dvander@591: /**
ds@653:  * vim: set ts=4 :
ds@1330:  * =============================================================================
dvander@1231:  * SourceMod
dvander@1231:  * Copyright (C) 2004-2007 AlliedModders LLC.  All rights reserved.
ds@1330:  * =============================================================================
dvander@591:  *
ds@1330:  * This program is free software; you can redistribute it and/or modify it under
ds@1330:  * the terms of the GNU General Public License, version 3.0, as published by the
ds@1330:  * Free Software Foundation.
dvander@1231:  * 
ds@1330:  * This program is distributed in the hope that it will be useful, but WITHOUT
ds@1330:  * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
ds@1330:  * FOR A PARTICULAR PURPOSE.  See the GNU General Public License for more
ds@1330:  * details.
dvander@1231:  *
ds@1330:  * You should have received a copy of the GNU General Public License along with
ds@1330:  * this program.  If not, see <http://www.gnu.org/licenses/>.
dvander@1231:  *
ds@1330:  * As a special exception, AlliedModders LLC gives you permission to link the
ds@1330:  * code of this program (as well as its derivative works) to "Half-Life 2," the
ds@1330:  * "Source Engine," the "SourcePawn JIT," and any Game MODs that run on software
ds@1330:  * by the Valve Corporation.  You must obey the GNU General Public License in
ds@1330:  * all respects for all other code used.  Additionally, AlliedModders LLC grants
ds@1330:  * this exception to all derivative works.  AlliedModders LLC defines further
ds@1330:  * exceptions, found in LICENSE.txt (as of this writing, version JULY-31-2007),
ds@1330:  * or <http://www.sourcemod.net/license.php>.
dvander@591:  *
dvander@591:  * Version: $Id$
dvander@591:  */
dvander@591: 
dvander@591: #include <stdio.h>
dvander@591: #include <ctype.h>
dvander@591: #include <wctype.h>
dvander@591: #include <string.h>
dvander@591: #include <stdlib.h>
dvander@591: #include <assert.h>
dvander@591: #include "TextParsers.h"
dvander@591: #include "ShareSys.h"
dvander@591: 
dvander@616: TextParsers g_TextParser;
dvander@591: 
dvander@591: static int g_ini_chartable1[255] = {0};
dvander@591: static int g_ws_chartable[255] = {0};
dvander@591: 
dvander@631: bool IsWhitespace(const char *stream)
dvander@631: {
dvander@631: 	return g_ws_chartable[(unsigned)*stream] == 1;
dvander@631: }
dvander@631: 
dvander@616: TextParsers::TextParsers()
dvander@591: {
dvander@591: 	g_ini_chartable1[(unsigned)'_'] = 1;
dvander@591: 	g_ini_chartable1[(unsigned)'-'] = 1;
dvander@591: 	g_ini_chartable1[(unsigned)','] = 1;
dvander@591: 	g_ini_chartable1[(unsigned)'+'] = 1;
dvander@591: 	g_ini_chartable1[(unsigned)'.'] = 1;
dvander@591: 	g_ini_chartable1[(unsigned)'$'] = 1;
dvander@591: 	g_ini_chartable1[(unsigned)'?'] = 1;
dvander@591: 	g_ini_chartable1[(unsigned)'/'] = 1;
dvander@591: 	g_ws_chartable[(unsigned)'\n'] = 1;
dvander@591: 	g_ws_chartable[(unsigned)'\v'] = 1;
dvander@591: 	g_ws_chartable[(unsigned)'\r'] = 1;
dvander@591: 	g_ws_chartable[(unsigned)'\t'] = 1;
dvander@591: 	g_ws_chartable[(unsigned)'\f'] = 1;
dvander@591: 	g_ws_chartable[(unsigned)' '] = 1;
dvander@591: }
dvander@591: 
dvander@616: void TextParsers::OnSourceModAllInitialized()
dvander@591: {
dvander@591: 	g_ShareSys.AddInterface(NULL, this);
dvander@591: }
dvander@591: 
dvander@616: unsigned int TextParsers::GetUTF8CharBytes(const char *stream)
dvander@591: {
dvander@591: 	return _GetUTF8CharBytes(stream);
dvander@591: }
dvander@591: 
dvander@591: /**
dvander@591:  * Character streams
dvander@591:  */
dvander@591: 
dvander@591: struct CharStream
dvander@591: {
dvander@591: 	const char *curpos;
dvander@591: };
dvander@591: 
dvander@591: bool CharStreamReader(void *stream, char *buffer, size_t maxlength, unsigned int *read)
dvander@591: {
dvander@591: 	CharStream *srdr = (CharStream *)stream;
dvander@591: 
dvander@591: 	const char *ptr = srdr->curpos;
dvander@591: 	for (size_t i=0; i<maxlength; i++)
dvander@591: 	{
dvander@591: 		if (*ptr == '\0')
dvander@591: 		{
dvander@591: 			break;
dvander@591: 		}
dvander@591: 		*buffer++ = *ptr++;
dvander@591: 	}
dvander@591: 
dvander@591: 	*read = ptr - srdr->curpos;
dvander@591: 
dvander@591: 	srdr->curpos = ptr;
dvander@591: 
dvander@591: 	return true;
dvander@591: }
dvander@591: 
dvander@616: SMCParseError TextParsers::ParseString_SMC(const char *stream, 
dvander@591: 					 ITextListener_SMC *smc,
dvander@591: 					 unsigned int *line,
dvander@591: 					 unsigned int *col)
dvander@591: {
dvander@591: 	CharStream srdr = { stream };
dvander@591: 
dvander@591: 	return ParseStream_SMC(&srdr, CharStreamReader, smc, line, col);
dvander@591: }
dvander@591: 
