|
|
65 Returns:
|
|
|
66 None
|
|
|
67 """
|
|
|
68 if not self.op: self.op = op
|
|
|
69
|
|
|
70 def __repr__(self, indent = "") -> str:
|
|
|
71 """
|
|
|
72 (Private) Returns a string representation of the current OpList instance.
|
|
|
73
|
|
|
74 Args:
|
|
|
75 indent (str): Indentation level . Defaults to "".
|
|
|
76
|
|
|
77 Returns:
|
|
|
78 str: A string representation of the current OpList instance.
|
|
|
79 """
|
|
|
80 nextIndent = indent + " "
|
|
|
81 return f"<{self.op}>[\n" + ",\n".join([
|
|
|
82 f"{nextIndent}{item.__repr__(nextIndent) if isinstance(item, OpList) else item}"
|
|
|
83 for item in self ]) + f"\n{indent}]"
|
|
|
84
|
|
|
85 class RuleStack:
|
|
|
86 """
|
|
|
87 FILO stack used during parsing to build nested OpLists; the top is the current level.
|
|
|
88 """
|
|
|
89 def __init__(self) -> None:
|
|
|
90 """
|
|
|
91 (Private) initializes an instance of RuleStack.
|
|
|
92
|
|
|
93 Returns:
|
|
|
94 None : practically, a RuleStack instance.
|
|
|
95 """
|
|
|
96 self.__stack = [OpList()] # the stack starts out with the result list already allocated
|
|
|
97 self.__updateCurrent()
|
|
|
98
|
|
|
99 def pop(self) -> None:
|
|
|
100 """
|
|
|
101 Removes the OpList on top of the stack, also flattening it once when possible.
|
|
|
102
|
|
|
103 Side Effects:
|
|
|
104 self : mut
|
|
|
105
|
|
|
106 Returns:
|
|
|
107 None
|
|
|
108 """
|
|
|
109 oldTop = self.__stack.pop()
|
|
|
110 if len(oldTop) == 1 and isinstance(oldTop[0], OpList): self.__stack[-1][-1] = oldTop[0]
|
|
|
111 self.__updateCurrent()
|
|
|
112
|
|
|
113 def push(self, operator = "") -> None:
|
|
|
114 """
|
|
|
115 Adds a new nesting level, in the form of a new OpList on top of the stack.
|
|
|
116
|
|
|
117 Args:
|
|
|
118 operator : the operator assigned to the new OpList.
|
|
|
119
|
|
|
120 Side Effects:
|
|
|
121 self : mut
|
|
|
122
|
|
|
123 Returns:
|
|
|
124 None
|
|
|
125 """
|
|
|
126 newLevel = OpList(operator)
|
|
|
127 self.current.append(newLevel)
|
|
|
128 self.__stack.append(newLevel)
|
|
|
129 self.__updateCurrent()
|
|
|
130
|
|
|
131 def popForward(self) -> None:
|
|
|
132 """
|
|
|
133 Moves the last "actual" item from the 2nd to last list to the beginning of the top list, as per
|
|
|
134 the example below:
|
|
|
135 stack : [list_a, list_b]
|
|
|
136 list_a : [item1, item2, list_b] --> [item1, list_b]
|
|
|
137 list_b : [item3, item4] --> [item2, item3, item4]
|
|
|
138
|
|
|
139 This is essentially a "give back as needed" operation.
|
|
|
140
|
|
|
141 Side Effects:
|
|
|
142 self : mut
|
|
|
143
|
|
|
144 Returns:
|
|
|
145 None
|
|
|
146 """
|
|
|
147 self.current.insert(0, self.__stack[-2].pop(-2))
|
|
|
148
|
|
|
149 def currentIsAnd(self) -> bool:
|
|
|
150 """
|
|
|
151 Checks if the current OpList's assigned operator is "and".
|
|
|
152
|
|
|
153 Returns:
|
|
|
154 bool : True if the current OpList's assigned operator is "and", False otherwise.
|
|
|
155 """
|
|
|
156 return self.current.op is RuleOp.AND
|
|
|
157
|
|
|
158 def obtain(self, err :Optional[utils.CustomErr] = None) -> Optional[OpList]:
|
|
|
159 """
|
|
|
160 Obtains the first OpList on the stack, only if it's the only element.
|
|
|
161
|
|
|
162 Args:
|
|
|
163 err : The error to raise if obtaining the result is not possible.
