parse.c

关系型数据库 Postgresql 6.5.2

/* Module:          parse.c
 *
 * Description:     This module contains routines related to parsing SQL statements.
 *                  This can be useful for two reasons:
 *
 *                  1. So the query does not actually have to be executed to return data about it
 *
 *                  2. To be able to return information about precision, nullability, aliases, etc.
 *                     in the functions SQLDescribeCol and SQLColAttributes.  Currently, Postgres
 *                     doesn't return any information about these things in a query.
 *
 * Classes:         none
 *
 * API functions:   none
 *
 * Comments:        See "notice.txt" for copyright and license information.
 *
 */

#include <stdio.h>
#include <string.h>
#include <ctype.h>

#include "statement.h"
#include "connection.h"
#include "qresult.h"
#include "pgtypes.h"

#ifndef WIN32
#ifndef HAVE_STRICMP
#define stricmp(s1,s2) 		strcasecmp(s1,s2)
#define strnicmp(s1,s2,n)	strncasecmp(s1,s2,n)
#endif
#endif

#define FLD_INCR	32
#define TAB_INCR	8
#define COL_INCR	16

char *getNextToken(char *s, char *token, int smax, char *delim, char *quote, char *dquote, char *numeric);
void getColInfo(COL_INFO *col_info, FIELD_INFO *fi, int k);
char searchColInfo(COL_INFO *col_info, FIELD_INFO *fi);

char *
getNextToken(char *s, char *token, int smax, char *delim, char *quote, char *dquote, char *numeric)
{
int i = 0;
int out = 0;
char qc, in_escape = FALSE;

	if (smax <= 1)
		return NULL;

	smax--;

	/* skip leading delimiters */
	while (isspace(s[i]) || s[i] == ',') {
		// mylog("skipping '%c'\n", s[i]);
		i++;
	}

	if (s[0] == '\0') {
		token[0] = '\0';
		return NULL;
	}

	if (quote) *quote = FALSE;
	if (dquote) *dquote = FALSE;
	if (numeric) *numeric = FALSE;

	/* get the next token */
	while ( ! isspace(s[i]) && s[i] != ',' && s[i] != '\0' && out != smax) {

		/*	Handle quoted stuff */
		if ( out == 0 && (s[i] == '\"' || s[i] == '\'')) {
			qc = s[i];
			if (qc == '\"') {
				if (dquote) *dquote = TRUE;
			}
			if (qc == '\'') {
				if (quote) *quote = TRUE;
			}

			i++;		/* dont return the quote */
			while (s[i] != '\0' && out != smax) {  
				if (s[i] == qc && ! in_escape) {
					break;
				}
				if (s[i] == '\\' && ! in_escape) {
					in_escape = TRUE;
				}
				else {
					in_escape = FALSE;	
					token[out++] = s[i];
				}
				i++;
			}
			if (s[i] == qc)
				i++;
			break;
		}

		/*	Check for numeric literals */
		if ( out == 0 && isdigit(s[i])) {
			if (numeric) *numeric = TRUE;
			token[out++] = s[i++];
			while ( isalnum(s[i]) || s[i] == '.')
				token[out++] = s[i++];

			break;
		}

		if ( ispunct(s[i]) && s[i] != '_') {
			mylog("got ispunct: s[%d] = '%c'\n", i, s[i]);

			if (out == 0) {
				token[out++] = s[i++];
				break;
			}
			else
				break;
		}

		if (out != smax)
			token[out++] = s[i];

		i++;
	}

	// mylog("done -- s[%d] = '%c'\n", i, s[i]);

	token[out] = '\0';

	/*	find the delimiter  */
	while ( isspace(s[i]))
		i++;

	/*	return the most priority delimiter */
	if (s[i] == ',')  {
		if (delim) *delim = s[i];
	}
	else if (s[i] == '\0') {
		if (delim) *delim = '\0';
	}
	else  {
		if (delim) *delim = ' ';
	}