dvander@591: /**
dvander@591:  * File streams
dvander@591:  */
dvander@591: 
dvander@591: bool FileStreamReader(void *stream, char *buffer, size_t maxlength, unsigned int *read)
dvander@591: {
dvander@591: 	size_t num = fread(buffer, 1, maxlength, (FILE *)stream);
dvander@591: 
dvander@591: 	*read = static_cast<unsigned int>(num);
dvander@591: 
dvander@591: 	if (num == 0 && feof((FILE *)stream))
dvander@591: 	{
dvander@591: 		return true;
dvander@591: 	}
dvander@591: 
dvander@591: 	return (ferror((FILE *)stream) == 0);
dvander@591: }
dvander@591: 
dvander@616: SMCParseError TextParsers::ParseFile_SMC(const char *file, ITextListener_SMC *smc, unsigned int *line, unsigned int *col)
dvander@591: {
dvander@591: 	FILE *fp = fopen(file, "rt");
dvander@591: 
dvander@591: 	if (!fp)
dvander@591: 	{
dvander@591: 		return SMCParse_StreamOpen;
dvander@591: 	}
dvander@591: 
dvander@591: 	SMCParseError result = ParseStream_SMC(fp, FileStreamReader, smc, line, col);
dvander@591: 
dvander@591: 	fclose(fp);
dvander@591: 
dvander@591: 	return result;
dvander@591: }
dvander@591: 
dvander@591: /** 
dvander@591:  * Raw parsing of streams with helper functions
dvander@591:  */
dvander@591: 
dvander@591: struct StringInfo
dvander@591: {
dvander@591: 	StringInfo() : quoted(false), ptr(NULL), end(NULL), special(false) { }
dvander@591: 	bool quoted;
dvander@591: 	char *ptr;
dvander@591: 	char *end;
dvander@591: 	bool special;
dvander@591: };
dvander@591: 
dvander@591: const char *FixupString(StringInfo &data)
dvander@591: {
dvander@591: 	if (!data.ptr)
dvander@591: 	{
dvander@591: 		return NULL;
dvander@591: 	}
dvander@591: 
dvander@591: 	if (data.quoted)
dvander@591: 	{
dvander@591: 		data.ptr++;
dvander@591: 	}
dvander@591: #if defined _DEBUG
dvander@591: 	else {
dvander@591: 		/* A string will never have beginning whitespace because we ignore it in the stream.
dvander@591: 		 * Furthermore, if there is trailing whitespace, the end ptr will point to it, so it is valid
dvander@591: 		 * to overwrite!  Lastly, the last character must be whitespace or a comment/invalid character.
dvander@591: 		 */
dvander@591:  	}
dvander@591: #endif
dvander@591: 
dvander@1073: 	/* Do some extra work on strings that have special quoted characters. */
dvander@591: 	if (data.special)
dvander@591: 	{
dvander@1073: 		char *outptr = data.ptr;
dvander@1073: 		size_t len = data.end - data.ptr;
dvander@1073: 		if (len >= 2)
dvander@1073: 		{
dvander@1073: 			for (size_t i=0; i<len; i++)
dvander@1073: 			{
dvander@1073: 				if (data.ptr[i] == '\\' && i < len - 1)
dvander@1073: 				{
dvander@1073: 					/* Resolve the next character. */
dvander@1073: 					i++;
dvander@1073: 					if (data.ptr[i] == 'n')
dvander@1073: 					{
dvander@1073: 						data.ptr[i] = '\n';
dvander@1073: 					} else if (data.ptr[i] == 't') {
dvander@1073: 						data.ptr[i] = '\t';
dvander@1073: 					} else if (data.ptr[i] == 'r') {
dvander@1073: 						data.ptr[i] = '\r';
dvander@1073: 					} else if (data.ptr[i] != '\\'
dvander@1073: 						&& data.ptr[i] != '"') {
dvander@1073: 						/* This character is invalid, so go back one */
dvander@1073: 						i--;
dvander@1073: 					}
dvander@1073: 				}
dvander@1073: 				*outptr++ = data.ptr[i];
dvander@1073: 			}
dvander@1073: 			*outptr = '\0';
dvander@1073: 		}
dvander@591: 	}
dvander@591: 
dvander@1260: 	if (data.end)
dvander@1260: 	{
dvander@1260: 		*(data.end) = '\0';
dvander@1260: 	}
dvander@591: 
dvander@591: 	return data.ptr;
dvander@591: }
dvander@591: 
dvander@591: const char *rotate(StringInfo info[3])
dvander@591: {
dvander@591: 	if (info[2].ptr != NULL)
dvander@591: 	{
dvander@591: 		return info[2].ptr;
dvander@591: 	}
dvander@591: 
dvander@591: 	if (info[0].ptr != NULL)
dvander@591: 	{
dvander@591: 		info[2] = info[1];
dvander@591: 		info[1] = info[0];
dvander@591: 		info[0] = StringInfo();
dvander@591: 	}
dvander@591: 	
dvander@591: 	return NULL;
dvander@591: }
dvander@591: 
dvander@591: void scrap(StringInfo info[3])
dvander@591: {
dvander@591: 	info[2] = StringInfo();
dvander@591: 	info[1] = StringInfo();
dvander@591: 	info[0] = StringInfo();
dvander@591: }
dvander@591: 
dvander@591: void reloc(StringInfo &data, unsigned int bytes)
dvander@591: {
dvander@591: 	if (data.ptr)
dvander@591: 	{
dvander@591: 		data.ptr -= bytes;
dvander@591: 	}
dvander@591: 	if (data.end)
dvander@591: 	{
dvander@591: 		data.end -= bytes;
dvander@591: 	}
dvander@591: }
dvander@591: 
dvander@591: char *lowstring(StringInfo info[3])
dvander@591: {
dvander@591: 	for (int i=2; i>=0; i--)
dvander@591: 	{
dvander@591: 		if (info[i].ptr)
dvander@591: 		{
dvander@591: 			return info[i].ptr;
dvander@591: 		}
dvander@591: 	}
dvander@591: 
dvander@591: 	return NULL;
