From e4b4c3e37e16e18ef0c6d8cafdbbb03a95feacf5 Mon Sep 17 00:00:00 2001 From: Peter Eisentraut Date: Mon, 2 Dec 2024 10:35:37 +0100 Subject: [PATCH v1 14/19] jsonpath scanner: reentrant scanner Note: The parser was already pure. --- src/backend/utils/adt/jsonpath_gram.y | 2 + src/backend/utils/adt/jsonpath_internal.h | 9 +- src/backend/utils/adt/jsonpath_scan.l | 114 ++++++++-------------- 3 files changed, 50 insertions(+), 75 deletions(-) diff --git a/src/backend/utils/adt/jsonpath_gram.y b/src/backend/utils/adt/jsonpath_gram.y index 8733a0eac66..de5a455c96d 100644 --- a/src/backend/utils/adt/jsonpath_gram.y +++ b/src/backend/utils/adt/jsonpath_gram.y @@ -60,8 +60,10 @@ static bool makeItemLikeRegex(JsonPathParseItem *expr, %name-prefix="jsonpath_yy" %parse-param {JsonPathParseResult **result} %parse-param {struct Node *escontext} +%parse-param {yyscan_t yyscanner} %lex-param {JsonPathParseResult **result} %lex-param {struct Node *escontext} +%lex-param {yyscan_t yyscanner} %union { diff --git a/src/backend/utils/adt/jsonpath_internal.h b/src/backend/utils/adt/jsonpath_internal.h index 6cd6d8b652d..71f885475dd 100644 --- a/src/backend/utils/adt/jsonpath_internal.h +++ b/src/backend/utils/adt/jsonpath_internal.h @@ -22,17 +22,22 @@ typedef struct JsonPathString int total; } JsonPathString; +typedef void *yyscan_t; + #include "utils/jsonpath.h" #include "jsonpath_gram.h" #define YY_DECL extern int jsonpath_yylex(YYSTYPE *yylval_param, \ JsonPathParseResult **result, \ - struct Node *escontext) + struct Node *escontext, \ + yyscan_t yyscanner) YY_DECL; extern int jsonpath_yyparse(JsonPathParseResult **result, - struct Node *escontext); + struct Node *escontext, + yyscan_t yyscanner); extern void jsonpath_yyerror(JsonPathParseResult **result, struct Node *escontext, + yyscan_t yyscanner, const char *message); #endif /* JSONPATH_INTERNAL_H */ diff --git a/src/backend/utils/adt/jsonpath_scan.l b/src/backend/utils/adt/jsonpath_scan.l index f5a85de36f5..700c17712d0 100644 --- a/src/backend/utils/adt/jsonpath_scan.l +++ b/src/backend/utils/adt/jsonpath_scan.l @@ -30,18 +30,13 @@ } %{ -static JsonPathString scanstring; - -/* Handles to the buffer that the lexer uses internally */ -static YY_BUFFER_STATE scanbufhandle; -static char *scanbuf; -static int scanbuflen; +static JsonPathString scanstring; /* FIXME */ static void addstring(bool init, char *s, int l); static void addchar(bool init, char c); static enum yytokentype checkKeyword(void); -static bool parseUnicode(char *s, int l, struct Node *escontext); -static bool parseHexChar(char *s, struct Node *escontext); +static bool parseUnicode(char *s, int l, struct Node *escontext, yyscan_t yyscanner); +static bool parseHexChar(char *s, struct Node *escontext, yyscan_t yyscanner); /* Avoid exit() on fatal scanner errors (a bit ugly -- see yy_fatal_error) */ #undef fprintf @@ -65,6 +60,7 @@ fprintf_to_ereport(const char *fmt, const char *msg) %option noyywrap %option warn %option prefix="jsonpath_yy" +%option reentrant %option bison-bridge %option noyyalloc %option noyyrealloc @@ -160,23 +156,23 @@ hex_fail \\x{hexdigit}{0,1} \\v { addchar(false, '\v'); } {unicode}+ { - if (!parseUnicode(yytext, yyleng, escontext)) + if (!parseUnicode(yytext, yyleng, escontext, yyscanner)) yyterminate(); } {hex_char} { - if (!parseHexChar(yytext, escontext)) + if (!parseHexChar(yytext, escontext, yyscanner)) yyterminate(); } {unicode}*{unicodefail} { - jsonpath_yyerror(NULL, escontext, + jsonpath_yyerror(NULL, escontext, yyscanner, "invalid Unicode escape sequence"); yyterminate(); } {hex_fail} { - jsonpath_yyerror(NULL, escontext, + jsonpath_yyerror(NULL, escontext, yyscanner, "invalid hexadecimal character sequence"); yyterminate(); } @@ -184,20 +180,20 @@ hex_fail \\x{hexdigit}{0,1} {unicode}+\\ { /* throw back the \\, and treat as unicode */ yyless(yyleng - 1); - if (!parseUnicode(yytext, yyleng, escontext)) + if (!parseUnicode(yytext, yyleng, escontext, yyscanner)) yyterminate(); } \\. { addchar(false, yytext[1]); } \\ { - jsonpath_yyerror(NULL, escontext, + jsonpath_yyerror(NULL, escontext, yyscanner, "unexpected end after backslash"); yyterminate(); } <> { - jsonpath_yyerror(NULL, escontext, + jsonpath_yyerror(NULL, escontext, yyscanner, "unterminated quoted string"); yyterminate(); } @@ -223,7 +219,7 @@ hex_fail \\x{hexdigit}{0,1} \* { } <> { - jsonpath_yyerror(NULL, escontext, + jsonpath_yyerror(NULL, escontext, yyscanner, "unexpected end of comment"); yyterminate(); } @@ -313,22 +309,22 @@ hex_fail \\x{hexdigit}{0,1} } {realfail} { - jsonpath_yyerror(NULL, escontext, + jsonpath_yyerror(NULL, escontext, yyscanner, "invalid numeric literal"); yyterminate(); } {decinteger_junk} { - jsonpath_yyerror(NULL, escontext, + jsonpath_yyerror(NULL, escontext, yyscanner, "trailing junk after numeric literal"); yyterminate(); } {decimal_junk} { - jsonpath_yyerror(NULL, escontext, + jsonpath_yyerror(NULL, escontext, yyscanner, "trailing junk after numeric literal"); yyterminate(); } {real_junk} { - jsonpath_yyerror(NULL, escontext, + jsonpath_yyerror(NULL, escontext, yyscanner, "trailing junk after numeric literal"); yyterminate(); } @@ -356,8 +352,11 @@ hex_fail \\x{hexdigit}{0,1} void jsonpath_yyerror(JsonPathParseResult **result, struct Node *escontext, + yyscan_t yyscanner, const char *message) { + struct yyguts_t * yyg = (struct yyguts_t *) yyscanner; /* needed for yytext macro */ + /* don't overwrite escontext if it's already been set */ if (SOFT_ERROR_OCCURRED(escontext)) return; @@ -470,44 +469,6 @@ checkKeyword() return res; } -/* - * Called before any actual parsing is done - */ -static void -jsonpath_scanner_init(const char *str, int slen) -{ - if (slen <= 0) - slen = strlen(str); - - /* - * Might be left over after ereport() - */ - yy_init_globals(); - - /* - * Make a scan buffer with special termination needed by flex. - */ - - scanbuflen = slen; - scanbuf = palloc(slen + 2); - memcpy(scanbuf, str, slen); - scanbuf[slen] = scanbuf[slen + 1] = YY_END_OF_BUFFER_CHAR; - scanbufhandle = yy_scan_buffer(scanbuf, slen + 2); - - BEGIN(INITIAL); -} - - -/* - * Called after parsing is done to clean up after jsonpath_scanner_init() - */ -static void -jsonpath_scanner_finish(void) -{ - yy_delete_buffer(scanbufhandle); - pfree(scanbuf); -} - /* * Resize scanstring so that it can append string of given length. * Reinitialize if required. @@ -556,20 +517,27 @@ JsonPathParseResult * parsejsonpath(const char *str, int len, struct Node *escontext) { JsonPathParseResult *parseresult; + yyscan_t scanner; + + if (jsonpath_yylex_init(&scanner) != 0) + elog(ERROR, "yylex_init() failed: %m"); + + if (len <= 0) + len = strlen(str); - jsonpath_scanner_init(str, len); + jsonpath_yy_scan_bytes(str, len, scanner); - if (jsonpath_yyparse(&parseresult, escontext) != 0) - jsonpath_yyerror(NULL, escontext, "invalid input"); /* shouldn't happen */ + if (jsonpath_yyparse(&parseresult, escontext, scanner) != 0) + jsonpath_yyerror(NULL, escontext, scanner, "invalid input"); /* shouldn't happen */ - jsonpath_scanner_finish(); + jsonpath_yylex_destroy(scanner); return parseresult; } /* Turn hex character into integer */ static bool -hexval(char c, int *result, struct Node *escontext) +hexval(char c, int *result, struct Node *escontext, yyscan_t yyscanner) { if (c >= '0' && c <= '9') { @@ -586,7 +554,7 @@ hexval(char c, int *result, struct Node *escontext) *result = c - 'A' + 0xA; return true; } - jsonpath_yyerror(NULL, escontext, "invalid hexadecimal digit"); + jsonpath_yyerror(NULL, escontext, yyscanner, "invalid hexadecimal digit"); return false; } @@ -666,7 +634,7 @@ addUnicode(int ch, int *hi_surrogate, struct Node *escontext) * src/backend/utils/adt/json.c */ static bool -parseUnicode(char *s, int l, struct Node *escontext) +parseUnicode(char *s, int l, struct Node *escontext, yyscan_t yyscanner) { int i = 2; int hi_surrogate = -1; @@ -680,7 +648,7 @@ parseUnicode(char *s, int l, struct Node *escontext) { while (s[++i] != '}' && i < l) { - if (!hexval(s[i], &si, escontext)) + if (!hexval(s[i], &si, escontext, yyscanner)) return false; ch = (ch << 4) | si; } @@ -690,7 +658,7 @@ parseUnicode(char *s, int l, struct Node *escontext) { for (j = 0; j < 4 && i < l; j++) { - if (!hexval(s[i++], &si, escontext)) + if (!hexval(s[i++], &si, escontext, yyscanner)) return false; ch = (ch << 4) | si; } @@ -714,12 +682,12 @@ parseUnicode(char *s, int l, struct Node *escontext) /* Parse sequence of hex-encoded characters */ static bool -parseHexChar(char *s, struct Node *escontext) +parseHexChar(char *s, struct Node *escontext, yyscan_t yyscanner) { int s2, s3, ch; - if (!hexval(s[2], &s2, escontext)) + if (!hexval(s[2], &s2, escontext, yyscanner)) return false; - if (!hexval(s[3], &s3, escontext)) + if (!hexval(s[3], &s3, escontext, yyscanner)) return false; ch = (s2 << 4) | s3; @@ -733,13 +701,13 @@ parseHexChar(char *s, struct Node *escontext) */ void * -jsonpath_yyalloc(yy_size_t bytes) +jsonpath_yyalloc(yy_size_t bytes, yyscan_t yyscanner) { return palloc(bytes); } void * -jsonpath_yyrealloc(void *ptr, yy_size_t bytes) +jsonpath_yyrealloc(void *ptr, yy_size_t bytes, yyscan_t yyscanner) { if (ptr) return repalloc(ptr, bytes); @@ -748,7 +716,7 @@ jsonpath_yyrealloc(void *ptr, yy_size_t bytes) } void -jsonpath_yyfree(void *ptr) +jsonpath_yyfree(void *ptr, yyscan_t yyscanner) { if (ptr) pfree(ptr); -- 2.47.1