|
|
|
164
|
|
|
165 Side Effects:
|
|
|
166 self : mut
|
|
|
167
|
|
|
168 Raises:
|
|
|
169 err: If given, otherwise None is returned.
|
|
|
170
|
|
|
171 Returns:
|
|
|
172 Optional[OpList]: The first OpList on the stack, only if it's the only element.
|
|
|
173 """
|
|
|
174
|
|
|
175 if len(self.__stack) == 1: return self.__stack.pop()
|
|
|
176 if err: raise err
|
|
|
177 return None
|
|
|
178
|
|
|
179 def __updateCurrent(self) -> None:
|
|
|
180 """
|
|
|
181 (Private) Updates the current OpList to the one on top of the stack.
|
|
|
182
|
|
|
183 Side Effects:
|
|
|
184 self : mut
|
|
|
185
|
|
|
186 Returns:
|
|
|
187 None
|
|
|
188 """
|
|
|
189 self.current = self.__stack[-1]
|
|
|
190
|
|
|
191 def parseRuleToNestedList(rule :str) -> OpList:
|
|
|
192 """
|
|
|
193 Parse a rule string into an OpList, making operator precedence explicit via nesting.
|
|
|
194
|
|
|
195 Args:
|
|
|
196 rule: Rule string to parse (supports parentheses, 'and', 'or').
|
|
|
197
|
|
|
198 Raises:
|
|
|
199 RuleErr: If the rule is malformed (e.g., mismatched parentheses or misplaced operators).
|
|
|
200
|
|
|
201 Returns:
|
|
|
202 OpList: Parsed rule as an OpList structure.
|
|
|
203 """
|
|
|
204 source = iter(rule
|
|
|
205 .replace("(", "( ").replace(")", " )") # single out parentheses as words
|
|
|
206 .strip() # trim edges
|
|
|
207 .split()) # split by spaces
|
|
|
208
|
|
|
209 stack = RuleStack()
|
|
|
210 nestingErr = RuleErr(rule, "mismatch between open and closed parentheses")
|
|
|
211 try:
|
|
|
212 while True: # read until source ends
|
|
|
213 while True:
|
|
|
214 operand = next(source, None) # expect operand or '('
|
|
|
215 if operand is None: raise RuleErr(rule, "found trailing open parentheses")
|
|
|
216 if operand in ("and", "or", ")"): # unexpected operator position
|
|
|
217 raise RuleErr(rule, f"found \"{operand}\" in unexpected position")
|
|
|
218
|
|
|
219 if operand != "(": break # got a name
|
|
|
220
|
|
|
221 # found rule opening: add a new nesting level
|
|
|
222 stack.push()
|
|
|
223
|
|
|
224 stack.current.append(operand)
|
|
|
225
|
|
|
226 while True: # read until operator found or source ends
|
|
|
227 operator = next(source, None) # expect operator or ')'
|
|
|
228 if operator and operator != ")": break # got operator
|
|
|
229
|
|
|
230 if stack.currentIsAnd(): stack.pop() # close current AND chain
|
|
|
231
|
|
|
232 if not operator: break
|
|
|
233 stack.pop() # close parentheses
|
|
|
234
|
|
|
235 if not operator: break
|
|
|
236
|
|
|
237 if not RuleOp.isOperator(operator): raise RuleErr(
|
|
|
238 rule, f"found \"{operator}\" in unexpected position, expected operator")
|
|
|
239
|
|
|
240 operator = RuleOp(operator)
|
|
|
241 if operator is RuleOp.OR and stack.currentIsAnd():
|
|
|
242 stack.pop()
|
|
|
243
|
|
|
244 elif operator is RuleOp.AND and not stack.currentIsAnd():
|
|
|
245 stack.push(operator)
|
|
|
246 stack.popForward()
|
|
|
247
|
|
|
248 stack.current.setOpIfMissing(operator)
|
|
|
249
|
|
|
250 except RuleErr as err: raise err # bubble up proper errors
|
|
|
251 except: raise nestingErr # everything else is interpreted as a nesting error.
|
|
|
252
|
|
|
253 parsedRule = stack.obtain(nestingErr)
|
|
|
254 return parsedRule[0] if len(parsedRule) == 1 and isinstance(parsedRule[0], list) else parsedRule |