dvander@591: }
dvander@591: 
dvander@616: SMCParseError TextParsers::ParseStream_SMC(void *stream, 
dvander@591: 								   STREAMREADER srdr, 
dvander@591: 								   ITextListener_SMC *smc, 
dvander@591: 								   unsigned int *line, 
dvander@591: 								   unsigned int *col)
dvander@591: {
dvander@591: 	char in_buf[4096];
dvander@591: 	char *reparse_point = NULL;
dvander@591: 	char *parse_point = in_buf;
dvander@591: 	char *line_begin = in_buf;
dvander@591: 	unsigned int read;
dvander@591: 	unsigned int curline = 1;
dvander@591: 	unsigned int curtok = 0;
dvander@591: 	unsigned int curlevel = 0;
dvander@591: 	bool in_quote = false;
dvander@591: 	bool ignoring = false;
dvander@591: 	bool eol_comment = false;
dvander@591: 	bool ml_comment = false;
dvander@591: 	unsigned int i;
dvander@591: 	SMCParseError err = SMCParse_Okay;
dvander@591: 	SMCParseResult res;
dvander@591: 	char c;
dvander@591: 
dvander@591: 	StringInfo strings[3];
dvander@591: 	StringInfo emptystring;
dvander@591: 
dvander@591: 	smc->ReadSMC_ParseStart();
dvander@591: 
dvander@591: 	while (srdr(stream, parse_point, sizeof(in_buf) - (parse_point - line_begin) - 1, &read))
dvander@591: 	{
dvander@591: 		if (!read)
dvander@591: 		{
dvander@591: 			break;
dvander@591: 		}
dvander@591: 
dvander@591: 		/* :TODO: do this outside of the main loop somehow
dvander@591: 		 * This checks for BOM markings
dvander@591: 		 */
dvander@591: 		if (curline == 1 && 
dvander@591: 			in_buf[0] == (char)0xEF && 
dvander@591: 			in_buf[1] == (char)0xBB && 
dvander@591: 			in_buf[2] == (char)0xBF)
dvander@591: 		{
dvander@1143: 			/* Move EVERYTHING down :\ */
dvander@1143: 			memmove(in_buf, &in_buf[3], read - 3);
dvander@1143: 			read -= 3;
dvander@591: 		}
dvander@591: 
dvander@591: 		if (reparse_point)
dvander@591: 		{
dvander@591: 			read += (parse_point - reparse_point);
dvander@591: 			parse_point = reparse_point;
dvander@591: 			reparse_point = NULL;
dvander@591: 		}
dvander@591: 
dvander@591: 		for (i=0; i<read; i++)
dvander@591: 		{
dvander@591: 			c = parse_point[i];
dvander@591: 			if (c == '\n')
dvander@591: 			{
dvander@591: 				/* If we got a newline, there's a lot of things that could have happened in the interim.
dvander@591: 				 * First, let's make sure the staged strings are rotated.
dvander@591: 				 */
dvander@591: 				if (strings[0].ptr)
dvander@591: 				{
dvander@591: 					strings[0].end = &parse_point[i];
dvander@591: 					if (rotate(strings) != NULL)
dvander@591: 					{
dvander@591: 						err = SMCParse_InvalidTokens;
dvander@591: 						goto failed;
dvander@591: 					}
dvander@591: 				}
dvander@591: 
dvander@591: 				/* Next, let's clear some line-based values that may no longer have meaning */
dvander@591: 				eol_comment = false;
dvander@591: 				in_quote = false;
dvander@591: 				if (ignoring && !ml_comment)
dvander@591: 				{
dvander@591: 					ignoring = false;
dvander@591: 				}
dvander@591: 
dvander@591: 				/* Pass the raw line onto the listener */
dvander@591: 				if ((res=smc->ReadSMC_RawLine(line_begin, curline)) != SMCParse_Continue)
dvander@591: 				{
dvander@591: 					err = (res == SMCParse_HaltFail) ? SMCParse_Custom : SMCParse_Okay;
dvander@591: 					goto failed;
dvander@591: 				}
dvander@591: 
dvander@591: 				/* Now we check the sanity of our staged strings! */
dvander@591: 				if (strings[2].ptr)
dvander@591: 				{
dvander@591: 					if (!curlevel)
dvander@591: 					{
dvander@591: 						err = SMCParse_InvalidProperty1;
dvander@591: 						goto failed;
dvander@591: 					}
dvander@591: 					/* Assume the next string is a property and pass the info on. */
dvander@591: 					if ((res=smc->ReadSMC_KeyValue(
dvander@591: 						FixupString(strings[2]),
dvander@591: 						FixupString(strings[1]),
dvander@591: 						strings[2].quoted,
dvander@591: 						strings[1].quoted)) != SMCParse_Continue)
dvander@591: 					{
dvander@591: 						err = (res == SMCParse_HaltFail) ? SMCParse_Custom : SMCParse_Okay;
dvander@591: 						goto failed;
dvander@591: 					}
dvander@591: 					scrap(strings);
dvander@591: 				}
dvander@591: 
dvander@591: 				/* Change the states for the next line */
dvander@591: 				curtok = 0;
dvander@591: 				curline++;
dvander@591: 				line_begin = &parse_point[i+1];		//Note: safe because this gets relocated later
dvander@591: 			} else if (ignoring) {
dvander@591: 				if (in_quote)
dvander@591: 				{
dvander@591: 					/* If i was 0, this case is impossible due to reparsing */
dvander@591: 					if ((i != 0) && c == '"' && parse_point[i-1] != '\\')
dvander@591: 					{
dvander@591: 						/* If we reached a quote in an ignore phase,
dvander@591: 						 * we're staging a string and we must rotate it out.
