--- a/src/eric7/Utilities/ModuleParser.py Thu Jan 04 10:43:29 2024 +0100
+++ b/src/eric7/Utilities/ModuleParser.py Thu Jan 04 17:07:38 2024 +0100
@@ -23,6 +23,8 @@
from functools import reduce
+from PyQt6.QtCore import QRegularExpression
+
from eric7 import Utilities
__all__ = [
@@ -66,7 +68,7 @@
return -1
-_py_getnext = re.compile(
+_py_getnext = QRegularExpression(
r"""
(?P<Comment>
\# .*? $ # ignore everything in comments
@@ -150,7 +152,7 @@
^
(?P<MethodIndent> [ \t]* )
(?: async [ \t]+ )? (?: cdef | cpdef | def) [ \t]+
- (?P<MethodName> \w+ )
+ (?P<MethodName> [\pL_] \w* )
(?: [ \t]* \[ [^\]]+ \] )?
[ \t]* \(
(?P<MethodSignature> (?: [^)] | \)[ \t]*,? )*? )
@@ -164,7 +166,7 @@
(?P<ClassIndent> [ \t]* )
(?: cdef [ \t]+ )?
class [ \t]+
- (?P<ClassName> \w+ )
+ (?P<ClassName> [\pL_] \w* )
(?: [ \t]* \[ [^\]]+ \] )?
[ \t]*
(?P<ClassSupers> \( [^)]* \) )?
@@ -175,23 +177,30 @@
^
(?P<AttributeIndent> [ \t]* )
self [ \t]* \. [ \t]*
- (?P<AttributeName> \w+ )
- (?: [ \t]* : [^=\n]+ )?
+ (?P<AttributeName> [\pL_] \w* )
[ \t]* =
)
+| (?P<TypedAttribute>
+ ^
+ (?P<TypedAttributeIndent> [ \t]* )
+ self [ \t]* \. [ \t]*
+ (?P<TypedAttributeName> [\pL_] \w* )
+ [ \t]* : [ \t]+
+ )
+
| (?P<Variable>
^
(?P<VariableIndent> [ \t]* )
- (?P<VariableName> \w+ )
+ (?P<VariableName> [\pL_] \w* )
[ \t]* = [ \t]* (?P<VariableSignal> (?:pyqtSignal)? )
)
| (?P<TypedVariable>
^
(?P<TypedVariableIndent> [ \t]* )
- (?P<TypedVariableName> \w+ )
- [ \t]* :
+ (?P<TypedVariableName> [\pL_] \w* )
+ [ \t]* : [ \t]+
)
| (?P<Main>
@@ -226,10 +235,12 @@
(?: (?: if | elif ) [ \t]+ [^:]* | else [ \t]* ) :
(?= \s* (?: async [ \t]+ )? def)
)""",
- re.VERBOSE | re.MULTILINE,
-).search
+ QRegularExpression.PatternOption.MultilineOption
+ | QRegularExpression.PatternOption.ExtendedPatternSyntaxOption
+ | QRegularExpression.PatternOption.UseUnicodePropertiesOption,
+).match
-_rb_getnext = re.compile(
+_rb_getnext = QRegularExpression(
r"""
(?P<Docstring>
=begin [ \t]+ edoc (?P<DocstringContents> .*? ) =end
@@ -364,8 +375,10 @@
end \b [^_]
)
)""",
- re.VERBOSE | re.MULTILINE,
-).search
+ QRegularExpression.PatternOption.MultilineOption
+ | QRegularExpression.PatternOption.ExtendedPatternSyntaxOption
+ | QRegularExpression.PatternOption.UseUnicodePropertiesOption,
+).match
_hashsub = re.compile(r"""^([ \t]*)#[ \t]?""", re.MULTILINE).sub
@@ -613,25 +626,25 @@
modifierIndent = -1
while True:
m = self._getnext(src, i)
- if not m:
+ if not m.hasMatch():
break
- start, i = m.span()
+ start, i = m.capturedStart(), m.capturedEnd()
- if m.start("MethodModifier") >= 0:
- modifierIndent = _indent(m.group("MethodModifierIndent"))
- modifierType = m.group("MethodModifierType")
+ if m.captured("MethodModifier"):
+ modifierIndent = _indent(m.captured("MethodModifierIndent"))
+ modifierType = m.captured("MethodModifierType")
- elif m.start("Method") >= 0:
+ elif m.captured("Method"):
# found a method definition or function
- thisindent = _indent(m.group("MethodIndent"))
- meth_name = m.group("MethodName")
- meth_sig = m.group("MethodSignature")
+ thisindent = _indent(m.captured("MethodIndent"))
+ meth_name = m.captured("MethodName")
+ meth_sig = m.captured("MethodSignature")
meth_sig = meth_sig.replace("\\\n", "")
