我正在編寫一個(gè)語(yǔ)法來(lái)解析HTSQL語(yǔ)法,并堅(jiān)持如何處理段和除法運(yùn)算符的/字符重用.

described grammar并不是非常正式,所以我一直在關(guān)注Python實(shí)現(xiàn)的確切輸出,從粗略的一瞥似乎是一個(gè)手寫的解析器,而不是使用解析器生成器 – 作為參考解析器生成器我目前使用是

CL-YACC和

CL-LEX.(FWIW完整的東西

is here,雖然可能有點(diǎn)過(guò)時(shí).)

由于“/ 1”被解析為'(：COLLECT(：INTEGER“1”))’,但是“/ 1/2”被解析為'(：COLLECT(：OPERATOR /) (：INTEGER“1”)(：INTEGER“2”)))’,即一個(gè)是段分隔符,另一個(gè)是分區(qū); “/ 1 // 2”再次是'(：COLLECT(：OPERATOR /(：INTEGER“1”)(：COLLECT(：INTEGER“2”))))’.

問(wèn)題是,如何在不借助切換到手動(dòng)解析器的情況下在語(yǔ)法規(guī)范中處理這個(gè)問(wèn)題？切換到不同的解析器生成器類(而不是LALR(1))會(huì)有幫助嗎？

到目前為止,我已經(jīng)嘗試了不同的語(yǔ)法變體,但是對(duì)于整個(gè)語(yǔ)法來(lái)說(shuō),優(yōu)先級(jí)是固定的這一事實(shí)也會(huì)干擾斜杠的兩種解釋.我嘗試的另一種方法是在詞法分析器中消除歧義,即以不同方式處理第一個(gè)斜杠(在每個(gè)“組”中)并返回不同的符號(hào),例如, DIV – 然而,我找不到一個(gè)好的規(guī)則,并以某種方式懷疑它純粹是通過(guò)查看詞匯結(jié)構(gòu)而存在.

最后,我很想通過(guò)完全脫離給定的解析器來(lái)解決這個(gè)問(wèn)題,以使我的生活更輕松.從某種意義上說(shuō),更容易理解可預(yù)測(cè)的語(yǔ)法,你會(huì)認(rèn)為這更令人滿意嗎？

圖1,用于檢查解析樹(shù)的Python腳本：

import htsql

application = htsql.HTSQL("sqlite:///htsql_demo.sqlite")

global y

y = None

def p(string):

global y

with application:

y = htsql.core.syn.parse.parse(string)

return y

def l(name):

result = []

for c in name:

if c.isupper() and result:

result.append("-")

result.append(c)

return "".join(result)

def keyword(name):

return ":{}".format(name.upper())

def n(expression):

name = expression.__class__.__name__

name = name[:name.find("Syntax")]

return keyword(l(name))

def t(expression):

arguments = [n(expression)]

d = expression.__dict__

if "identifier" in d:

arguments.append(t(expression.identifier))

if "text" in d:

arguments.append("\"{}\"".format(expression.text))

if "symbol" in d:

if not isinstance(expression, (htsql.core.syn.syntax.ProjectSyntax, htsql.core.syn.syntax.FilterSyntax, htsql.core.syn.syntax.CollectSyntax, htsql.core.syn.syntax.DetachSyntax)):

arguments.append(expression.symbol)

if "arm" in d:

arguments.append(t(expression.arm))

if "larm" in d:

arguments.append(t(expression.larm))

if "rarm" in d:

arguments.append(t(expression.rarm))

if "arms" in d:

arguments.extend(t(x) for x in expression.arms)

if "rarms" in d:

arguments.extend(t(x) for x in expression.rarms)

return "({})".format(" ".join(arguments))

# t(p("/school"))

# '(:COLLECT (:IDENTIFIER "school"))

# t(p("/'school'"))

# '(:COLLECT (:STRING "school"))

圖2,我當(dāng)前的解析器,沒(méi)有正確處理這個(gè)問(wèn)題：

(defpackage #:cl-htsql

(:use #:cl #:alexandria #:cl-lex #:yacc)

(:import-from #:arnesi #:with-collector))

(eval-when (:compile-toplevel :load-toplevel :execute)

(defun maybe-intern (name &optional (package NIL package-p))

"If NAME is a SYMBOL, return it, otherwise INTERN it."

(cond

((symbolp name)

name)

(package-p

(intern name package))

(T

(intern name))))

(defmacro define-lexer (name &body patterns)

"Shortcut for DEFINE-STRING-LEXER."

`(define-string-lexer ,name

,@(mapcar

(lambda (pattern)

(etypecase pattern

((or symbol string)

(let ((symbol (maybe-intern pattern))

(pattern (string pattern)))

`(,pattern (return (values ',symbol ',symbol)))))

(list

(destructuring-bind (pattern &optional symbol value) pattern

(let* ((symbol (or symbol (intern pattern)))

(value (or value symbol)))

(etypecase symbol

(list

`(,pattern ,symbol))

(symbol

`(,pattern (return (values ',symbol ',value))))))))))

patterns))))

;; parser are results are to be treated immutable

(define-lexer string-lexer

/

("\\|" \|)

("\\&" &)

<=

>=

==

=

!==

!=

!~

!

~

<

>

@

("\\?" ?)

("\\." \.)