dvander@591: 						 */
dvander@591: 						in_quote = false;
dvander@591: 						ignoring = false;
dvander@591: 						/* Set our info */
dvander@591: 						strings[0].end = &parse_point[i];
dvander@591: 						strings[0].quoted = true;
dvander@591: 						if (rotate(strings) != NULL)
dvander@591: 						{
dvander@591: 							/* If we rotated too many strings, there was too much crap on one line */
dvander@591: 							err = SMCParse_InvalidTokens;
dvander@591: 							goto failed;
dvander@591: 						}
dvander@1073: 					} else if (c == '\\') {
dvander@591: 						strings[0].special = true;
dvander@1073: 						if (i == (read - 1))
dvander@1073: 						{
dvander@1073: 							reparse_point = &parse_point[i];
dvander@1073: 							break;
dvander@1073: 						}
dvander@591: 					}
dvander@591: 				} else if (ml_comment) {
dvander@591: 					if (c == '*')
dvander@591: 					{
dvander@591: 						/* Check if we need to get more input first */
dvander@591: 						if (i == read - 1)
dvander@591: 						{
dvander@591: 							reparse_point = &parse_point[i];
dvander@591: 							break;
dvander@591: 						}
dvander@591: 						if (parse_point[i+1] == '/')
dvander@591: 						{
dvander@591: 							ml_comment = false;
dvander@591: 							ignoring = false;
dvander@591: 							/* We should not be staging anything right now. */
dvander@591: 							assert(strings[0].ptr == NULL);
dvander@591: 							/* Advance the input stream so we don't choke on this token */
dvander@591: 							i++;
dvander@591: 							curtok++;
dvander@591: 						}
dvander@591: 					}
dvander@591: 				}
dvander@591: 			} else {
dvander@591: 				/* Check if we're whitespace or not */
dvander@591: 				if (!g_ws_chartable[(unsigned)c])
dvander@591: 				{
dvander@591: 					bool restage = false;
dvander@591: 					/* Check various special tokens:
dvander@591: 					 * ;
dvander@591: 					 * //
dvander@591: 					 * / *
dvander@591: 					 * {
dvander@591: 					 * }
dvander@591: 					 */
dvander@591: 					if (c == ';' || c == '/')
dvander@591: 					{
dvander@591: 						/* If it's a line-based comment (that is, ; or //)
dvander@591: 						 * we will need to scrap everything until the end of the line.
dvander@591: 						 */
dvander@591: 						if (c == '/')
dvander@591: 						{
dvander@591: 							if (i == read - 1)
dvander@591: 							{
dvander@591: 								/* If we reached the end of the look-ahead, we need to re-check our input.
dvander@591: 								 * Breaking out will force this to be the new reparse point!
dvander@591: 								 */
dvander@1109: 								reparse_point = &parse_point[i];
dvander@591: 								break;
dvander@591: 							}
dvander@591: 							if (parse_point[i + 1] == '/')
dvander@591: 							{
dvander@591: 								/* standard comment */
dvander@591: 								ignoring = true;
dvander@591: 								eol_comment = true;
dvander@591: 								restage = true;
dvander@591: 							} else if (parse_point[i+1] == '*') {
dvander@591: 								/* inline comment - start ignoring */
dvander@591: 								ignoring = true;
dvander@591: 								ml_comment = true;
dvander@591: 								/* yes, we restage, meaning that:
dvander@591: 								 * STR/ *stuff* /ING  (space because ml comments don't nest in C++)
dvander@591: 								 * will not generate 'STRING', but rather 'STR' and 'ING'.
dvander@591: 								 * This should be a rare occurrence and is done here for convenience.