	/* skip trailing blanks  */
	while ( isspace(s[i])) {
		i++;
	}

	return &s[i];
}


/*
QR_set_num_fields(stmt->result, 14);
QR_set_field_info(stmt->result, 0, "TABLE_QUALIFIER", PG_TYPE_TEXT, MAX_INFO_STRING);
QR_set_field_info(stmt->result, 1, "TABLE_OWNER", PG_TYPE_TEXT, MAX_INFO_STRING);
QR_set_field_info(stmt->result, 2, "TABLE_NAME", PG_TYPE_TEXT, MAX_INFO_STRING);
QR_set_field_info(stmt->result, 3, "COLUMN_NAME", PG_TYPE_TEXT, MAX_INFO_STRING);
QR_set_field_info(stmt->result, 4, "DATA_TYPE", PG_TYPE_INT2, 2);
QR_set_field_info(stmt->result, 5, "TYPE_NAME", PG_TYPE_TEXT, MAX_INFO_STRING);
QR_set_field_info(stmt->result, 6, "PRECISION", PG_TYPE_INT4, 4);
QR_set_field_info(stmt->result, 7, "LENGTH", PG_TYPE_INT4, 4);
QR_set_field_info(stmt->result, 8, "SCALE", PG_TYPE_INT2, 2);
QR_set_field_info(stmt->result, 9, "RADIX", PG_TYPE_INT2, 2);
QR_set_field_info(stmt->result, 10, "NULLABLE", PG_TYPE_INT2, 2);
QR_set_field_info(stmt->result, 11, "REMARKS", PG_TYPE_TEXT, 254);
//	User defined fields
QR_set_field_info(stmt->result, 12, "DISPLAY_SIZE", PG_TYPE_INT4, 4);
QR_set_field_info(stmt->result, 13, "FIELD_TYPE", PG_TYPE_INT4, 4);
*/

void
getColInfo(COL_INFO *col_info, FIELD_INFO *fi, int k)
{
	if (fi->name[0] == '\0')
		strcpy(fi->name, QR_get_value_manual(col_info->result, k, 3));

	fi->type = atoi( QR_get_value_manual(col_info->result, k, 13));
	fi->precision = atoi( QR_get_value_manual(col_info->result, k, 6));
	fi->length = atoi( QR_get_value_manual(col_info->result, k, 7));
	fi->nullable = atoi( QR_get_value_manual(col_info->result, k, 10));
	fi->display_size = atoi( QR_get_value_manual(col_info->result, k, 12));
}

char
searchColInfo(COL_INFO *col_info, FIELD_INFO *fi)
{
int k;
char *col;


	for (k = 0; k < QR_get_num_tuples(col_info->result); k++) {
		col = QR_get_value_manual(col_info->result, k, 3);
		if ( ! strcmp(col, fi->name)) {
			getColInfo(col_info, fi, k);

			mylog("PARSE: searchColInfo: \n");
			return TRUE;
		}
	}

	return FALSE;
}


char
parse_statement(StatementClass *stmt)
{
static char *func="parse_statement";
char token[256];
char delim, quote, dquote, numeric, unquoted;
char *ptr;
char in_select = FALSE, in_distinct = FALSE, in_on = FALSE, in_from = FALSE, in_where = FALSE, in_table = FALSE;
char in_field = FALSE, in_expr = FALSE, in_func = FALSE, in_dot = FALSE, in_as = FALSE;
int j, i, k = 0, n, blevel = 0;
FIELD_INFO **fi;
TABLE_INFO **ti;
char parse;
ConnectionClass *conn = stmt->hdbc;
HSTMT hcol_stmt;
StatementClass *col_stmt;
RETCODE result;


	mylog("%s: entering...\n", func);

	ptr = stmt->statement;
	fi = stmt->fi;
	ti = stmt->ti;

	stmt->nfld = 0;
	stmt->ntab = 0;

	while ((ptr = getNextToken(ptr, token, sizeof(token), &delim, &quote, &dquote, &numeric)) != NULL) {

		unquoted = ! ( quote || dquote );

		mylog("unquoted=%d, quote=%d, dquote=%d, numeric=%d, delim='%c', token='%s', ptr='%s'\n", unquoted, quote, dquote, numeric, delim, token, ptr);

