From 245f44b284eea86888597080460935c8057a7a03 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Hugo=20H=C3=B6rnquist?= <hugo@lysator.liu.se>
Date: Wed, 20 Sep 2023 23:41:34 +0200
Subject: [PATCH] Cleanup repeated code.
---
muppet/puppet/format/parser.py | 127 +++++++++++++++++++--------------
1 file changed, 75 insertions(+), 52 deletions(-)
diff --git a/muppet/puppet/format/parser.py b/muppet/puppet/format/parser.py
index 1a960b8..0607448 100644
--- a/muppet/puppet/format/parser.py
+++ b/muppet/puppet/format/parser.py
@@ -57,6 +57,7 @@ from muppet.puppet.ast import (
from muppet.parser_combinator import (
MatchObject,
ParseDirective,
+ Parser,
ParseError,
ParserCombinator,
char,
@@ -107,6 +108,56 @@ else:
return f
+def heredoc_declarator(syntax: str | None) -> Parser:
+ """
+ Parser for the start of heredoc.
+
+ .. code-block::
+ :caption: Sample Heredoc beginning form.
+
+ @("EOF":json/$)
+
+ :param syntax:
+ The expected syntax of this entry, `json` in the example above.
+
+ This can technically be ignored, since it always end with '/' or ')'.
+ """
+ escape_switches = s('/') & many(s('n') | 'r' | 't' | 's' | '$' | 'u' | 'L')
+ switches = optional(ws & escape_switches)
+
+ syntax_decl = optional(ws & ':' & ws & syntax)
+
+ unquoted_heredoc = many(complement('):/'))
+ quoted_heredoc = s('"') & many(complement('"')) & '"'
+
+ return ws & '@(' & ws & (quoted_heredoc | unquoted_heredoc) \
+ & syntax_decl & switches & ws & ')'
+
+
+heredoc_end = many(space) & optional(s('|')) & ws & optional(s('-')) \
+ & ws & many1(all_(~ space, char))
+"""Parser for the end of a heredoc."""
+
+
+def interpolated_form(inner: Parser) -> Parser:
+ """
+ Parse a form interpolated into a string.
+
+ .. code-block:: puppet
+
+ "Hello ${inner}"
+
+ In the above example, parse from the ``${`` until ``}``.
+
+ :param inner:
+ The parser for whatever is inside the braces.
+ """
+ return (many(space)
+ & '$' & optional(s('{'))
+ & inner
+ & ws & optional(s('}')))
+
+
logger = logging.getLogger(__name__)
@@ -359,21 +410,11 @@ class ParserFormatter(Serializer[ParseDirective]):
parser = ws & '"'
for fragment in it.fragments:
match fragment:
- case PuppetVar(x):
- f = (ws
- & '$'
- & optional(s('{'))
- & ws
- & optional(s('$'))
- & tag('var', x)
- & ws
- & optional(s('}')))
- parser &= f
case PuppetString(st):
for c in st:
parser &= rich_char(c)
case _:
- parser &= many(space) & "${" & ws & self.s(fragment) & ws & "}"
+ parser &= interpolated_form(self.s(fragment))
parser &= s('"') & ws
return parser
@@ -486,28 +527,27 @@ class ParserFormatter(Serializer[ParseDirective]):
@override
def _puppet_heredoc(self, it: PuppetHeredoc) -> ParseDirective:
- # TODO The header and footer is mostly shared wiht literal
- # heredoc. Merge these
- parser = ws & '@(' & ws & '"' & ws & many(complement('"')) & '"'
- parser &= optional(ws & ':' & ws & it.syntax)
- switches = name('switches', many(s('n') | 'r' | 't' | 's' | '$' | 'u' | 'L'))
- parser &= optional(ws & '/' & switches)
- parser &= ws & ')'
-
- # TODO the contents shares muttch with concat, merge these
+ # This uses ``many(space)`` rather than ``ws``. Our ``ws``
+ # implementation also "eats" comments, but inside heredoc's
+ # there are no comments.
+
+ # This is technically incorrect, since heredoc_declarator
+ # always accepts non-quoted starts. This however doesn't
+ # matter.
+ parser = ws & heredoc_declarator(it.syntax)
+
for fragment in it.fragments:
match fragment:
- case PuppetVar(x):
- f = (many(space)
- & '$'
- & optional(s('{'))
- & ws
- & optional(s('$'))
- & tag('var', x)
- & ws
- & optional(s('}')))
- parser &= f
case PuppetString(st):
+ # This discards the given whitespace, and
+ # re-applies arbitrary whitespace between.
+ # This is for two reasons:
+ # 1. Leading indentation whitespace is ignored up
+ # to the '|' line, meaning that we still have
+ # to check for whitespace before each string.
+ # 2. If the '-' ending modifier is used then we
+ # DON'T have a newline at the end, confusing
+ # matters.
word = many(space)
for line in st.split('\n'):
word &= many(space)
@@ -515,37 +555,20 @@ class ParserFormatter(Serializer[ParseDirective]):
word &= rich_char(c)
word &= optional(s('\n'))
parser &= name(repr(st), word)
- # parser &= many(space) & st.strip()
case _:
- parser &= many(space) & "${" & ws & self.s(fragment) & ws & "}"
- parser &= many(space) & optional(s('|')) & ws & optional(s('-'))
- parser &= ws & many1(all_(~ space, char))
+ parser &= interpolated_form(self.s(fragment))
+
+ parser &= heredoc_end
return parser
@override
def _puppet_literal_heredoc(self, it: PuppetLiteralHeredoc) -> ParseDirective:
- escape_switches = s('/') & many(s('n') | 'r' | 't' | 's' | '$' | 'u' | 'L')
- switches = optional(ws & escape_switches)
-
- if it.syntax:
- syntax = ws & ':' & ws & it.syntax
- else:
- syntax = nop
-
- unquoted_heredoc = many(complement('):/')) & syntax & switches
- quoted_heredoc = s('"') & many(complement('"')) & '"' & syntax & switches
-
- heredoc_declarator = ws & '@(' & ws & (quoted_heredoc | unquoted_heredoc) & ws & ')'
-
- # delim = stringify_match(delim_parts)
-
- parser = heredoc_declarator
+ parser = ws & heredoc_declarator(it.syntax)
for line in it.content.split('\n'):
parser &= many(space) & line.strip()
- parser &= many(space) & optional(s('|')) & ws & optional(s('-'))
- parser &= ws & many1(all_(~ space, char))
+ parser &= heredoc_end
return parser
@override
--
GitLab