dvander@591: 								 */
dvander@591: 								restage = true;
dvander@591: 							}
dvander@591: 						} else {
dvander@591: 							ignoring = true;
dvander@591: 							eol_comment = true;
dvander@591: 							restage = true;
dvander@591: 						}
dvander@591: 					} else if (c == '{') {
dvander@591: 						/* If we are staging a string, we must rotate here */
dvander@591: 						if (strings[0].ptr)
dvander@591: 						{
dvander@591: 							/* We have unacceptable tokens on this line */
dvander@591: 							if (rotate(strings) != NULL)
dvander@591: 							{
dvander@591: 								err = SMCParse_InvalidSection1;
dvander@591: 								goto failed;
dvander@591: 							}
dvander@591: 						}
dvander@591: 						/* Sections must always be alone */
dvander@591: 						if (strings[2].ptr != NULL)
dvander@591: 						{
dvander@591: 							err = SMCParse_InvalidSection1;
dvander@591: 							goto failed;
dvander@591: 						} else if (strings[1].ptr == NULL) {
dvander@591: 							err = SMCParse_InvalidSection2;
dvander@591: 							goto failed;
dvander@591: 						}
dvander@591: 						if ((res=smc->ReadSMC_NewSection(FixupString(strings[1]), strings[1].quoted))
dvander@591: 							!= SMCParse_Continue)
dvander@591: 						{
dvander@591: 							err = (res == SMCParse_HaltFail) ? SMCParse_Custom : SMCParse_Okay;
dvander@591: 							goto failed;
dvander@591: 						}
dvander@591: 						strings[1] = emptystring;
dvander@591: 						curlevel++;
dvander@591: 					} else if (c == '}') {
dvander@591: 						/* Unlike our matching friend, this can be on the same line as something prior */
dvander@591: 						if (rotate(strings) != NULL)
dvander@591: 						{
dvander@591: 							err = SMCParse_InvalidSection3;
dvander@591: 							goto failed;
dvander@591: 						}
dvander@591: 						if (strings[2].ptr)
dvander@591: 						{
dvander@591: 							if (!curlevel)
dvander@591: 							{
dvander@591: 								err = SMCParse_InvalidProperty1;
dvander@591: 								goto failed;
dvander@591: 							}
dvander@591: 							if ((res=smc->ReadSMC_KeyValue(
dvander@591: 											FixupString(strings[2]),
dvander@591: 											FixupString(strings[1]),
dvander@591: 											strings[2].quoted,
dvander@591: 											strings[1].quoted))
dvander@591: 								!= SMCParse_Continue)
dvander@591: 							{
dvander@591: 								err = (res == SMCParse_HaltFail) ? SMCParse_Custom : SMCParse_Okay;
dvander@591: 								goto failed;
dvander@591: 							}
dvander@591: 						} else if (strings[1].ptr) {
dvander@591: 							err = SMCParse_InvalidSection3;
dvander@591: 							goto failed;
dvander@591: 						} else if (!curlevel) {
dvander@591: 							err = SMCParse_InvalidSection4;
dvander@591: 							goto failed;
dvander@591: 						}
dvander@591: 						/* Now it's safe to leave the section */
dvander@591: 						scrap(strings);
dvander@591: 						if ((res=smc->ReadSMC_LeavingSection()) != SMCParse_Continue)
dvander@591: 						{
dvander@591: 							err = (res == SMCParse_HaltFail) ? SMCParse_Custom : SMCParse_Okay;
dvander@591: 							goto failed;
dvander@591: 						}
dvander@591: 						curlevel--;
dvander@591: 					} else if (c == '"') {
dvander@591: 						/* If we get a quote mark, we always restage, but we need to do it beforehand */
dvander@591: 						if (strings[0].ptr)
dvander@591: 						{
dvander@591: 							strings[0].end = &parse_point[i];
dvander@591: 							if (rotate(strings) != NULL)
dvander@591: 							{
dvander@591: 								err = SMCParse_InvalidTokens;
dvander@591: 								goto failed;
dvander@591: 							}
dvander@591: 						}
dvander@591: 						strings[0].ptr = &parse_point[i];
dvander@591: 						in_quote = true;
dvander@591: 						ignoring = true;
dvander@591: 					} else if (!strings[0].ptr) {
dvander@591: 						/* If we have no string, we must start one */
dvander@591: 						strings[0].ptr = &parse_point[i];
dvander@591: 					}
dvander@591: 					if (restage && strings[0].ptr)
dvander@591: 					{
dvander@591: 						strings[0].end = &parse_point[i];
dvander@591: 						if (rotate(strings) != NULL)
dvander@591: 						{
dvander@591: 							err = SMCParse_InvalidTokens;
dvander@591: 							goto failed;
dvander@591: 						}
dvander@591: 					}
dvander@591: 				} else {
dvander@591: 					/* If we're eating a string and get whitespace, we need to restage.
dvander@591: 					 * (Note that if we are quoted, this is being ignored)
dvander@591: 					 */
dvander@591: 					if (strings[0].ptr)
dvander@591: 					{
dvander@591: 						/*
dvander@591: 						 * The specification says the second string in a pair does not need to be quoted.
dvander@591: 						 * Thus, we check if there's already a string on the stack.
dvander@591: 						 * If there's a newline, we always rotate so the newline has an empty starter.
dvander@591: 						 */
dvander@591: 						if (!strings[1].ptr)
dvander@591: 						{
dvander@591: 							/* There's no string, so we must move this one down and eat up another */
dvander@591: 							strings[0].end = &parse_point[i];
dvander@591: 							rotate(strings);
dvander@591: 						} else if (!strings[1].quoted) {
dvander@591: 							err = SMCParse_InvalidTokens;
dvander@591: 							goto failed;
dvander@591: 						}
dvander@591: 					}
dvander@591: 				}
dvander@591: 			}
dvander@591: 
dvander@591: 			/* Advance which token we're on */
dvander@591: 			curtok++;
dvander@591: 		}
dvander@591: 
dvander@591: 		if (line_begin != in_buf)
dvander@591: 		{
dvander@591: 			/* The line buffer has advanced, so it's safe to copy N bytes back to the beginning.