		if ( unquoted && ! stricmp(token, "select")) {
			in_select = TRUE;

			mylog("SELECT\n");
			continue;
		}

		if ( unquoted && in_select && ! stricmp(token, "distinct")) {
			in_distinct = TRUE;

			mylog("DISTINCT\n");
			continue;
		}

		if ( unquoted && ! stricmp(token, "into")) {
			in_select = FALSE;

			mylog("INTO\n");
			continue;
		}

		if ( unquoted && ! stricmp(token, "from")) {
			in_select = FALSE;
			in_from = TRUE;

			mylog("FROM\n");
			continue;
		}

		if ( unquoted && (! stricmp(token, "where") ||
			! stricmp(token, "union") || 
			! stricmp(token, "order") || 
			! stricmp(token, "group") || 
			! stricmp(token, "having"))) {

			in_select = FALSE;
			in_from = FALSE;
			in_where = TRUE;

			mylog("WHERE...\n");
			break;
		}

		if (in_select) {

			if ( in_distinct) {
				mylog("in distinct\n");

				if (unquoted && ! stricmp(token, "on")) {
					in_on = TRUE;
					mylog("got on\n");
					continue;
				}
				if (in_on) {
					in_distinct = FALSE;
					in_on = FALSE;
					continue;		/*	just skip the unique on field */
				}
				mylog("done distinct\n");
				in_distinct = FALSE;
			}

			if ( in_expr || in_func) {		/* just eat the expression */
				mylog("in_expr=%d or func=%d\n", in_expr, in_func);
				if (quote || dquote)
					continue;

				if (in_expr && blevel == 0 && delim == ',') {
					mylog("**** in_expr and Got comma\n");
					in_expr = FALSE;
					in_field = FALSE;
				}

				else if (token[0] == '(') {
					blevel++;
					mylog("blevel++ = %d\n", blevel);
				}
				else if (token[0] == ')') {
					blevel--;
					mylog("blevel-- = %d\n", blevel);
					if (delim==',') {
						in_func = FALSE;
						in_expr = FALSE;
						in_field = FALSE;
					}
				}
				continue;
			}

			if ( ! in_field) {

				if ( ! token[0])
					continue;

				if ( ! (stmt->nfld % FLD_INCR)) {
					mylog("reallocing at nfld=%d\n", stmt->nfld);
					fi = (FIELD_INFO **) realloc(fi, (stmt->nfld + FLD_INCR) * sizeof(FIELD_INFO *));
					if ( ! fi) {
						stmt->parse_status = STMT_PARSE_FATAL;
						return FALSE;
					}
					stmt->fi = fi;
				}

				fi[stmt->nfld] = (FIELD_INFO *) malloc( sizeof(FIELD_INFO));
				if (fi[stmt->nfld] == NULL) {
					stmt->parse_status = STMT_PARSE_FATAL;
					return FALSE;
				}

				/*	Initialize the field info */
				memset(fi[stmt->nfld], 0, sizeof(FIELD_INFO));

				/*	double quotes are for qualifiers */
				if (dquote)
					fi[stmt->nfld]->dquote = TRUE;

				if (quote) {
					fi[stmt->nfld++]->quote = TRUE;
					continue;
				}
				else if (numeric) {
					mylog("**** got numeric: nfld = %d\n", stmt->nfld);
					fi[stmt->nfld]->numeric = TRUE;
				}
				else if (token[0] == '(') {		/* expression */
					mylog("got EXPRESSION\n");
					fi[stmt->nfld++]->expr = TRUE;
					in_expr = TRUE;
					blevel = 1;
					continue;
				}

				else {
					strcpy(fi[stmt->nfld]->name, token);
					fi[stmt->nfld]->dot[0] = '\0';
				}
				mylog("got field='%s', dot='%s'\n", fi[stmt->nfld]->name, fi[stmt->nfld]->dot);

				if (delim == ',') {
					mylog("comma (1)\n");
				}
				else {
					in_field = TRUE;
				}
				stmt->nfld++;
				continue;
			}