- meth_ret = m.group("MethodReturnAnnotation")
+ meth_ret = m.captured("MethodReturnAnnotation")
meth_ret = meth_ret.replace("\\\n", "")
- if m.group("MethodPyQtSignature") is not None:
+ if m.captured("MethodPyQtSignature"):
meth_pyqtSig = (
- m.group("MethodPyQtSignature")
+ m.captured("MethodPyQtSignature")
.replace("\\\n", "")
.split("result")[0]
.split("name")[0]
@@ -640,6 +653,13 @@
else:
meth_pyqtSig = None
lineno += src.count("\n", last_lineno_pos, start)
+ def_lineno = (
+ lineno + src.count(
+ "\n", start, m.capturedStart("MethodIndent")
+ )
+ if m.capturedStart("MethodIndent") > start
+ else lineno
+ )
last_lineno_pos = start
if modifierType and modifierIndent == thisindent:
if modifierType == "@staticmethod":
@@ -719,7 +739,7 @@
)
self.__py_setVisibility(f)
self.addFunction(meth_name, f)
- endlineno = calculateEndline(lineno, srcLines, thisindent)
+ endlineno = calculateEndline(def_lineno, srcLines, thisindent)
f.setEndLine(endlineno)
cur_obj = f
classstack.append((None, thisindent)) # Marker for nested fns
@@ -728,49 +748,49 @@
modifierType = Function.General
modifierIndent = -1
- elif m.start("Docstring") >= 0:
- contents = m.group("DocstringContents3")
- if contents is not None:
+ elif m.captured("Docstring"):
+ contents = m.captured("DocstringContents3")
+ if contents:
contents = _hashsub(r"\1", contents)
else:
if self.file.lower().endswith(".ptl"):
contents = ""
else:
- contents = m.group("DocstringContents1") or m.group(
+ contents = m.captured("DocstringContents1") or m.captured(
"DocstringContents2"
)
if cur_obj:
cur_obj.addDescription(contents)
- elif m.start("String") >= 0:
+ elif m.captured("String"):
if modulelevel and (
src[start - len("\r\n") : start] == "\r\n"
or src[start - len("\n") : start] == "\n"
or src[start - len("\r") : start] == "\r"
):
- contents = m.group("StringContents3")
- if contents is not None:
+ contents = m.captured("StringContents3")
+ if contents:
contents = _hashsub(r"\1", contents)
else:
if self.file.lower().endswith(".ptl"):
contents = ""
else:
- contents = m.group("StringContents1") or m.group(
+ contents = m.captured("StringContents1") or m.captured(
"StringContents2"
)
if cur_obj:
cur_obj.addDescription(contents)
- elif m.start("Class") >= 0:
+ elif m.captured("Class"):
# we found a class definition
- thisindent = _indent(m.group("ClassIndent"))
+ thisindent = _indent(m.captured("ClassIndent"))
lineno += src.count("\n", last_lineno_pos, start)
last_lineno_pos = start
# close all classes indented at least as much
while classstack and classstack[-1][1] >= thisindent:
del classstack[-1]
- class_name = m.group("ClassName")
- inherit = m.group("ClassSupers")
+ class_name = m.captured("ClassName")
+ inherit = m.captured("ClassSupers")
if inherit:
# the class inherits from other classes
inherit = inherit[1:-1].strip()
@@ -819,13 +839,16 @@
# add nested classes to the module
classstack.append((cur_class, thisindent))
- elif m.start("Attribute") >= 0:
+ elif m.captured("Attribute") or m.captured("TypedAttribute"):
+ if m.captured("Attribute"):
+ attrName = m.captured("AttributeName")
+ else:
+ attrName = m.captured("TypedAttributeName")
lineno += src.count("\n", last_lineno_pos, start)
last_lineno_pos = start