dvander@591: 			 * What's N?  N is the lowest point we're currently relying on.
dvander@591: 			 */
dvander@591: 			char *stage = lowstring(strings);
dvander@591: 			if (!stage || stage > line_begin)
dvander@591: 			{
dvander@591: 				stage = line_begin;
dvander@591: 			}
dvander@591: 			unsigned int bytes = read - (stage - parse_point);
dvander@591: 
dvander@591: 			/* It is now safe to delete everything before the staged point */
dvander@591: 			memmove(in_buf, stage, bytes);
dvander@591: 
dvander@591: 			/* Calculate the number of bytes in the new buffer */
dvander@591: 			bytes = stage - in_buf;
dvander@591: 			/* Relocate all the cached pointers to our new base */
dvander@591: 			line_begin -= bytes;
dvander@591: 			reloc(strings[0], bytes);
dvander@591: 			reloc(strings[1], bytes);
dvander@591: 			reloc(strings[2], bytes);
dvander@591: 			if (reparse_point)
dvander@591: 			{
dvander@591: 				reparse_point -= bytes;
dvander@591: 			}
dvander@591: 			if (parse_point)
dvander@591: 			{
dvander@591: 				parse_point = &parse_point[read];
dvander@591: 				parse_point -= bytes;
dvander@591: 			}
dvander@591: 		} else if (read == sizeof(in_buf) - 1) {
dvander@591: 			err = SMCParse_TokenOverflow;
dvander@591: 			goto failed;
dvander@591: 		}
dvander@591: 	}
dvander@591: 
dvander@591: 	/* If we're done parsing and there are tokens left over... */
dvander@591: 	if (curlevel)
dvander@591: 	{
dvander@591: 		err = SMCParse_InvalidSection5;
dvander@591: 		goto failed;
dvander@591: 	} else if (strings[0].ptr || strings[1].ptr) {
dvander@591: 		err = SMCParse_InvalidTokens;
dvander@591: 		goto failed;
dvander@591: 	}
dvander@591: 	
dvander@591: 	smc->ReadSMC_ParseEnd(false, false);
dvander@591: 
dvander@591: 	return SMCParse_Okay;
dvander@591: 
dvander@591: failed:
dvander@591: 	if (line)
dvander@591: 	{
dvander@591: 		*line = curline;
dvander@591: 	}
dvander@591: 
dvander@591: 	smc->ReadSMC_ParseEnd(true, (err == SMCParse_Custom));
dvander@591: 
dvander@591: 	if (col)
dvander@591: 	{
dvander@591: 		*col = curtok;
dvander@591: 	}
dvander@591: 
dvander@591: 	return err;
dvander@591: }
dvander@591: 
dvander@591: 
dvander@591: /**
dvander@591:  * INI parser 
dvander@591:  */
dvander@591: 
dvander@616: bool TextParsers::ParseFile_INI(const char *file, ITextListener_INI *ini_listener, unsigned int *line, unsigned int *col)
dvander@591: {
dvander@591: 	FILE *fp = fopen(file, "rt");
dvander@591: 	unsigned int curline = 0;
dvander@591: 	unsigned int curtok;
dvander@591: 	size_t len;
dvander@591: 
dvander@591: 	if (!fp)
dvander@591: 	{
dvander@591: 		if (line)
dvander@591: 		{
dvander@591: 			*line = 0;
dvander@591: 		}
dvander@591: 
dvander@591: 		return false;
dvander@591: 	}
dvander@591: 
dvander@591: 	char buffer[2048];
dvander@591: 	char *ptr, *save_ptr;
dvander@591: 	bool in_quote;
dvander@591: 
dvander@591: 	while (!feof(fp))
dvander@591: 	{
dvander@591: 		curline++;
dvander@591: 		curtok = 0;
dvander@591: 		buffer[0] = '\0';
dvander@591: 		if (fgets(buffer, sizeof(buffer), fp) == NULL)
dvander@591: 		{
dvander@591: 			break;
dvander@591: 		}
dvander@591: 
dvander@591: 		//:TODO: this will only run once, so find a nice way to move it out of the while loop
dvander@591: 		/* If this is the first line, check the first three bytes for BOM */
dvander@591: 		if (curline == 1 && 
dvander@591: 			buffer[0] == (char)0xEF && 
dvander@591: 			buffer[1] == (char)0xBB && 
dvander@591: 			buffer[2] == (char)0xBF)
dvander@591: 		{
dvander@591: 			/* We have a UTF-8 marked file... skip these bytes */
dvander@591: 			ptr = &buffer[3];
dvander@591: 		} else {
dvander@591: 			ptr = buffer;
dvander@591: 		}
dvander@591: 
dvander@591: 		/***************************************************
dvander@591: 		 * We preprocess the string before parsing tokens! *
dvander@591: 		 ***************************************************/
dvander@591: 
dvander@591: 		/* First strip beginning whitespace */
dvander@591: 		while (*ptr != '\0' && g_ws_chartable[(unsigned)*ptr] != 0)
dvander@591: 		{
dvander@591: 			ptr++;
dvander@591: 		}
dvander@591: 
dvander@591: 		len = strlen(ptr);
dvander@591: 
dvander@591: 		if (!len)
dvander@591: 		{
dvander@591: 			continue;