("\\(" \()

("\\)" \))

("\\+" +)

-

("\\*" *)

\:

("-?0|[1-9][0-9]*(\\.[0-9]*)?([eE][+-]?[0-9]+)?"

(return (cond

((find #\e $@)

(values 'float $@))

((find #\. $@)

(values 'decimal $@))

(T

(values 'integer $@)))))

("([^\"\\.\\?~\'=<>\\(\\)@\\|\\&/:])+" (return (values 'name $@)))

("\'([^\\\']|\\.)*?\'" (return (values 'string (string-trim "\'" $@))))

("\"([^\\\"]|\\.)*?\"" (return (values 'string (string-trim "\"" $@)))))

(define-parser *expression-parser*

(:muffle-conflicts (44 0))

(:start-symbol query)

(:terminals (|\|| #+(or)div & ! |.| ? / = != !== !~ ~ < > == <= >= \( \) + - * @ name integer decimal float string))

(:precedence ((:left @) (:left ~) (:left |.|) (:left + -) (:left * div) (:left = != == !== ~ !~ < <= > >=) (:left !) (:left &) (:left |\||) (:left ?) (:left /)))

(query

segment)

(segment

(/ segment (lambda (x y) (declare (ignore x)) (if (eq y :skip) '(:skip) `(:collect ,y))))

skip

group)

(skip

((constantly :skip)))

(group

(\( segment \) (lambda (x y z) (declare (ignore x z)) `(:group ,y)))

sieve)

(sieve

(segment ? segment (lambda (x y z) (declare (ignore y)) `(:filter ,x ,z)))

or)

(or

(segment |\|| segment (lambda (x y z) `(:operator ,y ,x ,z)))

and)

(and

(segment & segment (lambda (x y z) `(:operator ,y ,x ,z)))

not)

(not

(! segment (lambda (x y) `(:prefix ,x ,y)))

comparison)

(comparison

(segment = segment (lambda (x y z) `(:operator ,y ,x ,z)))

(segment != segment (lambda (x y z) `(:operator ,y ,x ,z)))

(segment ~ segment (lambda (x y z) `(:operator ,y ,x ,z)))

(segment !~ segment (lambda (x y z) `(:operator ,y ,x ,z)))

(segment == segment (lambda (x y z) `(:operator ,y ,x ,z)))

(segment !== segment (lambda (x y z) `(:operator ,y ,x ,z)))

(segment < segment (lambda (x y z) `(:operator ,y ,x ,z)))

(segment <= segment (lambda (x y z) `(:operator ,y ,x ,z)))

(segment > segment (lambda (x y z) `(:operator ,y ,x ,z)))

(segment >= segment (lambda (x y z) `(:operator ,y ,x ,z)))

addition)

(addition

(segment + segment (lambda (x y z) `(:operator ,y ,x ,z)))

(segment - segment (lambda (x y z) `(:operator ,y ,x ,z)))

multiplication)

(multiplication

(segment * segment (lambda (x y z) `(:operator ,y ,x ,z)))

(segment / segment (lambda (x y z) (declare (ignore y)) `(:operator / ,x ,z)))

composition)

(composition

(segment |.| segment (lambda (x y z) (declare (ignore y)) `(:compose ,x ,z)))

attach)

(attach

(segment @ segment (lambda (x y z) (declare (ignore y)) `(:attach ,x ,z)))

detach)

(detach

(@ segment (lambda (x y) (declare (ignore x)) `(:detach ,y)))

term)

(term

(name (lambda (x) `(:identifier ,x)))

(string (lambda (x) `(:string ,x)))

(number (lambda (x) `(:integer ,x)))

(integer (lambda (x) `(:integer ,x)))

(decimal (lambda (x) `(:decimal ,x)))

(float (lambda (x) `(:float ,x)))))

(defun make-lexer-for-source (source)

"Make a lexer for the SOURCE, either a STRING or a STREAM."

(etypecase source

(string (string-lexer source))

(stream

(flet ((ignore (c)

(declare (ignore c))))

(stream-lexer #'read-line #'string-lexer #'ignore #'ignore)))))

(defun lex-source (source)

"Debug helper to lex a SOURCE into a list of tokens."

(let ((lexer (make-lexer-for-source source)))

(loop

for (x y) = (multiple-value-list (funcall lexer))

while x

collect (list x y))))

(define-condition htsql-parse-error (simple-error) ())

(defun translate-yacc-error (error)

(make-condition

'htsql-parse-error

:format-control "Couldn't parse HTSQL query: ~A."

:format-arguments (list error)))

(defun parse-htsql-query (source)

"Parse SOURCE into a syntax tree. The SOURCE may be either a STRING or

a STREAM."

(handler-case

(parse-with-lexer

(make-lexer-for-source source)

*expression-parser*)

(yacc-parse-error (error)

(error (translate-yacc-error error)))))

;; > (parse-htsql-query "/1/")

;; (:OPERATOR / (:COLLECT (:INTEGER "1")) :SKIP)

;; > (parse-htsql-query "/1/2")

;; (:OPERATOR / (:COLLECT (:INTEGER "1")) (:INTEGER "2"))

總結(jié)

以上是生活随笔為你收集整理的python中ht_python – 解析HTSQL时处理语法歧义的全部?jī)?nèi)容，希望文章能夠幫你解決所遇到的問(wèn)題。

如果覺(jué)得生活随笔網(wǎng)站內(nèi)容還不錯(cuò)，歡迎將生活随笔推薦給好友。