index = -1
while index >= -len(classstack):
if classstack[index][0] is not None:
- attrName = m.group("AttributeName")
attr = Attribute(self.name, attrName, self.file, lineno)
self.__py_setVisibility(attr)
classstack[index][0].addAttribute(attrName, attr)
@@ -833,19 +856,28 @@
else:
index -= 1
- elif m.start("Main") >= 0:
+ elif m.captured("Main"):
# 'main' part of the script, reset class stack
lineno += src.count("\n", last_lineno_pos, start)
last_lineno_pos = start
classstack = []
- elif m.start("Variable") >= 0:
- thisindent = _indent(m.group("VariableIndent"))
- variable_name = m.group("VariableName")
- isSignal = m.group("VariableSignal") != ""
+ elif m.captured("Variable") or m.captured("TypedVariable"):
+ if m.captured("Variable"):
+ thisindent = _indent(m.captured("VariableIndent"))
+ variable_name = m.captured("VariableName")
+ isSignal = m.captured("VariableSignal") != ""
+ else:
+ thisindent = _indent(m.captured("TypedVariableIndent"))
+ variable_name = m.captured("TypedVariableName")
+ isSignal = False
+ if keyword.iskeyword(variable_name):
+ # only if the determined name is not a keyword
+ # (e.g. else, except)
+ continue
lineno += src.count("\n", last_lineno_pos, start)
last_lineno_pos = start
- if thisindent == 0:
+ if thisindent == 0 or not classstack:
# global variable
attr = Attribute(
self.name, variable_name, self.file, lineno, isSignal=isSignal
@@ -872,39 +904,11 @@
classstack[index][0].addGlobal(variable_name, attr)
break
- elif m.start("TypedVariable") >= 0:
- thisindent = _indent(m.group("TypedVariableIndent"))
- variable_name = m.group("TypedVariableName")
- if not keyword.iskeyword(variable_name):
- # only if the determined name is not a keyword (e.g. else, except)
- lineno += src.count("\n", last_lineno_pos, start)
- last_lineno_pos = start
- if thisindent == 0:
- # global variable
- attr = Attribute(self.name, variable_name, self.file, lineno)
- self.__py_setVisibility(attr)
- self.addGlobal(variable_name, attr)
- else:
- index = -1
- while index >= -len(classstack):
- if classstack[index][1] >= thisindent:
- index -= 1
- else:
- if classstack[index][0] is not None and isinstance(
- classstack[index][0], Class
- ):
- attr = Attribute(
- self.name, variable_name, self.file, lineno
- )
- self.__py_setVisibility(attr)
- classstack[index][0].addGlobal(variable_name, attr)
- break
-
- elif m.start("Import") >= 0:
+ elif m.captured("Import"):
# - import module
names = [
n.strip()
- for n in "".join(m.group("ImportList").splitlines())
+ for n in "".join(m.captured("ImportList").splitlines())
.replace("\\", "")
.split(",")
]
@@ -912,11 +916,11 @@
[name for name in names if name not in self.imports]
)
- elif m.start("ImportFrom") >= 0:
+ elif m.captured("ImportFrom"):
# - from module import stuff
- mod = m.group("ImportFromPath")
+ mod = m.captured("ImportFromPath")
namesLines = (
- m.group("ImportFromList")
+ m.captured("ImportFromList")
.replace("(", "")
.replace(")", "")
.replace("\\", "")
@@ -931,9 +935,9 @@
[name for name in names if name not in self.from_imports[mod]]
)
- elif m.start("ConditionalDefine") >= 0:
+ elif m.captured("ConditionalDefine"):
# a conditional function/method definition
- thisindent = _indent(m.group("ConditionalDefineIndent"))