dvander@591: 		}
dvander@591: 
dvander@591: 		/* Now search for comment characters */
dvander@591: 		in_quote = false;
dvander@591: 		save_ptr = ptr;
dvander@591: 		for (size_t i=0; i<len; i++,ptr++)
dvander@591: 		{
dvander@591: 			if (!in_quote)
dvander@591: 			{
dvander@591: 				switch (*ptr)
dvander@591: 				{
dvander@591: 				case '"':
dvander@591: 					{
dvander@591: 						in_quote = true;
dvander@591: 						break;
dvander@591: 					}
dvander@591: 				case ';':
dvander@591: 					{
dvander@591: 						/* Stop the loop */
dvander@591: 						len = i;
dvander@591: 						/* Terminate the string here */
dvander@591: 						*ptr = '\0';
dvander@591: 						break;
dvander@591: 					}
dvander@591: 				}
dvander@591: 			} else {
dvander@591: 				if (*ptr == '"')
dvander@591: 				{
dvander@591: 					in_quote = false;
dvander@591: 				}
dvander@591: 			}
dvander@591: 		}
dvander@591: 
dvander@591: 		if (!len)
dvander@591: 		{
dvander@591: 			continue;
dvander@591: 		}
dvander@591: 
dvander@591: 		ptr = save_ptr;
dvander@591: 
dvander@591: 		/* Lastly, strip ending whitespace off */
dvander@591: 		for (size_t i=len-1; i>=0 && i<len; i--)
dvander@591: 		{
dvander@591: 			if (g_ws_chartable[(unsigned)ptr[i]])
dvander@591: 			{
dvander@591: 				ptr[i] = '\0';
dvander@591: 				len--;
dvander@591: 			} else {
dvander@591: 				break;
dvander@591: 			}
dvander@591: 		}
dvander@591: 
dvander@591: 		if (!len)
dvander@591: 		{
dvander@591: 			continue;
dvander@591: 		}
dvander@591: 
dvander@591: 		if (!ini_listener->ReadINI_RawLine(ptr, &curtok))
dvander@591: 		{
dvander@591: 			goto event_failed;
dvander@591: 		}
dvander@591: 
dvander@591: 		if (*ptr == '[')
dvander@591: 		{
dvander@591: 			bool invalid_tokens = false;
dvander@591: 			bool got_bracket = false;
dvander@591: 			bool extra_tokens = false;
dvander@591: 			char c;
dvander@591: 			bool alnum;
dvander@591: 			wchar_t wc;
dvander@591: 
dvander@591: 			for (size_t i=1; i<len; i++)
dvander@591: 			{
dvander@591: 				c = ptr[i];
dvander@591: 				alnum = false;
dvander@591: 
dvander@591: 				if (c & (1<<7))
dvander@591: 				{
dvander@591: 					if (mbtowc(&wc, &ptr[i], len-i) != -1)
dvander@591: 					{
dvander@591: 						alnum = (iswalnum(wc) != 0);
dvander@591: 						i += _GetUTF8CharBytes(&ptr[i]) - 1;
dvander@591: 					}
dvander@591: 				} else {
dvander@591: 					alnum = (isalnum(c) != 0) || (g_ini_chartable1[(unsigned)c] != 0);
dvander@591: 				}
dvander@591: 				if (!alnum)
dvander@591: 				{
dvander@591: 					/* First check - is this a bracket? */
dvander@591: 					if (c == ']')
dvander@591: 					{
dvander@591: 						/* Yes! */
dvander@591: 						got_bracket = true;
dvander@591: 						/* If this isn't the last character... */
dvander@591: 						if (i != len - 1)
dvander@591: 						{
dvander@591: 							extra_tokens = true;
dvander@591: 						}
dvander@591: 						/* terminate */
dvander@591: 						ptr[i] = '\0';
dvander@591: 						break;
dvander@591: 					} else {
dvander@591: 						/* n...No! Continue copying. */
dvander@591: 						invalid_tokens = true;
dvander@591: 					}
dvander@591: 				}
dvander@591: 			}
dvander@591: 
dvander@591: 			/* Tell the handler */
dvander@591: 			if (!ini_listener->ReadINI_NewSection(&ptr[1], invalid_tokens, got_bracket, extra_tokens, &curtok))
dvander@591: 			{
dvander@591: 				goto event_failed;
dvander@591: 			}
dvander@591: 		} else {
dvander@591: 			char *key_ptr = ptr;
dvander@591: 			char *val_ptr = NULL;
dvander@591: 			char c;
dvander@591: 			size_t first_space = 0;
dvander@591: 			bool invalid_tokens = false;
dvander@591: 			bool equal_token = false;
dvander@591: 			bool quotes = false;
dvander@591: 			bool alnum;
dvander@591: 			wchar_t wc;
dvander@591: 
dvander@591: 			for (size_t i=0; i<len; i++)
dvander@591: 			{
dvander@591: 				c = ptr[i];
dvander@591: 				alnum = false;
dvander@591: 				/* is this an invalid char? */
dvander@591: 				if (c & (1<<7))
dvander@591: 				{
dvander@591: 					if (mbtowc(&wc, &ptr[i], len-i) != -1)
dvander@591: 					{
dvander@591: 						alnum = (iswalnum(wc) != 0);
dvander@591: 						i += _GetUTF8CharBytes(&ptr[i]) - 1;
dvander@591: 					}
dvander@591: 				} else {
dvander@591: 					alnum = (isalnum(c) != 0) || (g_ini_chartable1[(unsigned)c] != 0);
dvander@591: 				}
dvander@591: 
dvander@591: 				if (!alnum)
dvander@591: 				{
dvander@591: 					if (g_ws_chartable[(unsigned)c])
dvander@591: 					{
dvander@591: 						/* if it's a space, keep track of the first occurring space */
dvander@591: 						if (!first_space)
dvander@591: 						{
dvander@591: 							first_space = i;
dvander@591: 						}
dvander@591: 					} else {
dvander@591: 						if (c == '=')
dvander@591: 						{
dvander@591: 							/* if it's an equal sign, we're done with the key */
dvander@591: 							if (first_space)
dvander@591: 							{
dvander@591: 								/* remove excess whitespace */
dvander@591: 								key_ptr[first_space] = '\0';
dvander@591: 							} else {
dvander@591: 								/* remove the equal sign */
dvander@591: 								key_ptr[i] = '\0';
dvander@591: 							}
dvander@591: 							if (ptr[++i] != '\0')
dvander@591: 							{
dvander@591: 								/* If this isn't the end, set next pointer */
dvander@591: 								val_ptr = &ptr[i];
dvander@591: 							}
dvander@591: 							equal_token = true;
dvander@591: 							break;
dvander@591: 						} else {
dvander@591: 							/* Mark that we got something invalid! */
dvander@591: 							invalid_tokens = true;
dvander@591: 							first_space = 0;
dvander@591: 						}
dvander@591: 					}
dvander@591: 				}
dvander@591: 			}
dvander@591: 
dvander@591: 			/* Now we need to parse the value, if any */
dvander@591: 			if (val_ptr)
dvander@591: 			{
dvander@591: 				/* eat up spaces! there shouldn't be any h*/
dvander@591: 				while ((*val_ptr != '\0') && g_ws_chartable[(unsigned)*val_ptr] != 0)
dvander@591: 				{
dvander@591: 					val_ptr++;
dvander@591: 				}
dvander@591: 				if (*val_ptr == '\0')
dvander@591: 				{
dvander@591: 					val_ptr = NULL;
dvander@591: 					goto skip_value;
dvander@591: 				}
dvander@591: 				/* Do we have an initial quote? If so, the parsing rules change! */
dvander@591: 				if (*val_ptr == '"' && *val_ptr != '\0')
dvander@591: 				{
dvander@591: 					len = strlen(val_ptr);
dvander@591: 					if (val_ptr[len-1] == '"')
dvander@591: 					{
dvander@591: 						/* Strip quotes! */
dvander@591: 						val_ptr[--len] = '\0';
dvander@591: 						val_ptr++;
dvander@591: 						quotes = true;
dvander@591: 					}
dvander@591: 				}
dvander@591: 			}
dvander@591: skip_value:
dvander@591: 			/* We're done! */
dvander@591: 			curtok = val_ptr - buffer;
dvander@591: 			if (!ini_listener->ReadINI_KeyValue(key_ptr, val_ptr, invalid_tokens, equal_token, quotes, &curtok))
dvander@591: 			{
dvander@591: 				curtok = 0;
dvander@591: 				goto event_failed;
dvander@591: 			}
dvander@591: 		}
dvander@591: 	}
dvander@591: 
dvander@591: 	if (line)
dvander@591: 	{
dvander@591: 		*line = curline;
dvander@591: 	}
dvander@591: 
dvander@591: 	fclose(fp);
dvander@591: 
dvander@591: 	return true;
dvander@591: 
dvander@591: event_failed:
dvander@591: 	if (line)
dvander@591: 	{
dvander@591: 		*line = curline;
dvander@591: 	}
dvander@591: 
dvander@591: 	if (col)
dvander@591: 	{
dvander@591: 		*col = curtok;
dvander@591: 	}
dvander@591: 
dvander@591: 	fclose(fp);
dvander@591: 
dvander@591: 	return false;
dvander@591: }
dvander@591: 
dvander@616: const char *TextParsers::GetSMCErrorString(SMCParseError err)
dvander@591: {
dvander@591: 	static const char *s_errors[] = 
dvander@591: 	{
dvander@591: 		NULL,
dvander@591: 		"Stream failed to open",
dvander@591: 		"Stream returned read error",
dvander@591: 		NULL,
dvander@591: 		"Un-quoted section has invalid tokens",
dvander@591: 		"Section declared without header",
dvander@591: 		"Section declared with unknown tokens",
dvander@591: 		"Section ending without a matching section beginning",
dvander@591: 		"Section beginning without a matching ending",
dvander@591: 		"Line contained too many invalid tokens",
dvander@591: 		"Token buffer overflowed",
dvander@591: 		"A property was declared outside of a section",
dvander@591: 	};
dvander@591: 
dvander@591: 	if (err < SMCParse_Okay || err > SMCParse_InvalidProperty1)
dvander@591: 	{
dvander@591: 		return NULL;
dvander@591: 	}
dvander@591: 
dvander@591: 	return s_errors[err];
dvander@591: }