+ thisindent = _indent(m.captured("ConditionalDefineIndent"))
while conditionalsstack and conditionalsstack[-1] >= thisindent:
del conditionalsstack[-1]
if deltastack:
@@ -941,7 +945,7 @@
conditionalsstack.append(thisindent)
deltaindentcalculated = 0
- elif m.start("Comment") >= 0 and modulelevel:
+ elif m.captured("Comment") and modulelevel:
continue
modulelevel = False
@@ -967,16 +971,16 @@
break
start, i = m.span()
- if m.start("Method") >= 0:
+ if m.captured("Method"):
# found a method definition or function
thisindent = indent
indent += 1
meth_name = (
- m.group("MethodName")
- or m.group("MethodName2")
- or m.group("MethodName3")
+ m.captured("MethodName")
+ or m.captured("MethodName2")
+ or m.captured("MethodName3")
)
- meth_sig = m.group("MethodSignature")
+ meth_sig = m.captured("MethodSignature")
meth_sig = meth_sig and meth_sig.replace("\\\n", "") or ""
lineno += src.count("\n", last_lineno_pos, start)
last_lineno_pos = start
@@ -1034,14 +1038,14 @@
cur_obj = f
classstack.append((None, thisindent)) # Marker for nested fns
- elif m.start("Docstring") >= 0:
- contents = m.group("DocstringContents")
- if contents is not None:
+ elif m.captured("Docstring"):
+ contents = m.captured("DocstringContents")
+ if contents:
contents = _hashsub(r"\1", contents)
if cur_obj:
cur_obj.addDescription(contents)
- elif m.start("Class") >= 0:
+ elif m.captured("Class"):
# we found a class definition
thisindent = indent
indent += 1
@@ -1055,8 +1059,8 @@
# record the end line of this class, function or module
classstack[-1][0].setEndLine(lineno - 1)
del classstack[-1]
- class_name = m.group("ClassName") or m.group("ClassName2")
- inherit = m.group("ClassSupers")
+ class_name = m.captured("ClassName") or m.captured("ClassName2")
+ inherit = m.captured("ClassSupers")
if inherit:
# the class inherits from other classes
inherit = inherit[1:].strip()
@@ -1089,7 +1093,7 @@
acstack.append(["public", thisindent])
# default access control is 'public'
- elif m.start("Module") >= 0:
+ elif m.captured("Module"):
# we found a module definition
thisindent = indent
indent += 1
@@ -1103,7 +1107,7 @@
# record the end line of this class, function or module
classstack[-1][0].setEndLine(lineno - 1)
del classstack[-1]
- module_name = m.group("ModuleName")
+ module_name = m.captured("ModuleName")
# remember this class
cur_class = RbModule(self.name, module_name, self.file, lineno)
# add nested Ruby modules to the file
@@ -1122,21 +1126,9 @@
acstack.append(["public", thisindent])
# default access control is 'public'
- elif m.start("AccessControl") >= 0:
- aclist = m.group("AccessControlList")
- if aclist is None:
- index = -1
- while index >= -len(acstack):
- if acstack[index][1] < indent:
- actype = (
- m.group("AccessControlType")
- or m.group("AccessControlType2").split("_")[0]
- )
- acstack[index][0] = actype.lower()
- break
- else:
- index -= 1
- else:
+ elif m.captured("AccessControl"):
+ aclist = m.captured("AccessControlList")
+ if aclist:
index = -1
while index >= -len(classstack):
if (
@@ -1146,8 +1138,8 @@
):
parent = classstack[index][0]
actype = (
- m.group("AccessControlType")
- or m.group("AccessControlType2").split("_")[0]
+ m.captured("AccessControlType")
+ or m.captured("AccessControlType2").split("_")[0]
)
actype = actype.lower()
for name in aclist.split(","):
@@ -1165,8 +1157,20 @@
break
else:
index -= 1
+ else:
+ index = -1
+ while index >= -len(acstack):
+ if acstack[index][1] < indent:
+ actype = (
+ m.captured("AccessControlType")
+ or m.captured("AccessControlType2").split("_")[0]
+ )
+ acstack[index][0] = actype.lower()
+ break
+ else:
+ index -= 1
- elif m.start("Attribute") >= 0:
+ elif m.captured("Attribute"):
lineno += src.count("\n", last_lineno_pos, start)
last_lineno_pos = start
index = -1
@@ -1176,7 +1180,7 @@
and not isinstance(classstack[index][0], Function)
and classstack[index][1] < indent
):
- attrName = m.group("AttributeName")
+ attrName = m.captured("AttributeName")
attr = Attribute(self.name, attrName, self.file, lineno)
if attrName.startswith("@@") or attrName[0].isupper():
classstack[index][0].addGlobal(attrName, attr)
@@ -1186,7 +1190,7 @@
else:
index -= 1
else:
- attrName = m.group("AttributeName")
+ attrName = m.captured("AttributeName")
if attrName[0] != "@":
attr = Attribute(self.name, attrName, self.file, lineno)
self.addGlobal(attrName, attr)
@@ -1194,7 +1198,7 @@
lastGlobalEntry.setEndLine(lineno - 1)
lastGlobalEntry = None
- elif m.start("Attr") >= 0:
+ elif m.captured("Attr"):
lineno += src.count("\n", last_lineno_pos, start)
last_lineno_pos = start
index = -1
@@ -1205,25 +1209,9 @@
and classstack[index][1] < indent
):
parent = classstack[index][0]
- if m.group("AttrType") is None:
- nv = m.group("AttrList").split(",")
- if not nv:
- break
- # get rid of leading ':'
- name = nv[0].strip()[1:]
- attr = (
- parent.getAttribute("@" + name)
- or parent.getAttribute("@@" + name)
- or Attribute(self.name, "@" + name, self.file, lineno)
- )
- if len(nv) == 1 or nv[1].strip() == "false":
- attr.setProtected()
- elif nv[1].strip() == "true":
- attr.setPublic()
- parent.addAttribute(attr.name, attr)
- else:
- access = m.group("AttrType")
- for name in m.group("AttrList").split(","):
+ if m.captured("AttrType"):
+ access = m.captured("AttrType")
+ for name in m.captured("AttrList").split(","):
# get rid of leading ':'
name = name.strip()[1:]
attr = (
@@ -1241,15 +1229,31 @@
elif attr.isProtected():
attr.setPublic()
parent.addAttribute(attr.name, attr)
+ else:
+ nv = m.captured("AttrList").split(",")
+ if not nv:
+ break
+ # get rid of leading ':'
+ name = nv[0].strip()[1:]
+ attr = (
+ parent.getAttribute("@" + name)
+ or parent.getAttribute("@@" + name)
+ or Attribute(self.name, "@" + name, self.file, lineno)
+ )
+ if len(nv) == 1 or nv[1].strip() == "false":
+ attr.setProtected()
+ elif nv[1].strip() == "true":
+ attr.setPublic()
+ parent.addAttribute(attr.name, attr)
break
else:
index -= 1
- elif m.start("Begin") >= 0:
+ elif m.captured("Begin"):
# a begin of a block we are not interested in
indent += 1
- elif m.start("End") >= 0:
+ elif m.captured("End"):
# an end of a block
indent -= 1
if indent < 0:
@@ -1261,10 +1265,10 @@
indent = 0
elif (
- m.start("String") >= 0
- or m.start("Comment") >= 0
- or m.start("ClassIgnored") >= 0
- or m.start("BeginEnd") >= 0
+ m.captured("String")
+ or m.captured("Comment")
+ or m.captured("ClassIgnored")
+ or m.captured("BeginEnd")
):
pass
